Menopause: Biology and Pathobiology

Contributors Irina D. Burd (383) Department of Obstetrics, Gynecology, and Reproductive Sciences, Robert Wood Johnson ...

Author: Rogerio A. Lobo | Jennifer Kelsey | Robert Marcus

104 downloads 1338 Views 80MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

Contributors

Irina D. Burd (383) Department of Obstetrics, Gynecology, and Reproductive Sciences, Robert Wood Johnson Medical School, University of Medicine and Dentistry of New Jersey, New Brunswick, New Jersey 08901

Numbers in parentheses indicate the pages on which the authors' contributions begin.

Sanjay K. Agarwal (591) Division of Reproductive Medicine, Department of Obstetrics and Gynecology, Cedars-Sinai Burns and Allen Research Institute, Cedars-Sinai Medical Center, Los Angeles, California 90048; and Department of Obstetrics and Gynecology, University of California, Los Angeles, School of Medicine, Los Angeles, California 90095 Deborah J. Anderson (353) Fearing Laboratory, Department of Obstetrics, Gynecology, and Reproductive Biology, Brigham and Women's Hospital, Harvard Medical School, Boston, Massachusetts 02115 Nancy E. Avis (339) Institute for Women's Research, New England Research Institutes, Watertown, Massachusetts 02472 Gloria A. Bachmann (383) Department of Obstetrics, Gynecology, and Reproductive Sciences, Robert Wood Johnson Medical School, University of Medicine and Dentistry of New Jersey, New Brunswick, New Jersey 08901 John A. Baron (583) Section of Biostatistics and Epidemiology, Dartmouth Medical School, Hanover, New Hampshire 03755 Steven Birken (61) Department of Medicine, Columbia University College of Physicians and Surgeons, New York, New York 10032 Julia E. Bradsher (203) Abt Associates, Inc., Cambridge, Massachusetts 02138 M. Brincat (261) Department of Obstetrics and Gynecology, St. Luke's Hospital Medical School, Gwardamangia MSD 07, Malta

Henry G. Burger (147) Prince Henry's Institute of Medical Research, Clayton, Victoria 3168, Australia John E. Buster (625) Department of Obstetrics and Gynecology, Baylor College of Medicine, Houston, Texas 77030 Peter R. Casson* (625) Department of Obstetrics and Gynecology, Baylor College of Medicine, Houston, Texas 77030

Sybil L. Crawford (159, 175) New England Research Institutes, Watertown, Massachusetts 02472 Susan R. Davis (445) The Jean Hailes Foundation, Clayton, Victoria 3168, Australia; and Department of Epidemiology and Preventive Medicine, Monash University, Melbourne, Victoria 3004, Australia Carol A. Derby (229) New England Research Institutes, Watertown, Massachusetts 02472 Christine Draper (287) Department of Medicine, University of Western Australia, and Department of Endocrinology and Diabetes, Sir Charles Gairdner Hospital, Nedlands, Western Australia 6009, Australia Gary A. Ebert (383) Department of Obstetrics, Gynecology, and Reproductive Sciences, Robert Wood Johnson Medical School, University of Medicine and Dentistry of New Jersey, New Brunswick, New Jersey 08901

* Current address: Department of Obstetrics and Gynecology, Ottawa Hospital, Ottawa, Canada

xiii

xiv

Gregory F. Erickson (13) Department of Obstetrics and Gynecology, University of California, San Diego, La Jolla, California 92093 Denis Evans (175) Rush Institute on Aging, Chicago, Illinois 60612 Patricia D. Finn (33) Ligand Pharmaceuticals, Inc., San Diego, California 92121 Jeanne Franck (309) Department of Dermatology, Cornell Medical College and New York Presbyterian Hospital, New York, New York, 10021 Robert R. Freedman (215) Departments of Psychiatry and Behavioral Neurosciences and Obstetrics and Gynecology, Wayne State University, Detroit, Michigan 48201 R. Galea (261) Department of Obstetrics and Gynecology, St. Luke's Hospital Medical School, Gwardamangia MSD 07, Malta Ellen B. Gold (175, 189) Department of Epidemiology and Preventive Medicine, School of Medicine, University of California, Davis, Davis, California 95616 Joseph W. Goldzieher (397) Department of Obstetrics and Gynecology, Texas Tech University Health Sciences Center, Amarillo, Texas 79106 Gail A. Greendale (175, 639) Departments of Medicine and Obstetrics and Gynecology, University of California, Los Angeles, School of Medicine, Los Angeles, California 90024 Francine Grodstein (543) Channing Laboratory, Department of Medicine, Brigham and Women's Hospital, Harvard Medical School, Boston, Massachusetts 02115 Robert E Heaney (481) Creighton University, Omaha, Nebraska 68131 Victor W. Henderson (315) Department of Neurology, University of Southern California, Los Angeles, California 90089 Victoria Hendrick ( l l l ) Department of Psychiatry, University of California, Los Angeles, School of Medicine, Los Angeles, California 90095 Howard L. Judd (591) Department of Obstetrics and Gynecology, Olive View/ UCLA Medical Center, Sylmar, California 91342; and Department of Obstetrics and Gynecology, University of California, Los Angeles, School of Medicine, Los Angeles, California 90095 Margaret R. Karagas (359) Section of Biostatistics and Epidemiology, Dartmouth Medical School, Lebanon, New Hampshire 03756

CONTRIBUTORS

F. S. J. Keating (509) Endocrinology and Metabolic Medicine, Imperial College School of Medicine, St. Mary's Hospital Medical School, London W2 1PG, United Kingdom Jennifer Kelsey (175, 359, 405) Division of Epidemiology, Department of Health Research & Policy, Stanford University School of Medicine, Stanford, California 94305 Stanley G. Korenman (111) Department of Medicine, University of California, Los Angeles, School of Medicine, Los Angeles, California 90095 Galina Kovalevskaya (61) Irving Center for Clinical Research, Columbia University, College of Physicians and Surgeons, New York, New York 10032 Mark A. Lawson (33) Ligand Pharmaceuticals, Inc., San Diego, California 92121 Annie Lo (175) Westat, Inc., Rockville, Maryland 20850 Leslie Lobel (61) Department of Obstetrics and Gynecology, Columbia University, College of Physicians and Surgeons, New York, New York 10032 Rogerio A. Lobo (429) Department of Obstetrics and Gynecology, Columbia University, College of Physicians and Surgeons, New York, New York 10032 Francisco Jos~ L6pez (33) Ligand Pharmaceuticals, Inc., San Diego, California 92121 Cecilia Magnusson (583) Department of Medical Epidemiology, Karolinska Institutet, S- 171 77 Stockholm, Sweden

N. Manassiev (509) Endocrinology and Metabolic Medicine, Imperial College School of Medicine, St. Mary's Hospital Medical School, London W2 1PG, United Kingdom Robert Marcus (405, 495) Department of Medicine, Stanford University School of Medicine, Geriatrics Research, Education & Clinical Center, Veterans Affairs Medical Center, Palo Alto, California 94304 Karen Matthews (175) Department of Psychiatry, University of Pittsburgh, Pittsburgh, Pennsylvania 15261 Donald P. McDonnell (3) Department of Pharmacology and Cancer Biology, Duke University Medical Center, Durham, North Carolina 277 l0

CONTRIBUTORS

Valerie McGuire (359) Division of Epidemiology, Department of Health Research & Policy, Stanford University School of Medicine, Stanford, California 94305 Sonja M. McKinlay (203) New England Research Institutes, Watertown, Massachusetts 02172 Arshag D. Mooradian (111) Department of Internal Medicine, Saint Louis University School of Medicine, Saint Louis, Missouri 63104 David Morganstein (175) Westat, Inc., Rockville, Maryland 20850 Robert Neer (175) Division of Endocrinology, Department of Medicine, Massachusetts General Hospital, Boston, Massachusetts 02114 AndrOs Negro-Vilar (33) Ligand Pharmaceuticals, Inc., San Diego, California 92121 John O'Connor (61) Department of Pathology and Irving Center for Clinical Research, Columbia University, College of Physicians and Surgeons, New York, New York 10032 Richard L. Prince (287) Department of Medicine, University of Western Australia, and Department of Endocrinology and Diabetes, Sir Charles Gairdner Hospital, Nedlands, Western Australia 6009, Australia Janet H. Prystowsky (309) Department of Surgery, Columbia Presbyterian Medical Center, Columbia University and New York Presbyterian Hospital, New York, New York 10032 Russalind H. Ramos (459) Center for Menopause, Hormonal Disorders, and Women's Health, Sloane Hospital for Women, Columbia Presbyterian Medical Center, New York, New York, 10032 Nancy E. Reame (95) Center for Nursing Research and Reproductive Sciences Program, The University of Michigan, Ann Arbor, Michigan 48109 Robert W. Rebar (135) Department of Obstetrics and Gynecology, University of Cincinnati Medical Center, Cincinnati, Ohio 45267; and the American Society for Reproductive Medicine, Birmingham, Alabama 35216 Clifford J. Rosen (271) St. Joseph Hospital, Maine Center for Osteoporosis Research and Education, Bangor, Maine 04401 Giiran Samsioe (327) Department of Obstetrics and Gynecology, Lund University Hospital, S-221 85 Lund, Sweden Sherry Sherman (175) NIH/NIA, Bethesda, Maryland 20892

XV

Barbara B. Sherwin (617) Department of Psychology and Department of Obstetrics and Gynecology, McGill University, Montreal, H3A 1B 1 Canada Joe Leigh Simpson (77) Departments of Obstetrics and Gynecology and Molecular and Human Genetics, Baylor College of Medicine, Houston, Texas 77030 MaryFran Sowers (175, 245, 535) Department of Epidemiology, School of Public Health, University of Michigan, Ann Arbor, Michigan 48109 Leon Speroff (553) Department of Obstetrics and Gynecology, Oregon Health Sciences University, Portland, Oregon 97201 Margaret G. Spinelli (563) Department of Clinical Psychiatry, Columbia University, College of Physicians and Surgeons; and The New York State Psychiatric Institute, New York, New York 10032 Meir J. Stampfer (543) Departments of Nutrition and Epidemiology, Harvard School of Public Health; and Channing Laboratory, Department of Medicine, Brigham and Women's Hospital, Harvard Medical School, Boston, Massachusetts 02115 Frank Z. Stanezyk (421) Department of Obstetrics and Gynecology, University of Southern California School of Medicine, Los Angeles, California 90033 Barbara Sternfeld (175, 495) Department of Epidemiology and Biostatistics, Division of Research, Kaiser Permanente, Oakland, California 94611 J. C. Stevenson (509) Endocrinology and Metabolic Medicine, Imperial College School of Medicine, St. Mary's Hospital Medical School, London W2 1PG, United Kingdom Jennifer Tiseh (245) Department of Epidemiology, School of Public Health, University of Michigan, Ann Arbor, Michigan 48109 Anna N. A. Tosteson (649) Clinical Research Section, Department of Medicine and the Center for the Evaluative Clinical Sciences, Department of Community and Family Medicine, Dartmouth Medical School, Hanover, New Hampshire 03755 Michelle P. Warren (459) Department of Obstetrics and Gynecology and Medicine, Columbia University, College of Physicians and Surgeons, New York, New York 10032; and Center for Menopause, Hormonal Disorders, and Women's Health, Sloane Hospital for Women, Columbia Presbyterian Medical Center, New York, New York, 10032

xvi Gerson Weiss (175) Department of Obstetrics and Gynecology, University of Medicine and Dentistry of New Jersey, Newark, New Jersey 07103 Carolyn Westhoff (607) Columbia University, College of Physicians and Surgeons, New York, New York 10032

CONTRIBUTORS

Preface

lished in recent years, this work represents what we feel will be the first compilation of the entire subject, from basic biology to medical issues and therapeutic considerations. The text has been divided into several areas: basic biology, epidemiology, pathophysiology, and interventions. The chapters in each section are authored by outstanding investigators in the field who are recognized for their expertise. Accordingly, this text will be useful to all who are interested in the field, including basic and clinical investigators, students, residents, fellows, and clinicians. We are indebted to our friends, the contributing authors, and the tireless work of Jenny Wrenn and Jasna Markovac of Academic Press, who have shepherded this project from its inception. Without their help and persistence, this volume would not have been completed so efficiently and professionally, and in a timely manner.

Menopause is defined as the cessation of menstrual flow. Because the age of menopause is largely genetically determined, the average age at which it occurs, approximately 51 years, has not changed over many centuries. However, life expectancy has increased substantially, and the current life expectancy for women is 80 years. If a woman reaches age 54, she can expect to reach the age of 84.3 years. Thus, the years after menopause may account for as much as 40% of a woman's life. Currently in the United States, there are approximately 31 million women over age 55, with estimates of 38 million in 2010 and 46 million in 2020. Similar trends will occur in many parts of the world. Thus, there is a large and ever-increasing population of women in this very important time of life. Many women look forward to this time, even viewing this signal of a change in their reproductive lives as an opportunity for change and for instituting preventive health care. Nevertheless, for many years menopause has not been well understood. Although numerous books on the clinical aspects and the management of menopause have been pub-

Rogerio A. Lobo Jennifer Kelsey Robert Marcus

xvii

2HAPTER

Molecular Pharmacology of Estrogen and

Progesterone Receptors DONALD P.

MCDONNELL

Department of Pharmacology and Cancer Biology, Duke University Medical Center, Durham, North Carolina 27710

I. Introduction II. Estrogen and Progesterone Receptors III. Established Models of Estrogen and Progesterone Action IV. Estrogen and Progesterone Receptor Isoforms and Subtypes V. Regulation of Estrogen and Progesterone Receptor Function by Ligands

I. I N T R O D U C T I O N The steroid hormones estrogen and progesterone are small molecular weight lipophilic hormones that, through their action as modulators of distinct signal transduction pathways, are involved in the regulation of reproductive function [1, 2]. These hormones have also been shown to be important regulators in bone, the cardiovascular system, and the central nervous system [3-5]. Despite their different roles in these systems, however, it has become apparent that estrogens and progestins are mechanistically similar [6]. Insights gleaned from the study of each hormone, therefore, have advanced our understanding of this class of molecules as a whole. This review highlights some of the recent mechanistic discoveries that have occurred in the field, and e x -

MENOPAUSE: B I O L O G Y AND PATHOBIOLOGY

VI. Estrogen and Progesterone Receptor Associated Proteins VII. An Updated Model of Estrogen and Progesterone Receptor Action References

plores the subsequent changes in our understanding of the pharmacology of this class of steroid hormones.

II. E S T R O G E N AND PROGESTERONE RECEPTORS The estrogen receptor (ER) and progesterone receptor (PR) cDNAs have been cloned and used to develop specific ligand-responsive transcription systems in heterologous cells, permitting the use of reverse genetic approaches to define the functional domains within each of the receptors [6]. A schematic that outlines the organization of the major functional domains within these two steroid receptors (SR) proteins is shown in Fig. 1. The largest domain (~--300 amino acids) that

Copyright 9 2000 by Academic Press. All rights of reproduction in any form reserved.

4

DONALD P. MCDONNELL An additional activation function (AF-1) is located within the amino terminus of each receptor [ 11 ]. The DNA-binding domain (DBD) is a short region (--~70 amino acid residues) located in the center of the receptor protein [12]. Thispermits the receptor to bind as a dimer to target genes. Within the DBD there are nine cysteine residues, eight of which can chelate two zinc atoms, thereby forming two fingerlike structures that allow the receptor to interact with DNA [13]. All of the information required to permit target gene identification by ligand-activated ERs and PRs is contained within this region.

III. ESTABLISHED OF ESTROGEN PROGESTERONE

FIGURE 1 Establishedmodels of estrogen and progesteroneaction. The classic models of estrogen and progesterone action suggested that, in the absence of ligand, the steroid receptor (SR) exists in the nuclei of target cells in an inactive form. On binding an agonist, the receptorwould undergo an activating transformation event that displaces inhibitory heat-shock proteins (HSP) and facilitates the interaction of the receptorwith specific DNA steroid response elements (SRE) within target gene promoters. The activated receptor dimer could then interact with the general transcription machinery and positively or negatively regulate target gene transcription. In this model the role of the agonist is that of a "switch" that merely converts the ER or PR from an inactive to an active form. Thus, when corrected for affinity, all agonists would be qualitatively the same and evoke the same phenotypic response. By inference,antagonists,compounds that oppose the actions of agonists, would competitivelybind to their cognate receptors and freeze them in an inactive form. As with agonists, this model predicted that all antagonists are qualitatively the same. Within the confines of this classic model it was difficult to explain the molecularpharmacologyof the known ER and PR agonists and antagonists. GTA, General transcriptional activity.

is responsible for ligand binding is located at the carboxyl terminus of each receptor. Crystallographic analysis of the agonist-bound forms of ERs and PRs has indicated that this domain consists of 12 short a-helical structures that fold to provide a complex ligand-binding pocket [7, 8]. The ligandbinding domain also contains sequences that facilitate receptor homodimerization and permit the interaction of apo receptors with inhibitory heat-shock proteins. An activation function (AF-2) required for receptor transcriptional activity is also contained within the ligand-binding domain [9, 10].

MODELS

AND ACTION

The steroid hormones estrogen and progesterone are representative members of a larger family of steroid hormones, all of which appear to share a common mechanism of action. It is generally believed that steroid hormones enter cells from the bloodstream by simple passive diffusion, exhibiting activity only in cells in which they encounter a specific high-affinity receptor protein [14]. These receptor proteins are transcriptionally inactive in the absence of ligand, sequestered in a large oligomeric heat-shock protein complex within target cells [15]. On binding ligand, however, the receptors undergo an activating conformational change that promotes the dissociation of inhibitory proteins [16]. This event permits the formation of receptor homodimers that are capable of interacting with specific high-affinity DNAresponse elements located within the regulatory regions of target genes (Fig. 2) [17]. The DNA-bound receptor can then exert a positive or negative influence on target g e n e transcription. In the classic models of steroid hormone action, it was proposed that progestins and estrogens function merely as switches that, on binding to their cognate receptor, permit

FIGURE 2 The domain structures of the estrogen and progesterone receptors are similar.

CHAPTER 1 Estrogen and Progesterone Receptors conversion of the receptor in an all or nothing manner from an inactive to an active state [ 18]. This implied that ER and PR pharmacology was very simple, and that when corrected for affinity all progestins and estrogens were qualitatively the same. Furthermore, it suggested that antihormones (antagonists) function simply as competitive inhibitors of agonist binding, freezing the target receptor in an inactive state within the cell. Under most experimental conditions this simple model was sufficient to explain the observed biology of known PR and ER agonists and antagonists. However, systems were discovered that did not fit this simple model, indicating that the pharmacology of these receptor systems is more complex than originally believed. Studies probing the complex pharmacology of the antiestrogen tamoxifen have been very informative with respect to understanding the inadequacies of the classic model. Tamoxifen is widely used as a breast cancer chemotherapeutic and has been approved for use as a breast cancer chemopreventive in high-risk patients [ 19, 20]. In ER-positive breast cancers, tamoxifen opposes the mitogenic action of estrogen(s) by binding to the receptor and competitively blocking agonist access. However, it has become clear in recent years that tamoxifen is not a pure antagonist, because in some target organs it can exhibit estrogen-like activity. This is most apparent in both the skeletal system, where tamoxifen, like estrogen, increases lumbar spine bone mineral density, and the cardiovascular system, where both tamoxifen and estradiol have been shown to decrease low-density liproprotein (LDL) cholesterol [21, 22]. These in vivo properties of tamoxifen led to its being reclassified as a selective estrogen receptor modulator (SERM) rather than an antagonist. The observation that different ligand-receptor complexes were not recognized in the same manner in all cells was at odds with the established models of ER action. From a clinical perspective this was an important finding, because it suggested for the first time the possibility of developing compounds that, acting through their cognate receptor, could manifest different activities in different cells. From a molecular point of view, however, the observed pharmacology of tamoxifen begged a reevaluation of the classic model of ER action, and initiated the search for the cellular systems that enable ER-ligand complexes to manifest different biologies in different cells. These ongoing investigations have also provided significant insight into PR action, and have demonstrated that, as in the case of ERs, it will be possible to develop compounds that manifest PRagonist activity in a tissue-selective manner.

IV. ESTROGEN AND PROGESTERONE RECEPTOR ISOFORMS AND SUBTYPES One mechanism to explain the cell-selective action of steroid receptor ligands is the likelihood that they may activate different receptor isoforms (derived from the same gene) or

5

Estrogen Receptor Subtypes 1

595

NH21

hERc~

1

530

NH21

hER[3

I

Progesterone Receptor Subtypes 1

933

1

hPR-A

NH~I

769

I~1

~

I

FIGURE 3 At least two distinct forms of the estrogen and progesterone receptors exist in target cells. DBD, DNA-binding domain; LBD, ligandbinding domain.

subtypes (derived from similar genes). This concept has been well established for the ce- and fi-adrenergic systems, where it has been shown that different receptor subtypes have distinct ligand preferences, and that selectivity can be explained by differences in the expression of these subtypes. Until recently, the parallel between this system and that of the nuclear receptors was not obvious. However, the identification and characterization of ER and PR isoforms and subtypes has shed new light on this issue (Fig. 3).

A. ProgesteroneReceptor Isoforms The progesterone receptor was the first receptor for which bona fide isoforms were shown to exist. Human PRs can exist within target cells in either of two distinct forms, hPR-A (94 kDa) or hPR-B (114 kDa) [23]. These proteins, differing only in that the hPR-B isoform contains an additional 164-amino acid extension at its amino terminus, are produced from distinct mRNAs that are derived from different promoters within the same gene [9]. In most progesterone-responsive tissues these two receptor isoforms are expressed in equimolar amounts. This apparent 1:1 relationship is so widespread that until about 10 years ago the hPR-A isoform was thought to be merely an artifact derived from hPR-B by proteolysis during biochemical fractionation. It has now been established that these two proteins are produced in a deliberate manner by the cell, and that they are not functionally equivalent [23-25]. The first evidence in support of this hypothesis came following the cloning and subsequent functional analysis of the chicken progesterone receptor (cPR) cDNA [26]. Specifically, on expression in

6

DONALD P. MCDONNELL

heterologous cells, it was found that although the A and B forms of cPRs display identical ligand binding preferences, they activate different target genes [26]. It was subsequently shown that the amino-terminal sequences, which distinguish cPR-B from cPR-A, are important in determining target gene selectivity. This concept was reaffirmed when the cloned hPR-B and hPR-A were analyzed in a similar manner [24]. In the systems examined thus far, with few exceptions, it has been observed that hPR-B alone functions as a transcriptional activator in response to progesterone, whereas hPR-A displays minimal or no activity. Further analysis has revealed that hPR-A functions primarily as a ligand-dependent transdominant modulator of the transcriptional activity of hPR-B, the ability of hPR-B to activate target gene transcription being influenced by the cellular concentration of hPR-A [24, 27]. Surprisingly, it was also determined that ligandactivated hPR-A can inhibit the transcriptional activity of agonist-activated ERs, androgen receptors (ARs), and mineralocorticoid receptors (MRs) [24]. Thus, by virtue of having two functionally different receptor isoforms, a single hormone such as progesterone can have completely different functions in target cells.

B. E s t r o g e n R e c e p t o r S u b t y p e s The identification of functionally distinct PR isoforms introduced a new dimension to progesterone action, although it was not until a second estrogen receptor was cloned in 1995 that the general significance of isoforms (or subtypes) in steroid receptor signaling was established [28]. Unlike the case of PRs, ERa and ERfl are encoded by different genes, and although they share significant amino acid homology in their ligand-binding domains, they are not pharmacologically equivalent. Both receptors bind the endogenous estrogen, 17fl-estradiol, with equivalent affinity [29]. However, when binding analysis was extended to additional compounds, significant differences in ligand preferences were noted. The biological and pharmacological consequences of these differences remain to be determined. Although the discovery of ERfl has occurred relatively recently, significant progress has been made in elucidating its role in estrogen signaling. It has been determined that the expression pattern of ERfl does not mirror that of ERce [30, 31]. Expression of both isoforms is found in some tissues, whereas ER/3 alone occurs in others, such as the lung, the urogenital tract, and the colon [29]. The distinct roles of these two receptors in the endocrinology of estrogen have been confirmed by the generation of mice whose ER/3 has been genetically disrupted [32]. The phenotype of these mice is different from that of ERa knockout mice [32]. The specific role for ER/3 in tissues in which it is the only estrogen receptor expressed has not yet been identified. The high degree of amino acid homology between ERa and

ERfl within the DNA-binding domain suggests, but does not prove, that these receptors may regulate the same genes. It is also possible that ER/3 may interact with target genes in a manner that does not require direct contact with DNA regulatory elements within target genes. The identification of estrogen and progesterone receptor isoforms and subtypes and the definition of specific functions that they modulate have introduced a new dimension in steroid hormone action. Understanding the regulatory mechanisms that control the expression levels of the individual forms of each receptor is likely to provide novel targets for pharmaceutical intervention.

V. R E G U L A T I O N

OF ESTROGEN

AND PROGESTERONE RECEPTOR FUNCTION

BY LIGANDS

The finding that ERs and PRs could exist in multiple forms within target cells suggests that some of the tissueselective actions of their cognate agonists and antagonists can be explained by their ability to regulate differentially the action of one specific receptor isoform or subtype. Although a specific example of a receptor subtype-selective steroid receptor ligand has not yet emerged, the fact that such ligands for the retinoic acid receptor(s) have been generated makes the discovery of the ER and PR subtype-selective ligands more likely. Regardless, however, it has become apparent from the study of antiestrogens that the identical ligand operating through the same receptor can manifest different biological activities in different target cells [33]. In breast tissue, for instance, where ERa predominates, all of the known antiestrogens oppose the mitogenic actions of estrogen [34]. In the endometrium, however, where ERa also predominates, it has been found that tamoxifen functions as a partial estrogen mimetic [35, 36], whereas compounds such as raloxifene, GW5638, and ICI182,780 function as pure antiestrogens. Thus, the same compounds, acting through ERa, manifest different biological activities in the breast and the endometrium. This finding is not in agreement with the classic models of ER action that indicate that ligands basically fall into two classes, agonists and antagonists. This paradox has been the subject of much investigation leading to the observation that different compounds can induce different alterations in ER structure and that not all structures are functionally identical. It is implied, therefore (discussed in more detail below) that the cell possesses the cellular machinery to distinguish between these dissimilar complexes and that the identification and characterization of the specific components of these systems are the keys to the development of the next generation of tissue-selective ER and PR modulators. Much of what we know about the effect of ligands on ste-

CHAPTER 1 Estrogen and Progesterone Receptors roid receptor structure has come from studies of different ER-ligand complexes. Initially, using differential sensitivity to proteases, it was demonstrated that the hormone-binding domain within the ER adopts different shapes on binding estradiol and tamoxifen, and that these structures are dissimilar to that of the apo receptor [33, 37]. Thus, receptor conformation is affected by the nature of the bound ligand. This relationship between structure and function was later confirmed by the observation that agonists and antagonists induce different alterations in PR structure [38, 39]. Further analysis has revealed that the majority of the structural changes that occur in the PR are located at the extreme carboxyl tail of the receptor, and that removal of the carboxylterminal 42 amino acids of hPR-B permit the antagonist RU486 to function as an agonist [38]. Interestingly, a similarly positioned domain enables the ER to discriminate between different compounds and, not surprisingly, removal of 35 amino acids from the C-terminal tail of the ER abolishes its ability to distinguish between agonists and antagonists [40]. The recent determination of the crystalline structures of the ER-estradiol and ER-tamoxifen complexes confirmed the important role of the carboxyl tail in determining the pharmacology of steroid receptors [7, 41, 42]. This new structural information has also revealed that agonist activation of the ER permits the formation of a unique surface (or pocket) on the receptor that allows it to interact with the general transcription machinery through the mediation of adaptor or coactivator proteins. In the presence of the antagonist tamoxifen, however, the carboxyl tail of the ER is positioned in such a manner that it occludes this coactivator binding pocket, preventing a productive association with the cellular transcription apparatus. In addition to tamoxifen there are several additional SERMs that manifest distinct activities in vivo. One of these compounds, raloxifene, has been approved as a SERM for the treatment of osteoporosis [43]. This compound distinguishes itself from tamoxifen in that it does not exhibit estrogenic action in the postmenopausal endometrium [44, 45]. Although clearly different biologically, the crystal structures of the ER-tamoxifen and ER-raloxifene complexes were shown to be virtually indistinguishable. Although these results appear to be at odds with the hypothesis that links receptor structure to function, some data from our group have reconciled these potential discrepancies. We have used phage display technology to identify small peptides, the ability of which to bind ERs is affected differentially by the nature of the ligand bound to the receptor [46-48]. The rationale behind this approach is that because of the vast complexity of the peptides available in these libraries, it may be possible to find peptides that have the ability to distinguish between two very similar receptorligand complexes. This approach has led to the identification of a series of high-affinity peptide probes that, in addition to

7 being able to distinguish between ER-estradiol and E R tamoxifen complexes, are also able to distinguish among several different E R - S E R M complexes (Fig. 4). This approach has been extended to the study of the PR and it was similarly observed that various PR ligands manifest different

FIGURE 4 Fingerprinting the surfaces of different ER-ligand complexes using conformation-sensitive peptide probes. (A) Random peptide libraries were constructed in an M13 bacteriophage; each of the resulting bacteriophages expressed a unique random peptide on its surface pilus. Screens were subsequently performed to identify specific peptides (bacteriophage) whose interaction with the ER was influenced by the nature of the bound ligand. The bacteriophages identified in this manner were used to develop an enzyme-linked immunoassay to monitor changes that occur in the ER on its interaction with different ligands. Specifically, a biotinylated estrogen response element (ERE) was used to immobilize recombinant ERs on streptavidin-coated plates. After incubation of this complex with the ligand to be tested, to each well was added an aliquot of a different class of ERinteracting bacteriophage. Binding of the bacteriophage was assessed enzymatically using an anti-M 13 antibody coupled to horseradish peroxidase (HRP). (B) Fingerprint analysis of ER conformation in the presence of different ER ligands. Immobilized ER was incubated in the presence of saturating concentrations of the indicated ligands, and the resulting complexes were incubated with aliquots of bacteriophage expressing eight different peptides. Tam, Tamoxifen; DES, diethylstilbestrol; Prog, progesterone. This figure has been published previously in a similar form [47] and is reproduced and presented here with permission (Copyright 1999 National Academy of Sciences, U.S.A.).

8

DONALD P. MCDONNELL

biologies in different cells, allowing the identification of peptide probes whose interaction with the receptor is influenced by the nature of the ligand bound to the PR. All these findings establish a firm relationship between the structure of a receptor-ligand complex and biological activity, and suggest that novel ER and PR ligands with unique pharmaceutical properties may be developed by exploiting this observation.

VI. ESTROGEN AND PROGESTERONE RECEPTOR

ASSOCIATED

PROTEINS

The estrogen and progesterone receptors are liganddependent transcription factors that, on activation by ligands, associate with specific DNA response elements located within the regulatory regions of target genes [ 14]. The DNA-bound receptor can then positively or negatively influence gene transcription by altering RNA polymerase II activity. However, because RNA polymerase does not appear to interact directly with the steroid receptors, there must be additional factors that allow these two proteins to communicate [ 14]. In the past few years it has become clear that there are at least two functional classes of proteins that are involved in recognizing the activated receptor. One class includes components of the basic transcription machinery, the general transcription factors, whose expression levels are generally invariant from cell to cell. The second class of proteins, "cofactors," is not a part of the general transcription machinery, and can exert either a positive or a negative influence on SR transcriptional activity [49]. Those cofactors that interact with agonist-activated SRs have been called coactivators, whereas those that interact with apo receptors or antagonist-activated receptors have been called corepressors. Interestingly, it has become apparent that differences in the relative expression levels of coactivators and corepressors can have a profound effect on the pharmacology of estrogen and progesterone receptor ligands [25, 50, 51].

have revealed that (1) the expression levels of these coactivators vary from cell to cell, (2) coactivators demonstrate specific receptor preferences, (3) a given receptor can interact with more than one type of coactivator, and (4) the conformation of the receptor adopted in the presence of a specific ligand can determine which coactivators are engaged. These findings strongly support the hypothesis that differential cofactor expression is the most important determinant of estrogen and progesterone receptor pharmacology. With the discovery of the nuclear receptor coactivators and the characterization of their biochemical properties has come a new understanding of the mechanism by which differently conformed receptor-ligand complexes are recognized in the cell. The studies that have been performed with ERs are the most informative. As described previously, it has been shown that ERs in the presence of estradiol undergo a conformational change that allows the presentation of surfaces on the receptors, permitting them to interact with coactivators. Because estradiol induces the same conformational change within ERs in all cells, the phenotypic consequence of the exposure of a cell to estradiol will depend on the properties of the coactivators expressed in that cell (Fig. 5). The

A. C o a c t i v a t o r P r o t e i n s One of the most well-characterized coactivator proteins, steroid receptor coactivator 1 (SRC-1), was identified in a yeast two-hybrid screen as a protein that interacted with agonist-activated PR [52]. Subsequently, this protein has been shown to also interact with the estrogen, glucocorticoid, and androgen receptors. It appears that SRC-1 increases target gene transcription by linking the hormone-activated receptor with the general transcription machinery, stabilizing the transcription preinitiation complex, and nucleating a large complex of proteins that together have the ability to acetylate histones and facilitate chromatin decondensation [53, 54]. In addition to SRC-1, over 30 additional coactivators have been identified and characterized [49]. Cumulatively, these studies

FIGURE 5 A molecular explanation for the tissue-selective agonist/antagonist activity of the SERM tamoxifen. The estrogen receptor undergoes different conformational changes on binding the full agonistestrogen or the SERM tamoxifen. The estradiol-induced conformation allowsthe ER to interact with any coactivator protein expressed in target cells, and thus it can activate transcription. The tamoxifen-induced conformational change, on the other hand, is more restrictive and allows the interaction of the ER with only a subset of available coactivators. In those cells in which the tamoxifen-ER complex can engage a coactivator, this compound can manifest agonist activity.In other contexts tamoxifen functions as an antagonist.

CHAPTER1 Estrogen and Progesterone Receptors situation gets more complicated, however, when considering the role of coactivators in mediating the cell-selective action of SERMs such as tamoxifen. It has been shown that the tamoxifen-induced conformational change within the ER does not allow the coactivator binding pocket to form properly, preventing or hindering the interaction of those coactivators that require the coactivator binding pocket in order to interact with the E R [46]. In cells in which this type of coactivator is important, therefore, tamoxifen can function as an antagonist. It is becoming clear, however, that not all coactivators rely on the coactivator binding pocket to the same degree. Thus, the relative agonist-antagonist activity of tamoxifen depends on the ability of the t a m o x i f e n - E R complex to engage a compatible coactivator in target cells [33]. As the repertoire of cofactors increases, we are likely to find that targeting specific c o f a c t o r - r e c e p t o r complexes will yield pharmaceuticals that manifest their activities in a cell- or tissuerestricted manner.

B.

Corepressor

Proteins

Two nuclear corepressor proteins that appear to be important in ER and PR action have thus far been identified. These proteins, N C o R and SMRT, initially found as proteins that interact with D N A - b o u n d thyroid hormone or retinoid X receptors, repress basal transcription in the absence of hormone [55, 56]. However, it has now been shown that these proteins can interact with either PR or ER, either in the absence of h o r m o n e or in the presence of antagonists [50, 51]. Under these conditions, the corepressors nucleate a large protein complex, which represses target gene transcription by deacetylating histones and facilitating chromatin condensation. The physiological importance of corepressors in ER pharmacology was suggested by the studies of Lavinsky and co-workers, who found that passage of breast tumors in mice from a state of tamoxifen sensitivity to an insensitive state was accompanied by a decrease in the expression level of the corepressor [57]. A similar process, if occurring in humans, could explain how cells become resistant to the antiestrogenic actions of tamoxifen.

AN UPDATED M O D E L O F E S T R O G E N AND P R O G E S T E R O N E RECEPTOR ACTION VII.

On ligand binding, the activated receptor (ER or PR) can interact as a dimer with specific D N A response elements within target genes. It is now apparent that the conformation of the resulting receptor is influenced by the nature of the bound ligand and that the shape of the resulting receptor-ligand complex is a critical determinant of whether it can activate transcription. In the presence of a full agonist, the conformation adopted by the receptor facilitates the displacement

9 of corepressor proteins and recruitment of coactivator proteins, permitting the assembly of a histone-acetylating complex on D N A with a concomitant increase in target gene transcription. Pure antagonists, on the other hand, drive their cognate receptor into a conformation that favors corepressor interaction. The activity of mixed agonists/antagonists appears to relate to their ability to alter receptor conformation differentially, and the ability of corepressor and coactivator proteins within a given target cell to recognize these complexes. Clearly, the classic models of estrogen and progesterone action need to be updated to accommodate the insights that have emerged from the study of the genetics and molecular pharmacology of these two receptors.

References 1. Clark, J. H., and Markaverich, B. M. (1988). Actions of ovarian steroid hormones. In "The Physiology of Reproduction," E. Knobil and J. Neill, eds., pp. 675-724. Raven Press, New York. 2. Clarke, C. L., and Sutherland, R. L. (1990). Progestin regulation of cellular proliferation. Endocr. Rev 11, 266-301. 3. Colditz, G. A., Hankinson, S. E., Hunter, D. J., Willett, W. C., Manson, J. E., Stampfer, M. J., Hennekens, C., Rosner, B., and Speizer, E E. (1995). The use of estrogens and progestins and the risk of breast cancer in postmenopausal women. N. Engl. J. Med. 332, 1589-1593. 4. Barzel, U. S. (1988). Estrogens in the prevention and treatment of postmenopausal osteoporosis. Am. J. Med. 85, 847-850. 5. Grodstein, F., Stampfer, M. J., Manson, J. E., Colditz, G. A., Willett, W. C., Rosner, B., Speizer, E E., and Hennekens, C. H. (1996). Postmenopausal estrogen and progestin use and the risk of cardiovascular disease. N. Engl. J. Med. 335, 453-461. 6. McDonnell, D. P., Vegeto, E., and Gleeson, M. A. G. (1993). Nuclear hormone receptors as targets for new drug discovery. Bio/Technology 11, 1256-1260. 7. Brzozowski, A. M., Pike, A. C., Dauter, Z., Hubbard, R. E., Bonn, T., Engstr6m, O., Ohman, L., Greene, G. L., Gustafsson, J. A., and Carlquist, M. (1997). Molecular basis of agonism and antagonism in the oestrogen receptor. Nature (London) 389, 753-758. 8. Williams, S. P., and Sigler, P. B. (1998). Atomic structure of progesterone complexed with its receptor. Nature (London) 393, 392-396. 9. Giangrande, P. H., and McDonnell, D. P. (1999). The A and B isoforms of the human progesterone receptor: Two functionally different transcription factors encoded by a single gene. Recent Prog. Horm. Res. 54, 291-314. 10. Tzukerman, M. T., Esty, A., Santiso-Mere, D., Danielian, P., Parker, M. G., Stein, R. B., Pike, J. W., and McDonnell, D. P. (1994). Human estrogen receptor transactivational capacity is determined by both cellular and promoter context and mediated by two functionally distinct intramolecular regions. Mol. Endocrinol. 8, 21-30. 11. Gronemeyer, H. (1992). Control of transcription activation by steroid hormone. FASEB J. 6, 2524-2529. 12. Kumar, V., Green, S., Stack, G., Berry, M., Jin, J.-R., and Chambon, P. (1987). Functional domains of the human estrogen receptor. Cell (Cambridge, Mass.) 51, 941-951. 13. Schwabe, J. W. R., Chapman, L., Finch, J. T., and Rhodes, D. (1993). The crystal structure of the estrogen receptor DNA-binding domain bound to DNA: How receptors discriminate between their response elements. Cell (Cambridge, Mass.) 75, 567-578. 14. O'Malley, B. W., Tsai, S. Y., Bagchi, M., Weigel, N. L., Schrader, W. T., and Tsai, M.-J. (1991). Molecular mechanism of action of a steroid hormone receptor. Recent Prog. Horm. Res. 47, 1-26.

10 15. Bagchi, M. K., Tsai, S. Y., Tsai, M.-J., and O'Malley, B. W. (1990). Identification of a functional intermediate in receptor activation in progesterone-dependent cell-free transcription. Nature (London) 345, 547-550. 16. Allan, G. E, Tsai, S. Y., Tsai, M.-J., and O'Malley, B. W. (1992). Ligand-dependent conformational changes in the progesterone receptor are necessary for events that follow DNA binding. Proc. Natl. Acad. Sci. US.A. 89, 11750-11754. 17. Bagchi, M. K., Elliston, J. F., Tsai, S. Y., Edwards, D. P., Tsai, M.-J., and O' Malley, B. W. (1988). Steroid hormone-dependent interaction of human progesterone receptor with its target enhancer element. Mol. Endocrinol. 2, 1221-1229. 18. Clark, J. H., and Peck, E. J. (1979). "Female Sex Steroids: Receptors and Function," 1st ed., Vol. 14. Springer-Verlag, New York. 19. Jordan, V. C., ed. (1996). "Tamoxifen: A Guide for Clinicians and Patients." PRR, Huntington, N.Y. 20. Fisher, B., Costantino, J. P., Wickerham, D. L., Redmond, C. K., Kavanah, M., Cronin, W. M., Vogel, V., Robidoux, A., Dimitrov, N., Atkins, J., Daly, M., Wieand, S., Tan-Chiu, E., Ford, I., and Wolmark, N. (1998). Tamoxifen for prevention of breast cancer: Report of the National Surgical Adjuvant Breast and Bowel Project P-1 Study. J. Natl. Cancer Inst. 90, 1371-1388. 21. Love, R. R., Mazess, R. B., Barden, H. S., Epstein, S., Newcomb, P. A., Jordan, V. C., Carbone, P. P., and DeMets, D. L. (1992). Effects of tamoxifen on bone mineral density in postmenopausal women with breast cancer. N. Engl. J. Med. 326, 852-856. 22. Love, R. R., Wiebe, D. A., Newcombe, P. A., Cameron, L., Levanthal, H., Jordan, V. C., Feyzi, J., and DeMets, D. L. (1991). Effects of tamoxifen on cardiovascular risk factors in postmenopausal women. Ann. Intern. Med. 115, 860-864. 23. Lessey, B. A., Alexander, P. S., and Horwitz, K. B. (1983). The subunit characterization of human breast cancer progesterone receptors: Characterization by chromatography and photoaffinity labelling. Endocrinology (Baltimore) 112, 1267-1274. 24. Vegeto, E., Shahbaz, M. M., Wen, D. X., Goldman, M. E., O'Malley, B. W., and McDonnell, D. P. (1993). Human progesterone receptor A form is a cell- and promoter-specific repressor of human progesterone receptor B function. Mol. Endocrinol. 7, ! 244-1255. 25. Jackson, T. A., Richer, J. K., Bain, D. L., Takimoto, G. S., Tung, L., and Horwitz, K. B. (1997). The partial agonist activity of antagonistoccupied steroid receptors is controlled by a novel hinge domainbinding coactivator L7/SPA and the corepressors NCoR or SMRT. Mol. Endocrinol. 11, 693-705. 26. Conneely, O. M., Dobson, A. D. W., Huckaby, C., Carson, M. A., Maxwell, B. L., Toft, D. O., Tsai, M.-J., Schrader, W. T., and O' Malley, B. W. (1988). Structure-function relationships of the chicken progesterone receptor. In "Steroid Hormone Action," G. Ringold, ed., pp. 227-236. Alan R. Liss, New York. 27. Wen, D. X., Xu, Y.-E, Mais, D. E., Goldman, M. E., and McDonnell, D. P. (1994). The A and B isoforms of the human progesterone receptor operate through distinct signaling pathways within target cells. Mol. Cell. Biol. 14, 8356-8364. 28. Kuiper, G. G. J. M., Enmark, E., Pelto-Huikko, M., Nilsson, S., and Gustafsson, J.-A. (1996). Cloning of a novel estrogen receptor expressed in rat prostate and ovary. Proc. Natl. Acad. Sci. U.S.A. 93, 5925 -5930. 29. Kuiper, G. G. J. M., Carlsson, B., Grandien, K., Enmark, E., H~iggblad, J., Nilsson, S., and Gustafsson, J.-A. (1997). Comparison of the ligand binding specificity and transcript tissue distribution of estrogen receptors ce and ft. Endocrinology (Baltimore) 138, 863-870. 30. Shughrue, P. J., Komm, B., and Merchenthaler, I. (1996). The distribution of estrogen receptor-fl mRNA in the rat hypothalamus. Steroids 61,678-681. 31. Shughrue, P. J., Lane, M. V., and Merchenthaler, I. (1997). Comparative

DONALD P. MCDONNELL

32. 33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

distribution of estrogen receptor-ce and fi mRNA in the rat central nervous system. J. Comp. Neurol. 388, 507-525. Couse, J. F., and Korach, K. S. (1999). Estrogen receptor null mice: What have we learned and where will they lead us? Endocr. Rev. 20, 358-417. McDonnell, D. P., Clemm, D. L., Hermann, T., Goldman, M. E., and Pike, J. W. (1995). Analysis of estrogen receptor function in vitro reveals three distinct classes of antiestrogens. Mol. Endocrinol. 9, 659-668. Jordan, V. C. (1992). The strategic use of antiestrogen to control the development and growth of breast cancer. Cancer (Philadelphia) 70, 977-982. Kilackey, M. A., Hakes, T. B., and Pierce, V. K. (1985). Endometrial adenocarcinoma in breast cancer patients receiving antiestrogens. Cancer Treat. Rep. 69, 237-238. Gottardis, M. M., Robinson, S. P., Satyaswaroop, P. G., and Jordan, V. C. (1988). Contrasting actions of tamoxifen on endometrial and breast tumor growth in the athymic mouse. Cancer Res. 48, 812-815. Allan, G. F., Leng, X., Tsai, S. Y., Weigel, N. L., Edwards, D. P., Tsai, M.-J., and O'Malley, B. W. (1992). Hormone and antihormone induce distinct conformational changes which are central to steroid receptor activation. J. Biol. Chem. 267, 19513-19520. Vegeto, E., Allan, G. F., Schrader, W. T., Tsai, M.-J., McDonnell, D. P., and O'Malley, B. W. (1992). The mechanism of RU486 antagonism is dependent on the conformation of the carboxy-terminal tail of the human progesterone receptor. Cell (Cambridge, Mass.) 69, 703-713. Wagner, B. L., Pollio, G., Leonhardt, S., Wani, M. C., Lee, D. Y.-W., Imhof, M. O., Edwards, D. P., Cook, C. E., and McDonnell, D. P. (1996). 16a-Substituted anologs of the antiprogestin RU486 induce a unique conformation in the human progesterone receptor resulting in mixed agonist activity. Proc. Natl. Acad. Sci. U.S.A. 93, 87398744. Mahfoudi, A., Roulet, E., Dauvois, S., Parker, M. G., and Wahli, W. (1995). Specific mutations in the estrogen receptor change the properties of antiestrogens to full agonists. Proc. Natl. Acad. Sci. U.S.A. 92, 4206-4210. Tanenbaum, D. M., Wang, Y., Williams, S. P., and Sigler, E B. (1998). Crystallographic comparison of the estrogen and progesterone receptor's ligand binding domains. Proc. Natl. Acad. Sci. U.S.A. 95, 59986003. Shiau, A. K., Barstad, D., Loria, P. M., Cheng, L., Kushner, E J., Agard, D. A., and Greene, G. L. (1998). The structural basis of estrogen receptor/coactivator recognition and the antagonism of this interaction by tamoxifen. Cell (Cambridge, Mass.) 95, 927-937. Delmas, P. D., Bjarnason, N. H., Mitlak, B. H., Ravoux, A. C., Shah, A. S., Huster, W. J., Draper, M., and Christiansen, C. (1997). Effects of raloxifene on bone mineral density, serum cholesterol concentrations, and uterine endometrium in postmenopausal women. N. Engl. J. Med. 337, 1641 - 1647. Sato, M., Rippy, M. K., and Bryant, H. U. (1996). Raloxifene, tamoxifen, nafoxidine, or estrogen effects on reproductive and nonreproductive tissues in ovariectomized rats. FASEB J. 10, 905-912. Ashby, J., Odum, and Foster, J. R. (1997). Activity of raloxifene in immature and ovariectomized rat uterotrophic assays. Regul. Toxicol. Pharmacol. 25, 226-231. Norris, J. D., Paige, L. A., Christensen, D. J., Chang, C.-Y., Huacani, M. R., Fan, D., Hamilton, P. T., Fowlkes, D. M., and McDonnell, D. E (1999). Peptide antagonists of the human estrogen receptor. Science 285, 744-746. Paige, L. A., Christensen, D. J., Gron, H., Norris, J. D., Gottlin, E. B., Padilla, K. M., Chang, C.-Y., Ballas, L. M., Hamilton, E T., and McDonnell, D. E (1999). Estrogen receptor (ER) modulators each induce distinct conformational changes in ERa and ERfl. Proc. Natl. Acad. Sci. U.S.A. 96, 3999-4004.

11

CHAPTER 1 Estrogen and Progesterone Receptors 48. Wijayaratne, A. L., Nagel, S. C., Paige, L. A., Christensen, D. J., Norris, J. D., Fowlkes, D. M., and McDonnell, D. P. (1999). Comparative analyses of the mechanistic differences among antiestrogens. Endocrinology 140, 5828-5840. 49. Horwitz, K. B., Jackson, T. A., Bain, D. L., Richer, J. K., Takimoto, G. S., and Tung, L. (1996). Nuclear receptor coactivators and corepressors. Mol. Endocrinol. 10, 1167-1177. 50. Smith, C. L., Nawaz, Z., and O'Malley, B. W. (1997). Coactivator and corepressor regulation of the agonist/antagonist activity of the mixed antiestrogen, 4-hydroxytamoxifen. Mol. Endocrinol. 11, 657- 666. 51. Wagner, B. L., Norris, J. D., Knotts, T. A., Weigel, N. L., and McDonnell, D. P. (1998). The nuclear corepressors NCoR and SMRT are key regulators of both ligand- and 8-bromo-cyclic AMP-dependent transcriptional activity of the human progesterone receptor. Mol. Cell. Biol. 18, 1369-1378. 52. Onate, S. A., Tsai, S., Tsai, M.-J., and O'Malley, B. W. (1995). Sequence and characterization of a coactivator for the steroid hormone receptor superfamily. Science 270, 1354-1357. 53. Spencer, T. E., Jenster, G., Burcin, M. M., Allis, C. D., Zhou, J., Mizzen, C. A., McKenna, N. J., Onate, S. A., Tsai, S. Y., Tsai, M.-J., and

54.

55.

56.

57.

O'Malley, B. W. (1997). Steroid receptor coactivator- 1 is a histone acetyltransferase. Nature (London) 389, 194-198. Shibata, H., Spencer, T. E., Onate, S. A., Jenster, G., Tsai, S. Y., Tsai, M.-J., and O'Malley, B. W. (1997). Role of co-activators and corepressors in the mechanism of steroid/thyroid receptor action. Recent Prog. Horm. Res. 52, 141-165. H6rlein, A. J., N~iar, A. M., Heinzel, T., Torchia, J., Gloss, B., Kurokawa, R., Ryan, A., Kamei, Y., S6derstr6m, M., Glass, C. K., and Rosenfeld, M. G. (1995). Ligand-independent repression by the thyroid hormone receptor mediated by a nuclear receptor co-repressor. Nature (London) 377, 397-403. Chen, J. D., and Evans, R. M. (1995). A transcriptional co-repressor that interacts with nuclear hormone receptors. Nature (London) 377, 454-457. Lavinsky, R. M., Jepsen, K., Heinzel, T., Torchia, J., Mullen, T.-M., Schiff, R., Del-Rio, A. L., Ricote, M., Ngo, S., Gemsch, J., Hilsenbeck, S. G., Osborne, C. K., Glass, C. K., Rosenfeld, M. G., and Rose, D. W. (1998). Diverse signaling pathways modulate nuclear receptor recruitment of N-CoR and SMRT complexes. Proc. Natl. Acad. Sci. U.S.A. 95, 2920-2925.

~HAPTER

Ovarian Anatomy and Physiology GREGORY F.

ERICKSON

Department of Obstetrics and Gynecology, University of California, San Diego, La Jolla, California 92093

VI. Accelerated Loss in OR: Activin Hypothesis VII. New Data on the Effects of Activin VIII. Statement of Conclusion References

I. Introduction II. Statement of the P r o b l e m

III. The Primordial Follicle IV. The Adult Ovary V. Ovary Reserve

I. INTRODUCTION

II. STATEMENT OF THE P R O B L E M

A characteristic feature of reproductive capacity in women is its cyclical nature, a feature that is strikingly reflected by the growth and development of oocytes and follicles within the ovaries [1]. Typically, the tissues of the adult ovaries undergo dramatic cyclical changes, which in turn are reflected in cyclical changes in the production of the steroid hormones, estradiol (E 2) and progesterone (p4), as well as key regulatory proteins, such as inhibin. The regulated expression of these ligands is critically important in the mechanisms that underlie fertility in women, including the ovulation of an oocyte with the potential for producing a normal embryo. The process begins at puberty with the first cycle and ends at the menopause with the permanent cessation of menses. The causative event in the menopause is the disappearance of primordial follicles in the ovaries, i.e., the loss of ovary reserve (OR). Consequently, the size of a woman's OR (number of primordial follicles) has great impact on when the menopause begins. Therefore, to understand the menopause from either a physiological or a pathophysiological perspective, we must understand the interactions between OR, folliculogenesis, and the menstrual cycle during aging. This chapter reviews what is currently known about these relationships. MENOPAUSE: BIOLOGY AND PATHOBIOLOGY

Today, the menopause occurs in most women at --~51 years of age. Demographic studies have demonstrated that the mean life expectancy of women in developed countries [1] has increased from - 4 5 years in 1850 to - 8 2 years in 1998 (Fig. 1) [ l a]. This is an important observation because it indicates that most women today will live one-third to onehalf of their lives postmenopausally, i.e., they will live - 30 years after the menopause. Clinicians can therefore expect to extend care increasingly to large numbers of women of advanced reproductive age in whom ovarian dysfunction will be a major cause of infertility and morbidity. If one considers that the vast majority of fertility and gynecologic problems in the aging woman are a direct consequence of the age-related decrease in ovary reserve, it becomes apparent that the disappearance of primordial follicles is one of the critical events in the life of all women. One female function most adversely affected by the agerelated decrease in OR is decreased fecundity. The basis for this age-related change is the failure of dominant follicles to release eggs that can undergo normal embryonic development [2-5]. Results from in vitro fertilization (IVF) [6,7] have demonstrated that this decrease becomes particularly evident in patients after 36 years of age, when pregnancy 13


1

4

G

R

E

G

FIGURE 1 Changesin the life expectancy and age of the menopause in women over the past 150 years. From Nachtigall [la], with permission.

rates with self-produced oocytes decline sharply by --~65%, i.e., from ---25% per transfer in women ~35 years of age [16]. Thus, one is led to the conclusion that (1) the endocrine and gametogenic function of dominant follicles can become dissociated in women after --~36 years of age and (2) the aberrant expression of cellular responses in the egg would appear to be the basis for the age-related decrease in fecundity. Understanding how the developmental potential of the

O

R

Y

F. ERICKSON

aged oocyte is altered independently of changes in the granulosa and theca cells is a fundamental question in ovary research. Although relatively little is known about this problem, an interesting role for ovary reserve (the number of primordial follicles) has been suggested from analysis of folliclestimulating hormone (FSH) and inhibin levels and pregnancy rates. An important point to emerge from the clinical studies is that the number of follicles within the ovaries, not oocyte age, is the main determinant predicting pregnancy in older women [9]. That is, the incidence of pregnancy with self-produced oocytes is highest in older women with adequate OR, i.e., those with no significant elevation in plasma FSH levels and whose ovaries contain a comparatively large number of mature Graafian follicles [6,13,17]. Given this relationship, it is not unreasonable to propose that the selective deteriorative changes that occur in aging oocytes are either correlated with or causally connected to a significant decrease in OR, rather than aging itself. There is a fundamental question: How does this occur?

IIl. THE PRIMORDIAL

FOLLICLE

Before addressing this question, we must understand some basic biology of the primordial follicles or OR. The primordial follicles represent a pool of nongrowing follicles from which all dominant preovulatory follicles are selected [1]. Thus, primordial follicles are, in a real sense, the fundamental reproductive units of the ovary. Morphologically, each primordial follicle is composed of an outer single layer of squamous epithelial cells that are termed granulosa or follicle cells, and a small (approximately 15/xm in diameter) immature oocyte arrested in the dictyotene stage of meiosis; both cell types are enveloped by a thin, delicate membrane called the basal lamina (Fig. 2). By virtue of the basal lamina, the granulosa and the oocyte exist in a microenvironment in which direct contact with other cells does not occur. Although small capillaries are occasionally observed in proximity to primordial follicles, these follicles do not have an independent blood supply [1]. The mean diameter of a nongrowing primordial follicle is 29/xm [18]. All the primordial follicles present in a woman's ovaries are formed before birth. Developmentally, the primordial follicles are formed in the cortical cords of the fetal ovaries between the sixth and ninth months of gestation [1]. During this period, all the germ cells are stimulated to initiate meiosis. Because the oocytes in the primordial follicles have entered meiosis, all oocytes that are capable of participating in reproduction during a woman's life are formed at birth, i.e., the human ovaries acquire a lifetime quota of eggs before birth. Soon after primordial follicle formation, some are recruited (activated) to initiate growth. As successive recruitment proceeds over time, the size of the pool of primordial follicles becomes progressively smaller (Fig. 3) [ 18a]. Between the times of birth and menarche the number

CHAPTER2 Ovarian Anatomy and Physiology

15 of primordial follicles (and thus oocytes) decreases from several million to several hundred thousand (Fig. 3). As a woman ages, the number of primordial follicles (OR) continues to decline, until at menopause they are difficult to find (Fig. 4) [ 18b].

IV. THE ADULT OVARY

FIGURE 2 Electron micrograph of a human primordial follicle showing oocyte nucleus (N), Balbiani body (.), and granulosa or follicle cells (arrowheads). From [ 1], Erickson, G. F. (1995). The Ovary: Basic Principles and Concepts. In "Endocrinology and Metabolism" (Felig, E, Baxter, J. D., Broadus, A. E., Frohman, L. A., Eds.), 3rd ed., pp. 973-1015. McGrawHill, with permission.

7.0'

6.0-

_o ; 5.0"

~ U

4.0'

1 ~

:3.0'

N 2.0" ~ Z

1.0" 0.6" 0.3.

g

9i :t 5 ~b

AGE(months pc) ]

2"o

-3"o-

AGE(yeors)

BIRTH

FIGURE 3 Changes in the total number of oocytes in human ovaries during aging. In the embryo at early to midgestation, the number of oocytes increases to almost 7 million. Shortly thereafter, the number falls sharply to about 2 million at birth. The enormous loss (--~70%) of oocytes in the embryo between 6 and 9 months is caused by apoptosis. The store of eggs continues to diminish with age until no oocytes are detected in the ovaries at about 50 years of age. From [18a], Baker, T. G. (1971). Radiosensitivity of mammalian oocytes with particular reference to the human female. Am. J. Obstet. Gynecol. 110, 746-761, with permission.

In this section, we will deal with the anatomy and physiology of folliculogenesis as it occurs in normal women during the reproductive years. We will focus our attention on the manner in which the developmental program is expressed in a recruited primordial follicle as it develops to the ovulatory stage or dies by atresia. An underlying principle of the human ovaries is that of the 2 million primordial follicles, only 400 of so will complete their development and undergo ovulation and corpus luteum formation; all others (99.9%) will die by atresia after recruitment (Fig. 5) [ 18c]. Therefore, the very essence of folliculogenesis is selection.

A. Anatomy The adult human ovary is a mass of follicles, luteal tissue, blood vessels, nerves, and connective tissue elements, all of which form a relatively heterogeneous assemblance of histological units. It is the continuous and progressive change in the follicles and corpora lutea that gives rise to the cyclical changes in the menstrual cycle. During the reproductive years, the normal human ovaries are oval-shaped bodies that each measure 2.5-5.0 cm in length, 1.5-3 cm in width, and 0.6-1.5 cm in thickness [1]. The medial edge of the ovary is attached by the mesovarium to the broad ligament, which extends from the uterus laterally to the wall of the pelvic cavity. The surface of the ovary is covered by an epithelial layer of cuboidal cells resting on a basement membrane. This layer, termed the germinal or serous epithelium, is continuous with the peritoneum. Underlying the serous epithelium is a layer of dense connective tissue termed the tunica albuginea. The ovary is organized into two principal parts: a central zone, the medulla, which is surrounded by a particularly prominent peripheral zone, the cortex (Fig. 6). Embedded in the connective tissue of the cortex are the follicles containing the female gametes, oocytes. The number and size of the follicles vary depending on the age and reproductive state of the female. The existence of follicles of different sizes (primordial, primary, secondary, tertiary, Graafian, and atretic)reflects specific changes associated with their growth, development, and fate. At the end of the follicular phase of the menstrual cycle, the Graafian follicle that reaches maturity secretes its ovum into the peritoneal cavity (Fig. 6). After ovulation, the wall of the ovulating follicle develops into an endocrine structure, the corpus luteum. If implantation does

16

GREGORY F. ERICKSON

FIGURE 4 Photomicrographs of the cortex of human ovaries from birth to 50 years of age. Arrowheads indicate small, nongrowing primordial follicles with a single layer of squamous granulosa cells. From Erickson [ 18b], with permission.

not occur, the corpus l u t e u m deteriorates and eventually bec o m e s a n o d u l e of dense connective tissue t e r m e d the c o r p u s a l b i c a n s . O t h e r cells in the cortex are the s t e r o i d o g e n i c cells t e r m e d s e c o n d a r y interstitial cells. T h e s e cells, which are derived f r o m the theca c o m p a r t m e n t of atretic follicles, are found in nests or cords t h r o u g h o u t the life of the female. At the m e d i a l b o r d e r of the cortex is a mass of loose connective tissue, the medulla. This tissue contains a n e t w o r k of

c o n v o l u t e d b l o o d vessels and associated nerves that pass through the connective tissue toward the cortex (Fig. 6). A distinct group of t e s t o s t e r o n e - p r o d u c i n g cells, the hilus cells, lie in the m e d u l l a at the hilum of the ovary [19]. The arterial supply to the ovary originates from two prin-

Cumulus with egg

Ovulation

2,000,000 Total

........:." ...~

Corpus L-uteum

"i~."..' ': : ": i:' ".: ". :7~.: : ..'! ~..

(~

Primordial Follicles -

ATRESIA

o,oo,,o~

_./

,o,,,.,.

400 1

,F S H

Ovulation--~Corpus Luteum

FIGURE 5 Folliculogenesis is a highly selective process. Of the 2 million primordial follicles at birth, only 400 or so are brought to ovulation and luteinization by FSH and LH. From [18c], Soules, M. R., and Bremmer, W. J. (1982). The menopause and climacteric: Endocrinologic basis and associated symptomatology. J. Am. Geriatr. Soc. 30, 547-561, with permission.

Ear ly Tertiary Follicle

Inter|titiol Cells

\ ry

Follicle

Cortex

Primordial Follicles

FIGURE 6 Diagram summarizing the anatomy and histology of the human ovary during the reproductive years. Development of the follicles and

corpus luteum occurs within the cortex, whereas the spiral arteries, hilus cells, and autonomic nerves are located in the medulla. From [1], Erickson, G. F. (1995). The Ovary: Basic Principles and Concepts. In "Endocrinology and Metabolism" (Felig, P., Baxter, J. D., Broadus, A. E., Frohman, L. A., Eds.), 3rd ed., pp. 973-1015. McGraw-Hill, with permission.

CHAPTER2 Ovarian Anatomy and Physiology cipal sources: one, the ovarian artery, arises from the abdominal aorta; the other is derived from the uterine artery [ 1]. These two vessels, which enter the mesovarium from opposite directions, form an anastomotic trunk and become a common vessel called the r a m u s o v a r i c u s artery. At frequent intervals this artery gives rise to a series of primary branches, which enter the hilum like teeth on a rake. In the hilum, numerous secondary and tertiary branches are given off to supply the medulla (Fig. 6).

17 THECA INTERNA

Luteinizing HormoneI

f

Cholesterol

~lic I..... ! l A"P

,

I Androstenedionel

1I

B. P h y s i o l o g y Typically, the human ovaries produce a single dominant follicle that secretes into the oviduct a mature egg that is ready to be fertilized at the end of the follicular phase of the menstrual cycle. Each dominant follicle begins with the recruitment of a primordial follicle into the pool of growing follicles. It is not known exactly how recruitment occurs, but it appears to be controlled by one or more local ovarian regulatory factors by autocrine/paracrine mechanisms. In a broad sense, all growing follicles can be divided into two groups, healthy and atretic, according to whether apoptosis (programmed cell death) is present in granulosa cells [20,21 ]. As a consequence of successive recruitments, the ovaries appear to always contain a pool of small Graafian follicles from which a prospective dominant follicle can be selected. Once selected, a dominant follicle typically grows and develops to the preovulatory state. Those follicles that are not selected become committed to die by atresia. 1. ENDOCRINOLOGY OF FOLLICULOGENESIS

Regardless of age, follicle growth and development are brought about by the combined action of FSH and luteinizing hormone (LH) on the follicle cells. FSH and LH bind to specific high-affinity receptors on the membranes of the granulosa and theca interstitial cells, respectively. The interaction of these ligands with their receptors activates signal transduction pathways that stimulate mitosis and differentiation responses in the granulosa and theca cells [22,23]. Physiologically, these signaling pathways act in parallel to regulate the expression of specific genes in a precise quantitative and temporal fashion. There are two major endocrine responses associated with folliculogenesis. The first is that the actions of FSH and LH stimulate the production of large amounts of estradiol by the dominant follicle. This important gonadotropin-dependent response is called the twogonadotropin/two-cell concept Fig. 7) [23a]. Because the estradiol response appears to be specific to a dominant follicle, the levels of plasma estradiol can be used as a marker for the status of the dominant follicle. The second endocrine response is the marked increase in the production of inhibin by FSH [24]. With respect to aging, observations support the possibility that localized changes in inhibin production may

1 J~ Androstenedione J~-/ m ~--~"~Basement Lami n

a

~

'

~

j/ ~

~

"

r">! Estrogen l //,~,~ . . . . ', tc, irculatlon/ ~

[Androstenedionel ~Aromatization

Induction

Follicle

Stimulating Hormone

l

( Follicular fluid )

GRANULOSA CELLS

FIGURE 7 The two-gonadotropin/two-cell concept of follicle estrogen production. From Erickson [23a], with permission.

play a role in the accelerated loss of OR. We will return to this issue later. 2. CHRONOLOGY

In women, folliculogenesis is a very long process [22]. In each menstrual cycle, the dominant follicle that is selected to ovulate originates from a primordial follicle that was recruited to grow about 1 year earlier. Whatever the course of development or the final destiny (atresia or ovulation), all follicles undergo various progressive changes (Fig. 8). The very early stages of folliculogenesis (class 1, primary and secondary; class 2, early tertiary) proceed very slowly. Consequently, it requires -->300 days for a recruited primordial follicle to complete the preantral or hormone-independent period. The basis for the slow growth is the very long doubling time (---250 hr) of the granulosa cells. When follicular fluid begins to accumulate at the class 2 stage, the Graafian follicle begins to expand relatively rapidly (Fig. 8). As the antral (hormone-dependent) period proceeds, the Graafian follicle passes through the small, (classes 3, 4, and 5), medium (classes 6 and 7), and large (class 8) stages. A dominant follicle that survives to the ovulatory stage requires about 4 0 - 5 0 days to complete the whole antral period. Selection of the dominant follicle is one of the last steps in

18

GREGORYF. ERICKSON

co~)'-"

'3

c

.LASS 4

L

c Ass 8

ZO

oV/; /

ol ~ ....;I / _ ~

; '--/#l

/ # .~f

~\

-

,~-

'

~

oA -~\+. "~ V-~X

%\

'

\

_.

-

~

,'

\

,1t~ # -o.\ ~

u'24%~ " ATRESIL/ ' 5 0 .... 9 mm in diameter, regardless of the stage in the cycle. FSH is obligatory for follicle selection, and no other ligand by itself can serve in this regulatory capacity. Physiologically, the mechanism of selection is causally connected to the secondary rise in plasma FSH (the primary FSH rise being the midcycle preovulatory surge of FSH and LH). The secondary rise in plasma FSH begins a few days before the progesterone and estradiol concentrations reach baseline values at the end of luteal phase, and it continues through the first week of the follicular phase (Fig. 10) [24c]. The importance of the secondary rise in FSH is demonstrated by the fact that the dominant follicle will undergo atresia if the FSH levels are decreased. Consequently, the secondary rise in FSH is obligatory for the selection of a dominant follicle that will ovulate in the next cycle. One of the major conse-

~

THECA INTERSTITIAL

CELLS

('FOLLICULAR) ANTRUM \ FLUID J

LOOSECONNECTIVE CORONARADIATA GRANULOSACELLS-

~ ~ ~'B '

CAPILLARIES ZONAPELLUCIDA CUMULUSOOPHOROUS GRANULOSACELLS THECAEXTERNA

FIGURE 9 Diagrammatic representation of the histologic architecture of a Graafian follicle. Reprinted from [24b], Mol. Cell. Endocrinol. 29, G. F. Erickson. Primary cultures of ovarian cells in serum-free medium as models of hormone-dependent differentiation, 2149. Copyright 1983 with permission from Elsevier Science.

CHAPTER2 Ovarian Anatomy and Physiology

19

quences of the secondary FSH rise is that a critical threshold level of FSH accumulates in the follicular fluid of the chosen follicle. In normal class 5 to class 8 follicles, the mean concentration of follicular fluid FSH increases from --~1.3 mIU/ ml (--~58 ng/ml) to --~3.2 mIU/ml (--~143 ng/ml) through the follicular phase [25]. In contrast, FSH concentrations are low or undetectable in the microenvironment of the nondominant cohort follicles. Thus, the selection and the continued growth of a dominant follicle involve a progressive increase in the concentration of FSH within its microenvironment. Once activated, the dominant follicle becomes dependent on FSH for its survival. The regulation of FSH levels in follicular fluid is totally obscure. FSH triggers a marked activation of mitosis and differentiation of the granulosa cells, which in turn is reflected in a progressive increase in estradiol and inhibin synthesis and follicular fluid accumulation (Fig. 10). One of the effects of the increased estradiol and inhibin production is that the secondary rise in FSH is suppressed (Fig. 10). When this oc-

curs, the concentration of FSH falls below threshold levels and the development of the nondominant follicles stops. It is noteworthy that mitosis in these atretic follicles can be markedly stimulated by treatment with human menopausal gonadotropin (hMG) during the early follicular phase. Thus, if FSH levels within the microenvironment are increased, the nondominant follicle could perhaps be rescued from atresia.

C. The Role of FSH The major FSH-dependent changes that occur during the development of the dominant follicle are summarized in Fig. 11 25a]. The granulosa cells are the only cell types known to express FSH receptors. It follows, therefore, that FSH-mediated effects in the dominant follicle are at the levels of the granulosa cells. In dominant follicles, the FSHinduced differentiation of the granulosa cells involves three major responses, increased steroidogenic potential, mitosis, and LH receptors. 1. S T E R O I D O G E N I C P O T E N T I A L

a A

75

xt

"//,. LH

,,ro~ +

Postmenopausal W o m e n POF

I I

~r

GnRH X

F

~FSH~,~ "~tLH~V~ ~Estrogen

@

9

@

z

GnRH X-

LH

~(Estrogen

FIGURE 7 Summary illustration of hormonal changes during reproductive aging of women. Key: e, Dominant follicle; 1", level elevated in proportion to length of arrow (bold indicates significant elevation); $, level reduced in proportion to length of arrow (bold indicates significant reduction); arrows that are N-shaped indicate that a level is not always elevated; zig-zag lines indicate oscillating levels and pulsations (width and height symbolize frequency and amplitude, respectively); slanted lines crossed by two parallel bars indicate a disruption of an effect. (A) Premenopause: hFSH under control of GnRH (+ effect, bold arrow signifies strong effect), inhibins A and B from developing follicles ( - effect, dashed bold arrow signifies strong negative effect), and estrogen ( - effect). (B) Early perimenopause: hFSH begins to be variable while inhibin declines (although the decline in inhibin is somewhat questionable). Estrogen concentrations rise and hFSH is variably higher resulting in more follicle development and higher estrogen levels. (C) Late perimenopause: hFSH levels high, inhibin levels lower, and HLH variably higher. A dominant follicle still may develop. (D) Postmenopause: hFSH levels high with pulsatile pattern similar to that in a younger woman but with greater amplitude (note: pulsating hormone levels shown only for hFSH and hLH in postmenopausal women); same for hLH. GnRH patterns variable, inhibin absent, and estrogen very low due to absence of dominant follicle; no follicle development. (E) Postmenopausal elderly: hFSH levels high but pulsation rate slower; same for hLH. GnRH low and quite variable, no inhibin, and very low estrogen. (F) Postmenopause, premature ovarian failure (POF): high hFSH levels but pulsation pattern similar to that of a younger woman, as is hLH. No inhibin. GnRH levels and patterns similar to those of younger women. A portion of this illustration is based on Figure 7 from [20], Prior, J. (1998). Perimenopause: the complex endocrinology of the menopausal transition. Endocrine Reviews 19, 397-428. 9 The Endocrine Society.

Wise et al. support the neuroendocrine hypothesis [15,16]. Soules and colleagues have made a number of observations that are not consistent with either the ovarian hypothesis, in terms of inhibin levels in aging women, or the neuroendocrine hypothesis, in terms of similar GnRH pulsation patterns during aging. Nonetheless, they reach the overall conclusion that the ovarian hypothesis can better predict the changes in the gonadotropin levels and the number of follicles that are a precursor to menopause. Burger and others [79,80] have presented interpretations for the different stages during the transition to menopause. Prior has presented a synthesis of these various proposals

[20]. We have attempted to simplify this summary in Fig. 7. In this illustration, we present six panels illustrating the changing gonadotropin patterns from young premenopausal women to early and late perimenopausal, to postmenopausal, and to elderly postmenopausal women. Also included are young women experiencing premature ovarian failure. In many perimenopausal patients hLH and hFSH ovarian receptors were very low and such receptors were absent in postmenopausal patients [81 ]. This led Vihko to suggest that high serum gonadotropin levels act in concert with low or absent ovarian receptors for these hormones. Santoro demonstrated that there is also an age-related

70

BIRKEN ET AL.

alteration in hypothalamic or pituitary function that acts to decrease what would be even higher hLH and hFSH levels. In studies of women with premature ovarian failure compared to normal age menopausal women, it was apparent that hLH secretion was greater in the younger women along with greater pulse amplitudes, although pulse frequency was similar in both groups [19]. Therefore, the changes in gonadotropin secretion as women age demonstrate a decline in capabilities of the pituitary to produce gonadotropins in response to GnRH and/or a decline in GnRH during aging. These capacity differences in gonadotropin secretion in women with premature ovarian failure as compared to normal age menopausal women are also apparent in the gonadotropin fragment analysis patterns, which will be reviewed later. Gonadotropin bioactivity also increases in menopause along with the total immunoreactive concentration ofgonadotropins [82]. Changes in sialylation (increased sialic acid content) with decreasing circulating steroids contribute to the increase in bioactivity of the gonadotropins after menopause. Studies of biological to immunological gonadotropin ratios show that postmenopausal women display nearly twice the ratio value than do normal young cycling women or older perimenopausal women [82]. Although the high concentrations of hFSH and hLH after menopause are apparent, these vary greatly during the transition to menopause, the perimenopause. Santoro provides an excellent illustration of the changes in gonadotropin and steroid levels among 11 midreproductive age women and 11 perimenopausal women during the menstrual cycle [18] (Fig. 8). The rise in overall hFSH concentrations during the

,50

§ .... ii , ,',

peri I

40

LH

FSH

40 =_

a0

C:~ 30

20

2o

10

I0

E

E

150

PDG

;F i (-~

20

100

E

"~

so

__ 0

I

J

-,s -io

I

-s

=

o

~

DAYS

,~

i~

~-;s

-,o

~--"-

-s

-

~

~

I

1o

I

,~

,

~o

I0

DAYS

FIGURE 8 Daily hLH, hFSH, estrone conjugates (El), and PDG excretions (mean _ SEM) corrected for creatinine and standardized to the day of presumed ovulation (day 0) in 11 regularly menstruating perimenopausal women (open circles) compared with 11 younger women (closed circles). E] was higher in the perimenopausal women (P = 0.023) and integrated PDG was lower (P = 0.015). Reproduced from [18], Santoro, N., Brown, J. R., Adel, T., and Skurnick, J. H. (1996). Characterization of reproductive hormonal dynamics in the perimenopause. J. Clin. Endocrinol. Metab. 81, 1495-1501. 9 The Endocrine Society.

early follicular phase and an overall rise in estrogen in perimenopausal women are apparent in this illustration. Although hFSH increases prior to a rise in hLH levels during perimenopause, a great range of possible values makes such gradual increases of little diagnostic value in determining the perimenopausal stage [20,79,80]. Likewise, estrogen values are quite variable during perimenopause, being even higher in some perimenopausal women than in young, normal cycling women. Despite the use of day-3 serum FSH and/or circulating estrogen concentrations as indicative of the perimenopausal state, these single point measurements are far from conclusive; further exploration should provide new and improved markers that can indicate proximity to menopause.

VI. G O N A D O T R O P I N AS URINARY

FRAGMENTS

ANALYTES

The gonadotropins, being heterodimeric hormones whose subunits are noncovalently bound together, are unstable in urine. Although it is possible to decrease their instability by addition of additives such as glycerol and by minimizing freeze-thaw cycles, intact gonadotropins are not ideal molecules for accurate quantification in urine [83-86]. The gonadotropins undergo proteolytic cleavages during transit through the kidney and their stability in urine is likely to be further reduced by such damage [57,59,64,68,87]. Most commercial gonadotropin assays are certified as quantitatively accurate in blood but only qualitatively accurate in urine due to such problems [88,89]. It is easier to quantify glycoprotein hormones in blood, but each sampling requires a technician or physician or even extremes, such as an indwelling catheter for multiple samplings. Large-scale studies using blood specimens cannot be conducted for this reason, but such studies can readily be performed in urine if an appropriate, highly stable analyte can be identified. The/3 core fragments of the gonadotropins are ideal molecules for urinary measurement. The two gonadotropins of significance to menopause are hFSH and hLH, with the former being the most relevant. Although we have evidence for the presence of a urinary hFSHflcf, only/3 core fragments of hCG and hLH have been isolated, characterized, and measured thus far. The hCGflcf has been used extensively in urinary measurement systems for the past decade. It is extremely stable, as attested to by a variety of groups [72], and is generally present in higher concentrations than heterodimeric hCG in urine. Measurement applications include certain cancers and various problem pregnancies, including Down syndrome. The utility of hCGflcf in the cancer marker field (used only informally for this purpose) is limited to monitoring the recurrence of hCGsecreting tumors after therapy. The higher molar concentration of this fragment as compared to hCG in urine leads to greater sensitivity of detection. The hLHflcf epitope is similarly stable in urine. Current assays for the hLHflcf are based

CHAPTER4 Gonadotropins and Menopause" New Markers on the isoform isolated from pituitary tissue, the urinary form of this molecule has not yet been purified [59,60,63]. hLH is more difficult to measure accurately in urine than is hCG because it is less stable than hCG and frequently displays isoforms not recognized by many monoclonal antibody-based immunochemical measuring systems [63,71,9093] (see Fig. 6). In addition, hLH may appear to be absent from urine after having completely dissociated into subunits or after having been completely metabolized to smaller fragments. FSH also presents measuring problems due to dissociation of subunits in urine. This problem has occurred in analyses of FSH in nonhuman primates, which are used as models for studies of human reproduction. One lab has recently proposed boiling all monkey urine samples to cause complete dissociation of FSH into subunits and then measuring the released fl subunit [94]. However, if any heterodimeric hormone is proteolytically cleaved, such boiling may cause irreversible loss of some epitopes and total fl concentration may not represent total FSH originally present in the specimen. In all of these situations,/3 core fragment measurement provides the most stable and consistent reflection of gonadotropins in circulation. These fragments have reached an end point in proteolysis and are not further degraded in urine as long as microbial growth is inhibited. While acknowledging that FSHflcf would most likely provide the best menopausal marker, because we have already developed an immunochemical system for measurement of hLHflcf, we have applied these measurements first to assess their utility in studies of women in menopausal transition. As previously described, we demonstrated that hLHflcf appeared in high concentrations in women during the menstrual cycle starting on the day of the LH surge, peaking 1-2 days after the surge [63]. Figure 9 shows the hormone profiles of 15 normal cycling premenopausal women and the temporal relationship in the appearance of hLH, hLHfl, hCG, and hLHflc. The X axis is normalized to day 0, this being the zenith of the hLH surge appearance in urine. This delayed appearance of hLHflcf after the hLH surge suggests that circulating hLH is sequestered in a body compartment (likely kidney tissue) and is excreted after 2 4 - 4 8 hr of proteolytic processing, hLHflcf can be easily measured in urine but not in serum. This is shown in Figure 10 and implies that the kidney creates the fragment by absorbing hLH from the blood and excreting it later into the urine. It is also possible that some of the hLHcf is directly secreted by the pituitary into the bloodstream but is cleared so rapidly that its circulating concentration remains very low. Our original hypothesis was that postmenopausal women, who do not experience the large midcycle surge of hLH but instead maintain a continuous pulsatile high concentration of hLH, would tend to display a high plateau level of hLHflcf. Essentially, this would result from an integration of the multitude of hLH pulses in the time delay during the proteolytic processing steps. We tested this hypothesis by examining first morning void urines for 10 consecutive days in a series

71

100 -

hLH

~[

9

hCGI3cf

hLH~l

4 -

80(o r

600

~

S Eo

4020-

32-

1 -

0

I

I

I

0 i i i i 1 6 0 - S t e r o i d h o r m o n e ratio

I

600 o

120O m 400 800

E 200

oo

-

-

40-

LU ,,_..

0-

0

I

I

I

I

80-

I

I

I

I

-2

0

2

4

day ~o 6 0 E J)

E

E

4020-

0

I

I

I

I

-2

0

2

4

day

FIGURE 9 Hormone profiles in the urine of normally cycling women (n = 15). Concentrations were presented as mean _+ SE, femtomoles/milligram creatinine (fmol/mg C). hLH concentration has been measured using two different immunoradiometric assays (n = 8 for the hLH-2 assay). Steroid hormone ratios are calculated using estrone-3-glucuronide (E1-3-G) and pregnandiol-3-glucuronide (Pd-3-G) (• 103). Day 0 is the day of hLH surge. Reproduced from [63], O'Connor, J. F., Kovalevskaya, G., Birken, S., Schlatterer, J. E, Schechter, D., McMahon, D. J., and Canfield, R. E. (1998). The expression of the urinary forms of human luteinizing hormone beta fragment in various populations as assessed by a specific immunoradiometric assay. Hum. Reprod. 13, 826-835, with permission of the authors and Human Reproduction.

of postmenopausal women. The resultant data did not agree with our hypothesis. Instead, very large-amplitude pulses of hLHflcf, which could not readily be correlated with hLH surges, appeared in the urine of these women. One such pattern of hLHflcf excretion in a postmenopausal woman is illustrated in Fig. 11. Note the very large fluctuations of hLHflcf (between 0 and 600 fmol/mg creatinine), which do not correlate with urinary hLH, even considering the 24to 48-hr time delay in the appearance of the core after an hLH surge (Fig. 9). Analysis of 10 consecutive first morning void urine samples from cycling women of reproductive age (< 35 yr) during the follicular phase (day 1 was the first day of menses) indicated that much shallower fluctuations occurred in these women, prior to the hLH surge at the time of ovulation. Indeed, by integrating the area under the peaks of graphs of hLHflcf in femtomoles/milligram of creatinine versus day of collection, it was possible to differentiate young cycling women from postmenopausal women, even with the occasional high spikes in some young women

72

BIRKEN ET AL.

120 Serum lOO -

~0 80o

~

I

60-

_J e-

~ "r" _1 '-

40 20 o 200

~

I

l

I

1200 e

-

1000

150 800

0

-~

0

~100 Z

- 600

:f

-

o -1-

~

400

t-

50 - 200

0

i

i

i

1

0

1

2

3

0

day

FIGURE 10 hLH and hLHflcf in serum and urine of the same patient, o, hLH- 1 ; r~, hLH2; A, hLHfl; e, hLHflcf. The serum levels of intact hLH (o) and hLHflcf (e) indicate that there is an insignificant amount of hLHflcf detected in the blood. The lower panel illustrates the urinary values for hLH and hLHflcf for the same days of collection. The surge of hLH (day 0) and the surge of hLHflcf (1-2 days later) are detected in urine, but the peak of hLHflcf lags that of the intact hLH by 1-2 days, suggesting that urinary hLHflcf is a consequence of the peripheral or renal metabolic processing of intact hLH. Reproduced from [63], O'Connor, J. E, Kovalevskaya, G., Birken, S., Schlatterer, J. P., Schechter, D., McMahon, D. J., and Canfield, R. E. (1998). The expression of the urinary forms of human luteinizing hormone beta fragment in various populations as assessed by a specific immunoradiometric assay. Hum. Reprod. 13, 826-835, with permission of the authors and Human Reproduction.

during the follicular phase. We selected 10-day intervals as a convenient research set, and specimens could be easily collected by volunteer subjects and stored frozen until samples were brought to the laboratory. In a future marker assay, the collection protocol is unlikely to consist of such a large sample set. When samples are analyzed from women still experiencing regular menstrual cycles, the 10-day interval collection provides a convenient starting point within the cycle and encompasses the follicular phase that most closely corresponds to the postmenopausal state of relatively low circulating steroids. For regularly cycling women, the mean of the areas under the peaks of 10 subjects was 278 with a median area of 169. The mean of the areas among postmenopausal subjects ranged between approximately 1000-4000 with medians in the ranges of 900-3000. The postmenopausal subjects differed significantly from the population of normal cycling women by the amplitude and area under the peaks of the daily fluctuations of this fragment. We com-

pared the areas under the peak of the lowest hLHflcf levels of postmenopausal women with the highest core levels of premenopausal women and could statistically differentiate the two groups even in this worst-scenario sampling situation. Perimenopausal women fall in between, with some clearly in the postmenopausal pattern and some in a premenopausal pattern. Figures 12 and 13 illustrate analysis of two perimenopausal women. The patterns of hLHflcf shown in Fig. 12 are typical of those seen in normal cycling women of midreproductive age (see Fig. 9) [63]. A peak hLHflcf appears 1-2 days after the hLH surge, after the presumed metabolic breakdown of circulating hLH into hLHflcf within a tissue compartment. Figure 13 depicts a perimenopausal woman with a typical postmenopausal pattern (see Fig. 11), with many very large hLHflcf peaks, not coordinated to particular hLH peaks in the urine. However, we must keep in mind the earlier discussion of the problems of accurately measuring hLH, in urine and the phenomenon of "invisible" urinary hLH, which we have encountered several times in our own laboratories [63,71 ]. We are in the process of developing assays for a similar hFSHflcf. With the combination of the two fragment assays,

600

500

400 (

o C:~

300

0

E 2oo 0 _J t-

0

10

0

1'0

20

30

40

50

60

70

20

30

4'0

50

60

70

20

~" 0

15

g Z) E "l_J t-

10

5

D a y of U r i n e C o l l e c t i o n

FIGURE 11 The 60-day patterns of hLHflcf and hLH in first morning void urine collections from a postmenopausal woman. In the upper panel the hLHflcf is normalized to creatinine, whereas in the lower panel the hLH is measured by the DELFIA assay on glycerol-preserved urines.

73

CHAPTER 4 Gonadotropins and Menopause: New Markers

800

600

5O0

t

t 400 t_ o 300 E

600

1

i,..

o

E 400 O

O

E 200 tO

~- 100 zIZ

~" 200 -.I r

._1

"

0 10

20

30

40

50

I

14 I 12 -~ 10 E -r

._J

.c

10

20

30

40

50

60

60

T

I

_

0

60

20 18 16

,

8

liI

50

T

JI II li

40

I..

oO') 30 E D 20 E -110 ....I

6 4

t-

O

0

10

20

30

40

50

60

Day of Urine Collection FIGURE 12 The 60-day patterns of hLH/3cf and hLH in first morning void urine collections from a perimenopausal woman. In the upper panel the hLH/3cf is normalized to creatinine, whereas in the lower panel the hLH is measured by the DELFIA assay on glycerol-preserved urines. The pattern of hLH/3cf resembles that of midreproductive age women (see Fig. 9).

hLH/3cf and hFSH/3cf, it may be possible to develop urinary assays that will define the stage of m e n o p a u s a l transition. Because changes in h F S H concentrations usually precede those of hLH, it is envisioned that the hFSH/3cf assay, when developed, m a y provide a stronger discriminant function than the hLH/3cf assay. In summary, the gonadotropin patterns in b l o o d and urine undergo significant alterations during the m e n o p a u s a l transition. The d e v e l o p m e n t of sensitive assays for stable proteolytically derived fragments of h L H and h F S H in urine will provide new markers to determine the phases of the menopausal transition. The standardization of these assays with stable h o r m o n a l metabolites should obviate the p r o b l e m s often encountered in c o m p a r i s o n and analysis of data collected from different laboratories and thereby ease collection of data in large-scale epidemiological studies. Furthermore, implementation of assays based on stable h o r m o n a l metabolites will heighten awareness in the general clinical c h e m i s t r y c o m m u n i t y o f the utility of metabolic by-products as markers for other physiologically relevant proteins.

0

10

20

30

40

50

60

70

Day of Urine.Collection FIGURE 13 The 60-day patterns of hLHflcf and hLH in first morning void urine collections from a perimenopausal woman. In the upper panel the hLI-I/3cfis normalized to creatinine, whereas in the lower panel the hLH is measured by the DELFIA assay on glycerol-preserved urines. The pattern of hLHflcf resembles that of a postmenopausal woman. Part of this figure is reproduced from Birken, S., Santoro, I. V., Maydelman, Y., Kovalevskaya, G., Lobo, R., Freeman, E. W., Warren, M., McMahon, D., and O'Connor, J. (1999). Differences in urinary excretion patterns of the hLH beta core fragment in premenopausal, perimenopausal, and postmenopausal women. Menopause 6(4), with permission of the authors and Lippincott Williams & Wilkins.

Acknowledgments This work was supported by NIH grants RO1-AG 13783 and ROlES07589. We wish to express appreciation to Nanette Santoro for thoughtful advice.

References 1. Pierce, J. G., and Parsons, T. E (1981). Glycoprotein hormones: Structure and function. Annu. Rev. Biochem. 50, 465-495. 2. Birken, S., Maydelman, Y., Gawinowicz, M. A., Pound, A., Liu, Y., and Hartree, A. S. (1996). Isolation and characterization of human pituitary chorionic gonadotropin. Endocrinology (Baltimore) 137, 14021411.

74 3. Sawitzke, A. L., Griffin, J., and Odell, W. D. (1991). Purified preparations of human luteinizing hormone are contaminated with small amounts of a chorionic gonadotropin-like material. J. Clin. Endocrinol. Metab. 72, 841-846. 4. Hammond, E., Griffin, J., and Odell, W. D. (1991). A chorionic gonadotropin-secreting human pituitary cell. J. Clin. Endocrinol. Metab. 72, 747-754. 5. Odell, W. D., Griffin, J., Bashey, H. M., and Snyder, E J. (1990). Secretion of chorionic gonadotropin by cultured human pituitary cells. J. Clin. Endocrinol. Metab. 71, 1318-1321. 6. Odell, W. D., and Griffin, J., (1989). Pulsatile secretion of chorionic gonadotropin during the normal menstrual cycle. J. Clin. Endocrinol. Metab. 69, 528-532. 7. Moyle, W. R., Myers, R. V., Wang, Y., Han, Y., Lin, W., Kelley, G. L., Ehrlich, P. H., Rao, S. N., and Bernard, M. P. (1998). Functional homodimeric glycoprotein hormones: Implications for hormone action and evolution. Chem. Biol. 5, 241-254. 8. Talmadge, K., Vamvakopoulos, N. C., and Fiddes, J. C. (1984). Evolution of the genes for the beta subunits of human chorionic gonadotropin and luteinizing hormone. Nature (London) 307, 37-40. 9. Manzella, S. M., Dharmesh, S. M., Beranek, M. C., Swanson, P., and Baenziger, J. U. (1995). Evolutionary conservation of the sulfated oligosaccharides on vertebrate glycoprotein hormones that control circulatory half-life. J. Biol. Chem. 270, 21665-21671. 10. Manzella, S. M., Hooper, L. V., and Baenziger, J. U. (1996). Oligosaccharides containing beta- 1,4-1inked N-acetylgalactosamine, a paradigm for protein-specific glycosylation. J Biol Chem 271, 12117-12120. 11. Fiddes, J. C., and Goodman, H. M. (1980). The cDNA for the betasubunit of human chorionic gonadotropin suggests evolution of a gene by readthrough into the 3'-untranslated region. Nature (London) 286, 684-687. 12. Furuhashi, M., Shikone, T., Fares, F. A., Sugahara, T., Hsueh, A. J., and Boime, I. (1995). Fusing the carboxy-terminal peptide of the chorionic gonadotropin (CG) beta-subunit to the common alpha-subunit: Retention of O-linked glycosylation and enhanced in vivo bioactivity of chimeric human CG. Mol. Endocrinol. 9, 54-63. 13. Knobil, E. (1988). The hypothalamic gonadotrophic hormone releasing hormone (GnRH) pulse generator in the rhesus monkey and its neuroendocrine control. Hum. Reprod. 3, 29-31. 14. Knobil, E. (1989). The electrophysiology of the GnRH pulse generator in the rhesus monkey. J. Steroid Biochem. 33, 669- 671. 15. Wise, P. M., Krajnak, K. M,. and Kashon, M. L. (1996). Menopause: The aging of multiple pacemakers Science 273, 67-70. 16. Wise, P. M., Kashon, M. L., Krajnak, K. M., Rosewell, K. L., Cai, A., Scarbrough, K., Harney, J. P., McShane, T., Lloyd, J. M., and Weiland, N. G. (1997). Aging of the female reproductive system: A window into brain aging. Recent. Prog. Horm. Res. 52, 279-303. 17. Brown, J. R., Skurnick, J. H., Sharma, N., Adel, T., and Santoro, N. (1993). Frequent intermittent ovarian function in women with premature menopause: A longitudinal study. Endocr. J. 1,467-474. 18. Santoro, N., Brown, J. R., Adel, T., and Skurnick, J. H. (1996). Characterization of reproductive hormonal dynamics in the perimenopause. J. Clin. Endocrinol. Metab. 81, 1495-1501. 19. Santoro, N., Banwell, T., Tortoriello, D., Lieman, H., Adel, T., and Skurnick, J. (1998). Effects of aging and gonadal failure on the hypothalamic-pituitary axis in women. Am. J. Obstet. Gynecol. 178, 732741. 20. Prior, J. C. (1998). Perimenopause: The complex endocrinology of the menopausal transition. Endocr. Rev. 19, 397-428. 21. Zeleznik, A. J., and Hillier, S. G. (1984). The role of gonadotropins in the selection of the preovulatory follicle. Clin. Obstet. Gynecol. 27, 927-940. 22. Hillier, S. G., Zeleznik, A. J., Knazek, R. A. and Ross, G. T. (1980). Hormonal regulation of preovulatory follicle maturation in the rat. J. Reprod. Fertil. 60, 219-229.

BIRKEN ET AL. 23. Reame, N. E., Wyman, T. L., Phillips, D. J., de Kretser, D. M., and Padmanabhan, V. (1998). Net increase in stimulatory input resulting from a decrease in inhibin B and an increase in activin A may contribute in part to the rise in follicular phase follicle-stimulating hormone of aging cycling women. J. Clin. Endocrinol. Metab. 83, 3302-3307. 24. Santoro, N., Adel, T., and Skurnick, J. (1999). Decreased inhibin tone and increased activin A secretion characterize reproductive aging in women. Fertil. Steril. 71,658-662. 25. Speroff, L., Glass, R. H., and Kase, N. G. (1994). Regulation of the menstrual cycle. In "Clinical Gynecologic Endocrinology and Infertility" (C. Mitchell, ed.), pp. 183-230. Williams & Wilkins, Baltimore, MD. 26. Carr, B. R. (1998). The ovary. In "Textbook of Reproductive Medicine" (B. R. Carr and R. E. Blackwell, eds.), pp. 207-243. Appleton & Lange, Stanford, CT. 27. Singer, D. B. and Haning, R. V. (1995). The menstrual cycle. In "Human Reproduction, Growth and Development" (D. R. Coustan, R. V. Hanning, and D. B. Singer, eds.), pp. 3-3, Little, Brown, New York. 28. Pantel, J., Robert, E, Troalen, E, Kujas, M., Bellet, D., and Bidart, J. M. (1998). Characterization of human lutropin carboxyl-terminus isoforms. Endocrinology (Baltimore) 139, 527-533. 29. Lapthorn, A. J., Harris, D. C., Littlejohn, A., Lustbader, J. W., Canfield, R. E., Machin, K. J., Morgan, F. J., and Isaacs, N. W. (1994). Crystal structure of human chorionic gonadotropin Nature (London) 369, 455461. 30. Lustbader, J. W., Wu, H., Birken, S., Pollak, S., Gawinowicz, K. M., Pound, A. M., Austen, D., Hendrickson, W. A., and Canfield, R. E. (1995). The expression, characterization, and crystallization of wildtype and selenomethionyl human chorionic gonadotropin. Endocrinology (Baltimore) 136, 640-650. 31. Wu, H., Lustbader, J. W., Liu, Y., Canfield, R. E., and Hendrickson, W. A. (1994). Structure of human chorionic gonadotropin at 2.6 A resolution from MAD analysis of the selenomethionyl protein. Structure 2, 545-558. 32. Matzuk, M. M., Keene, J. L., and Boime, I. (1989). Site specificity of the chorionic gonadotropin N-linked oligosaccharides in signal transduction. J. Biol. Chem. 264, 2409-2414. 33. Baenziger, J. U. (1996). Glycosylation: To what end for the glycoprotein hormones? Endocrinology (Baltimore) 137, 1520-1522. 34. Fiete, D., and Baenziger, J. U. (1997). Isolation of the SO4-4-GalNAcbetal,4GlcNAcbetal,2Manalpha-specific receptor from rat liver. J. Biol. Chem. 272, 14629-14637. 35. Baenziger, J. U., and Green, E. D. (1988). Pituitary glycoprotein hormone oligosaccharides: Structure, synthesis and function of the asparagine-linked oligosaccharides on lutropin, follitropin and thyrotropin. Biochim. Biophys. Acta 947, 287-306. 36. Baenziger, J. U., Kumar, S., Brodbeck, R. M., Smith, P. L., and Beranek, M. C. (1992). Circulatory half-life but not interaction with the lutropin/chorionic gonadotropin receptor is modulated by sulfation of bovine lutropin oligosaccharides. Proc. Natl. Acad. Sci. U.S.A. 89, 334-338. 37. Baenziger, J. U. (1994). Protein-specific glycosyltransferases: How and why they do it! FASEB J. 8, 1019-1025. 38. Wide, L., and Hobson, B. (1987). Some qualitative differences of hCG in serum from early and late pregnancies and trophoblastic diseases. Acta Endocrinol. (Copenhagen) 116, 465-472. 39. Gudermann, T., Brockmann, H., Simoni, M., Gromoll, J., and Nieschlag, E. (1994). In vitro bioassay for human serum follicle-stimulating hormone (FSH) based on L cells transfected with recombinant rat FSH receptor: Validation of a model system. Endocrinology (Baltimore) 135, 2204-2213. 40. Christin-Maitre, S., Taylor, A. E., Khoury, R. H., Hall, J. E., Martin, K. A., Smith, E C., Albanese, C., Jameson, J. L., Crowley, W. F., Jr., and Sluss, E M. (1996). Homologous in vitro bioassay for folliclestimulating hormone (FSH) reveals increased FSH biological signal

CHAPTER 4 Gonadotropins and Menopause: New Markers

41. 42.

43.

44. 45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

55.

56.

during the mid- to late luteal phase of the human menstrual cycle. J. Clin. Endocrinol. Metab. 81, 2080-2088. Christin-Maitre, S., and Bouchard, P. (1996). Bioassays of gonadotropins based on cloned receptors. Mol. Cell. Endocrinol. 125, 151-159. Padmanabhan, V., Chappel, S. C., and Beitins, I. Z. (1987). An improved in vitro bioassay for follicle-stimulating hormone (FSH): Suitable for measurement of FSH in unextracted human serum. Endocrinology (Baltimore) 121, 1089-1098. Diaz-Cueto, L., Mendez, J. P., Barrios-de-Tomasi, J., Lee, J. Y., Wide, L., Veldhuis, J. D., and Ulloa-Aguirre, A. (1994). Amplitude regulation of episodic release, in vitro biological to immunological ratio, and median charge of human chorionic gonadotropin in pregnancy. J. Clin. Endocrinol. Metab. 78, 890-897. Nisula, B. C. (1993). Measurement of human chorion gonadotropin by biological methods. Scand. J. Clin. Lab. Invest., Suppl. 216, 114-117. Schneyer, A. L., Sluss, P. M., Whitcomb, R. W., Hall, J. E., Crowley, W. F., Jr., and Freeman, R. G. (1991). Development of a radioligand receptor assay for measuring follitropin in serum: Application to premature ovarian failure. Clin. Chem. (Winston-Salem, N.C.) 37, 508514. Whitcomb, R. W., and Schneyer, A. L. (1990). Development and validation of a radioligand receptor assay for measurement of luteinizing hormone in human serum. J. Clin. Endocrinol. Metab. 71, 591-595. Ravindranath, N., Srilatha, N. S., Sairam, M. R., and Moudgal, N. R. (1992). Ability of deglycosylated human chorionic gonadotropin (dghCG) to block luteal function and establishment of pregnancy in bonnet monkeys (Macaca radiata). Indian J. Exp. Biol. 30, 982-986. Larsen, P. R. (1976). Quantitation of triiodothyronine and thyroxine in human serum by radioimmunoassay. In "Hormones in Human Blood: Detection and Assay" (H. N. Antonaides, ed.), pp. 679-697. Harvard University Press, Cambridge, MA. O'Connor, J. F., Schlatterer, J. P., Birken, S., Krichevsky, A., Armstrong, E. G., McMahon, D., and Canfield, R. E. (1988). Development of highly sensitive immunoassays to measure human chorionic gonadotropin, its beta-subunit, and beta core fragment in the urine: Application to malignancies. Cancer Res. 48, 1361-1366. Alfthan, H., Haglund, C., Roberts, P., and Stenman, U. H. (1992). Elevation of free beta subunit of human choriogonadotropin and core beta fragment of human choriogonadotropin in the serum and urine of patients with malignant pancreatic and biliary disease. Cancer Res. 52, 4628-4633. Spencer, K., Aitken, D. A., Macri, J. N., and Buchanan, P. D. (1996). Urine free beta hCG and beta core in pregnancies affected by Down's syndrome. Prenat. Diagn. 16, 605-613. Cole, L. A., Wang, Y. X., Elliott, M., Latif, M., Chambers, J. T., Chambers, S. K., and Schwartz, P. E. (1988). Urinary human chorionic gonadotropin free beta-subunit and beta- core fragment: A new marker of gynecological cancers. Cancer Res. 48, 1356-1360. Diaz-Cueto, L., Barrios-de-Tomasi, J., Timossi, C., Mendez, J. P., and Ulloa-Aguirre, A. (1996). More in vitro bioactive, shorter-lived human chorionic gonadotrophin charge isoforms increase at the end of the first and during the third trimesters of gestation. Mol. Hum. Reprod. 2, 643-650. Ho, H.-H., O'Connor, J. F., Tieu, J., Overstreet, J. W., and Lasley, B. L. (1997). Characterization of hCG in normal and abnormal pregnancies. Early Pregnancy: Biol. Med. 3, 213-224. Romani, P., Robertson, D. M., and Diczfalusy, E. D. (1977). Biologically active luteinizing hormone (LH) in plasma: II. Comparison with immunologically active LH levels throughout the human menstrual cycle. Acta Endocrinol. (Copenhagen) 84, 697-712. Reame, N. E., Kelche, R. P., Beitins, I. Z., Yu, M. Y., Zawacki, C. M., and Padmanabhan, V. (1996). Age effects of follicle-stimulating hormone and pulsatile luteinizing hormone secretion across the menstrual cycle of premenopausal women. J. Clin. Endocrinol. Metab. 81, 15121518.

75 57. Birken, S., Armstrong, E. G., Kolks, M. A., Cole, L. A., Agosto, G. M., Krichevsky, A., Vaitukaitis, J. L., and Canfield, R. E. (1988). Structure of the human chorionic gonadotropin beta-subunit fragment from pregnancy urine. Endocrinology (Baltimore) 123, 572-583. 58. Birken, S., Chen, Y., Gawinowicz, M. A., Agosto, G. M., Canfield, R. E., and Hartree, A. S. (1993). Structure and significance of human luteinizing hormone-beta core fragment purified from human pituitary extracts. Endocrinology (Baltimore) 133, 985-989. 59. Birken, S., Kovalevskaya, G., and O'Connor, J. (1996). Metabolism of hCG and hLH to multiple urinary forms. Mol. Cell. Endocrinol. 125, 121-131. 60. Kovalevskaya, G., Birken, S., O'Connor, J., Schlatterer, J., Maydelman, Y., and Canfield, R. (1995). HLH beta core fragment in the urine of ovulating women: A sensitive and specific immunometric assay for its detection. Endocrine 3, 881-887. 61. Blithe, D. L., Akar, A. H., Wehmann, R. E., and Nisula, B. C. (1988). Purification of beta-core fragment from pregnancy urine and demonstration that its carbohydrate moieties differ from those of native human chorionic gonadotropin-beta. Endocrinology (Baltimore) 122, 173-180. 62. Nisula, B. C., Blithe, D. L., Akar, A., Lefort, G., and Wehmann, R. E. (1989). Metabolic fate of human choriogonadotropin. J. Steroid Biochem. 33, 733-737. 63. O'Connor, J. F., Kovalevskaya, G., Birken, S., Schlatterer, J. P., Schechter, D., McMahon, D. J., and Canfield, R. E. (1998). The expression of the urinary forms of human luteinizing hormone beta fragment in various populations as assessed by a specific immunoradiometric assay. Hum. Reprod. 13, 826-835. 64. Birken, S., Gawinowicz, M. A., Kardana, A., and Cole, L. A. (1991). The heterogeneity of human chorionic gonadotropin (hCG). II. Characteristics and origins of nicks in hCG reference standards. Endocrinology (Baltimore) 129, 1551-1558. 65. Cole, L. A., Kardana, A., Ying, F. C., and Birken, S. (1991). The biological and clinical significance of nicks in human chorionic gonadotropin and its free beta-subunit. Yale J. Biol. Meal. 64, 627-637. 66. Krichevsky, A., Birken, S., O'Connor, J., Bikel, K., Schlatterer, J., Yi, C., Agosto, G., and Canfield, R. (1991). Development and characterization of a new, highly specific antibody to the human chorionic gonadotropin-beta fragment. Endocrinology (Baltimore) 128, 12551264. 67. Krichevsky, A., Birken, S., O'Connor, J., Bikel, K., Schlatterer, J., and Canfield, R. (1994). The development of a panel of monoclonal antibodies to human luteinizing hormone and its application to immunological mapping and two-site assays. Endocrine 2, 511-520. 68. O'Connor, J. F., Birken, S., Lustbader, J. W., Krichevsky, A., Chen, Y., and Canfield, R. E. (1994). Recent advances in the chemistry and immunochemistry of human chorionic gonadotropin: Impact on clinical measurements. Endocr. Rev. 15, 650-683. 69. Pettersson, K., Ding, Y. Q., and Huhtaniemi, I. (1991). Monoclonal antibody-based discrepancies between two-site immunometric tests for lutropin. Clin. Chem. (Winston-Salem, N.C.) 37, Pt 1):1745-1748. 70. Pettersson, K., Ding, Y. Q., and Huhtaniemi, I. (1992). An immunologically anomalous luteinizing hormone variant in a healthy woman. J. Clin. Endocrinol. Metab. 74, 164-171. 71. Martin-Du-Pan, R. C., Horak, M., and Bischof, P. (1994). Clinical significance of invisible or partially visible luteinizing hormone. Hum. Reprod. 9, 1987-1990. 72. de Medeiros, S. F., Amato, F., and Norman, R. J. (1991). Stability of immunoreactive beta-core fragment of hCG. Obstet. Gynecol. 77, 53-59. 73. Klein, N. A., Illingworth, P. J., Groome, N. P., McNeilly, A. S., Battaglia, D.E., and Soules, M. R. (1996). Decreased inhibin B secretion is associated with the monotropic FSH rise in older, ovulatory women: A study of serum and follicular fluid levels of dimeric inhibin A and B in spontaneous menstrual cycles. J. Clin. Endocrinol. Metab. 81, 27422745.

76 74. Welt, C. K., McNicholl, D. J., Taylor, A. E., and Hall, J. E. (1999). Female reproductive aging is marked by decreased secretion of dimeric inhibin. J. Clin. Endocrinol. Metab. 84, 105 - 111. 75. Burger, H. G., Cahir, N., Robertson, D. M., Groome, N. P., Dudley, E., Green, A., and Dennerstein, L. (1998). Serum inhibins A and B fall differentially as FSH rises in perimenopausal women. Clin. Endocrinol. (Oxford) 48, 809-813; published erratum: Ibid., 49(4), 550. 76. Matt, D. W., Kauma, S. W., Pincus, S. M., Veldhuis, J. D., and Evans, W. S. (1998). Characteristics of luteinizing hormone secretion in younger versus older premenopausal women. Am. J. Obstet. Gynecol. 178, 504-510. 77. Klein, N. A., and Soules, M. R. (1998). Endocrine changes of the perimenopause. Clin. Obstet. Gynecol. 41,912-920. 78. Soules, M. R., Battaglia, D. E., and Klein, N. A. (1998). Inhibin and reproductive aging in women. Maturitas 30, 193-204. 79. Burger, H. G. (1996). The endocrinology of the menopause. Maturitas 23, 129-136. 80. Burger, H. G. (1996). The menopausal transition. Bailliere's Clin. Obstet. Gynecol. 10, 347-359. 81. Vihko, K. K. (1996). Gonadotropins and ovarian gonadotropin receptors during the perimenopausal transition period. Maturitas 23 (Suppl.), S19-$22. 82. Schmidt, P. J., Gindoff, P. R., Baron, D. A., and Rubinow, D. R. (1996). Basal and stimulated gonadotropin levels in the perimenopause. Am. J. Obstet. Gynecol. 175, 643-650. 83. Saketos, M., Sharma, N., Adel, T., Raghuwanshi, M., and Santoro, N. (1994). Evalution of time-resolved immunofluorometric assay and specimen storage conditions for measuring gonadotropins. Clin. Chem. (Winston-Salem, N.C.) 40, 749-753. 84. Kesner, J. S., Knecht, E. A., and Krieg, E. F., Jr. (1995). Stability of urinary female reproductive hormones stored under various conditions. Reprod. Toxicol. 9, 239-244. 85. Livesey, J. H., Hodgkinson, S. C., Roud, H. R., and Donald, R. A. (1980). Effect of time, temperature and freezing on the stability of im-

BIRKEN ET AL.

86.

87.

88. 89.

90.

91.

92.

93.

94.

munoreactive LH, FSH, TSH, growth hormone, prolactin and insulin in plasma. Clin. Biochem. 13, 151-155. Livesey, J. H., Roud, H. K., Metcalf, M. G., and Donald, R. A. (1983). Glycerol prevents loss of immunoreactive follicle-stimulating hormone and luteinizing hormone from frozen urine. J. Endocrinol. 98, 381-384. Kardana, A., Elliott, M. M., Gawinowicz, M. A., Birken, S., and Cole, L. A. (1991). The heterogeneity of human chorionic gonadotropin (hCG). I. Characterization of peptide heterogeneity in 13 individual preparations ofhCG. Endocrinology (Baltimore) 129, 1541-1550. Sturgeon, C. M., and McAllister, E. J. (1998). Analysis ofhCG: Clinical applications and assay requirements. Ann. Clin. Biochem. 35,460- 491. Cole, L. A. (1997). Immunoassay of human chorionic gonadotropin, its free subunits, and metabolites Clin. Chem. (Winston-Salem, N.C.) 43, 2233-2243. Pettersson, K. S., and Soderholm, J. R. (1991). Individual differences in lutropin immunoreactivity revealed by monoclonal antibodies. Clin. Chem. (Winston-Salem, N.C.) 37, 333-340. Barbe, E, Legagneur, H., Watrin, V., Klein, M., and Badonnel, Y. (1995). Undetectable luteinizing hormone levels using a monoclonal immunometric assay. J. Endocrinol. Invest. 18, 806-808. Mitchell, R., Hollis, S., Crowley, V., McLoughlin, J., Peers, N., and Robertson, W. R. (1995). Immunometric assays of luteinizing hormone (LH): Differences in recognition of plasma LH by anti-intact and betasubunit-specific antibodies in various physiological and pathophysiological situations. Clin. Chem. (Winston-Salem, N.C.) 41, (Pt. 1), 11391145. Costagliola, S., Niccoli, P., and Carayon, P. (1994). Glycoprotein hormone isomorphism and assay discrepancy: The paradigm of luteinizing hormone (LH). J. Endocrinol. Invest. 17, 291-299. Qiu, Q., Kuo, A., Todd, H., Dias, J. A., Gould, J. E., Overstreet, J. W., and Lasley, B. L. (1998). Enzyme immunoassay method for total urinary follicle-stimulating hormone (FSH) beta subunit and its application for measurement of total urinary FSH. Fertil. Steril. 69, 278-285.

~HAPTER

Genetic Programming In Ovarian

Development

and Oogenesis JOE LEIGH SIMPSON

Departments of Obstetrics and Gynecology and Molecular and Human Genetics, Baylor College of Medicine, Houston, Texas 77030

I. II. III. IV.

Ovarian Differentiation Requires Only One X (Constitutive) Polygenic and Stochastic Control over Oocyte Number Monosomy X X Chromosomal Mosaicism: 45,X/46,XX and 45,X/47,XXX V. Pitfalls in Localizing Ovarian Maintenance Genes to Specific Regions of the X

VI. VII. VIII. IX. X. XI.

I. O V A R I A N

Failure of germ cell development is associated with complete ovarian failure, resulting in lack of secondary sexual pubertal development (primary amenorrhea). A decreased number but not a total absence of germ cells is more likely associated with premature ovarian failure, presenting with infertility or secondary amenorrhea (see Chapter 8). Yet complete and premature ovarian failure may be different manifestations of the same underlying pathogenic and etiologic processes. Many different genetic mechanisms are pertinent to the processes m chromosomal abnormalities, Mendelian mutations of autosomal or X-linked genes, and polygenic/ multifactorial factors. In this contribution, we enumerate clinical disorders associated with germ cell abnormalities, deducing etiologic factors responsible for ovarian differentiation and oogenesis in normal females.

MENOPAUSE:

BIOLOGY AND PATHOBIOLOGY

Genes on the X Short Arm Genes on the X Long Arm Nature of X Ovarian Maintenance Determinants Autosomal Chromosomal Abnormalities Autosomal Genes (Mendelian) To What Extent Is Premature Ovarian Failure Genetic? References

REQUIRES

DIFFERENTIATION

ONLY ONE X

(CONSTITUTIVE) In the absence of the Y chromosome, the indifferent embryonic gonad always develops into an ovary. Germ cells exist in 45,X human fetuses [ 1]. Oocyte development initially exists even in 46,XY phenotypic females, such as in infants with XY gonadal dysgenesis [2] or the genito-palatocardiac syndrome [3]. Oocyte development in the presence of a Y chromosome is also well documented in mice [4]. Thus, the pathogenesis of germ cell failure in humans can be deduced to be increased germ cell attrition. If two intact X chromosomes are not present, ovarian follicles in 45,X individuals usually degenerate by birth. Genes on the second

77


78

JOE LEIGH SIMPSON

X chromosome are thus responsible for ovarian maintenance, rather than for ovarian differentiation.

II. P O L Y G E N I C CONTROL

OVER

AND STOCHASTIC OOCYTE

NUMBER

It is to be expected that oocyte number (reservoir) will be low in some women simply on statistical (stochastic) grounds. Normal distribution exists for all common anatomic traits (e.g., height), and this principle should apply to oocyte number and reservoir at birth. That a normal distribution of germ cell number exists in ostensibly normal females is well established in animals but difficult to prove in humans. Different rodent strains show characteristic breeding duration, implying genetic control over either the rate of oocyte depletion or the number of oocytes initially present. It follows that some ostensibly normal (menstruating) women may have decreased oocyte reservoir or increased oocyte attrition on a genetic basis, analogous to animal models. In humans a genetic basis for the above can be presumed by analogy to the heritability of age at human menopause, a characteristic that clearly shows familiar tendencies. Assessing heritability of age at menopause is complicated because iatrogenic behavior (e.g., hysterectomy) and other confounding factors (e.g., leiomyomata or uterine cancer) must be taken into account. However, several studies within the past decade have directly addressed the issue. Cramer et al. [5] performed a case control study on 10,606 United States women who were between 45 and 54 years of age. Women with an early menopause ( 4 0 - 4 5 years) were age-matched with controls who were either still menstruating or had experienced menopause after age 45 years. Of 129 early menopause cases (age 65 years), although the late phase remains intact. This could be related to changes in blood-brain transport of cortisol [72]. The elderly have a higher serum cortisol response to perioperative stress or during depression and the dexamethasone suppression test often fails to yield the expected drop in plasma cortisol levels [75]. It is tempting to speculate that this relative increase in cortisol level with age may contribute to some age-related changes in body composition, notably osteopenia. 2.

MINERALOCORTICOIDS

Aldosterone secretion decreases with age, probably because of reduced plasma renin activity [76]. This reduction is evident both at basal conditions and during stimulation with salt restriction, upright posture, or ACTH [76]. An

CHAPTER7 Changes in Aging Men age-related reduction in aldosterone clearance partially offsets the decreased aldosterone production rate. The reduced plasma renin activity and aldosterone production may contribute to the orthostatic hypotension commonly found in the elderly and may expose them to a higher risk of developing hyperkalemia following administration of angiotensinconverting enzyme inhibitors. 3. ADRENAL ANDROGENS

One of the most dramatic age-related changes in the hormonal system is adrenal androgen production. In men, dehydroepiandrosterone (DHEA) secretion declines progressively between 20 and 96 years of age [77]. The serum level of DHEA in older men is approximately 5 - 3 0 % of that seen in young men [77]. This "adrenopause" is probably the result of reduced 17,20-desmolase activity with age. Weight loss in overweight men but not women is associated with a 125% rise in serum DHEA sulfate, which suggests that agerelated increase in adiposity or insulin resistance may contribute to reduced DHEA levels in aging men [78]. The biological implications of the decline of DHEA in aging humans are still not clear. Experiments in animals who do not secret DHEA suggest that DHEA may be implicated in longevity and have protective effects in tumorigenesis, atherosclerosis, and age-related memory disturbances [79]. In one epidemiological study, death from cardiovascular disease in men over the age of 50 years was inversely related to DHEA sulfate levels [80]. DHEA administration reduces plasma plasminogen activator inhibitor type 1 and tissue plasminogen activator concentrations in men [81]. DHEA also inhibits platelet activity [82]. These effects may help prevent heart disease in men. More interventional studies are needed to establish the clinical relevance of DHEA in the biology of aging. 4. ADRENAL MEDULLA

Elevated plasma levels of epinephrine and norepinephrine (NE) have been found in healthy octogenarians compared to younger subjects [83,84]. Plasma dopamine levels do not change with age. The increased NE levels are due to an increased production and decreased clearance rate. This is accompanied by a decrease in platelet az-adrenergic receptors [85] and cardiac fl-adrenergic transmission [86]. The NE response to upright posture, during the cold pressor test, following glucose ingestion, and during insulin tolerance testing is increased in the elderly, whereas the NE and epinephrine response to exercise may be reduced in healthy elderly men [84]. The clinical consequences of these changes are not apparent but they may contribute to orthostatic or postprandial hypotension [87]. They could also be related to increased vascular resistance and therefore contribute to hypertension and the need for after-load reduction, especially in those with congestive heart failure.

117 D. C a l c i u m a n d B o n e M e t a b o l i s m Whereas age-related bone loss is a common phenomenon in both men and women, the process is accelerated by coexisting hormonal deficiencies, notably estrogen deficiency during the menopause and androgen deficiency in men. In healthy men, radial bone mineral content decreases by 1% per year whereas vertebral bone mineral content decreases by 2.3% per year [88]. Parathyroid hormone (PTH) secretion increases with age, as production of 1,25-dihydroxycholecalciferol (calcitriol) and intestinal calcium absorption are reduced [89]. Nutritional deficiency vitamin D and limited exposure to sunlight, coupled with reduced conversion of 25a-hydroxycholecalciferol to calcitriol, contribute to the reduced calcitriol levels seen in elderly men [89,90] Age-related osteopenia is the result of multiple factors, including altered dynamics of bone cell populations inherent to aging p e r s e aggravated by multiple nutritional and hormonal changes including deficiencies of sex steroids, GH, IGF-I, and calcitriol.

E. C a r b o h y d r a t e M e t a b o l i s m One of the major consequences of the age-related hormonal changes is the emergence of type 2 diabetes. This is the result of both altered insulin secretion and action with age [91]. These changes may be partly due to decreased physical activity and altered body composition favoring accumulation of central adiposity. The incidence of type 2 diabetes increases progressively with age starting at about age 40. Approximately 20% of the population in the United States over the age of 65 years has type 2 diabetes mellitus and at least 40% have glucose intolerance [91,92]. A decline in insulin secretory capacity with age along with reduced insulin sensitivity is common [91]. The plasma levels of glucagon and its clearance rate remain unaltered. Clinical diabetes, especially when poorly managed, causes a variety of complications that are associated with a poor quality of life. Increased glycation of various proteins and enhanced lipid peroxidation accelerate the age-related deterioration of various organ systems. In particular, body composition and vigor are adversely affected. Older subjects with diabetes are at increased risk of dehydration and malnutrition. Optimization of blood glucose control reverses most of the short-term and possibly long-term complications of diabetes.

E Water Metabolism The age-related changes in water and electrolyte homeostasis are summarized in Table IV. The total body water and

118

KORENMAN

TABLE IV

Biological Changes with Age

Body composition Increased: body weight until the sixth decade, thereafter declines; central adiposity Normal:extracellular fluid volume Decreased: lean body mass, bone mass, muscle mass; intracellular and total body water. Cardiovascular system Increased: stroke volume, end-diastolic volume, systolic blood pressure Normal: cardiac output, myocardial contractility Decreased: heart rate, ejection fraction Pulmonary function Increased: residual volume, closing volume Normal: total lung capacity Decreased: vital capacity, arterial PaO2, elastic recoil, maximum expiratory flow rate, maximum voluntary ventilation Digestive system Increased: frequency of teritary contractions in esophagus Normal: motility of stomach and intestine Decreased: salivation, taste, peristalsis of esophagus, acid/pepsin production; colonic motility Hormonal system See Tables I-III Central nervous system Increased: incidence of neurodegenerative diseases and depression, difficulty in learning new tasks Normal: overall intelligence Decreased: speed of cognitive processing, memory I-Iematopoietic/immune system Increased: incidence of anemia, certain hematological malignancies Normal: complete blood count, B cells, macrophages Decreased: progenitor cells, certain components of complement, T cells, intracellular bactericidal activity

intracellular fluid volume are decreased with age while extracellular blood volume is maintained. Elderly men have reduced thirst perception [93]. Basal arginine vasopressin (AVP) secretion may increase with age [94]. The osmolar threshold (the level of plasma osmolarity that will initiate AVP secretion) is lower in the elderly. However, AVP responses to volume and pressure changes are reduced, and the renal response to AVP is also blunted with age [94]. This results in a reduced capacity to conserve water that predisposes the elderly to dehydration, especially when water access is limited or during excess water losses as a result of intercurrent illness. A reduced capacity to generate angiotensin, a potent stimulator of AVP and thirst, also limits the ability of older subjects to maintain water homeostasis. The ability to maintain salt and water balance is further compromised by changes in atrial natriuretic peptide (ANP) secretion [95]. Baseline ANP is increased in elderly subjects and the expected reduction in ANP following dehydration is blunted. The natriuretic effect of ANP is probably preserved, although hemodynamic responses to ANP may be reduced

ET AL.

[95]. The increased ANP secretion with age may contribute to the suppression of plasma renin activity and aldosterone secretion.

IV. O T H E R RELATED

CHANGES

WITH

TO HORMONAL

AGING

FACTORS

A variety of other physiological and structural changes occur with age [96] (Table IV). Some of these are probably related to processes inherent to aging per se, whereas others are secondary to lifestyle changes or nutritional and hormonal alterations. Consequent to the loss of muscle mass, the basal metabolic rate is reduced with age [96]. The reduced skeletal muscle mass with age, along with changes in cardiovascular and pulmonary physiology, results in reduced exercise capacity and low maximum oxygen consumption (VO 2 max). It appears that the changes in pulmonary function are more important than the changes in the cardiovascular system in limiting exercise capacity in the elderly [97]. The partial pressure of arterial oxygen (PaO2) declines steadily with age while P a C O 2 is not significantly altered [97]. Weak respiratory muscles, decreased lung compliance, and increased chest wall stiffness account for most of the age-related changes in pulmonary functions, some of which may be related to cigarette smoking. The other components of age-related loss of lean body mass are bone loss and altered body water content. Hormonal factors, such as loss of androgens, GH, and IGF-I and nutritional factors such as calcium and vitamin D deficiency, along with genetic factors, account for the bone loss (Table V) [98].

TABLE V

Epidemiological Correlates with Erectile Dysfunction a

Positively associated with ED Age Cigarette smoking Depression, inward looking or with expressed anger Diabetes, treated with medications (more severe) Cardiovascular disease, treated Use of vasodilators (however they were defined) Not grossly associated with ED Hypertension Alcohol intake over a wide range Allergies Serum cortisol or DHT level Inversely associated with ED Dominance DHEA level in blood HDL cholesterol level in blood a Adapted from Feldman et al. [98].

CHAPTER7 Changes in Aging Men V. A N D R O G E N

EFFECTS

AND REPRODUCTIVE

119 ON SEXUAL

FUNCTION

The biological actions of androgens are far reaching, and reduced androgen availability in aging men contributes to a host of biological changes. Testosterone acts on most body tissues. However, the biological effects in aging tissue do not always correlate with plasma concentrations, because local tissue factors such as conversion to DHT or E 2, or metabolism to glucuronides, modulate its activity. Thus, although T concentrations may decrease with age, some T-sensitive organs, especially the prostate, commonly undergo hyperplasia. Although androgens have a permissive role for prostatic tissue growth and development, their precise role in benign prostatic hyperplasia (BPH) is not clear. This may be related to altered T metabolism in aging prostate such that in B PH the ratio of DHT to 3ce-androstanediol is increased compared to normal prostate [16]. The androgen dependency of the other accessory sex organs is also well established. Secretory epithelium of epididymus regresses following castration. Exogenous androgen treatment restores some but not complete secretory function [16]. The seminal vesicles, in particular the epithelial component, are androgen dependent and so is spermatogenesis. However, these organs change modestly with age, suggesting that the age-related reduction in T availability is not sufficient to result in a clinically relevant change in these organs. Androgens do not appear to alter erections in response to erotic films [99,100]. To what extent do androgens play a role in erectile function in the adult male and what are the effects of declining androgen availability with aging on sexual function? Of course androgens are necessary both in utero and after birth for the proper development of the male external and internal genitalia. Growth of the penis and testes during puberty is also totally dependent on adequate androgen availability. It has already been noted that adequate T concentrations are necessary for normal libido [ 101 ]. Pharmacological reduction of circulating T in young men reduced sex drive and responses although the erectile response to erotic stimuli was unchanged [ 102]. It has been reported that severe hypogonadism is responsible for less than 7% of cases of erectile dysfunction (ED) [103,104]. An important role for T is suggested by studies demonstrating in other species that castration reduces erectile capacity, which can be preserved with dihydrotestosterone (DHT) [105]. In rats, T facilitates centrally mediated erections and yawning (a sexual response) [106]. Androgens stimulate the sexually dimorphic brain nuclei and increase the size and dendritic spread of the spinal cord motor neurons innervating the bulbospongiosus and ischiocavernosus muscles [107]. They may also affect the penile vascular re-

sponse mechanism [ 108]. In men, androgens are responsible for normal seminal fluid and prostatic secretions and the frequency of nonerotic or nocturnal erections [99]. The frequency of nocturnal penile erections correlates with circulating testosterone [109]. The reduction in testosterone status after age 45 is generally paralleled by a gradual decline in sexual desire, arousal, and activity [110]. The magnitude of this decline varies widely [ 111] and is frequently the result not of aging but of other factors, including medications, depressed mood, alcohol use, obesity, and chronic illnesses (e.g., diabetes, vascular disease). Despite reduced libido, sexual enjoyment and satisfaction do not decline with age [112]. An important predictive factor for sexual enjoyment in aging men, as in younger men, is the quality of the marital relationship [ 112]. As distinguished from libido, the role of a decline in T in ED in men as they age has not been well characterized, but in one study there was no relationship between the mild hypogonadism of aging and ED. Both conditions were common, but independently segregated [21 ]. Many men seek androgen replacement to improve erectile function, but the use of testosterone for this purpose in eugonadal men is usually unsuccessful [20,113]. Testosterone supplementation does, however, increase sexual interest [ 114].

VI. NONERECTILE SEXUAL DYSFUNCTIONS Sexual dysfunction in men consists of a small group of problems including early or premature ejaculation, retarded or lack of ejaculation, loss of libido, and erectile d y s function.

A. E j a c u l a t o r y D y s f u n c t i o n The mechanism of ejaculation encompasses seminal emission, ejaculation, and bladder neck closure. Afferent stimuli include activation of higher centers of sexual response to reach a threshold. Sympathetic nerves then cause smooth muscle contraction in the epididymis, vas, seminal vesicle, and prostate to produce filling of the prostatic urethra with the seminal emission. Finally, contraction of the bulbospongiosus and ischiocavernosus muscles and contraction of the bladder neck lead to propulsion of the semen out of the penis while the sensation of orgasm is experienced. 1. PREMATURE EJACULATION Early or premature ejaculation is perhaps the most common disorder of sexual function in men, affecting at least a third. Ejaculation usually occurs very close to the time of

120

K O R E N M A N ET AL.

vaginal penetration, in the most severe cases before vaginal penetration. Milder cases are associated with ejaculation after a few seconds of thrusting. This condition improves with sexual experience and age but persists in a substantial number of men well into the fourth and fifth decade. It is thought to be due to anxiety associated with sexual activity. For men with premature ejaculation, sex therapy behavioral techniques are beneficial but usually do not suffice [115]. Pharmacological interventions targeted at augmentation of serotonin function have been reported in numerous studies to be highly effective for this condition. In particular, the use of standard doses of the serotoninergic agents fluoxetine, paroxetine, sertraline, and clomipramine have been found to prolong significantly latency to ejaculation [116-121 ], with improvement noted as early as 1 week following initiation of medication [120]. One study found that clomipramine produced the greatest increase in latency time, although it was associated with more side effects than the serotonin reuptake inhibitors (fluoxetine, paroxetine, and sertraline). Following discontinuation of the serotonin reuptake inhibitors, premature ejaculation has been observed to recur in 90% of treated men [ 117]. 2. RETARDED EJACULATION Retarded ejaculation, which is unusual, is sometimes found in association with the use of antipsychotic drugs and with certain antidepressives [ 122,123]. Sometimes removal or a change of medication will reverse the condition. Retrograde ejaculation is one of the consequences of prostate surgery. Most commonly this is due to surgical damage to the vesical-urethral sphincters, making it easier to pass the ejaculate into the bladder than through the urethra. This may also occur with diabetic autonomic neuropathy, in which the same sphincters become dysfunctional. Retrograde ejaculation is usually treated with reassurance but many men complain that their sensation of orgasm and release is substantially reduced in the absence of an ejaculate.

VII. ERECTILE

DYSFUNCTION

Erectile dysfunction, on the other hand, is progressively common with age and has many etiologies and risk factors. There have been extensive publications and reviews of the field [ 124-127] and we will not try to recapitulate here the history of research in the area. In the past year the problem of ED has mutated from an underdiagnosed disorder managed by a few physicians to a public phenomenon characterized by media frenzy, numerous jokes, and an intense debate featuring patients, health care providers, and government, regarding whether treatment of ED with sildenafil or other oral agents should remain covered by insurance. What kind of problem is ED and how often does it occur?

A. E p i d e m i o l o g y At the 1993 National Institutes of Health Consensus Development conference on ED [ 128], erectile dysfunction was formally defined as "an inability of the male to achieve an erect penis as part of the overall multifaceted process of male sexual function." Although the adopted definition seemed refreshingly simple and useful at the time, in practice it has become too vague. For example, how erect? Does a nonrigid but usable erection count as ED or normal erectile function? How is a full erection for masturbation and on awakening, but no erection in the presence of a partner, to count? What should we call variable erectile response? These questions pertain not only to research determining the prevalence of the condition but also to difficult questions as to support of the treatment of ED by health insurers. Feldman and colleagues in the Massachusetts Male Aging Study (MMAS) [98] developed a nine-point questionnaire and divided ED into three levels of dysfunction, with the most severe being a complete absence of sexual response and "mild" being an occasional failure of certain aspects of response. By these criteria, in a community-based group of men from ages 40 to 70 years, 9.6% had complete ED, 25.2% had moderate ED, 17.2% had minimal Ed, and 48% had no ED. Although this approach engendered a degree of criticism, until the advent of sildenafil this partition was effective in the selection of patients for treatment, because men usually wished to be treated only if they were seriously affected by the problem. With the advent of sildenafil, the target population could conceivably become 52% of men ages 4 0 - 7 0 years and much higher percentages of older men [129]. If, of the 110 million American males over age 40 years, 50 million had ED, and they wanted to have sex once weekly, that would require 52 • 50 million pills at $8.00/pill or nearly $21 billion/year for this single indication. Obviously, a more precise, medically determined objective diagnosis of ED is required. Table VI lists factors found in the MMAS [98] to be associated with an increase of ED and factors decreasing ED. Factors inhibiting sexual function include coronary artery disease and diabetes, especially if treated (more severe) and if associated with smoking cigarettes. Depression and anger, whether internalized or expressed, were highly associated with ED. This is of particular importance because depression is very strongly associated with loss of libido (see above) and is greatly underdiagnosed in men, especially in middleaged men (see below). Studies of populations in the medical system, however, although biased because of the patterns of referral and the expertise of the practitioner, demonstrate a high prevalence of hypertension, coronary heart disease, and diabetes, as well as treatment for each, associated with ED [103,104]. Gener-

CHAPTER 7 Changes in Aging Men

121

TABLE VI Condition

Criteria for Depressive C o n d i t i o n s Symptoms

Duration

Major depressive episode

Five or more of the following symptoms present for the same 2-week period (must include symptom 1 or 2): 1. Depressed mood most of the day nearly every day 2. Markedly diminished interest or pleasure in almost all activities 3. Significant weight loss when not dieting, or weight gain 4. Insomnia or hypersomnia 5. Psychomotor agitation or retardation 6. Fatigue or loss of energy 7. Feelings of worthlessness or excessive/inappropriate guilt 8. Diminished ability to think/concentrate, or indecisiveness 9. Recurrent thoughts of death

2 weeks

Dysthymic disorder

Depressed mood most of the day more days than not for at least 2 years (two or more of the following symptoms while depressed): 1. Poor appetite or overeating 2. Insomnia or hypersomnia 3. Low energy or fatigue 4. Low self-esteem 5. Poor concentration/indecisiveness 6. Feelings of hopelessness

At least 2 years

Adjustment disorder with depressed mooda

Development of emotional or behavioral symptoms in response to an identifiable stressor occurring within 3 months of the onset of the stressor; predominant manifestations are depressed mood, tearfulness, or feelings of hopelessness (these symptoms are clinically significant as evidenced by either of the following criteria): 1. Marked distress in excess of what would be expected from exposure to the stressor 2. Significant impairment in social/occupational functioning

Subclinical depression

Depressive symptoms that do not meet criteria for major depression, dysthymia, or adjustment disorder with depressed mood

a

Occurs within 3 months of stressor and does not persist beyond 6 months after stressor terminates

a During the 2-year disturbance, the person has never had a major depressive episode and has never been without the above symptoms for more than 2 months. Symptoms must cause significant distress or impairment in functioning and are not due to the effects of a substance or general medical condition. b The disturbance should not meet criteria for another psychiatric disorder and does not represent bereavement. Once the stressor has terminated, the symptoms do not persist more than 6 months.

ally, in those studies, the definition was limited to those with a c o m p l e t e inability to c o m p l e t e sexual i n t e r c o u r s e for at least 3 m o n t h s . Other clinical associations with E D i n c l u d e d pelvic surgical p r o c e d u r e s and m a j o r p e r i p h e r a l vascular disease as well as n e u r o l o g i c a l disorders s u c h as m u l t i p l e sclerosis. T h e drugs associated with E D i n c l u d e d p r i m a r i l y vasodilators, t r e a t m e n t s for d e p r e s s i o n and psychosis, and h o r m o n e s or drugs affecting the reproductive e n d o c r i n e syst e m [98,122,123].

B. Erectile Mechanism W h e n at rest, the penis m a i n t a i n s a state of flaccidity t h r o u g h a - a d r e n e r g i c a l l y m e d i a t e d c o n t r a c t i o n of c a v e r n o s a l and vascular s m o o t h m u s c l e , inhibiting b l o o d flow into the organ [130,131 ]. As the result of an erotic stimulus, received t h r o u g h one or m o r e of the five senses or via m e m o r y (fan-

tasy), inhibition of the s y m p a t h e t i c d i s c h a r g e takes place and a p a r a s y m p a t h e t i c d i s c h a r g e is initiated, with p r e s y n a p t i c t e r m i n a l s in the pelvic p l e x u s [ 1 3 2 - 1 3 4 ] . Postsynaptically, the signals travel by n o n a d r e n e r g i c , n o n c h o l i n e r g i c ( N A N C ) nitric oxide (NO) nerves to t e r m i n a t e in the s m o o t h m u s c l e of the c a v e r n o s a l arteries and t r a b e c u l a r sinusoids [135]. (Fig. 3). T h e s e m u s c l e s relax w h e n N O stimulates g u a n y l y l cyclase to c o n v e r t G T P to cyclic GMP. In s m o o t h m u s c l e , cyclic G M P inhibits Ca entry and facilitates Ca loss [136, 137]. In the a b s e n c e of sufficient C a 2+ s m o o t h m u s c l e relaxes, allowing the heart to p u m p m u c h m o r e b l o o d into the corpora, i n d u c i n g penile swelling. O t h e r n e u r o t r a n s m i t t e r s that have b e e n related to erectile f u n c t i o n include prostaglandin E 1 (PGE1) and other stimulators of a d e n y l y l cyclase, vasoactive intestinal p e p t i d e (VIP), endothelin, calcitoninrelated peptide, and histamine. T h e y p r o b a b l y play a m i n o r role in the h u m a n u n d e r p h y s i o l o g i c a l conditions. I n c r e a s e d inflow of b l o o d alone will not result in an

122

KORENMAN ET AL.

FIGURE 3 Neurogenic mediation of penile vasodilatation via smooth muscle relaxation. On the left the postganglionic autonomic nerve is seen stimulating a smooth muscle cell, on the right. The group of cells in the left lower section represent endothelial cells. ARG, Arginine; CIT, citrulline, GC, guanylyl cyclase; DHT, 5a-dihydrotestosterone EFS, electric field stimulation (nerve discharge); ACH, acetylcholine; PGE~, prostaglandin E I.

erection; it requires nearly complete inhibition of venous return. That is accomplished passively by the unique anatomy of the penis, in which the expanding corporal sinusoids compress the subtunical plexus of veins draining the corpora cavernosa against the unyielding tunica albuginea [138]. The subtunical plexus, in turn, is drained through veins penetrating the tunica albuginea that are also compressed during stimulation, so that at maximum erection, penile blood flow is nearly zero [ 139,131 ]. Penile blood pressure may exceed systemic pressure at that time as a result of contraction of the ischiocavernosus muscle, which acts as a constriction ring at the base of the penis [ 140]. Ejaculation is neurally mediated in response to the filling of the prostatic urethra with semen and achievement of the "orgasmic plateau" of sexual stimulation, via contraction of the bulbospongiosus muscle [ 140].

C. Inadequate Erectile Function What happens to cause ED? First, there can be inhibition of the central nervous system centers mediating the response to erotic stimuli. Both testosterone deficiency (see above) and depression (see below) reduce libido substantially and together they are responsible for the majority of cases of reduced sexual interests in adult men. Second, the integrity of the neural pathways mediating an erection can be interrupted by spinal cord injury, pelvic surgery (usually due to resection of a prostate or colon cancer),

or autonomic neuropathy such as in Type I diabetes mellitus, or by primary neurological diseases such as multiple sclerosis [ 141 ]. A number of medications and recreational drugs affect the neural response at the periphery so as to contribute to ED [98,122,123,142,143]. Third, there may be a failure of response to the neural signals as a result of diminished NO synthesis, which has some relation to intact neural pathways, adequate androgen availability, and cavernosal smooth muscle integrity. This is commonly found in association with diabetic ED [105,144,145]. There is also evidence of enhanced contractility due to increased ce-adrenergic sensitivity in ED [146]. Fourth, ED is very commonly associated with abnormalities of the intrinsic tissues of the corpora cavernosa, including disrupted muscle fibers, an increase of dense connective tissue in the perisinusoidal area, and a reduction of tunical elastic fibers, which probably prevent adequate compression of the subtunical venous plexus [144,147-149]. This is commonly associated with atherosclerotic disease and diabetes and is thought to be due in part to ischemia. Fifth, the blood supply to the penis may be compromised by arterial atheromatous disease, a very low cardiac output, or arteriolar disease [ 150]. These conditions, once thought to be common and irreversible, probably account for a small proportion (less than 20%) of cases of ED. Failure of venous occlusion is very common in ED [ 151, 152], and although once it was considered to be a common etiologic factor, it is now believed to be largely a conse-

CHAPXER7 Changes in Aging Men quence of an inadequate filling rate and a degree of scarring in the perisinusoidal tissues except in cases of penile trauma, in which damage to large vessels is not uncommon, and in Peyronie's disease, in which peripheral fibrous plaque formation often inhibits venous compression [ 148]. Penile vein ablative surgery has been generally unsuccessful in the management of ED. How conditions associated with a high incidence of ED [98,124] produce the effects listed above is not fully understood, and until a minimally invasive acceptable method of biopsying the corpora cavernosa is developed we have no good way to correlate penile structure and ultrastructure with function and disease.

D. D i a g n o s t i c A s s e s s m e n t With the advent of sildenafil and other oral medications to come, the role of the health system in the diagnosis and treatment of ED has changed. No longer must physicians carefully elicit information about sexual function from reluctant patients. Rather, they often have to ensure that the patient requesting treatment indeed has an erectile problem as opposed to other sexual disorders or a desire for a somewhat enhanced lifestyle. Prior to initiating therapy, physicians must take a detailed sexual history that includes the nature of the dysfunction-weak or absent erection, erection of short duration, curved or distorted erection; the duration and progression of the condition; prior level of sexual activity, including repertoire and frequency as well as partner's interest, availability, and satisfaction. A careful review of the presence of nocturnal and especially morning erections gives considerable information about the erectile potential of the individual with simple therapies [153], although the long-standing attempt to differentiate psychogenic from organic ED by these means seems irrelevant. However, psychological elements, which should be elucidated, are present in virtually every case of ED. In this context, a simple questionnaire designed to evaluate male sexual dysfunction, e.g., the International Index of Erectile Function, can be used by clinicians [154]. Honest answers to this line of inquiry will provide an understanding of the degree of ED, relationship issues, and, in concert with the remainder of the medical assessment, medical and life style risk factors that contribute to the problem. Associated factors including the patient's endocrine status also need to be assessed during the history and physical examination. For laboratory testing, in patients who are regularly followed, we simply measure a TSH and bioavailable T, and if both are normal, we proceed. A low bioavailable T will precipitate measurement of prolactin and LH and a very low bioavailable T will precipitate an MRI of the pituitary gland, especially in younger men [155]. In the vast majority of men over the age of 40, and more particularly over the age

123 of 50, ED will be multifactorial in origin and susceptible to simple therapies (see below). For those with unusual problems, such as penile trauma or severe Peyronie's disease, evaluation by a skilled urologist is required.

E. T r e a t m e n t For men with bone fide ED, it is difficult to withhold initial therapy when a simple and effective agent such as sildenafil is available. The drug, a pill, is taken 1-2 hr prior to anticipated sexual activity [156]. It acts as an inhibitor of phosphodiesterase V. Phosphodiesterase V is responsible for degrading cyclic GMP to GMP, eliminating its biological effect (Fig. 3). As noted previously, cyclic GMP is the second messenger stimulated by NO in the corpus cavernosum. It is responsible for inhibiting Ca e+ intake and increasing Ca e+ egress from the smooth muscle of the corpora cavernosa, relaxing the arteries and sinusoids. Inhibition of phosphodiesterase V (PDE V) maintains the level of cyclic GMP for a much longer time, facilitating the erection. To be effective, the drug requires a substantial innervation of the penis. The drug does not affect libido. The side effects of sildenafil are attributed to its lack of perfect specificity. Patients may experience headaches, flushes, gastrointestinal distress, visual blurring, or a bluish haze during the 5 hr or so that the drug is active but they willingly accept the side effects if the primary effect is delivered (Fig. 4). In this study, note that with increased dosage, which produced increased efficacy, the discontinuation rate declined despite increased side effects. There are no reports of long-term consequences of sildenafil, which is taken only when intercourse is anticipated. It has been quite successful in restoring erectile function in over two-thirds of men with moderate ED [ 156]. The most significant issues are an absolute contraindication of sildenafil use

ADVERSE EVENTS Dose Response Study

20

Discontinuation Rate

10

7 o

7 c:~

0 4030-

Percent

20-

m~ 0

P

25

50

100

Dose of Sildenafil

H=headache

F=flushing

D=dyspepsia

V=visual disturbance

FIGURE 4 Adverse effects and discontinuation of sildenafil therapy. Derived from the data of Goldstein et al. [ 156].

124 in anyone taking nitrates in any form and an absolute contraindication of the use of nitrates in anyone who has recently used sildenafil. How recently? No one knows for sure. The drugs, in combination, can and have produced vascular collapse. A few deaths have been reported in association with the use of sildenafil in older men, by and large either using nitrates or with preexisting heart disease. It is likely that the death rate is not in excess of what is expected for this population. The next line of therapy in ED is based on the introduction of PGE 1 and or other agents, mainly papaverine and phentolamine, into the corpora cavernosa, by direct injection. Prostaglandin E l (alprostadil) is available in an injectable form (Caverject, Upjohn; EDEX, Schwarz Pharma) [157], and in a form that allows introduction through the urethra [158]. PGE 1 stimulates adenylyl cyclase to produce cyclic AMP. This second messenger inhibits Ca 2§ entry into smooth muscle cells, causing their relaxation. These agents affect the penis directly. They require neither an erotic stimulus nor an intact neural system. Thus, they are useful in spinal cord injury or after radical prostatectomy or colectomy. Their main problems are the necessity to introduce them directly into the penis and their propensity to produce hypotension in about 1% of patients, especially those with severe cardiovascular disease. This is particularly significant with the medicated urethral system for erection (MUSE), which requires up to 1 mg of PGE 1, whereas the injection therapies require only up to 20/zg to be introduced. Intracavernosal injection produces a satisfactory result for about two-thirds of men tested, whereas MUSE is effective for only about one-third. Vacuum tumescence devices can produce an erection in about 90% of the men who attempt them [ 159]. They consist of a plastic cylinder in which the pendulous penis is placed. The cylinder is connected to a vacuum pump, and on evacuation, blood is drawn into the penis. To keep the blood there, an obstructing band is slipped over the base of the penis when a full erection is achieved. These devices are efficient and inexpensive over time. They produce a somewhat abnormal erection in that all of the tissues, not only the corpora, are engorged, and obstruction of blood flow sometimes leads to cooling of the penis. This procedure does not interfere with a patient's medications, nor does it produce hypotension. However, there is a loss of spontaneity with sex, the erection is sometimes "on a hinge," sometimes ejaculation through the obstructing band is a problem, and, in some instances, application of excessive negative pressure produces petechiae. Also a degree of manual dexterity and skill is required to get the device to work. Penile prostheses were once the primary therapy for ED. The recent advent of much less costly and invasive approaches had made them more of a last resort. They produce a satisfactory erection in about 80% of the men who have them. Unfortunately, the rigid rod versions have a tendency

KORENMAN ET AL.

to explant and the versions including a pump mechanism have a tendency to undergo mechanical failure [ 160]. Vascular surgery is employed in specialized urology centers to establish appropriate blood flow to the penis. It is indicated only for specialized conditions such as penile or pelvic trauma to a young person. Considering that sildenafil is a small molecule, it is highly likely that other oral agents with the appropriate specificity will become available for therapy. With the cloning of penile inducible nitric oxide synthase [ 161 ], it may be possible to develop gene therapy or very specific small molecules that can enhance or preserve NO synthesis or concentrations. Numerous other components of this now well-understood system are susceptible to pharmacological attack as well.

VIII. MANOPAUSE AND MENTAL HEALTH A. D e p r e s s i o n in M i d d l e - A g e d and Elderly M e n ~ E p i d e m i o l o g y and R i s k Factors Although a number of epidemiological studies have assessed the relationship between mood and aging in women [ 162-164], the psychological changes accompanying aging in men have received little attention. Studies of mood in middle-aged men are particularly scarce. However, several authors have reported high rates of depressed mood, insomnia, mood swings, irritability, impotence, decreased libido, weakness, and lethargy in this population [165,166]. Because these symptoms may not meet criteria for major depression, as defined by the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) (see Table VII), they are likely to be missed in epidemiological studies of psychiatric disorders. Among the elderly (older than 65 years), epidemiological surveys report lower rates of major depression as compared to younger populations [ 167,168]. However, studies exploring the prevalence of subsyndromal depression, i.e., depressive symptoms not meeting criteria for major depression, have consistently found a high rate among elderly persons [169-171]. Despite the high prevalence of depressive symptomatology among older persons, the symptoms are seldom recognized or treated [ 169,172]. This undertreatment may reflect clinicians' attribution of the symptoms to physical illnesses or to understandable responses to adversity [ 173,174]. Also, the greater tendency among elderly patients to express psychological distress through somatic symptoms contributes to oversights in the diagnosis of depressive disorders [ 174]. Factors associated with depressive symptoms in older men include limited economic resources, poor health, Caucasian race, and impaired sexual functioning [165-177].

125

CHAPTER 7 C h a n g e s in A g i n g M e n

TABLE VII

Criteria for Anxiety and Panic Syndromes

Condition Panic disorder without agoraphobia

Symptoms Recurrent 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.

unexpected panic attacks consisting of the following symptoms a: Palpitations, pounding heart, or accelerated heart rate Sweating, trembling, or shaking Sensations of shortness of breath or smothering Feeling of choking, chest pain, or discomfort Nausea or abdominal stress Feeling dizzy, unsteady, lightheaded, or faint Derealization (feelings of unreality) or depersonalization (being detached from oneself) Fear of losing control or going crazy Fear of dying Numbness or tingling sensations Chills or hot flashes

At least one of the attacks has been followed by the following symptoms: 1. Persistent concern about having additional attacks 2. Worry about the implications of the attack or its consequences (e.g., losing control, having a heart attack, or going crazy) 3. A significant change in behavior related to the attacks Panic disorder with agoraphobia

Same as above, but with anxiety about being in places or situations from which escape might be difficult or embarrassing or in which help may be unavailable b

Generalized anxiety disorder

Excessive anxiety and worry occurring more days than not, about a number of events and activities. Difficulty in controlling the worry, and the anxiety and worry are associated with three or more of the following symptoms (with at least some symptoms present for more days than not for the past 6 months)c: 1. Restlessness or feeling keyed up or on edge 2. Being easily fatigued 3. Difficulty concentrating, mind going blank 4. Irritability 5. Muscle tension 6. Sleep disturbance (difficulty falling or staying asleep, or restless, unsatisfying sleep)

aThe panic attacks are not due to the direct physiological effects of a substance (e.g., a drug of abuse or a medication) or a general medical condition (e.g., hyperthyroidism). bThe situations are avoided or endured with distress or anxiety about having a panic attack or paniclike symptoms, or require the presence of a companion. CThe anxiety, worry, or physical symptoms cause clinically significant distress or impairment in social, occupational, or other important areas of functioning, and are not due to the direct physiological effects of a substance (e.g., a drug of abuse or a medication) or a general medical condition (e.g., hyperthyroidism), and do not occur exclusively during a mood disorder, a psychotic disorder, or a pervasive developmental disorder.

Poor physical function is also a risk factor for depressive symptoms and, conversely, depressive symptoms are associated with subsequent physical decline in elderly persons [169,178]. Being widowed, divorced, or separated are additional risk factors for depressive symptoms in this population [176,177]. Negative stereotypes of aging also are likely to impact a man's mood and self-image [179]. For many men, retirement produces a sense of letdown and can contribute to depressive symptoms [ 180]. Retirement frequently represents a loss of prestige, income, status, purpose, and workrelated friendships [ 181]. Depressive symptoms related to retirement occur most frequently in men whose lives and self-identity centered around their work or in men who have had to retire because of poor health or inability to maintain

their jobs [ 181 ]. Men who are healthy, active, and have adequate financial resources and extended social networks are least likely to experience difficulty with retirement [181].

B. E v a l u a t i o n and T r e a t m e n t o f D e p r e s s e d M o o d in A g i n g M e n Depressive symptoms in aging individuals are often missed by hospital physicians [ 182]. A careful assessment of mood is important in this population, particularly because somatic complaints may mask symptoms of depressed mood [183]. Any patient presenting with fatigue, changes in

126 appetite or sleep, and reduced libido should be evaluated for depressed mood. Depression rating scales such as the Geriatric Depression Scale [184] help screen for major depression. Unusual thought content should also be explored, because approximately 50% of depressed men over age 60 years experience delusional depression. Common presentations include delusions of being ill (somatic delusions) and of being followed or spied on (persecutory delusions) [183]. Criteria for depressive illnesses and anxiety syndromes are presented in Tables VI and VII. Patients who endorse depressive symptoms but do not meet full criteria for major depression or dysthymia may still benefit from treatment. The relationship between the depressive symptoms and psychosocial stressors (e.g., death of a family member, loss of a job, onset of an illness, or relocation) should be explored. Knowledge of factors that may have triggered the depressive mood changes is important in the choice of intervention. Interpersonal psychotherapy is particularly helpful for men who have undergone recent life transitions, because it focuses on strategies to cope with role changes or grief or to modify unrealistic expectations about relatives and other people in one's life. Cognitive-behavior therapy is an alternative approach based on training people to identify and challenge self-defeating thoughts, such as "I'm no longer working, therefore people will find me boring." Certain medications (e.g., antihypertensive agents) have been linked with depressive mood changes, thus a complete determination of the patient's medication usage should be obtained. Laboratory studies should include thyroid function testing and a bioavailable T level to rule out hypothyroidism and hypogonadism as contributing factors to the depressive symptoms. Antidepressant medications can be very helpful in promoting the recovery from major depression. Currently the most commonly used antidepressant medications are the serotonin reuptake inhibitors (fluoxetine, sertraline, and paroxetine), and they are generally well tolerated and are relatively safe in overdose. Typical side effects from these medications include gastrointestinal symptoms and impairment of sexual function, mostly libido. The tricyclic antidepressants (e.g., nortriptyline, desipramine, doxepin, and amitriptyline) have less effect on sexual function but can produce sedation, orthostatic hypotension, blurry vision, constipation, and EKG changes. Other available antidepressant medications include venlafaxine, nefazodone, bupropion, and the monoamine oxidase inhibitors (tranylcypromine and phenelzine). The antidepressants that are least likely to affect sexual function are bupropion and nefazodone. Monoamine oxidase inhibitors have the disadvantage of requiring very close attention to dietary guidelines and drug-drug interactions, to avoid the possibility of a "tyramine reaction," which can produce an abrupt and dangerous rise in blood pressure. When using antidepressant medications with men aged 65 years or older, the starting dose should be approximately half

K O R E N M A N ET AL.

that used for younger populations. Patients should be reminded that beneficial effects may not become fully apparent until after 4 - 6 weeks of treatment. Once a patient has experienced a positive response, he should be maintained on the same dose for a minimum of an additional 6 months. Long-term follow-up should continue after resolution of the depression, because the likelihood that a depressive episode will recur exceeds 50% for individuals aged 60 years or older [ 185]. Ideally, medications should be used in combination with psychotherapy and life style changes. Depressive symptoms may discourage men from maintaining healthy habits, such as exercising and not smoking, and following healthy diets. Depressed mood has also been significantly associated with a lower likelihood of engaging in walking, gardening, and exercise [186]. Unhealthy aspects of the patient's life style should therefore be explored, such as being sedentary, using alcohol and nicotine, and tendencies toward isolation, and the patient should be encouraged to exercise regularly and to be involved in stimulating activities. For men with limited social support networks, referrals to group therapy is beneficial. Screening and treatment of depressed mood in older people is a cost-effective intervention in terms of health and well being per dollar spent [ 187]. Appropriate treatment also appears to increase the number of years during which older persons are free of disability [ 169]. More research is needed, however, on specific screening and treatment interventions for middle-aged and elderly men, particularly because these populations are growing with the aging of the baby-boom generation.

C. A n d r o g e n s and the Central N e r v o u s S y s t e m Declines in bioavailable T may well account for the reduced libido with age. The latter is believed to be a central nervous system (CNS)-related effect of T that may be modulated through E 2 produced locally [18]. Androgens are necessary but not sufficient for maintaining normal libido. In older men, unlike young men, higher plasma T levels are associated with greater sexual activity [97,99] Also, latency to erection stimulated by erotic material correlates with T levels. In hypogonadal men, T replacement restores sexual interest and improves the latency, frequency, and magnitude of the nocturnal penile tumescence and the frequency of early morning erections [ 100,115]. The effect of T on the CNS extends beyond sexual behavior. T has been shown to alter mood, memory, ability to concentrate, and the overall sense of vigor and well being [ 117119]. A number of studies have examined the relationship between mood and levels of testosterone in men. However, most have included wide age ranges rather than focusing on middle-aged or elderly men. Some of these studies have

CHAPTER7 Changes in Aging Men

127

found testosterone levels in men with major depression to be lower [88], whereas others have found no significant difference from controls [189,190]. Methodological problems may explain the discrepant findings, including a lack of control for time of day of blood-drawing, total T versus free or bioavailable T, age distribution, body mass index (BMI; high B MI is associated with decreased T binding and thus lower total T values), cortisol levels (which may affect the hypothalamic-pituitary-gonadal axis), and medication use. One study that did control for medical illness, age, alcohol use, weight, and use of medications found no significant differences in free or total testosterone among 12 patients with major depression compared with 12 controls. It did identify a trend for lower testosterone levels (10% lower total testosterone and 20% lower free testosterone) in the depressed group [190]. Replication of this study in a population of middle-aged men and with a larger sample size would be of immense interest. In the only study of T levels in middleaged men, a high level of psychosocial stress was inversely related to free T levels in a sample of 439 men aged 51 years [191]. The authors concluded that psychosocial stress may be associated with premature aging in middle-aged men.

IX. P S Y C H O L O G I C A L

STATE

AND SEXUAL FUNCTION A. P s y c h o l o g i c a l C a u s e s o f E D Negative expectations of changes of sexual functioning with age may contribute to erectile difficulties [192]. Other potential causes include marital conflict, employment-related problems, family illnesses, boredom, poor communication of sexual needs, and lack of interest from one's spouse. An important and underrecognized etiology for sexual dysfunction is depressed mood. In a study of 1709 men, moderate to complete ED was found 1.82 times more in those men with depressive symptoms (as assessed by Center for Epidemiological Studies--Depression Scale) compared to those without symptoms, after controlling for age, health, medication use, demographic factors, and hormone levels [ 193]. Depressed mood has also been associated with reduced penile rigidity and nocturnal penile tumescence (NPT) time [194,195]. In the Massachusetts Male Aging Study [98], measures of depressed mood and anger were strongly correlated with ED, and were postulated to result from elevations in blood catecholamines, producing vasoconstriction and thereby inhibiting the physiological events necessary for normal sexual function. Because ED can dramatically affect mood and self-confidence [165], a vicious cycle may develop in which depressed mood and anxiety concerning sexual performance exacerbates erectile difficulties.

B. P s y c h o l o g i c a l E v a l u a t i o n a n d T r e a t m e n t of Sexual Dysfunction Premature ejaculation is the most common male sexual dysfunction, occurring in approximately 36-38% of men [ 196]. Psychological factors, including high levels of anxiety [197] and lack of intimacy with one's partner [198], are linked with this condition. Although psychogenic factors are also common in young men with ED, in men over age 50 years, most have organic etiologies for the sexual dysfunction [ 199] and often psychological issues exacerbated by ED [200]. An evaluation of sexual dysfunction, therefore, should include a psychological assessment. A review of situational factors associated with the dysfunction is essential in evaluating the extent to which the problem may have a psychogenic origin. For example, a man's sexual problems may arise only when he feels criticized or rejected, or only when he is under pressure at work. When present, alcohol and substance abuse will impair sexual function. Partners should be present during the evaluation, because they can provide useful observations and the relationship between the two can be explored. A tense or conflictual relationship is a major impediment to successful restoration of sexual function, and couples' therapy should be recommended as part of the treatment strategy. A woman's reaction to her partner's sexual difficulties should also be assessed, because she may view the man's sexual problems as a reflection on herself and feel hurt or angry. An open discussion, in which common reactions are described and normalized, can help bolster the couple's mutual trust and support. Even if an organic etiology for the sexual dysfunction is identified, psychological evaluation is still beneficial because the couple's emotional reactions may exacerbate the problem. Marital therapy may be necessary in cases in which either partner experiences persistent frustration or hostility toward the other. Sex therapy techniques can also be of great benefit, and include structured sexual exercises, psychodynamic exploration of emotional conflicts, and cognitivebehavioral strategies [ 199].

X. C O N C L U S I O N S In the 1960s, the claim was that we all began to go downhill after the age of 30, and we should never trust anyone over 30. Well, that crowd is all in its 50s now. What does happen to men as they age and what can we learn to make that inevitable process healthier and more enjoyable? Must the acquisition of wisdom invariably be associated with "settling" of the body? We really do not know and the information presented here provides only an antipasto to what should be a rich scientific repast. Only by much more intensive investigation of men as

128

KORENMAN ET AL.

t h e y p a s s t h r o u g h t h e i r 4 0 s a n d 5 0 s w i l l w e b e a b l e to r e c o m mend soundly behaviors and evaluations that will not only i m p r o v e h e a l t h a n d w e l l b e i n g , b u t , in t h e l o n g r u n , p e r h a p s

19.

r e d u c e h e a l t h c a r e c o s t s as w e l l .

References 1. Kaufman, J. M., and Vermeulen, A. (1997). Declining gonadal function in elderly men. BailliOres Clin. Endocrinol. Metab. 11,289-309. 2. Nowak, E V., and Mooradian, A. D. (1996). Endocrine function and dysfunction. Encycl. Gerontol. 1,477-491. 3. Mooradian, A. D. (1993). Mechanisms of age-related endocrine alterations. Part II. Drugs Aging 3, 131-146. 4. Mooradian, A. D. (1991). Geriatric sexuality and chronic diseases. Clin. Geriatr. Med. 7, 113-131. 5. Mooradian, A. (1993). Mechanisms of age-related endocrine alterations. Part I. Drugs Aging 3, 81-97. 6. Kovacs, K., Ryan, N., Horvath, E., Singer, W., and Ezrin, C. (1980). Pituitary adenomas in old age. J. Gerontol. 35, 16-22. 7. Sun, Y., Xi, Y. P., Fenoglio, C. M., Pushparaj, N., O'Toole, K. M., Kledizik, G. S., Nette, E. G., and King, D. W. (1984). The effect of age on the number of pituitary cells immunoreactive to growth hormone and prolactin. Hum. Pathol. 15(2), 169-180. 8. Zegarelli-Schmidt, E., Yu, X. R., Fenoglio-Preiser, C. M., O'Toole, K., Pushparaj, N., Kledzik G., and King, D. W. (1985). Endocrine changes associated with the human aging process: II. Effect of age on the number and size of thyrotropin immunoreactive cells in the human pituitary. Hum. Pathol. la(3), 277-286. 9. Matsumoto, A. M., and Brenner, W. J. (1984). Modulation of pulsatile gonadotropin secretion by testosterone in man. J. Clin. Endocrinol. Metab. 58, 609-614. 10. Tsuruhara, T., Dufau, M. L., and Cigorraga, S. (1977). Hormonal regulation of testicular luteinizing hormone receptors. Effects on cyclic AMP and testosterone responses in isolated Leydig cells. J. Biol. Chem. 252, 9002-9009. 11. Bremner, W. J., Vitiello, M. V., and Prinz, P. N. (1983). Loss of circadian rhythmicity in blood testosterone levels with aging in normal men. J. Clin. Endocrinol. Metab. 56, 1278-1281. 12. Manni, A., Pardridge, W. M., Cefalu, W., Nisula, B. C., Bardin, C. W., Santner, S. S., and Santen, R. J. (1985). Bioavailability of albuminbound testosterone. J. Clin. Endocrinol. Metab. 61,705-710. 13. Rittmaster, R. S., Zwicker, H., Thompson, D. L., Konok, G., and Norman, R. W. (1993). Androstandediol glucuronide production in human liver, prostate and skin. Evidence for the importance of the liver in 5alpha-reduced androgen metabolism. J. Clin. Endocrinol. Metab. 76, 977-982. 14. Moghissi, E., Ablan, E, and Horton, R. (1984). Origin of plasma androstanediol glucuronide in men. J. Clin. Endocrinol. Metab. 59, 417421. 15. Casey, R. W., and Wilson, J. D. (1984). Antiestrogenic action of dihydrotestosterone in mouse breast; competition with estradiol for binding to the estrogen receptor. J. Clin. Invest. 74, 2272-2278. 16. Mooradian, A. D., Morley, J. E., and Korenman, S. G. (1987). Biological actions of androgens. Endocr. Rev. 8(1), 1-28. 17. Morley, J. E., Kaiser, E E., Perry, H. M., III, and Patrick, P. (1996). Longitudinal changes in testosterone, luteinizing hormone and folliclestimulating hormone in healthy older men. Metab., Clin. Exp. 46, 410-413. 18. Morley, J. E., Kaiser, E E., Raum, W. J., Perry, H. M., III, Flood, J. E, Jensen, J., Silver, A. J., and Roberts, E. (1997). Potentially predictive and manipulable blood serum correlates of aging in the healthy human

20.

21.

22. 23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

male: Progressive decreases in bioavailable testosterone, dehydroepiandrosterone sulfate and the ratio of insulin-like growth factor 1 to growth hormone. Proc. Natl. Acad. Sci. U.S.A. 94, 7537-7542. Gray, A., Nunez, A. A., Siegel, L. I., and Wade, G. N. (1991). Age, disease, and changing sex hormone levels in middle-aged men: Resuits of the Massachusetts Male Aging Study. J. Clin. Endocrinol. Metab. 73, 1016-1025. Kaiser, F., Viosca, S. P., Morley, J. E., Mooradian, A. D., Davis, S. S., and Korenman, S. G. (1988). Impotence and Aging: Clinical and hormonal factors. J. Am. Geriatr. Soc. 36(6), 511-519. Korenman, S., Morley, J. E., Mooradian, A. D., Stanik-Davis, S., Kaiser, F. E., Silver, A. J., Viiosca, S. P., and Garza, D. (1990). Secondary hypogonadism in older men: Its relation to impotence. J. Clin. Endocrinol. Metab. 71,963-969. Vermeulen, A., and Kaufman, J. M. (1995). Aging of the hypothalamo-pituitary-testicular axis in men. Horm. Res. 43, 25-28. Deslypere, J., Kaufman, J. M., Vermeulen, T., Vogalaers, D., Vandalem, J. L., and Vermeulen, A. (1987). Influence of age on pulsatile luteinizing hormone release and responsiveness of gonadotrophs to sex hormone feedback in men. J. Clin. Endocrinol. Metab. 64, 68-73. Tenover, J. S., Matsumoto, A. M., Clifton, D. K., and Bremner, W. J. (1988). Age-related alterations in the circadian rhythms of pulsatile luteinizing hormone and testosterone secretion in healthy men. J. Gerontol. 43, M 163-M 169. Touitou, Y., F~vre, M., Lagoguey, M., Carayon, A., Bogdan, A., Reinberg, A., Beck, H., Cesselin, E, and Touitou, C. (1981). Age- and mental health-related circadian rhythms of plasma levels of melatonin, prolactin, luteinizing hormone and follicle-stimulating hormone in man. J. Endocrinol. 91(3), 467-475. Touitou, Y., Lagoguey, M., Bogdan, A., Reinberg, A., and Beck, H. (1983). Seasonal rhythms of plasma gonadotrophins: Their persistence in elderly men and women. J. Endocrinol. 96, 15-21. Winters, S. J., Sherins, R. J., and Troen, P. (1986). The gonadotropinsuppressive activity of androgen is increased in elderly men. Metab. Clin. Exp. 33, 1052-1059. Winters, J. J., and Troen, P. (1982). Episodic luteinizing hormone (LH) secretion and the response of LH and follicle-stimulating hormone to LH-releasing hormone in aged men. Evidence for co-existent primary testicular insufficiency and an impairment in gonadotropin secretion. J. Clin. Endocrinol. Metab. 55, 560-565. Muta, K., Kato, K. I., Akamine, Y., and Ibayashi, H. (1981). Agerelated changes in the feedback regulation of gonadotropin secretion by sex steroids in men. Acta Endocrinol. (Copenhagen)96, 154-162. Tenover, J. S., Matsumoto, A. M., Plymate, S. R., and Bremner, W. J. (1987). The effects of aging in normal men on bioavailable testosterone and luteinizing hormone secretion response to clomiphine citrate. J. Clin. Endocrinol. Metab. 65, 118-1126. Tenover, J. S., McLachlan, R. I., Dahl, K. D., Burger, H. G., de Kretser, D. M., and Bremner, W. J. (1988). Decreased serum inhibin levels in normal elderly men: Evidence for a decline in Sertoli cell function with aging. J. Clin. Endocrinol. Metab. 67(3), 455-459. Burger, H. G., and Robertson, D. M. (1997). Editorial: Inhibin in the male--Progress at last. Endocrinology (Baltimore) 138(4), 13611362. Harman, S. M., Tsitouras, P. D., Costa, P. T., and Blackman, M. R. (1982). Reproductive hormones in aging men. II. Basal pituitary gonadotropins and gonadotropin responses to luteinizing hormonereleasing hormone. J. Clin. Endocrinol. Metab. 54, 547-551. Harman, S. M., and Tsitouras, P. D. (1980). Reproductive hormones in aging men. I. Measurement of sex steroids, basal luteinizing hormone and Leydig cell response to human chorionic gonadotropin. J. Clin. Endocrinol. Metab. 51, 35-40.

129

CHAPTER 7 Changes in Aging Men 35. Ishimura, T., Pages, L., and Horton, R. (1977). Altered metabolism of androgens in elderly men with benign prostatic hyperplasia. J. Clin. Endocrinol. Metab. 45, 695-701. 36. Baker, H. W., Burger, H. G., de Krester, D. M., Hudson, B., O'Connor, S., Wang, C., Mirovics, A., Court, J., Dunlop, M., and Rennie, G. C. (1976). Changes in the pituitary-testicular system with age. Clin. Endocrinol. 5(4), 349-372. 37. Moore, R. J., and Wilson, J. D. (1973). The effect of androgenic hormones on the reduced nicotinamide dinucleotide phosphate 4-3-ketosteroid 5-alpha-oxidoreductase of rat ventral prostate. J. Clin. Endocrinol. Metab. 95, 581-592. 38. Korenman, S. G., Viosca, S., Garza, D., Guralnik, M., Place, V., Campbell, P., and Davis, S. S. (1987). Androgen therapy of hypogonadal men with transscrotal testosterone systems. Am. J. Med. 83(3), 471-478. 39. Tunn, S., Hochstrate, H., Grunwald, I., Fluchter, St. H., and Krieg, M. (1988). Effect of aging on kinetic parameters of 5c~-reductase in epithelium and stroma of normal and hyperplastic human prostate. J. Clin. Endocrinol. Metab. 67, 979-985. 40. Tunn, S., Haumann, R., Hey, J., Fluchter, St. H., and Krieg, M. (1990). Effect of aging on kinetic parameters of 3a (fl)-hydroxysteroid oxidoreductases in epithelium and stroma of human normal and hyperplastic prostate. J. Clin. Endocrinol. Metab. 71, 732-739. 41. Hardy, D. O., Scher, H. I., Bogenreider, T., Sabbatini, P., Zhang, Z. F., Nanus, D. M., and Catterall, J. F. (1996). Androgen receptor CAG repeat lengths in prostate cancer: Correlation with age of onset. J. Clin. Endocrinol. Metab. 81, 4400-4405. 42. Gardner, E H., and Besa, E. C. (1983). Physiologic mechanisms and the hematopoietic effects of the androstanes and their derivatives. Curr. Top Hematol. 4, 123-195. 43. Nunez, A. A. (1982). Brief report. Dose-dependent effects of testosterone on feeding and body weight in male rats. Behav. Neural Biol. 34, 445-449. 44. Gray, J. M., Nunez, A. A., Siegel, L. I., and Wade, G. N. (1979). Effects of testosterone on body weight and adipose tissue: Role of aromatization. Physiol. Behav. 23, 465-469. 45. Krotkiewsky, M., Kral, J. G., and Karlsson, J. (1980). Effects of castration and testosterone substitution on body composition and muscle metabolism in rats. Acta Physiol. Scand. 109, 233-237. 46. Urban, R. J., Bodenbury, Y. H., Gilkison, C., Foxworth, J., Coggan, A. K., Wolfe, R. R., and Ferrando, A. (1995). Testosterone administration to elderly men increases skeletal muscle strength and protein synthesis. Am. J. Physiol. 269, E820-E826. 47. Sih, R., Morley, J. E., Kaiser, F. E., Perry, H. M., 3rd, Patrick, P., and Ross, C. (1997). Testosterone replacement in older hypogonadal men: A 12-month randomized controlled trial. J. Clin. Endocrinol. Metab. 82(6), 1661-1667. 48. Morita, R., Yamamoto, I., Fukunaga, M., Dokoh, S., Konishi, J., Kousaka, T., Nakajima, K., Toizuka, K., Aso, T., and Motahashi, T. (1979). Changes in sex hormones and calcium regulating hormones with reference to bone mass associated with aging. Endocrinol. Jpn. Res. Soc. 1, 15-22. 49. Foresta, C., Busnardo, B., Ruzza, G., Zanatta, G., and Mioni, R. (1983). Lower calcitonin levels in young hypogonadal men with osteoporosis. Horm. Metab. Res. 15, 206-207. 50. Rudman, D., Kutner, M. H., Rogers, C. M., Lubin, M. F., Fleming, G. A., and Bain, R. P. (1981). Impaired growth hormone secretion in the adult population: Relation to age and adiposity. J. Clin. Invest. 67(5), 1361-1369. 51. Florini, J. R., Prinz, P. N., Vitiello, M. V., and Hintz, R. L. (1985). Somatomedin-C levels in healthy young and old men: Relationship to peak and 24-hour integrated levels of growth hormone. J. Gerontol. 40(1 ), 2-7. 52. Vermeulen, A. (1987). Nyctohemeral growth hormone profiles in

53.

54.

55.

56.

57.

58.

59.

60.

61.

62.

62a. 63. 64. 65. 66.

67. 68.

young and aged men: Correlation with somatomedin-C levels. J. Clin. Endocrinol. Metab. 64, 884-888. Ho, K. Y., Evans, W. S., Blizzard, R. M., Veldhuis, J. D., Merriam, G. R., Samojlik, E., Furlanetto, R., Rogol, A. D., Kaiser, D. L., and Thorner, M. O. (1987). Effects of sex and age on the 24-hour profile of growth hormone secretion in man: Importance of endogenous estradiol concentrations. J. Clin. Endocrinol. Metab. 64(1), 51-58. Shibasaki, T., Shizume, K., Nakahara, M., Masuda, A., Jibiki, K., Demura, H., Wakabayashi, I., and Ling, N. (1984). Age-related changes in plasma growth hormone response to growth hormone-releasing factor in man. J. Clin. Endocrinol. Metab. 58(1), 212-214. Pavlov, E. P., Harman, S. M., Merriam, G. R., Gelato, M. C., and Blackman, M. R. (1956). Responses of growth hormone (GH) and somatomedin-C to GH-releasing hormone in healthy aging men. J. Clin. Endocrinol. Metab. 62(3), 595-600. Muggeo, M., Fedele, D., Tiengo, A., Molinari, M., and Crepaldi, G. (1975). Human growth hormone and cortisol response to insulin stimulation in aging. J. Gerontol. 30(5), 546-551. Abribat, T., Deslauriers, N., Brazcau, P., and Gaudreau, P. (1991). Alterations of pituitary growth hormone-releasing factor binding sites in aging rats. Endocrinology (Baltimore) 128, 633-635. Sonntag, W. E., Forman, L. J., Miki, N., Steger, R. W., Ramos, T., Arimura, A., and Meites, J. (1981). Effects of CNS active drugs and somatostatin antiserum on growth hormone release in young and old male rats. Neuroendocrinology 33(2), 73-78. Blum, W. F., and Ranke, M. B. (1991). Plasma IGFBP-3 levels as clinical indicators. In "Modern Concepts of Insulin-Like Growth Factors" (M. Spencer, ed.), pp. 383-383. Elsevier, Amsterdam. Rutanen, E.-M., Karkkainen, T., Stenman, U.-H., and Yki-Jarvinen, H., (1993). Aging is associated with decreased suppression of insulinlike growth factor binding protein-1 by insulin. J. Clin. Endocrinol. Metab. 77, 1152-1155. Vidalon, C., Khurana, R. C., Chae, S., Gegick, C. G., Stephan, T., Nolan, S., and Danowski, T. S. (1973). Age-related changes in growth hormone in non-diabetic women. J. Amer. Geriat. Soc. 21(6), 235-255. Prinz, P. N., Weitzman, E. D., Cunningham, G. R., and Karacan, I. (1983). Plasma growth hormone during sleep in young and aged men. J. Gerontol. 38(5), 519-524. Mooradian, A., and Wong, N. C. (1994). Age-related changes in thyroid hormone action. Eur. J. Endocrinol. 131, 451. Mooradian, A. D. (1995). Normal age-related changes in thyroid hormone economy. Clin. Geriatr. Med. 11, 159-169. Robuschi, G., Safran, M., Braverman, L. E., Gnudi, A., and Roti, E. (1987). Hypothyroidism in the elderly. Endocr. Rev. 8, 142-153. Mooradian, A. D., Wong, N. C. W. (1995). Age-related changes in thyroid hormone action. Eur. J. Endocrinol. 131, 451-461. Sawin, C. T., Castelli, W. P., Hershamn, J. P., McNamara, P., and Bacharach, P. (1985). The aging thyroid. Thyroid deficiency in the Framingham Study. Arch. Intern. Med. 145, 1386-1388. Mooradian, A. D., (1990). Blood-brain barrier transport of triiodothyronine is reduced in aged rats. Mech. Ageing Dev. 52, 141-147. Mooradian, A. D. (1990). The hepatic transcellular transport of 3,5,3'triiodothyronine is reduced in aged rats. Biochem. Biophys. Acta

1054, 1-7. 69. Mooradian, A., Fox-Robichaud, A., Meijer, M. E., and Wong, N. C. W. (1994). Relationship between transcription factors and S 14 gene expression in response to thyroid hormone with age. Proc. Soc. Exp. Biol. Med. 207, 97-101. 70. Scarpace, P. J., Mooradian, A. D., and Morley, J. E. (1988). Ageassociated decrease in beta-adrenergic receptors and adenylate cyclase activity in rat brown adipose tissue. J. Gerontol. 43, B65-B70. 71. Wolfsen, A. R. (1982). Aging and the adrenals. In "Endocrine As-

130

72.

73.

74.

75.

76.

77.

78.

79.

80.

81.

82.

83.

84.

85.

86.

87.

88.

89.

90.

KORENMAN ET AL. pects of Aging" (S. G. Korenman, ed.), pp. 55-74. Elsevier NorthHolland, New York. Boscaro, M., Paoletta, A., Scarpa, E. Barzon, L., Fusaro, P., Fallo, F., and Sonino, N. (1998). Age-related changes in glucocorticoid fast feedback inhibition of adrenocorticotropin in man. J. Clin. Endocrinol. Metab. 83, 1380-1383. Pavlov, E. P., Harman, S. M., Chrousos, G. P., Loriaux, D. L., and Blackman, M. R. (1986). Responses of plasma adrenocorticotropin, cortisol, and dehydroepiandrosterone to ovine corticotropin-releasing hormone in healthy aging men. J. Clin. Endocrinol. Metab. 62(4), 767-772. Sherman, B., Wysham, C., and Pfohl, B. (1985). Age-related changes in the circadian rhythm of plasma cortisol in man. J. Clin. Endocrinol. Metab. 61,439-443. Davis, K. L., Davis, B. M. Math6, A. A., Mohs, R. C., Rothpearl, A. B., Levy, M. I., Gorman, L. K., and Berger, P. (1984). Age and the dexamethasone suppression test in depression. Am. J. Psychiatry 141(7), 872-874. Gergerman, R. I., and Bierman, E. L. (1981). Aging and hormones. In "Textbook of Endocrinology" (R. M. Williams, ed.), pp. 1192-1212. Saunders, Philadelphia. Parker, L. N., and Odell, W. D. (1978). Decline of adrenal androgen production as measured by radioimmunoassay of urinary conjugated dehydroepiandrosterone. J. Clin. Endocrinol. Metab. 47, 600-602. Jakubowicz, D. J., Beer, N. A., Beer, R. M., and Nestler, J. E. (1995). Disparate effects of weight reduction by diet on serum dehydroepiandrosterone-sulfate levels in obese men and women. J. Clin Endocrinol. Metab. 80, 3373-3376. Coleman, D. L., Leiter, E. M., and Applezweig, N. (1984). Therapeutic effects of dehydroepiandrosterone metabolites in diabetic mutant mice (C57 G L / K s - d b / d b ) . Endocrinology (Baltimore) 115, 239-243. Barrett-Connor, E., Khaw, K. T., and Yen, S. S. C. (1986). A prospective study of dehydroepiandrosterone sulfate, mortality, and cardiovascular disease. N. Engl. J. Med. 315, 1519-1524. Beer, N. A., Jakubowicz, D. J., Matt, D. W., Beer, R. M., and Nestler, J. E. (1996). Dehydroepiandroterone reduces plasma plasminogen activator inhibitory type 1 and tissue plasminogen activator antigen in men. Am. J. Med. Sci. 311,205-210. Jesse, R. L., Loesser, K., Eich, D. M., Qian, Y. A., Hess, M. L., and Nestler, J. E. (1995). Dehydroepiandrosterone inhibits human platelet aggregation in vitro and in vivo. Ann. N. Y. Acad. Sci. 774, 281-290. Schwartz, R. S., Jaeger, L. E, and Veith, R. C. (1987). The importance of body composition to the increase in plasma norepinephrine appearance rate in elderly men. J. Gerontol. 421, 546-551. Sowers, J. R., Rubenstein, L. Z., and Stern, N. (1983). Plasma norepinephrine responses to posture and isometric exercise increase with age in the absence of obesity. J. Gerontol. 38, 315-317. Brodde, O. E., Anluf, M., Graben, N., and Bock, K. D. (1982). Agedependent decrease of alpha 2-adrenergic receptor number in human platelets. Eur. J. Pharmacol. 81(2), 345-347. Vestal, R. E., Wood, A. J., and Shand, D. G. (1979). Reduced betaadrenoreceptor sensitivity in the elderly. Clin. Pharmacol. Ther. 26, 181-186. Lipsitz, L. A., Nyquist, R. P., Jr., Wei, J. Y., and Rowe, J. W. (1983). Postprandial reduction in blood pressure in the elderly. N. Engl. J. Med. 309(2), 81-83. Riggs, B. L., Wahner, H. W., Dunn, W. L., Mazess, R. B., Offord, K. P., and Melton, L. J., 3rd (1981). Differential changes in bone mineral density of the appendicular and axial skeleton with aging: Relationship to spinal osteoporosis. J. Clin. Invest. 67(2), 328-335. Avioli, L. V. (1982). Aging, bone and osteoporosis. In "Endocrine Aspects of Aging" (S. G. Korenman, ed.), pp. 199-229. Elsevier North-Holland, New York. Dandona, P., Menon, R. K., Shenoy, R., Houlder, S. Thomas, M., and

91. 92.

93.

94.

95.

96.

97. 98.

99.

100. 101.

102.

103.

104.

105.

106.

107.

108.

109.

110.

Mallinson, W. J. (1986). Low 1,25-dihydroxyvitamin D, secondary hyperparathyroidism, and normal osteocalcin in elderly subjects. J. Clin. Endocrinol. Metab. 63(2), 459-462. Mooradian, A. D. (1996). Drug therapy of non-insulin-dependent diabetes mellitus in the elderly. Drugs 51, 931-941. Campbell, S., and Mooradian, A. D. (1993). Diabetes mellitus. In "Geriatric Pharmacology" (R. Bressler and M. K. Katz, eds.), pp. 409-423. McGraw-hill, New York. Phillips, P. A., Rolls, B. J., Ledingham, J. G., Forsling, M. L., Morton, J. J., Crowe, M. J., and Wollner, L. (1984). Reduced thirst after water deprivation in healthy elderly men. N. Engl. J. Med. 311(12), 753-759. Helderman, J. H. (1982). The impact of normal aging on the hypothalamic-neurohypophyseal-renal axis. In "Endocrine Aspects of Aging" (S. G. Korenman, ed.), pp. 9-32. Elsevier North-Holland, New York. Ohashi, M., Fumio, N., Nawata, H. Kato, K., Ibayashi, H., Kangawa, K., and Matsuo, H. (1987). High plasma concentrations of human atrial natriuretic polypeptide in aged men. J. Clin. Endocrinol. Metab. 64(1), 81-85. Mooradian, A. (1994). Biology of aging. In "Rehabilitation of the Aging and the Elderly Patient" (S. J. G. G. Felsenthal and F. U. Steinberg, eds.), pp. 3-10. Williams & Wilkins, Baltimore, MD. Knudson, R. J. (1989). Aging of the respiratory system. Curr. Pulmonol. 10, 1-24. Feldman, H., Goldstein, I., Hatzichristou, D. G., Krane, R. I., and McKinlay, J. B. (1994). Impotence and its medical and psychosocial correlates: Results of the Massachusetts Male Aging Study." J. Urol. 151,54-61. Kwan, M., Greenleaf, W. J., Mann, J., Crapo, L., and Davidson, J. M. (1983). The nature of androgen action on male sexuality: A combined laboratory-self-report study on hypogonadal men. J. Clin. Endocrinol. Metab. 57(3), 557-562. Bancroft, J., and Wu, F. C. W. (1983). Changes in erectile responsiveness during androgen therapy. Arch. Sex. Behav. 12, 59-66. Davidson, J., Camargo, C., and Smith, E. (1979). Effects of androgen on sexual behavior in hypogonadal men. J. Clin. Endocrinol. Metab. 48, 955-958. Bagatell, C., Heiman, J., Rivier, J., and Bremner, W. Effects of endogenous testosterone and estradiol on sexual behavior in normal young men. J. Clin. Endocrinol. Metab. 78, 711-716. Slag, M. F., Morley, J. E., Elson, M. K., Trence, D. L., Nelson, C. J., Nelson, A. E., Kinlaw, W. B., Beyer, H. S. Nuttall, F. Q., and Shafer, R. B. (1983). Impotence in medical clinic outpatients. JAMA, J. Am. Med. Assoc. 249(13), 1736-1740. Davis, S. S., Viosca, S. P., Guralnik, M., Windsor, C., Buttiglieri, M. W., Baker, J. D., Mehta, A. J., and Korenman, S. G. (1985). Evaluation of impotence in older men. West. J. Med. 142(4), 499-505. Lugg, J., Rajfer, J., and Gonzalez-Cadavid, N. F. (1995). Dihydrotestosterone is the active androgen in the maintenance of nitric oxidemediated penile erection in the rat. Endocrinology (Baltimore) 136, 1495-1501. Heaton, J., and Varrin, S. (1994). Effects of castration and exogenous testosterone supplementation in an animal model of penile erection. J. Urol. 151 797-800. Kurz, E., Sengelaub, D., and Arnold, A. (1986). Androgens regulate the dendritic length of mammalian motoneurons in adulthood. Science 232, 395-396. Leipheimer, R., and Sachs, B. Relative androgen-sensitivity of the vascular and striated-muscle systems regulating penile erection in rats. Physiol. Behav. 54, 1085-1090. Shiavi, R., White, D., Mandeli, J., and Schreiner-Engel, P. (1992). Hormones and nocturnal penile tumescence in healthy aging men. Arch. Sex. Behav. 22, 207-215. Schiavi, R.C., Schreiner-Engel, P., White, D., and Mandeli, J. The re-

CHAPTER 7 Changes in Aging Men

111. 112. 113.

114.

115.

116.

117.

118.

119.

120.

121.

122. 123. 124.

125.

126. 127. 128. 129.

130. 131.

132.

133.

lationship between pituitary-gonadal function and sexual behavior in healthy aging men. Psychosom. Med. 53, 363-374. de Lignieres, B. (1993). Transdermal dihyrotestosterone treatment of "andropause." Ann. Med. 25, 235-241. Schiavi, R. C., Mandeli, J., and Schreiner-Engel, P. (1994). Sexual satisfaction in healthy aging men. J. Sex Marital Ther. 20, 3-13. Davidson, J., Chen, J. J., Crapo, L., Gray, G. D., Greenleaf, W. J., and Catania, J. A. (1983). Hormonal changes and sexual function in aging men. J. Clin. Endocrinol. Metab. 57, 71-77. O'Carroll, R., and Bancroft, J. (1984). Testosterone therapy for low sexual interest and erectile dysfunction in men: A controlled study. Br. J. Psychiatry 145, 146-151. Metz, M. E., P. J., Nesvacil, L. J., Abuzzahab, E, Sr., and Koznar, J. (1997). Premature ejaculation: A psychophysiological review. J. Sex Marital Ther. 23, 3-23. Waldinger, M. D., Hengeveld, M. W., and Zwinderman, A. H. Ejaculation-retarding properties of paroxetine in patients with primary premature ejaculation: A double-blind, randomized, dose-response study. Br. J. Urol. 79(4), 592-595. Ludovico, G. M., C. A., Pagliarulo, G., Cirillo-Marucco, E., Marano, A., and Pagliarulo, A. (1996). Paroxetine in the treatment of premature ejaculation. Br. J. Urol. 77, 881-882. Haensel, S. M., Klem, T. M., Hop, W. C., and Slob, A. K. (1998). Fluoxetine and premature ejaculation: A double-blind, crossover, placebo-controlled study. J. Clin. Psychopharmacol. 18(1), 72-77. Haensel, S. M., Rowland, D. L., and Kallan, K. T. (1996). Clomipramine and sexual function in men with premature ejaculation and controis. J. Urol. 156(4), 1310-1315. Balbay, M. D., Yildiz, M., Salvarci, A., Ozsan, O., and Ozbek, E. (1998). Treatment of premature ejaculation with sertralin. Int. Urol. Nephrol. 30(1), 81-83. Lee, H. S., Song, D. H., Kim, C. H., and Choi, H. K. (1996). An open clinical trial of fluoxetine in the treatment of premature ejaculation. J. Clin. Psychopharmacol. 16(5), 379-382. Buffum, J. (1982). Pharmacosexology: The effects of drugs on sexual function--A review. J Psychoactive Drugs 14, 5-44. Wein, A., and Van Arsdale, K. (1988). Drug-induced male sexual dysfunction. Urol. Clin. North Am. 15, 23-31. Korenman, S. (1998). Sexual function and dysfunction. In "Williams Textbook of Endocrinology" (J. Wilson and D. W. Foster, eds.), pp. 927-937. Saunders, Philadelphia. Korenman, S. (1995). Advances in the understanding and management of erectile dysfunction. J. Clin. Endocrinol. Metab. 80, 19851988. Korenman, S. (1998). New insights into erectile dysfunction: A practical approach. Am. J. Med. (in press). Krane, R., Goldstein, I., and Saenz de Tejada, I. (1989). Impotence. N. Engl. J. Med. 321, 1648-1659. National Institutes of Health Consensus Statement (1992). Impotence 10(1). JCnler, M., Moon, T., Brannan, W., Stone, N. N., Heisey, D., and Bruskewitz, R. C. (1995). The effect of age, ethnicity and geographical location on impotence and quality of life. Br. J. Urol. 75(5), 651-655. Anderson, K., and Wagner G. (1995). Physiology of penile erection. Physiol. Rev. 75, 191. Azadzoi, K. M., Vlachiotis, J., Pontari, M., and Siroky, M. B. (1995). Hemodynamics of penile erection: III. Measurement of deep intracavernosal and subtunical blood flow and oxygen tension. J. Urol. 153(2), 521-526. Lue, T., Zeineh, S. L., Schmidt, R. A., and Tanagho, E. A. (1984). Neuroanatomy of penile erection: Its relevance to iatrogenic impotence. J. Urol. 131, 273-280. De Groat, W., and Booth, A. Neural control of penile erection. In "The Autonomic Nervous System. Nervous Control of the Urogenital System" (C. Maggi, ed.), pp. 465-513. Harwood, London.

131 134. Steers, W. (1993). Neural control of penile erection. Semin. Urol. 8, 66 -79. 135. Raifer, J., Aronson, W. J., Bush, P. A., Dorey, E J., and Ignarro, L. J. (1992). Nitric oxide as a mediator of relaxation of the corpus cavernosum in response to nonadrenergic, noncholinergic neurotransmission. N. Engl. J. Med. 326(2), 90-94. 136. Saenz de Tejada, I., Goldstein, I., Azadzoi, K., Krane, R. J., and Cohen, R. A. (1989). Impaired neurogenic and endothelium-mediated relaxation of penile smooth muscle from diabetic men with impotence. N. Engl. J. Med. 320(16), 1025-1030. 137. Ignarro, L., Bush, P. A., Buga, G. M., Woods, K. S., Fukuto, J. M., and Raifer, J. (1990). Nitric oxide and cyclic-GMP formation upon electric field stimulation cause relaxation of corpus cavernosum smooth muscle. Biochem. Biophys. Res. Commun. 170, 843-850. 138. Saenz de Tejada, I., Moroukian, P., Tessier, I., Kim, J. L., Goldstein, I., and Frohrib, D. (1991). Trabecular smooth muscle modulates the capacitor function of the penis. Studies on a rabbit model. Am. J. Physiol. 260, H1590-H1595. 139. Fournier, G., Juenemann, K.-P., Lue, T. E, and Tanagho, E. A. (1987). Mechanisms of venous occlusion during canine penile erection: An anatomic demonstration. J. Urol. 137, 163-167. 140. Schmidt, M., and Schmidt, H. S. (1993). The ischiocavernosus and bulbospongiousus muscles in mammalian penile rigidity. Sleep 16, 171-183. 141. Mattson, D., Petrie, M., Srivastava, D. K., and McDermott, M. (1995). Multiple sclerosis. Sexual dysfunction and its response to medications. Arch. Neurol. (Chicago) 52(9), 862-868. 142. Jensen, S. (1984). Sexual function and dysfunction in younger married alcoholics. Acta Psychiatr. Scand. 69, 543-549. 143. Mannuino, D., Keuvens, R. M., and Flanders, W. D. (1994). Cigarette smoking: An independent risk factor for impotence? Am. J. Epidemiol. 140, 1003 - 1008. 144. Mersdorf, A., Goldsmith, P. C., Diederichs, W., Padula, C. A., Lue, T. E, Fishman, I. J., and Tanagho, E. A. (1991). Ultrastructural changes in impotent penile tissue: A comparison of 65 patients. J. Urol. 145(4), 749-758. 145. Pickard, R., Powell, P., and Zar, M. (1995). Nitric oxide and cyclic GMP formation following relaxant nerve stimulation in isolated human corpus cavernosum. Br. J. Urol. 75, 516-522. 146. Christ, G., Schwartz, C. B., Stone, B. A., Parker, M., Janis, M., Gondre, M., Valcic, M., and Melman, A. (1992). Kinetic characteristics of al-adrenergic contractions in human corpus cavernosum smooth muscle. Am. J. Physiol. 263, H15-H19. 147. Persson, C., Diederichs, W., Lue, T. E, Yen, T. S., Fishman, I. J., McLin, P. H., and Tanagho, E. A. (1989). Correlation of altered penile ultrastructure with clinical arterial evaluation. J. Urol. 142(6), 14621468. 148. Akkus, E., Carrier, S., Baba, K., Hus, G.-L., Padma-Nathan, H., Nunes, L., and Lue, T. E (1997). Structural alterations in the tunica albuginea of the penis: Impact of Peyronie's disease, ageing and impotence. Br. J. Urol. 79, 47-53. 149. Sattar, A., Wespes, E., and Schulman, C. (1994). Computerized measurement of penile elastic fibers in potent and impotent men. Eur. Urol. 25, 142-44. 150. Bookstein, J., and Valji, K. The arteriolar component in impotence: A possible paradigm shift. Am. J. Radiol. 157, 932-934. 151. Lue, T., Hricak, H., Schmidt, R. A., and Tanagho, E. A. (1986). Functional evaluation of penile veins by cavernosography in papaverineinduced erection. J. Urol. 135, 479-482. 152. Kaufman, J., Borges, F. D., Fitch, W. P., 3rd, Geller, R. A., Grubner, M. B., Hubbard, J. G., McKay, D. L., Jr., Tuttle, J. P., and Witten, E R. (1993). Evaluation of erectile dysfunction by dynamic infusion cavernosometry and cavernosography (DICC). Multi-institutional Study. Urology 41(5), 445-451. 153. Ackerman, M., D'Attilio, J. P., Antoni, M. H., Weinstein, D., Rhamy,

132

154.

155.

156.

157. 158.

159.

160. 161.

162.

163.

164.

165. 166.

167.

168.

169.

170.

171.

KORENMAN ET AL. R. K., and Politano, V. A. The predictive significance of patientreported sexual functioning in rigiscan sleep evaluations. J. Urol. 146, 1559-1563. Rosen, R. C., Riley, A., Wagner, G., Osterloh, I. H., Kirkpatrick, J., and Mishra, A. (1997). The international index of erectile function (IIEF): A multidimensional scale for assessment of erectile dysfunction. Urology 49(6), 822-830. Buvat, J., and Lemaire, A. (1997). Endocrine screening in 1,022 men with erectile dysfunction: Clinical significance and cost-effective strategy. J. Urol. 158(5), 1764-1767. Goldstein, I., Lue, T. F., Padma-Nathan, H., Rosen R. C., Steers, W. D., and Wicker, P. A. (1998). Oral sildenafil in the treatment of erectile dysfunction, Sildenafil Study Group. N. Engl. J. Med. 338(20), 1397-1404. Linet, O., and Neff, L. (1994). Intracavernous prostaglandin E1 in erectile dysfunction. Clin. Invest. 72, 139-149. Padma-Nathan, H., Hellstrom, W. J., Kaiser, F. E., Labasky R. E., Lue, T. E, Nolten, W. E., Norwood, P. C., Peterson, C. A. Shabsigh, R., and Tam, P. Y. (1997). Treatment of men with erectile dysfunction with transurethral alprostadil. Medicated Urethral System for Erection (MUSE) Study Group. N. Engl. J. Med. 336, 1-7. Korenman, S., Viosca, S. P., Kaiser, F. E., Mooradian, A. D., and Morley, J. E. (1990). Use of a vacuum tumescence device in the management of impotence. J. Am. Geriatr. Soc. 38, 217-220. Petrou, S., and Barrett, D. (1990). The use of penile prostheses in erectile dysfunction. Semin. Urol. 8, 138-152. Garban, H., Marquez, D., Magee, T., Moody, J., Rajavashisth, T., Ropdriguez, A., Hung, A., Vernet, D., Rajfer, J., and Gonzalez-Cadavid, N. E. (1997). Cloning of rat and human inducible penile nitric oxide synthase. Application for gene therapy of erectile dysfunction. Biol. Reprod. 56, 954-963. Kaufert, P., Gilblert, P., and Tate, R. (1992). The Manitoba Project: A reexamination of the link between menopause and depression. Maturitas 14, 143-155. Avis, N., and McKinlay, S. M. (1995). The Massachusetts Women's Health Study: An epidemiologic investigation of the menopause. J. Am. Med. Women's Assoc. 50, 45-49. Matthews, K., Wing, R. A., and Kuller, L. H. (1990). Influences of natural menopause on psychological characteristics and symptoms of middle-aged healthy women. J. Consult. Clin. Psychol. 58, 345-351. Schow, D., Redmon, B., and Pryor, J. L. (1997). Male menopause: How to define it, how to treat it. Postgrad. Med. 101, 62-79. McKinlay, J. (1989). Is there an epidemiologic basis for a male climacteric syndrome? The Massachusetts Male Aging Study. In "Menopause: Evaluation, Treatment, and Health Concerns" (Charles B. Hammond and Florence P. Haseltire, eds.), pp. 163-192. Alan R. Liss, New York. Gatz, M., and Hurwicz, M. L. (1990). Are old people more depressed? Cross-sectional data on Center for Epidemiological Studies Depression Scale Factors. Psychol. Aging 5, 284-290. Robins, L., and Regier, D. A. (1991). "Psychiatric Disorders in America: The Epidemiologic Catchment Area Study." Free Press, New York. Penninx, B., Guralnik, J. M., Ferrucci, L., Simonsick, E. M., Deeg, D. J. H., and Wallace, R. B. (1998). Depressive symptoms and physical decline in community-dwelling older persons. JAMA, J. Am. Med. Assoc. 279, 1720-1726. Beekman, A., Deeg, D. J., van Tilburg, T., Smit, J. H., Hooijer, C., and van Tilburg, W. (1995). Major and minor depression in later life: A study of prevalence and risk factors. J. Affective Disord. 36, 65-75. Unutzer, J., Patrick, D. L., Simon, G., Grembowski, D., Walker, E., Rutter, C., and Katon, W. (1997). Depressive symptoms and the cost of health services in HMO patients aged 65 years and older. JAMA, J. Am. Med. Assoc. 277, 1618-1623.

172. Hirschfeld, R., Keller, M. B., Panico, S., Arons, B. S., Barlow, D., Davidoff, F., Endicott, J., Froom, J., Goldstein, M., Gorman, J. M., Marek, R. G., Maurer, T. A., Meyer, R., Phillips, K., Ross, J., Schwenk, T. L., Sharfstein, S. S., Thase, M. E., and Wyatt, R. J. (1997). The National Depressive and Manic-Depressive Association consensus statement on the undertreatment of depression. JAMA, J. Am. Med. Assoc. 277, 333-340. 173. Knaupper, B., and Wittchen, H. U. (1994). Diagnosing major depression in the elderly: Evidence for response bias in standardized diagnostic interviews? J. Psychiat. Res. 28, 147-164. 174. Zisook, S. (1996). Depression in late life. Diagnosis, course, and consequences. Postrgrad. Med. 100, 143-148. 175. Tannock, C., and Katona, C. (1995). Minor depression in the aged. Concepts, prevalence and optimal management. Drugs Aging 6, 278-292. 176. Blazer, D., and Williams, C. D. (1980). Epidemiology of dysphoria and depression in an elderly population. Am. J. Psychiatry. 137, 439-444. 177. Murrell, S., Himmelfarb, S., and Wright, K. (1983). Prevalence of depression and its correlates in older adults. Am. J. Epidemiol. 117, 173-185. 178. Turner, R. J., and Noh, S. (1988). Physical disability and depression: A longitudinal analysis. J. Health Soc. Behav. 29(1), 23-37. 179. Featherstone, M., and Hepworth, M. (1985). The male menopause: Lifestyle and sexuality. Maturitas 7, 235-246. 180. Atchley, R. (1996). Retirement, In "Encyclopedia of Gerontology" (J. Birren, ed.), pp. 437-449. Academic Press, San Diego, CA. 181. Rybash, J., Roodin, P. A., and Hoyer, W. J., eds. (1995). "Adult Development and Aging," 3rd ed., pp. 276-290. Brown & Benchmark, Madison, WI. 182. Finch, E., Ramsay, R., and Katona, C. L. (1992). Depression and physical illness in the elderly. Clin. Geriatr. Med. 8, 275-287. 183. Baker, F. (1996). An overview of depression in the elderly: A U.S. perspective. J. Natl. Med. Assoc. 88, 178-184. 184. Yesavage, J. A. (1983). Development and validation of a Geriatric Depression Screening Scale: A preliminary report. J. Psychiat. Res. 17, 37-49. 185. Hirschfeld, R. (1994). Guidelines for the long-term treatment of depression. J. Clin. Psychiatry 55(12, Suppl.), 61-69. 186. Cornoni-Huntley, J., Ostfeld, A. M., Taylor, J. O., Wallace, R. B., Blazer, D., Berkman, L. F., Evans, D. A., Kohout, F. J., Lemke, J. H., Scherr, P. A. et al. (1993). Established populations for epidemiologic studies in the elderly: Study design and methodology. Aging 5, 27-37. 187. Sturm, R., and Wells, K. (1995). How can care for depression be more cost-effective? JAMA, J. Am. Med. Assoc. 273, 51-58. 188. Yesavage, J. A., Davidson, J., Widlow, L., and Berger, P. A. (1985). Plasma testosterone levels, depression, sexuality, and age. Biol. Psychiatry, 20, 222-225. 189. Davies, R., Harris, B., Thomas, D. R., Cook, N., Read, G., and RiadFahmy, D. (1992). Salivary testosterone levels and major depressive illness in men. Br. J. Psychiatry 161, 629-632. 190. Levitt, A. J., and Joffe, R. T. (1988). Total and free testosterone in depressed men. Acta Psychiatr. Scand. 77(3), 346-348. 191. Nilsson, P., Moiler, L., and Solstad, K. (1995). Adverse effects of psychosocial stress on gonadal function and insulin levels in middle-aged males. J. Intern. Med. 237, 479-486. 192. Schiavi, R., and Rehman, J. (1995). Sexuality and aging. Urol. Clin. North Amer. 4, 711-726. 193. Araujo, A., Durante, R. Feldman, H. A., Goldstein, I., and McKinlay, J. B. (1998). The relationship between depressive symptoms and male erectile dysfunction: Cross-sectional results from the Massachusetts Male Aging Study. Psychosom. Med. 194. Thase, M., Reynolds, C. F., Jennings, J. R., Frank, E., Howell, J. R., Houck, P. R., Berman, S., and Kupfer, D. I. (1988). Nocturnal penile

CHAPTER 7 Changes in Aging M e n tumescence is diminished in depressed men. Biol. Psychiatry 24(1), 33-46. 195. Thase, M., Reynolds, C. F., Glanz, L. M., Jennings, J. R., Sewitch, D. E., Kupfer, D. J., and Frank, E. (1987). Nocturnal penile tumescence in depressed men. Am. J. Psychiatry 144(1), 89-92. 196. Spector, I., and Carey, M. P. (1990). Incidence and prevalence of the sexual dysfunctions: A critical review of the empirical literature. Arch. Sex. Behav. 19, 389-408. 197. Cooper, A., Cernovsky, Z. A., and Colussi, K. (1993). Some clinical

133 and psychometric characteristics of primary and secondary premature ejaculators. J. Sex and Marital Ther. 19, 276-288. 198. McCabe, M. (1997). Intimacy and quality of life among sexually dysfunctional men and women. J. Sex Marital Ther. 23, 276-290. 199. Kaplan, H. (1994). Psychogenic impotence. Curr. Ther. Endocrinol. Metab. 5, 323-328. 200. Wylie, K. (1997). Male erectile disorder: Characteristics and treatment choice of a longitudinal cohort study of men. Int. J. Impotence Res. 9, 217-224.

2 H A P T E R {~

Premature Ovarian Failure ROBERT W. REBAR

I. II. III. IV.

Department of Obstetrics and Gynecology, University of Cincinnati Medical Center, Cincinnati, Ohio 45267; and the American Society for Reproductive Medicine, Birmingham, Alabama 35216

V. Evaluation of Patients with Hypergonadotropic Amenorrhea VI. Therapy References

Early Reports of Premature Ovarian Failure Clinical Features of Premature Ovarian Failure Prevalence of Premature Ovarian Failure Etiology of Premature Ovarian Failure

genes are important in controlling the number of oocytes ovulated and hence presumably the timing of the cessation of reproductive function [3]. Although these data are difficult to extrapolate to humans, given what is known about the control of ovarian function by the X chromosome [4], it is not difficult to believe that inherited tendencies are important. Any role for ovarian inhibin and its feedback action on pituitary FSH secretion also remains to be explored. Also potentially important in the regulation of the onset of menopause is the hypothalamic-pituitary axis. Although oocyte depletion may provide the major reason for the occurrence of menopause in humans, numerous animal studies document changes in neurotransmitter and in central nervous system (CNS) feedback responses to estrogen with aging. Of particular note is the observation that aging ovaries transplanted to young rodents cycle normally whereas young ovaries transplanted to aged animals do not function well [5]. Once more, however, extrapolation of such data to humans is most difficult. The concept that young women under the age of 40 with "hypergonadotropic" amenorrhea by definition should have depletion of their oocytes and premature ovarian failure was supported by the findings of Goldenberg and colleagues [6]. They reported in 1973 that women who had basal FSH concentrations greater than 40 mIU/ml [Second International Reference Preparationmhuman menopausal gonadotropin, (2nd IRP-hMG] without exception had no viable oocytes on ovarian biopsy.

I. EARLY REPORTS OF PREMATURE OVARIAN FAILURE Menopause, defined strictly as the last episode of menstrual bleeding, typically occurs around age 51 and is generally considered premature if it occurs before the age of 40 years. In fact, de Moraes and Jones [1] first defined premature menopause, or premature ovarian failure, as consisting of the triad of amenorrhea, hypergonadotropinism, and hypoestrogenism in women under the age of 40 years. Why the cessation of reproductive life should occur prematurely has been of great interest to clinicians and remains enigmatic in the majority of cases. How little is known about premature menopause is less surprising in view of how little is known about normal menopause. The events that signal menopause are unclear. Depletion of oocytes is obviously an important factor, and it has been documented that follicle depletion accelerates just prior to menopause [2]. Although a few follicles may be present at menopause, they do not respond to folliclestimulating hormone (FSH) and luteinizing hormone (LH). In an unsuccessful effort tastimulate follicular development and estradiol secretion, the hypothalamus signals the pituitary gland to secrete still more FSH and LH. Thus, an increase in serum FSH concentrations is an early sign heralding the cessation of ovarian function. Preliminary studies in strains of mice indicate that specific MENOPAUSE:


135


136

ROBERT W. REBAR

The belief that the ovarian "failure" observed in such young women was permanent was first questioned by a number of isolated case reports documenting the initiation or resumption of cyclic menses and/or pregnancy in affected women. Several large series have now confirmed these case reports [7]. In one of those reports, we documented pubertal progression in two young girls with elevated circulating FSH and multiple endocrine deficiencies (i.e., hypoparathyroidism and hypoadrenalism) and suggested that waxing and waning autoimmune dysfunction might account for the transient nature of the ovarian failure [8]. O'Herlihy and coworkers [9] reported that up to one-fourth of younger women with FSH values in the menopausal range will resume ovulation spontaneously and a few will even conceive. In 1982 we reported that 9 of 18 young women presumed to have ovarian failure had circulating estradiol typical of women with functioning ovarian follicles and that 4 of the 9 women who had ovarian biopsies had viable oocytes [10]. In addition, circulating concentrations of serum progesterone typical of ovulation were noted in 5 women, and a spontaneous pregnancy occurred in one. Aiman and Smentek [11] reported that 18% of 157 women reported in the literature who had ovarian biopsies had specimens containing apparently viable oocytes. They also noted that 14 of the women had conceived after the ovarian failure had been diagnosed. A number of recent series have confirmed ovarian follicular activity in many women with ovarian failure. Hague et al. [12] reported evidence of ovarian follicular activity in 12 (17.1%) of 93 women with amenorrhea and elevated FSH concentrations. By pelvic ultrasound Conway and colleagues [ 13] identified follicular activity in 65 of 109 women (60%) with "idiopathic" premature ovarian failure. Bone mineral density was lower in women in whom ovaries were not identified on ultrasound (n -- 26) than in those in whom TABLE I

follicles > 4 m m were identified (n = 57). Similarly, Nelson and colleagues [14] documented ovarian follicular activity by serum estradiol levels greater than 50 pg/ml in nearly half of 65 women with karyotypically normal spontaneous premature ovarian failure and imaged an antral follicle in over 40% [ 15] of the women. These observations led us to suggest that this disorder involved more than just the premature cessation of ovarian function and might more appropriately be termed "hypergonadotropic amenorrhea" [ 13 ] - - at least until such time as it was apparent that the premature loss of ovarian function was permanent.

II. CLINICAL

FEATURES

OF PREMATURE

OVARIAN

FAILURE

To define the clinical spectrum of women with hypergonadotropic amenorrhea, Rebar and Connolly [ 16] compiled data from 115 affected women seen sequentially between 1978 and 1988. Initial inclusion criteria were (1) amenorrhea of 3 or more months' duration, (2) age under 40 years at the onset of the amenorrhea, and (3) circulating FSH of more than 40 mlU/ml on at least two occasions. A number of interesting differences and similarities existed between those with primary and those with secondary amenorrhea and are summarized in Table I. In over 75% of the patients, symptoms of estrogen deficiency, most commonly hot flushes and/or dyspareunia, were evident, but these symptoms were far more common in those with secondary amenorrhea. Chromosomal abnormalities and failure to develop mature secondary sex characteristics were far more common in those with primary amenorrhea. Chromosomal abnormalities were present in over half

Features of Women with Primary and Secondary Amenorrheaa

Feature

Primary amenorrhea

Number of patients Symptoms of estrogen deficiency Incomplete sexual development Karyotypic abnormalities Immune abnormalities Spinal bone density 3mths (n=52)

F I G U R E 4 Changes in E 2 (top), INH (middle), and FSH (bottom), as a percentage of Group 1 women from the Melbourne Midlife Project who had experienced no change in their cycles. In each pair, the left column is unadjusted and the right is adjusted for age and body mass index.

levels relatively intact. When thie data were analyzed without reference to menstrual cycle status and purely as a function of age, a marked decline in INH was observed, inverse to rising FSH levels, whereas E 2 w a s relatively constant until the age of approximately 51 or 52 years, when it too declined steeply. A further cross-sectional analysis of data was undertaken on 110 subjects aged 4 8 - 5 9 years in the third year of the longitudinal-phase Melbourne Women's Midlife Health Project [ 11 ] (Fig.5). Subjects were divided into those calledpremenopausal, with no reported change in menstrual cycle pattern; early perimenopausal, with a reported change in cycle frequency in the preceding year but a bleed in the preceding 3 months; late perimenopausal, with no menses in the preceding 3-11 months; and postmenopausal, with no menses for more than 12 months. The hormone concentrations in the premenopausal subjects were used as reference points for the other groups. Early perimenopausal subjects had significantly lower levels of INH-B (13.5 ng/liter compared with 48 ng/liter) in the presence of a small, statistically nonsignificant rise in FSH (21.4 compared with 13.5 E2

1

5

2

H

E

N

-

!

60

7"

_F-L_F_,. . . . . . . . . . . . . . . . .~ 240q

a ....

b

b

-

0

i

~,, 10

I

a

I

III

a

10 nmol/liter) decreased from about 60% to less than 10% during the 6 years preceding the FMP. Ovulatory P concentrations were found in 62.2% of women 72 to 61 months premenopausal, and in 4.8% who were 6 to 0 months premenopausal, whereas all serum P measurements were less than 2 nmol/liter postmenopausally. There is some controversy regarding the maintenance of P secretion during the luteal phase in older regularly cycling women. Lee et al. [ 12] showed that P secretion was well preserved in a group of regularly cycling women aged 46 to 50 years, whereas Santoro et al. [14] showed decreased urinary pregnanediol excretion in a group of regularly cycling women aged 43 to 52 years, compared with women aged 19 to 38 years.

D. A n d r o g e n s Variable findings have been reported in regard to the changes in circulating androgens in relation to the FMP. Rannevik et al. [9] reported a small but significant decline in testosterone (T), androstenedione (A), and sex hormone binding globulin (SHBG) during the 2 years around the menopause. Thus T fell from 1.7 nmol 1 to 6 months before the FMP to 1.4 nmol 13 to 24 months afterward and 1.2 nmol 85 to 96 months afterward. SHBG fell from 4.0 mg/liter 1 to 6 months before the FMP to 3.5 mg/liter 85 to 96 months afterward but the ratio T/SHBG was unchanged over that period. The data for A were not specifically listed. Longcope et al. [7] did not see any change in T and A over 80 months from the FMP but noted that the mean concentrations of T in all their subjects, including those still having cyclic menses, were significantly less than those of a group of normal young women sampled on days 5 to 7 of the cycle, and suggested that there is a decrease in the ovarian secretion of T prior to the menopause. It is noteworthy that a recent report

tO L

9

9

~0 5 0 E 25t-"

9

E

|

9

L_

~1"

/

0

20

25

I

I

I

I

I

30

35

40

45

50

Age (years)

FIGURE 6 Plasma total testosterone (24-hr mean) plotted against age, in normal females. The regression equation was testosterone (nanomoles/ liter) = 37.8 • age (years) - 1.12 (r = -0.54; P < 0.003). From [33], Zumoff, B., Strain, G. W., Miller, L. K., and Rosner, W. (1995). Twentyfour hour mean plasma testosterone concentration declines with age in normal premenopausal women. J. Clin. Endocrinol. Metab. 80,1429-1430. 9 The Endocrine Society.

[33] found that there was a steep decline in total serum T concentration with age, such that the levels in a woman aged 40 were approximately 50% of those in a woman aged 21 (0.61 nmol compared with 1.3 nmol) (Fig. 6). Percentage of free T did not vary significantly with age but free T concentration clearly showed a steep decline. The ratio of dehydroepiandrosterone (DHEA) to T and dehydroepiandrosterone sulfate (DHEAS) to T were age invariant because of the declines of DHEA and DHEAS with age. Other studies have suggested that total T levels decrease by approximately 20% and A decreases by approximately 50% with natural menopause [34]. Vermeulen [35] showed that postmenopausal women aged 51 to 65 years had lower mean levels of T (1.03 nmol), A (3.45 nmol), and dihydrotestosterone (DHT) (0.33 nmol) in comparison with women aged 18 to 25 years, i.e., with T (1.53 nmol), A (5.80 nmol), and DHT (1.04 nmol). The effects of ovariectomy on androgen profiles were reported by Judd et al. [36] and Hughes et al. [37]. Before the menopause, oophorectomy results in a decrease of circulating A and T by about 50%, the decrease in the latter being due in large part to the decrease in A. Postmenopausally, removal of the ovaries results in a 50% decline in T and a much lesser decline in A. The postmenopausal ovary secretes more T but less A than its premenopausal counterpart. [34]. In light of the recent report of Zumoff et al. [33], and the difficulty in demonstrating a significant decline in T around the FMP, it may be that the apparent decline in T at the menopause is related as much to aging as to decreased ovarian function in those women with intact ovaries. In the Melbourne Women's Midlife Health Project, there was no significant change seen in total T or in the T/SHBG ratio as a function of changing menopausal status [ 10].

154

HENRY G. BURGER

V. C O N C L U S I O N S 7.

The perimenopause is a time of markedly fluctuating hormone levels. Attempts to define menopausal status purely on the basis of single measurements of FSH or E 2 are unlikely to yield useful information. Though E 2 concentrations appear to be preserved in regularly cycling women at least until the age of 50, INH-B declines and FSH rises. The establishment of menstrual irregularity is marked by a decrease in the follicular-phase concentrations of INH-B, an increase in FSH, but relative preservation of E 2 and INH-A until the time of the FMP. The frequency of anovulatory cycles increases markedly as the FMP approaches. It is difficult to demonstrate substantial changes in androgen concentrations in the immediate perimenopausal period, though levels postmenopausally appear to be lower than those of young regularly cycling women, perhaps as a function of increasing age rather than menopausal status. Hormonal measurements are of little diagnostic value during the perimenopause other than for the purposes of physiological study. The issue of the appropriate reference points for the study of the perimenopause remains unclear.

8.

9.

10.

11.

12.

13.

14.

Acknowledgments 15. The collaboration of my colleagues in the Melbourne Women's Midlife Health Project (Lorraine Dennerstein, Emma Dudley, John Hopper, Adele Green, John Wark, Peter Ebeling, and Janet Guthrie) is acknowledged. David Robertson and his staff at Prince Henry's Institute of Medical Research provided the INH-A and INH-B assays for which Nigel Groome, Oxford Brookes University, Oxford, UK, provided the reagents. Mr. N. Balazs and his staff in the Department of Chemical Pathology, Monash Medical Centre, provided the FSH and estradiol measurements. The Melbourne Women's Midlife Health Project is supported by grants from the Victorian Health Promotion Foundation and the Public Health Research and Development Committee of the Australian National Health and Medical Research Council. Support for the hormone assays has also been provided by Organon Australia Pty Ltd.

References 1. World Health Organization (1981). "Research on the Menopause. Report of a WHO Scientific Group," Tech. Rep. Ser. 670. WHO, Geneva. 2. World Health Organization (1996). "Research on the Menopause in the 1990's," Tech. Rep. Ser. 866. WHO, Geneva. 3. McKinlay, S. M., Brambilla, D. J., and Posner, J. G. (1992). The normal menopause transition. Maturitas 14, 103-115. 4. Sherman, B. M., West, J. H., and Korenman, S. G. (1976). The menopausal transition: Analysis of LH, FSH, estradiol, and progesterone concentrations during menstrual cycles of older women. J. Clin. Endocrinol. Metab. 42, 629-636. 5. Velasco, E., Malacara, J. M., Cervantes, E, Diaz de Le6n, J., Divalos, G., and Castillo, J. (1990). Gonadotropins and prolactin serum levels during the perimenopausal period: Correlation with diverse factors. Fertil. Steril. 53, 56-60. 6. Metcalf, M. G., Donald, R. A., and Livesey, J. H. (1981). Pituitary-

16.

17.

18.

19.

20.

21.

22.

23.

ovarian function in normal women during the menopause transition. Clin. Endocrinol. 14, 245-255. Longcope, C., Franz, C., Morello, C., Baker, R., and Conrad-Johnston, C., Jr. (1986). Steroid and gonadotropin levels in women during the perimenopausal years. Maturitas 8, 189-196. Rannevik, G., Caristr6m, K., Jeppsson, S., Bjerre, B., and Svanberg, L. (1986). A prospective long-term study in women from premenopause to postmenopause: Changing profiles of gonadotrophins, oestrogens and androgens. Maturitas 8, 297-307. Rannevik, G., Jeppsson, S., Johnell, 0., Bjerre, B., Laurell-Boruli, Y., and Svanberg, L. (1995). A longitudinal study of the perimenopausal transition: Altered profiles of steroid and pituitary hormones, SHBG and bone mineral density. Maturitas 21, 103-113. Burger, H. G., Dudley, E. C., Hopper, J. L., Shelley, J. M., Green, A., Smith, A., Dennerstein, L., and Morse, C. (1995). The endocrinology of the menopausal transition: A cross-sectional study of a populationbased sample. J. Clin. Endocrinol. Metab. 80, 3537-3545. Burger, H. G., Cahir, N., Robertson, D. M., Groome, N. P., Green, A., and Dennerstein, L. (1998). Serum inhibins A and B fall differentially as FSH rises in perimenopausal women. Clin. Endocrinol. 48, 809-813. Lee, S. J., Lenton, E. A., Sexton, L., and Cooke, I. D. (1988). The effect of age on the cyclical patterns of plasma LH, FSH, oestradiol and progesterone in women with regular menstrual cycles. Hum. Reprod. 3, 851-855. Groome, N. P., Illingworth, P. J., O' Brien, M., Rodger, P. A. L., Rodger, E E., Mather, J. E, and McNeilly, A. S. (1996). Measurement of dimeric inhibin B throughout the human menstrual cycle. J. Clin. Endocrinol. Metab. 81, 1401-1405. Santoro, N., Brown, J. R., Adel, T., and Skurnick, J. H. (1996). Characterization of reproductive hormonal dynamics in the perimenopause. J. Clin. Endocrinol. Metab. 81, 1495-1501. MacNaughton, J., Bangah, M., McCloud, P., Hee, J., and Burger, H. (1992). Age related changes in follicle stimulating hormone, luteinizing hormone, oestradiol and immunoreactive inhibin in women of reproductive age. Clin. Endocrinol. 36, 339-345. Klein, N. A., Illingworth, P. J., Groome, N. P., McNeilly, A. S., Battaglia, D. E., and Soules, M. R.(1996). Decreased inhibin B secretion is associated with the monotropic rise of FSH in older, ovulatory women: A study of serum and follicular fluid levels of dimeric inhibin A and B in spontaneous menstrual cycles. J. Clin. Endocrinol. Metab. 81, 27422745. Illingworth, P. J., Reddi, K., Smith, K. B., and Baird, D. T. (1991). The source of inhibin secretion during the human menstrual cycle. J. Clin. Endocrinol. Metab. 73, 667-673. Roberts, V. J., Barth, S., EI-Roeiy, A., and Yen, S. S. C. (1993). Expression of inhibin/activin subunits and follistatin messenger ribonucleic acids and proteins in ovarian follicles and the corpus luteum during the human menstrual cycle. J. Clin. Endocrinol. Metab. 77, 1402-1410. Richardson, S. J., Senikas, V., and Nelson, J. E (1987). Follicular depletion during the menopausal transition: Evidence for accelerated loss and ultimate exhaustion. J. Clin. Endocrinol. Metab. 65, 1231-1237. Hughes, E. G., Robertson, D. M., Handelsman, D. J., Haywood, S., Healy, D. I., and de Kretser, D. M. (1990). lnhibin and estradiol responses to ovarian hyperstimulation: Effects of age and predictive value for in vitro fertilization outcome. J. Clin. Endocrinol. Metab. 70, 358 -364. Seifer, D. B., Gardiner, A. C., Lambert-Messerlian, G., and Schneyer, A. L. (1996). Differential secretion of dimeric inhibin in cultured luteinized granulosa cells as a function of ovarian reserve. J. Clin. Endocrinol. Metab. 81,736-739. Burger, H. G. (1984). The physiological basis of the fertile period. In "Fertility and Sterility," R. F. Harrison and B. W. Thompson, eds., pp. 51-8. MTP Press, Lancaster, England. Longcope, C. (1990). Hormone dynamics at the menopause. Ann. N.Y. Acad. Sci. 592, 21-30.

155

CHAPTER 9 P e r i m e n o p a u s a l H o r m o n e Changes 24. Trevoux, R., De Brux, J., Castanier, M., Nahoul, K, Soule, J.-R, and Scholler, R. (1986). Endometrium and plasma hormone profile in the peri-menopause and postmenopause. Maturitas 8, 309-26. 25. Sherman, B. M., and Korenman, S. G. (1975). Hormonal characteristics of the human menstrual cycle throughout reproductive life. J. Clin. Invest. 55, 699-706. 26. Metcalf, M. G., and Donald, R. A. (1979). Fluctuating ovarian function in a perimenopausal woman. N.Z. Med. J. 89, 45-47. 27. Metcalf, M. G. (1988). The approach of menopause: A New Zealand study. N.Z. Med. J. 101, 103-106. 28. Hee, J., MacNaughton, J., Bangah, M., and Burger, H. G. (1993). Perimenopausal patterns of gonadotrophins, immunoreative inhibin, oestradiol and progesterone. Maturitas 18, 9-20. 29. Dennerstein, L., Smith, A.M., Morse, C., Burger, H. G., Green, A., Hopper, J., and Ryan, M. (1993). Menopausal symptoms in Australian women. Med. J. Aust. 259, 232-236. 30. Faddy, M. J., and Gosden, R. G. (1995). A mathematical model of follicle dynamics in the human ovary. Hum. Reprod. 10, 770-775. 31. Pellicer, A., Mari, M., de los Santos, M. J., Sim6n, C., Remohi, J., and Tarin, J. J. (1994). Effects of aging on the human ovary: The secre-

32. 33.

34. 35. 36.

37.

tion of immunoreactive (a-inhibin and progesterone. Fertil. Steril. 61, 663-668. Doring, G. K. (1969). The incidence of anovular cycles in women. J. Reprod. Fertil. 6, 77-81. Zumoff, B., Strain, G. W., Miller, L. K., and Rosner, W. (1995). Twenty-four hour mean plasma testosterone concentration declines with age in normal premenopausal women. J. Clin. Endocrinol. Metab. 80, 1429-30. Judd, H. L. (1976). Hormonal dynamics associated with the menopause. Clin. Obstet. Gynecol. 19, 775-788. Vermeulen, A. (1976). The hormonal activity of the postmenopausal ovary. J. Clin. Endocrinol. Metab. 42, 247-253. Judd, H. L., Lucas, W. E., and Yen, S. S. (1974). Effect of oophorectomy on circulating testosterone and androstenedione levels in patients with endometrial cancer. Am. J. Obstet. Gynecol. 118, 7 9 3 798. Hughes, C. L. Jr., Wall, L. L., and Creasman, W. R. (1991). Reproductive hormone levels in gynaecologic oncology patients undergoing surgical castration after spontaneous menopause. Gynecol. Oncol. 40, 42-45.

7 H A P T E R 1(

Epidemiology." Methodologic Challenges in the Study of Menopause SYBIL L.

I. II. III. IV.

CRAWFORD

New England Research Institutes, Watertown, Massachusetts 02472

Introduction Study Design Data Collection Methods Measurement Issues

V. Analytic Considerations

VI. Conclusion References

I. I N T R O D U C T I O N

care patterns, and a subject's own perceptions of menopause. Statistical data analyses in both cross-sectional and longitudinal studies often require handling issues such as biases of recall and selection and censored and missing data, and complicated data structures such as daily symptoms or reproductive hormone levels. Methodologic challenges in the study of menopause are discussed in this chapter. Topics covered include study design issues for both observational studies and clinical trials of hormone replacement therapy; types of data collection instruments; measurement issues, particularly ascertainment of menopause status; and analytic concerns, including choice of appropriate statistical techniques for different types of data.

The menopause is a complex and multifaceted phenomenon, one that is very challenging to study for a number of reasons. First, study design can be logistically and scientifically demanding. Community-based samples are critical for an examination of menopause in the general population, as opposed to nonrepresentative clinic-based samples, but the former are much more costly to obtain. Moreover, choice of the length of followup and eligibility criteria are complicated by variation within and across women in experiences of menopause transitions. In addition, the menopause involves multiple domains, including physiologic as well as psychologic and lifestyle changes. This has implications for selection of data collection methods, in order to obtain a complete picture. A number of measurement issues also arise in a study of menopause, including how to define menopause status. Many measures, particularly those derived from self-report, are affected by physiologic changes as well as by sociodemographic characteristics, psychologic factors, culture, healthMENOPAUSE: BIOLOGY AND PATHOBIOLOGY

II. STUDY DESIGN A number of issues involved in the design of studies of menopause are summarized in this section. The first topic considered is the choice of a sampling flame and the importance of a population-based sample. Issues relevant to 159


160

SYBIL L. CRAWFORD

observational studies include cross-sectional versus longitudinal design, choice of age range, and eligibility criteria for menopause status. For examining the role of hormone replacement therapy, the use of observational studies versus clinical trials is discussed, and eligibility criteria used in several large clinical trials are presented.

A. Choice of Sampling Frame Many early studies of menopause relied on clinic-based samples [1], such as patients at menopause clinics. Such samples are relatively cost-efficient to obtain and response rates are likely to be relatively high. These advantages are greatly offset, however, by the atypical nature of such groups. Women seeking health care, particularly those seeking treatment for menopausal symptoms, differ in a number of ways from other menopausal-aged women, with respect to characteristics likely to be related to menopausal and healthrelated factors under study. For example, they have greater access to health care and are more worried about their health, are more likely to report menopausal symptoms, are more likely to undergo a surgical menopause, are more likely to receive psychiatric treatment, and tend to experience more long-term ill health preceding menopause [2-9]. In short, a sample of patients, particularly those from a menopause clinic, is unlikely to provide adequate representative data on the experience of menopause in the general population. Moreover, the range of variability in characteristics of interest may be restricted, affecting the ability to detect associations with outcomes [10]. Thus, it is critical to sample from a frame including women in the general community. A related consideration is sufficient representation from traditionally understudied subgroups, particularly women of lower socioeconomic status (SES) and races and ethnicities other than non-Hispanic Caucasians [4,11-14]. Compared with non-Hispanic Caucasian women, particularly those with high levels of SES indicators such as education, we know relatively little about the menopausal experience for these women. Sampling frames used in past and current communitybased studies include census lists [11,15], driver's license lists [ 16], health maintenance organization patient lists [ 11 ], utility lists [ 11 ], telephone directories [ 17], and random-digit dialing [ 11 ]. Formal statistical generalization to the population of interest requires use of probability sampling from a sampling frame in which all members of the target population have a known nonzero probability of being sampled [18]. In practice, this may be costly or logistically difficult to achieve, particularly when attempting to sample hard-tofind subgroups less likely to be included on sampling frames. Studies may need to combine information from multiple sampling frames, as was done by several field sites involved in the Study of women's Health Across the Nation (SWAN) [ 11].

Obtaining sufficient numbers of racial/ethnic minorities or low-SES women may be particularly challenging, and may require techniques such as household enumeration, or "snowball" sampling in which potential subjects are referred to the study by already-identified subjects, both of which were employed by SWAN field sites [11]. Note that computation of sampling probabilities under this latter approach can be difficult or impossible, because the probability that a woman is "sampled" as a snowball is typically not known [ 11 ]. In turn, this affects the ability to use the sample to generalize to the population of interest. Thus, there can be trade offs between obtaining a probability sample and sampling sufficient numbers from hard-to-find subgroups such as low-SES subjects.

B. Issues in Observational Studies of Menopause This section considers study design issues relevant for observational studies of menopause, where the natural history of menopausal transitions m including surgical menopausem will be observed. I. CROSS-SECTIONAL VERSUS LONGITUDINAL DESIGN

This issue is particularly important for studies of menopause, due to the relatively lengthy duration of the entire menopausal experience for an individual woman. In a crosssectional study of menopause, one cannot observe withinsubject changes in health occurring concurrently with withinsubject menopause transitions. Thus, inferences regarding the role of menopause are based on between-woman comparisons, e.g., comparing age-matched pre- and postmenopausal women [ 19]. Such analyses assume that cross-sectional estimates of menopause status or reproductive age are equivalent to those estimated from longitudinal observations, an assumption that is not always correct [12,20-25]. In addition, retrospective data are more subject to recall error [26]. Accurate assessment of temporal sequences also is much more difficult with cross-sectional data, e.g., determining whether attitudes prior to the menopause transition predict subsequent menopausal experiences [2,24,27], and again requires assumptions regarding the applicability of between-woman differences. Choice of cross-sectional or longitudinal design, however, often is determined in large part by considerations of cost. The followup time necessary to capture the full menopausal period, from pre- to peri- to postmenopausal (definitions of these terms are presented in Section III), is relatively long. In Caucasians, the median time elapsed between the onset of perimenopause and the final menstrual period (FMP) has been estimated as 3.8 years [27]. Moreover, note that this interval does not include the length of time that a woman is in the study prior to reaching perimenopause. Thus a cross-sectional study can be a cost-efficient way to exam-

CHAPTER 10 Methodologic Challenges in Menopause Studies ine both menopausal transitions, pre- to perimenopause and peri- to postmenopause, in a short time. However, interpretation of results is subject to the caveat noted above regarding use of between-woman comparisons to make inferences regarding within-woman changes. Note that a median length of perimenopause of 3.8 years implies that a large percentage of women will transition from peri- to postmenopause in under 4 years. Length of this transition, however, is not entirely random but is associated with a number of characteristics, including smoking [27], which affects a number of important variables (e.g., cardiovascular disease risk factors), as well as the age at onset of perimenopause [27]. Women with a shorter perimenopause also are less likely to report menopausal symptoms or to be characterized as depressed [2,23]. Thus, care must be used when generalizing from women with a short perimenopause to the full population. An important advantage of following the same individuals through the entire menopausal periodmfrom pre- to peri- to postmenopause--is the enhanced ability to assess the presence of curvilinear (nonlinear) relationships between health outcomes and menopause status or reproductive age, measured as time before/after the final menstrual period [4,22,28]. For example, we can examine whether within-subject bone loss accelerates around the onset of perimenopause, and decelerates after FMP. Such analyses are more complicated and require assumptions when performed on cross-sectional data. 2. CHOICE OF AGE RANGE

A common goal of menopause studies is to distinguish the roles of menopause or reproductive age and chronologic age. Hence, the age range needs to be selected accordingly, with women observed to experience menopause transitions at different chronologic ages in order to reduce confounding between reproductive and chronologic age. The appropriate age range depends in part on whether the study is crosssectional or longitudinal; in general, the age range should be broader for a cross-sectional than for a longitudinal study, since the latter involves following subjects as they age, thereby widening the effective observed age range. For a cross-sectional study of natural menopause, the age range should include the full set of ages at which women typically become menopausal [26], approximately 40 years through 55 years, as was done in SWAN's cross-sectional phase [ 11]. A study of surgical menopause may need to include women younger than 40 years, as many women have a hysterectomy prior to age 40, particularly for diagnoses such as endometriosis [29]. African-American women also tend to undergo hysterectomy at a younger age than do Caucasians [29]. The age range should not be so large, however, as to involve cohort effects, where younger women are different from older women, e.g., with respect to characteristics such as use of oral contraceptives [28]. Another consideration for choice of age range is the

161 prevalence of smoking in the population under study. Smokers tend to experience menopause earlier than do nonsmokers, by 1-2 years on average [10,15,30,31]. Investigators may want to consider a lower age range for smokers than for nonsmokers, in order to capture the pre- to perimenopause transition in both groups. For a longitudinal study of initially premenopausal women, it is important to consider an upper age limit for recruitment. Later menopause has been linked to a number of observed characteristics, most importantly lower rates of smoking, and possibly higher body mass index [30,32,33], both of which are associated with key outcomes such as cardiovascular disease risk and bone density [34,35]. Thus, older women who are still premenopausal are atypical of premenopausal women in general, exhibiting "survivorship" bias [4]. In addition, these women tend to have a shorter perimenopause [27] and as a consequence may experience fewer menopausal symptoms [2,23]. In short, data from older premenopausal women may not be generalizable to the population at large. Note that this group's menopausal experience is important scientifically. A longitudinal study, however, will collect relevant data as women age into this group. For a longitudinal study with followup of subjects of under 4 - 5 years, where the full menopausal period may not be observed, choice of age range is determined in part by the stage of menopause of greatest interest. For example, a study focusing on the transition from pre- to perimenopause should sample primarily younger women, e.g., an age range centered around 4 7 - 4 8 years, the estimated median age for onset of perimenopause in Caucasians [27]. In contrast, a study of postmenopausal women should include primarily older women, e.g., aged 5 0 - 7 9 years, as in the Women's Health Initiative [36]. 3. MENOPAUSE STATUS ELIGIBILITY CRITERIA As with age range, the appropriate choice of eligibility criteria involving menopause status depends in part on whether the study is cross-sectional or longitudinal. For a cross-sectional design, one may want to be more inclusive, sampling women in a variety of menopause stages~including surgical menopause--in order to include data from all phases of menopause in the study. In fact, including only older postmenopausal women can be problematic when the outcome is age at menopause, due to recall bias [4,26]. For longitudinal studies, within-woman changes in menopause status or reproductive aging are observed directly, so that eligibility criteria regarding initial menopause status can be more restrictive than in a cross-sectional study. The aims of a study also affect the choice of status categories or reproductive age. Choice of status eligibility criteria which are appropriate for one set of scientific goals may not be adequate for others. For example, the Postmenopausal Estrogen-Progestin Intervention Study (PEPI) had as its primary goal to compare the impact of different hormone

162

SYBIL L. CRAWVORD

replacement therapy regimes on measures of cardiovascular disease risk [37,38]. To this end, the design employed sampling restrictions in terms of chronologic age and years since menopause (natural or surgical). A consequence of this design was an artificial collinearity between these two variables, i.e., a collinearity that does not exist in the general population, where age and years since menopause are not restricted. Thus, analyses attempting to identify predictors of age at menopause (not an original study goal) were hampered by the inability to distinguish covariates of age at menopausemthe outcome of interestmfrom correlates of chronologic age [38]. As this example demonstrates, careful thought should be given to potentially competing study aims when selecting eligibility criteria. Past studies of menopause vary with respect to inclusion or exclusion of surgically menopausal subjects. These women differ in a number of ways from women experiencing natural menopause, including better access to health care, and lower prior levels of health [4,24,39]. Their menopausal experiences and subsequent disease risk also are likely to diverge from those of naturally menopausal women, in part because of different characteristics prior to surgery, but also because their reproductive hormonal profile~e.g., rapidity and timing of changes in hormone levels w differs as well [24,40]. Thus, one cannot generalize results regarding menopause and health outcomes, such as cardiovascular disease risk factors or menopausal symptoms, from surgically menopausal women to naturally menopausal women. Because surgically menopausal women are not a random subsample of all women going through menopause, however, omitting them from analyses implies that the resulting description of the menopausal experience for the general population is in some sense incomplete. A common recommendation is to include these w o m e n ~ i n cross-sectional studies, include women with a prevalent hysterectomy or bilateral oophorectomy, and in longitudinal studies, continue to follow women with an incident surgical menopausembut study them separately from other women. It may be useful to consider surgically menopausal women as a separate stratum, both for sampling and for data analyses [4,22,36,37].

C. Issues in S t u d i e s o f H o r m o n e Replacement Therapy This section compares the use of observational studies and clinical trials in the study of hormone replacement therapy (HRT), and summarizes eligibility criteria used in several recent trials. 1. OBSERVATIONAL STUDIES VERSUS CLINICAL TRIALS

Clinical trials are preferable to observational studies of this topic due to selection bias [4,41-43]. Many studies have

found that prior to initiation of HRT, users are more likely to exhibit characteristics associated with better health, including higher SES (as indicated by income and education), higher use of health care, more exercise, lower body mass index, and a better risk profile for cardiovascular disease [41,44-49]. Thus part of the difference between users and nonusers in outcomes such as cardiovascular disease risk factors in observational studies is likely due to preexisting differences in health and related characteristics [50]. Comparing women prescribed and not prescribed HRT, many past studies took place when physicians were reluctant to prescribe HRT for women at high risk for cardiovascular disease (CVD) [47], reflected in higher observed CVD risk among nonusers. Even considering only ever-users, women who continue to use HRT differ from those prescribed HRT but who discontinue use. The former group is more tolerant of HRT and is less likely to experience adverse effects [4,51]. In fact, short-term users have been found to have greater subsequent cardiovascular disease risk compared to long-term users [52]. Sturgeon and colleagues also noted a "healthy user survivor effect," whereby women who developed an illness discontinued use of HRT [53]. In addition, long-term HRT users are by definition "compliant." In clinical trials, compliers on both study arms have been found to fare better than noncompliers, and the magnitude of the effect of compliance with placebo was similar to that found for HRT in two meta-analyses [54,55]. Thus, part of the difference in health outcomes or disease risk between users and nonusers may be due to a compliance effect. Moreover, it is difficult to control completely for all such biases. For example, it is not straightforward to adjust appropriately for education and SES when modeling health outcomes [4,43,45,46]. Even in a population that was relatively homogeneous with respect to SES, users differed from nonusers regarding behaviors affecting health promotion and disease prevention [41 ]. In addition, some biases involved in prescribing and compliance may not be observed or known [41,43,56]. In summary, estimates of the benefit of HRT obtained from observational studies are likely to be somewhat overstated [43,53,56]. To assess the impact of HRT on outcomes such as cardiovascular disease risk or events, it is important to look to results from clinical trials such as PEPI, the Women's Health Initiative, or the Heart and Estrogen/Progestin Replacement Study (HERS) [36,42,57]. Such trials avoid selection bias by employing random assignment to treatment arm, including a placebo arm [57].

2. ELIGIBILITY CRITERIA IN CLINICAL TRIALS o r H R T In the final stage of sample selection, both PEPI and HERS included only subjects with a high likelihood of complying with treatment arm, by requiring 80+ % compli-

CHAPTER 10 Methodologic Challenges in Menopause Studies ance during a run-in phase [57,58]. Current HRT users recruited to PEPI were required to stop treatment [58]. The trials also excluded women with contraindications for use of HRT [57,58]. Both restrictions are justified on logistical and ethical grounds. They may, however, limit generalizability of results somewhat to compliant subjects who are potentially able to be long-term HRT users. Moreover, short-term (3 months) cessation of HRT may be insufficient to achieve "wash-out" of its effects, e.g., on rate of bone turnover. McKinlay suggests that the most appropriate subjects for a clinical trial of HRT are those with no prior use [4]. Current users are less likely to report adverse effects for the HRT arm (because they have been taking HRT, and hence they can tolerate it), and are more likely to be unblinded on the placebo arm, whereas past u s e r s m w h o may have discontinued use due to adverse effects m may be less likely to be able to tolerate the HRT arm. Residual effects of HRT also may be an issue with current users. The availability of women with no prior HRT use varies by geographic region, however [59,60].

III. DATA COLLECTION

METHODS

As in all studies, there is typically a trade off between retrospective and prospective data collection in menopause studies, with a lower cost but also potentially lower accuracy for retrospective data collection. Moreover, as in studies of other topics, subject burden increases with the level of detail of data collection, and hence more detailed data collection regimes such as: daily hormone measurements tend to be employed with a selected subset (either by design, or by default due to subject nonresponse); consequently, the resulting data are less generalizable [4,24,61 ]. Thus, menopausal studies, which can involve relatively demanding data collection methods such as daily menstrual calendars, need to balance scientific rigor with participant burden. In addition, the menopause is a highly multifaceted phenomenon, involving changes in physiologic, epidemiologic, and psychosocial factors. Thus, investigators should consider collection of data in a number of domains [ 12]. These multidisciplinary data will provide better, more complete information for key study goals and will make efficient use of the large effort needed to recruit the participant sample. As an example, a recent study noted a link between depression and low bone density [62], possibly related to low estrogen concentrations, and another study found an association between bone density levels and breast cancer [63]; such findings are useful in a number of disciplines, including endocrinology, psychology, and oncology. This section presents a summary of data collection methods and types of instruments used in studies of menopause to collect information in various domains, moving from least to most detailed or demanding.

163 A. M a i l e d S u r v e y s This type of instrument has been used in past large epidemiologic cross-sectional surveys, such as the first phase of the Massachusetts Women's Health Study [15]. Such an instrument is fairly unobtrusive for subjects. It is also relatively inexpensive compared with other modes of data collection, although this advantage is offset somewhat by the need to conduct telephone followup of nonrespondents. For example, in the cross-sectional phase of the Massachusetts Women's Health Study, the initial response rate to the mailed survey was 57%; telephone followup of initial nonrespondents raised the combined final response rate to 77%. Moreover, the initial respondents to the mailed survey differed from nonrespondents to the mailed survey who subsequently responded by telephone, with higher levels of education and access to health care among the former [64]. Thus, it may be necessary to employ a mixed mode in order to reduce nonrespondent bias. By necessity, all data collected on a mailed pen-and-paper survey are self-reported. Thus, the investigators cannot verify data such as prescription medications, or anthropometric measures such as height or weight. Separate data coding and data entry also are required, unless scannable forms are used.

B. T e l e p h o n e S u r v e y s This mode has been used in a number of menopause studies, including the cross-sectional survey in the Melbourne Women's Midlife Health Project [ 17], the Healthy Women's Study [16], the longitudinal phase of the Massachusetts Women's Health Study [15] and in SWAN's cross-sectional phase [ 11 ]. Telephone surveys are useful not only for primary data collection, but for initial screening to identify and recruit specific cohorts. SWAN's cross-sectional telephone survey, for example, included questions used to assess cohort eligibility, particularly menopause status--which is not available on sampling frames; many field sites recruited eligible participants into the cohort at the conclusion of the telephone interview [11]. Depending on available technology, computeraided telephone interviewing can be used, which eliminates extra steps in data coding and data entry and leads to more accurate data collection. As with mailed surveys, however, there may be some bias, in this case due to subjects without telephone or with unlisted telephone numbers. Thus, a mixedmode approach may be needed, combining home visits with telephone interviews, as in SWAN [ 11 ].

C. I n - H o m e or I n - C l i n i c Visits Data collection has been carried out in subjects' homes or in local clinics in a number of recent menopause studies,

164

SYBILL. CRAWFORD

including the Healthy Women's Study, the Melbourne Women's Midlife Health Project, the second phase of the Massachusetts Women's Health Study, and SWAN's longitudinal phase. Many anthropometric measurements have been taken in a subject's home, including height, weight, blood pressure, and girth; blood and urine specimens also can be collected [65]. For longitudinal studies, it is important to employ similar collection methods for an individual subject at each visit in order to estimate within-woman changes, because measures such as blood pressure can vary by setting. Other physiologic data collected in clinics in studies of menopause include bone density and carotid ultrasound [ 11 ]. Collection of blood or urine specimens also can be done, as in the Melbourne Women's Midlife Health Project [17], the Massachusetts Women's Health Study [65], and SWAN [ 11 ]. This may be logistically difficult, however, depending on what is being measured. Specimens may need to be taken on a particular day of the menstrual cycle (e.g., days 2 - 5 in the early follicular phase for regularly cycling women), or at a certain time of day, for assessing concentrations of reproductive hormones. Fasting samples may be required for accurate measurements, e.g., of glucose. Studies examining the relationship of reproductive hormones to cardiovascular disease risk factors may impose multiple conditions on the specimen collection protocol. Consequently, the study should consider accommodation of subjects' schedules by allowing them to "drop in" for specimen collection on a different day from the rest of the data collection, as in SWAN [ 11 ]. This type of data collection, particularly when done at a clinic, involves considerably more participant burden than a mailed or telephone survey, and response rates are correspondingly lower. Comparing response rates for the crosssectional and longitudinal phases of SWAN, for example, the latter were substantially lower [ 11 ].

D. D a i l y C a l e n d a r s Daily collection of self-reported data is useful for measuring a series of similar, recurrent events such as menstrual bleeding or premenstrual or menopausal symptoms. Retrospective recall of these events is poor [22,66-69]. Several large menopause studies have employed calendars, including SWAN and the Massachusetts Women's Health Study. Response rates tend to be lower than for mailed or telephone surveys [67], with most of the dropouts occurring at the beginning of data collection. Depending on the length of the data collection period and on the amount of data women are asked to record, there may be fairly substantial data coding and data entry requirements. For a study of menopausal transitions or the perimenopausal period, a longer period of data collection may be required than for studying premenopausal women, e.g., 2 + years, in order to capture perimenopausal changes in bleeding over time.

The calendars are self-administered in a woman's home, which means that no review by study personnel is possible until after the calendar is sent back to the study site. Retrospective data recording is also an issue; data quality is less accurate when data are collected retrospectively rather than prospectively. Johannes and colleagues pilot-tested an electronic calendar for daily collection of menstrual bleeding and symptoms. The device recorded the time and date of data entry by the participant. In the pilot study comprising a month's data collection, all 23 subjects entered at least one day's data late, i.e.,after the day on which bleeding or symptoms occurred [70]. These results indicate that it is critical to take measures to ensure high-data quality, particularly when using traditional paper instruments. Calendar instruments should be very simple to understand, in order to minimize mistakes and respondent burden. Subjects should be asked to return completed calendars frequently, e.g., monthly, so that participation and data quality can be monitored, and to limit the amount of retrospective data recording. Completed calendarsmparticularly the first several calendars--should be examined in order to identify errors. Researchers may want to send a letter to participants noting commonly made errors, or even to make retraining calls to respondents whose cal' endars demonstrate a large number of problems.

E. D a i l y S p e c i m e n C o l l e c t i o n Daily collection of specimens such as serum or urine can be very informative, particularly in the study of perimenopause, during which hormone concentrations fluctuate widely even within an individual woman [71-76]. Thus such measures are highly superior to annual serum or urine samples, with respect to capturing within-woman variability. Such collection is expensive to conduct and requires a great deal of subject cooperation, however; consequently, sample sizes often are relatively small.

IV. M E A S U R E M E N T

ISSUES

A number of measurement issues arise in the study of menopause, particularly the determination of a woman's menopause status. Various researchers also have noted methodological difficulties in the measurement of menopausal symptoms, as well as cultural or ethnic differences in reporting of variables related to menopause.

A. D e f i n i t i o n s o f M e n o p a u s e Status Indicators used to define menopause status have included age, menstrual bleeding, levels of reproductive hormones, and a woman's self-report.

CHAPTER 10 Methodologic Challenges in Menopause Studies 1. CHRONOLOGIC AGE

Early studies of menopause used chronologic age as a proxy for postmenopause [12,24,77]. This is a very poor measure of menopause status, however, because the final menstrual period occurs over a wide age range [27]. A comparison of menses-based and age-based definitions using data from a case-control study of breast cancer [78] indicated that--using a menses-based definition as the "gold standard"msensitivity and specificity for the age-based definitions differed for cases and controls. In particular, there were more premenopausal women classified incorrectly as postmenopausal among cases than among controls, because age at menopause (by the menses-based definition) was later in cases than in controls. Thus studies of breast cancer employing age-based definitions of menopause status may suffer from this differential misclassification, which affects estimation of the association between menopause status and disease. 2. MENSTRUAL BLEEDING Past World Health Organization Working Group meetings [79,80] have recommended use of the following definitions for menopause status categories, based largely on observed menstrual cycle patterns, which are assumed to reflect underlying endocrinological changes or levels [81 ]: a. Natural menopause: the permanent cessation of menstruation, determined retrospectively after 12 consecutive months of amenorrhea without any other pathological or physiological cause. b. Perimenopause: the period just prior to the final menstrual period through the first year after the final menstrual period, beginning at the onset of endocrinologic and menstrual changes. c. Premenopause: the entire reproductive period prior to the FME d. Induced menopause: the cessation of menses due to removal of both ovaries with or without removal of the uterus, or iatrogenic ablation of ovarian function. e. Premature menopause: natural menopause occurring before age 40. Also known as premature ovarian failure. f. Postmenopause: dating from the FMP, including both natural and surgical menopause.

Note that the perimenopausal period as defined above overlaps with both the first 12 months of postmenopause after the FMP, and the premenopause. Metcalf [75] distinguishes premenopause as menstruating at regular intervals, whereas perimenopause begins with the onset of irregular cycling and continues after the FMP until hormone levels stabilize. Other uses of these terms in the literature [82-85] separate pre-, peri-, and postmenopause, with premenopause ending at the onset of endocrinologic or menstrual changes, and perimenopause concluding with the FMP.

165 3. ENDOCRINE MEASURES The decade prior to the FMP is characterized by an increase in variability in reproductive hormone concentrations, even though a woman may continue to have her normal menstrual bleeding. Abrupt changes in these hormones may occur, with values typical of postmenopause followed by those seen in younger reproductive-aged women [71-75,86]. Although hormone concentrations stabilize 1-2 years after the FMP [87-89], no sharp changes occur at the time of the FMP [74]. To categorize women regarding menopausal status, a cutoff of follicle-stimulating hormone (FSH) of 3 5 - 4 0 IU/liter is commonly used in clinical practice and in research studies [90,91 ]. Some studies have employed cutoffs of FSH greater than 10-20 IU/liter to indicate perimenopause [84,92]. The above discussion indicates, however, that endocrine variables cannot be considered reliable indicators of menopausal status [71,72,87], particularly based on a single serum or urine sample, because within-woman values fluctuate widely during perimenopause, and patterns are variable across women [76]. Hormone concentrations also vary by chronological age as well as by time before the FMP [71,76,93-97]. Although average values within and across women demonstrate general trends during this period, no single cutoff value is likely to be accurate as a predictor of status [98]. 4. SELF-DEFINITIONS

Women's perceptions of their own menopause status vary by culture and race/ethnicity [81], and do not agree completely with categorizations based on bleeding patterns [99]. Self-reported menopause status may not be appropriate for epidemiological purposes [81 ], but may be of interest in its own right or in studying women's experiences during the transition through menopause [99]. Assessment of years since the FMP from self-report on a cross-sectional survey can be inaccurate, because recall becomes increasingly unreliable with greater time elapsed since the FMP and there is evidence for digit preference [33,100-102] (see Section IV, C).

B. C h a r a c t e r i z i n g P e r i m e n o p a u s e The definition of natural menopause presented in Section IV,A,2 has become an accepted standard [82], and investigators have turned their attention to better characterizing perimenopause, for which no standard definition exists [5,82]. Treloar [103,104] defined the onset of perimenopause as the start of an increase in the variability in cycle length. Two later studies also identified menstrual irregularity as a perimenopausal indicator, using as a "gold standard" either an FSH level of at least 15 IU/liter [84] or subsequent transition to postmenopause within 3 years [82]. Menopausal symptoms, particularly hot flashes and night

166

SYBIL L. CRAWFORD

sweats, also were indicative of perimenopause. In the latter study, changes in menstrual flow were associated with status before controlling for irregularity, but were not independently related to status [82]. Further refinement of the characterization of perimenopause is needed [5,105]. Two studies have suggested a distinction between different stages of perimenopause, based on menstrual bleeding regularity [83,106]. Early perimenopause corresponds to self-reported changes in frequency of menstrual bleeding, whereas late perimenopause is defined by prolonged (more than 6 months) amenorrhea. Differentiation of these two stages appears to be informative in terms of summarizing the sequence of bleeding patterns, and for prediction of subsequent transition to postmenopause [83,106]. In addition, the simultaneous incorporation of multiple sources of data, e.g., symptoms, bleeding patterns, and reproductive hormone profiles, has been suggested as an area of future study [ 107].

C. R e l i a b i l i t y o f S e l f - R e p o r t e d D a t a Many epidemiologic studies of menopause ask a woman to report the date of her last menstrual period or the date of surgical menopause--used to estimate her age at menop a u s e m o r patterns of use of hormone replacement therapy. Several studies have investigated the reliability and reproducibility of such self-reported information, by comparison with medical records or by repeated interviewing of subjects over time. Considering menopause status, reliability and reproducibility tend to be relatively high for the type of menopause, i.e., natural versus surgical [108-110]. Reliability and reproducibility also are better for age at surgical menopause than for age at natural menopause [68,69,108-112]. In one study [108], women tended to underestimate age at menopause. This error has implications for estimating the association between age at menopause and disease. For example, breast cancer has been found to be positively related to later age at menopause, whereas osteoporosis and cardiovascular disease have been linked to an earlier menopause. If selfreported age at menopause is misclassified as compared with true age at menopause, the association of age at menopause with disease will be underestimated for breast cancer [ 108, 112] and overestimated for cardiovascular disease or osteoporosis [ 108]. Digit preference in reporting of age at menopause also was observed in several studies [102,111], particularly for ages ending in "0" and "5." For self-reported age at menopause, reliability tended to decrease as the time since the final menstrual period increased [ 109,111 ]. One study found, however, that reproducibility was higher as time since menopause increased, for women with an earlier menopause; thus it may be that women with a relatively young age at menopause (under 40) recall this event better [108].

Regarding recollections of patterns of HRT use, reliability was highest for ever-use [110,113-115]. For epidemiological studies, a single self-report question may be sufficiently accurate for this piece of information [ 113]. Selfreported details of use, including length of use, dose, and reasons for starting/stopping, however, were less reliable [110,113-114]. Lower reliability was related to subject characteristics such as higher age of the subject [114-115] and longer time elapsed since last use [ 113-115]. In summary, self-report may be adequate for basic data such as type of menopause or ever-use of HRT, but less appropriate for more detailed data such as the age at final menstrual period or length of HRT use. Thus, researchers may need to consider prospective designs or other sources of information, e.g., medical records abstraction, for these data.

D. M a s k i n g o f " N a t u r a l " M e n o p a u s e T r a n s i t i o n s Medical interventions, particularly exogenous reproductive hormone use and removal of the uterus and/or ovaries, affect characteristics commonly used to define menopause status [28]. Depending on the regime, use of HRT or oral contraceptives can alter menstrual bleeding, so that their use essentially masks menopausal status defined in terms of menstrual cycling [26,31,116]. Hormone use also can complicate classification based on endocrinological criteria, because it may affect reproductive hormone levels [116119]. Self-defined menopause status also varies by HRT use [99]. Surgical menopause, either from a hysterectomy or oophorectomy, obviously affects menstrual bleeding patterns. Moreover, women with a simple hysterectomy (ovaries not removed) may experience ovarian failure earlier than other women [ 120]. In short, straightforward determination of natural menopause status and timing of natural transitions that would have occurred in the absence of medical interventions is not possible. Implications for data analyses are discussed in Section V, and potential analytic strategies are presented.

E. M e a s u r e m e n t o f M e n o p a u s a l S y m p t o m s A number of methodological problems in past studies of menopausal symptoms have been identified. First, in order to avoid influencing a subject to associate certain symptoms with menopause, it is important to ask about general health and symptoms, including symptoms suspected of being related to menopause [69]; researchers should not identify symptoms a priori as being menopausal, or ask women what symptoms they experienced during menopause. Questions should include both positive and negative symptoms [3,5]. The reference period should be fairly short, e.g., 1 to 4 weeks, in order to minimize inaccuracies in recall [22,69]. Symptom reporting also is affected by cultural norms, as discussed in

CHAPTER 10 Methodologic Challenges in Menopause Studies the following subsection. In addition, researches should employ a standard scale, so that results across different studies can be compared [5,121 ]. A commonly used scale, the B lattKupperman index, is widely used but has been shown to be inadequate. Problems with this scale include development on a possibly unrepresentative sample of women and arbitrary weighting of items [122].

F. C u l t u r a l D i f f e r e n c e s in R e p o r t i n g Ethnicity and culture have been found to affect experiences and perceptions of menopause; this in turn leads to ethnic or cultural differences in reporting of menopauserelated data. For example, symptom reporting varies by culture and geographic region, with lower rates found in Asian and Central American populations than in the United States and Western Europe [ 123-125]. In Mayan Indians, one study [123] found no self-reported hot flashes, despite the occurrence of endocrine changes that were similar to those seen in women in the United States. Japanese women tend to report headaches, shoulder stiffness, and joint pain, whereas vasomotor symptoms appear to be rare and tend not to be associated with menopause status [126]. Ethnicity and culture also are related to sexual attitudes, values, and behavior [22], as well as to perceptions and reporting of menstrual bleeding [66,67]. These differences need to be accounted for in studies of menopause, by using culturally appropriate instruments [66,67,125].

V. A N A L Y T I C

CONSIDERATIONS

This section summarizes a variety of issues involved in analyses of data collected in studies of menopause.

A. M e t h o d s for E s t i m a t i n g A g e at N a t u r a l M e n o p a u s e Data from either cross-sectional (i.e., prevalence) or longitudinal (incidence) studies may be used for estimation of age at natural menopause. 1. PREVALENCE DATA Distributions of recalled age at menopause from crosssectional data can be analyzed using techniques such as Kaplan-Meier plots and Cox proportional hazards modeling [26]. Typically, prevalent cases of natural menopause are asked when their periods stopped, and data for pre- or perimenopausal women are censored at their current age. Techniques that do not account for censoring, e.g., histograms of age at menopause in the subset of prevalent naturally menopausal women, lead to estimates of age at menopause that are biased downward [26]. As noted in Section III, use of retrospective recall of age

167 at the FMP may be problematic; consequently, self-reporting of 12 + months amenorrhea (yes~no) at the time of the interview may be more accurate. Use of this outcome variable suggests estimation of a binary logistic regression of 12+ months of amenorrhea on chronologic age. Median age at menopause (or other percentiles) then can be estimated from the logistic regression model as a function of the intercept and slope [15,127]. Median age at menopause can be estimated for various subgroups, e.g., smokers and nonsmokers, by stratification on the characteristic of interest [15]. Note that a logistic regression analysis excludes women with a prevalent surgical menopause. For Caucasians, several studies suggest that this exclusion is appropriate, and that a competing risks model is not necessary [128,129]. It is unclear whether this holds for other racial/ethnic groups, however, particularly for African-Americans, who have a much higher rate of hysterectomy [29]. 2. INCIDENCE DATA As noted earlier, data from a prospective design, where information on menopause transitions is collected as they occur, is preferable to a retrospective design in terms of accuracy. Covariates also can be assessed prior to transitions, with less recall bias [ 10]. Longitudinal analyses of incidence data often employ hazards modeling, for example, estimating the probability that an event--such as the F M P - - occurs during the study, given that it has not occurred earlier. If the age at the FMP can be measured precisely, e.g., using menstrual calendars, then one can use techniques such as Cox proportional hazards modeling. If menopause status is ascertained only at intervals, e.g., at an annual interview, one can employ an approach used by Brambilla and McKinlay [ 10], which involves multinomial modeling of conditional probabilities of menopause status categories at each interview, including natural menopause, surgical menopause, and not yet menopausal. Analyses of age at natural menopause may need to take into account the competing risk of surgical menopause, although longitudinal analyses by Brambilla and McKinlay in non-Hispanic Caucasians suggest that this is not necessary for this racial/ethnic group [ 10], consistent with cross-sectional findings. B. A n a l y t i c M e t h o d s for M e n s t r u a l C a l e n d a r D a t a Goals of analyses of data from menstrual calendars typically include characterization of the distribution and patterns of menstrual segment lengths. Because bleeding may or may not correspond to a menstrual cycle, the term "segment" is sometimes used rather than "cycle" [67]. Many studies focus on segment length or bleeding length rather than on heaviness of menstrual flow [66,67,130,131 ]. The latter has been found to be less informative, e.g., in defining menopause status [82,83]. The majority of past studies utilizing menstrual calendars have been done in premenopausal women, with the exception of Treloar [103,104], who

168 followed women from menarche to the FME This subsection summarizes issues involved in the analysis of calendar data from premenopausal--regularly cycling--women, as well as analytic complications arising from the study of perimenopausal, i.e., irregular, bleeding. 1. ISSUES IN ANALYSIS OF PREMENOPAUSAL CALENDARS

Data can be analyzed using either a menstrual segment or an individual woman as the unit of analysis. Both approaches are useful, and the appropriate choice depends on the question of interest [132]. If the goal is inference regarding the distribution of segment length for an individual in the population of interest, then the woman should be used as the unit of analysis. An example is the reference period method, where each woman's bleeding patterns are summarized for a standard unit of time, typically 90 days [67]; this provides a cross-sectional summary for each subject. A related issue is length bias. In general, the observation period is fixed and the number of observations (segments) per woman varies. Consequently, women with shorter-and hence more--segments observed are overrepresented in analyses using the segment as the unit of analysis. Thus, using the segment as the unit of analysis can give estimates of segment length that are biased downward. In contrast, bywoman analyses give each woman the same weight. Because the observation period is usually defined in terms of calendar time rather than in terms of completion of a menstrual segment, the last segment is only partially observed. This censoring can cause bias in estimates, because the probability that its length is unknown is related to the length of the segment, with longer segments more likely to be censored [133]. The resulting bias may be small if the data collection period is relatively long. Belsey and Farley [67] summarize a number of analytic methods proposed to handle this censoring, including omitting the last segment from analyses; including it only in estimation of variability but not mean length; truncation, which affects variability estimates; and methods for handling right-censored survival data [ 133]. A number of statistical techniques have been employed in the analysis of menstrual segment lengths, all of which handle the correlation between multiple segments observed in the same subject. Methods also should account for length bias. Techniques such as growth curve modeling may not be particularly useful, because the number of observations (segments) is inherently part of the data to be observed. Methods that explicitly examine within-woman correlation between segments include estimation of segment-to-segment probabilities [106,130], e.g., whether long segments tend to be followed by shorter segments, as well as estimation of autocorrelation to assess the dependence between segment lengths as a function of the lag between segments [130]. Techniques used to model segment length include randomeffects modeling with a random intercept, or equivalently, a

SYBIL L. CRAWFORD

generalized estimating equation (GEE) approach with exchangeable correlation; this assumes that a woman's segment lengths vary randomly about her own mean [130]. Methods that incorporate covariates for segment length include GEE techniques [134,135], Poisson modeling [136], and autoregressive modeling [133]. Harlow and Zeger also employed a mixture model approach to characterize the distribution of segment lengths, whereby one component consisted of "normal" length segments and the other component included very long segments [130]. 2. ANALYTIC COMPLICATIONS FROM PERIMENOPAUSAL DATA Additional analytic issues arise in the study of perimenopausal menstrual segment lengths, due to increased menstrual irregularity. A key question is what constitutes a menstrual segment. World Health Organization definitions require at least one bleeding ~ not spotting~day followed by at least one bleed-free day [67]. Some analysts of premenopausal data omit spotting episodes from analyses entirely [67,137,138]. Spotting episodes, however, may be quite informative in the study of perimenopausal bleeding patterns. Johannes and colleagues, for example, found spotting episodes to be more common in the early perimenopause, indicating the utility of spotting episodes in distinguishing perimenopausal stages [106]. Thus a "conservative" definition of menstrual episodes or segments, whereby any spotting or bleeding is considered separately, may be in order. As noted in Section IV, perimenopause is characterized by within-woman changes in bleeding patterns, particularly an increase in irregularity. Hence the autocorrelation structure for segment lengths within an individual woman may be very different from the exchangeable correlation observed in studies of regularly cycling premenopausal women. To capture this, we may need a more complicated autocorrelation structure. Moreover, irregularity itself is not a welldefined concept. Bleeding patterns during the perimenopause may vary not only across women, but within women as well, and may depend on proximity to the final menstrual period [82,83,106]. Thus models of perimenopausal segment lengths need to allow autocorrelation structures to vary both within and across women. Finally, the complete interval of perimenopausal menstrual bleeding may not be observed during the period of study. Data may be subject to either left or right censoring, or both, depending on the woman's initial menopause status, the length of her perimenopause, and the length of the calendar data collection. Also, as just noted, bleeding patterns may change for an individual over time, depending on proximity to the final menstrual period [82,83,106]. Thus the menstrual segments observed during the study may not be representative of a woman's entire perimenopausal period, unlike segments observed in regularly cycling women. Analyses of perimenopausal segment length should account for this censoring.

CHAPTER 10 Methodologic Challenges in Menopause Studies C. M e t h o d s for C o m b i n i n g D a t a C o l l e c t e d at D i f f e r e n t F r e q u e n c i e s Menopause studies often involve different types of instruments, collecting data at varying frequencies, e.g., annual clinic visits, monthly symptom reporting collected via calendars, and daily urine specimens. Scientific questions of interest may require combining these data, as in assessment of the relationship between reproductive hormone levels measured annually and symptom patterns observed in monthly menstrual calendars, or in a comparison of daily menstrual calendar data to self-reported data on bleeding patterns from an annual interview. Possible approaches include "collapsing" the data measured at a higher frequency of measurement. For example, Johannes and colleagues summarized each woman's 12 months of bleeding patterns in terms of within-woman mean and variance of segment length [ 106]. The correlations of these summaries with annual reproductive hormone values then was computed. The time scale of one of the measures also can be adapted, as was done in several analyses of predictors of menstrual segment length [131,135]. Time-varying covariates such as weight were measured less frequently than monthly, and the schedule of measurement did not correspond to a woman's menstrual segments. To include these variables as predictors, the investigators defined the value of a covariate corresponding to a particular menstrual segment as the average of that covariate during the first 14 days of that segment or of the preceding segment. Another approach is to analyze only data measured concurrently, e.g., estimating the correlation between reproductive hormone concentrations measured at an annual clinic visit and characteristics of the menstrual segment occurring during that annual clinic visit. This has the disadvantage of ignoring other, possibly relevant, data, however.

D. M a s k i n g o f M e n o p a u s e S t a t u s D u e to U s e o f H R T As noted in Section IV, use of HRT prior to observation of 12 + months of amenorrhea results in an inability to assess a woman's "true" menopause status in the absence of HRT, or to determine her age at "natural" menopause. A variety of analytic techniques have been employed or proposed to handle these women in analyses where natural menopause status or transitions are variables of interest. A common approach is to omit ever-users or concurrent users from analyses [28,139-141]. However, as discussed in Section II, HRT users are not a random sample of all women traversing the menopause. Thus analyses that omit these women completely do not describe experiences in the overall

169 population. In longitudinal analyses in which baseline data are available prior to initiation of HRT, one can include some data from these women by censoring their observations at the time of HRT initiation, or omitting observations concurrent with HRT use [ 19]. Another technique is to analyze menopause status for non-HRT users, then compare users and nonusers, omitting menopause status as a variable [142]. This method has the advantage of including users in analyses. HRT users may be a mixture of "natural" menopause status categories, however, so that putting them in a single category may not be appropriate. For the same reason, treating HRT users as a separate menopause status category may distort the estimated association between menopause status and other variables. Other analyses have combined HRT users with postmenopausal women [85]. Many women, however, begin HRT use prior to permanent cessation of menses [49], and thus are likely to be dissimilar to naturally postmenopausal women. Inclusion of HRT users with this latter group could weaken or bias estimated associations of postmenopausal status with other characteristics. HRT use may be included in analyses as a covariate, and menopause status defined in terms of observed bleeding patterns or estrogen levels regardless of HRT use status [23]. For some outcomes, such as depression or sexual activity, the source of estrogen may be relatively unimportant to the question of interest, so that the outcome can be modeled as a function of a marker of the total estrogen exposure (endogenous and exogenous combined), and an indicator for HRT use (yes/no). For other outcomes, however, the distinction between endogenous and exogenous estrogen may be of greater relevance. For example, endogenous estrogen has little liver exposure compared with oral preparations of exogenous estrogen [40,143], and thus using total estrogen as a predictor may not be applicable for outcomes such as circulating lipids. Perhaps the most appropriate general approach is to consider "natural" menopause status as missing for HRT users, and employ techniques developed to handle missing data. Menopause status is unlikely to be missing completely randomly, so that analyses would require techniques that assume either data missing at random (related only to observed data) or nonignorable missingness (related to the unobserved "true" menopause status) [144,145]. Note that all approaches used in this situation necessarily rely on assumptions that are untestable, because "true" menopause status is not observable.

E. H a n d l i n g S u r g i c a l M e n o p a u s e in A n a l y s e s As with HRT use, surgical menopausemhysterectomy and/or bilateral oophorectomy--effectively masks or censors a woman's natural menopause status or transitions that

170 would have occurred in the absence of medical intervention. Often surgically menopausal women are omitted from data anlayses [28]. Similar to HRT users, however, surgically menopausal women are not a random subsample of all women, nor are they a small subsamplemover one-third of women in the United States undergo a hysterectomy by age 60 [ 146]. Thus results from analyses omitting these women will not be generalizable to the entire population [ 129]. Analyses of age at menopause as a potential risk factor for diseases such as breast cancer sometimes assign a "mean" or "typical" age at the FMP to surgically menopausal women. This imputation process, however, does not reflect the underlying distribution of the age at natural menopause, and thus distorts the associations of age at the FMP with disease outcomes [147]. Hysterectomy or oophorectomy status also can be included as a covariate or stratifying factor in analyses. Type of menopause can be included as a predictor, comparing naturally postmenopausal women to surgically menopausal women, as in PEPI [37]. Given the many differences between these two groups, it may be necessary to stratify data analyses on type of menopause [4,81 ]. Data may also be treated as censored for surgically menopausal women. For example, in survival analyses of age at natural menopause using prevalence data, data from surgically menopausal women may be censored at the time of surgery [31 ]. This approach assumes that a woman's experience in the absence of surgery is similar to that of women observed to have a natural menopause [129]. This may be accurate after controlling for predictors of type of menopause, such as access to medical care; that is, natural status may be missing at random. This assumption is essentially untestable, however. Finally, analyses may employ competing risks modeling. Such techniques can be used either with prevalence data, to estimate models for 12+ months of amenorrhea (yes~no) [128], or self-reported age at the FMP or surgery [129], or longitudinal incidence data [ 10].

E Assessing Associations between Menopause Status or Reproductive Age and Health Outcomes For cross-sectional data, any inferences regarding the role of menopause transition within a woman are done using between-women comparisons. Depending on the available data, analyses may employ categorized menopause status (pre, peri, post, surgical), reproductive age defined as time before/after the FMP, or levels of reproductive hormones. Analyses need to control for important confounders such as age, by including them as model predictors or by age matching [19]. For longitudinal data, one can directly examine withinwoman changes in outcomes concurrent with within-woman

SYBIL L. CRAWVORD

changes in menopause status. For example, one can model successive differences in the outcome, e.g., change in serum cholesterol or bone density from one annual visit to the next, as a function of corresponding successive changes in menopause status [23,142]. Longitudinal data also permit better assessment of temporal sequences of events, such as whether fluctuations in reproductive hormone levels precede or follow more overt signs of perimenopause such as increased menstrual irregularity. For cross-sectional data, statistical methods include linear and logistic regression, depending on the outcome variable, including concurrent menopause status as a covariate. Longitudinal analytic techniques must account for within-subject correlation of multiple observations. Approaches that consider each observation separatelymdata are not collapsed within a woman m include generalized estimating equation methods, repeated measures modeling, and random-effects modeling. This last method can be used, for example, to identify women with an extreme menopausal trajectory, e.g., "fast" losers of bone density. Other methods collapse data within a woman, e.g., a paired t-test of a subject's mean outcome level prior to the FMP versus the corresponding mean after the FMP [140]. Alternatively, one can apply spline analysis, fitting within-woman slopes before and after the FMP for a piecewise linear model, and compare pre-FMP and post-FMP slopes [ 139-141 ]. Choice of functional forms is key. For example, analysts of bone density data often use log of years since menopause as a predictor of bone density [92]; this functional form implies that bone loss is more rapid for women in early postmenopause. It is important to determine the presence or absence of curvilinear relationships, e.g., whether bone loss accelerates in perimenopause and levels off after the FME Depending on the outcome of interest, it is also critical to distinguish pre- from perimenopause rather than combining all observations prior to the FMP, because acceleration of changes due to menopause may occur well before the FMP [28]. One should also consider the amount of change in reproductive hormone concentrations in addition to absolute concentrations; rapid hormonal changes may be associated with outcomes such as symptoms [5,148-150].

G. Confounding, Effect Modification, and Stratification A key confounding factor related to both status and many health outcomes of interest is smoking. Smokers have an earlier menopause [10,15,30,31], higher levels of risk factors for cardiovascular disease [34], and a higher risk of low bone density and osteoporosis [35]. Without controlling for smoking status, the role of the menopause transition in changes in levels of disease risk is overstated [4,30,150]. Other potential

171

CHAPTER 10 Methodologic Challenges in Menopause Studies

confounders to consider include body mass index, possibly parity, and oral contraceptive use [30,31-33]. Relationships between menopause status and outcomes such as cardiovascular disease risk also may differ across subpopulations, e.g., smokers versus nonsmokers. In analyses of blood pressure and lipids, for example, menopause status played a larger role for nonsmokers [4]. Thus, analyses may need to include appropriate interaction terms between menopause status and smoking status, or even to stratify on smoking status. Type of menopause~surgical or naturalmis another potential stratification factor. As noted in Section II, surgically menopausal women have very different experiences before, during, and after the menopausal transitions. Consequently, including the type of menopause as a covariate may not be sufficiently complex to assess the role of surgical versus natural menopause; analyses may need to be stratified on type of menopause [4].

VI. CONCLUSION In closing, it is important to note comments by Lock [ 125]. The emphasis of much of the studies of menopause to datemparticularly in the United States and Europemhas been on its negative health consequences, e.g., experience of menopausal symptoms, loss of bone density, and increase in cardiovascular disease risk. Cross-cultural studies suggest, however, that the menopause is not universally a time of decline in health, and that influences other than biology, such as culture, are involved in women's menopausal transitions. Lock proposes that investigators identify factors that are associated with a positive menopausal experience, and that research take into account variables from a variety of domains including lifestyle and psychosocial as well as physiological.

9.

10.

11.

12.

13.

14.

15. 16.

17.

18. 19.

20. 21.

22.

References 1. McKinlay, S. M., and McKinlay, J. B. (1973). Selected studies of the menopause--A methodological critique. J. Biosoc. Sci. 5, 533-555. 2. Avis, N. E., Crawford, S. L., and McKinlay, S. M. (1997). Psychosocial, behavioral, and health factors related to menopause symptomatology. Women's Health: Res. Gender, Behav. Policy 3(2), 103-120. 3. McKinlay, J. B., McKinlay, S. M., and Brambilla, D. J. (1987). Health status and utilization behavior associated with menopause. Am. J. Epidemiol. 125, 110-121. 4. McKinlay, S. M. (1994). Issues in design, measurement, and analysis for menopause research. Exp. Gerontol. 29, 479-493. 5. Greendale, G. A., and Sowers, M. (1997). The menopause transition. Endocrinol. Metab. Clin. North Am. 26(2), 261-277. 6. Gath, D. (1998). The assessment of depression in peri-menopausal women. Maturitas 29, 33-39. 7. Morse, C. A., Smith, A., Dennerstein, L., Green, A., Hopper, J., and Burger, H. (1994). The treatment-seeking woman at menopause. Maturitas 29, 161-173. 8. Groeneveld, F. P. M. J., Bareman, F. P., Barensten, R., Dokter, H. J.,

23.

24.

25.

26. 27. 28. 29.

Drogendijk, A. C., and Hoes, A. W. (1993). Relationships between attitude towards menopause, well-being and medical attention among women aged 45-60 years. Maturitas 17, 77-88. Montero, I., Ruiz, I., and Hernandez, I. (1993). Social functioning as a significant factor in women's help-seeking behaviour during the climacteric period. Soc. Psychiatry Psychiatr. Epidemiol. 28, 178-183. Brambilla, D. J., and McKinlay, S. M. (1989). A prospective study of factors affecting age at menopause. J. Clin. Epidemiol. 42, 10311039. Sowers, M., Crawford, S., Sternfeld, B., Morganstein, D., Gold, E., Greendale, G., Evans, D., Neer, R., Matthews, K., Sherman, S., Lo, A., Weiss, G., and Kelsey, J. (2000). In "Menopause: Biology and Pathobiology" (R. A. Lobo, J. Kelsey, R. Marcus, eds.). Academic Press, San Diego. Rostosky, S. S., and Travis, C. B. (1996). Menopause research and the dominance of the biomedical model 1984-1994. Psychol. Woman Q. 20, 285-312. Miles, M. P., and Malik, K. C. (1994). Menopause and AfricanAmerican women: Clinical and research issues. Exp. Gerontol. 29, 511-518. Harlow, S. D., and Ephross, S. A. (1995). Epidemiology of menstruation and its relevance to women's health. Epidemiol. Rev. 17, 265286. McKinlay, S. M., Bifano, N. L., and McKinlay, J. B. (1985). Smoking and age at menopause in women. Ann. Intern. Med. 103, 350-356. Meilahn, E. N., Matthews, K. A., Egeland, G., and Kelsey, S. E (1989). Characteristics of women with hysterectomy. Maturitas 11, 319-329. Dennerstein, L., Smith, A. M. A., Morse, C., Burger, H., Green, A., Hopper, J., and Ryan, M. (1993). Menopausal symptoms in Australian women. Med. J. Aust. 159, 232-236. Kish, L. (1965). "Survey Sampling." Wiley, New York. Shelley, J. M., Green, A., Smith, A. M. A., Dudley, E., Dennerstein, L., Hopper, J., and Burger, H. (1998). Relationship of endogenous sex hormones to lipids and blood pressure in mid-aged women. Ann. Epidemiol. 8, 39-45. Ware, J. H. (1985). Linear models for the analysis of longitudinal studies. Am. Stat. 39, 95-101. Ware, J. H., Dockery, D., Louis, T. A., Xu, X. P., Ferris, B. G., Jr., and Speizer, F. E. (1990). Longitudinal and cross-sectional estimates of pulmonary function decline in never-smoking adults. Am. J. Epidemiol. 32, 685-700. McCoy, N. L. (1998). Methodological problems in the study of sexuality and the menopause. Maturitas 29, 51-60. Avis, N. E., Brambilla, D., McKinlay, S. M., and Vass, K. (1994). A longitudinal analysis of the association between menopause and depression: Results from the Massachusetts Women's Health Study. Ann. Epidemiol. 4, 214-220. Sowers, M. R., and La Pietra, M. T. (1995). Menopause: Its epidemiology and potential association with chronic diseases. Epidemiol. Rev. 17, 287-302. Davis, J. W., Ross, P. D., Wasnich, R. D., Maclean, C. J., and Vogel, J. M. (1989). Comparison of cross-sectional and longitudinal measurements of age-related changes in bone mineral content. J. Bone Miner. Res. 4, 351-357. Cramer, D. W., and Xu, H. (1996). Predicting age at menopause. Maturitas 23, 319-326. McKinlay, S. M., Brambilla, D. J., and Posner, J. G. (1992). The normal menopause transition. Am. J. Hum. Biol. 4, 37-46. Holte, A. (1998). Menopause, mood and hormone replacement therapy: Methodological issues. Maturitas 29, 5-18. Kjerulff, K. H, Guzinski, G. M., Langenberg, P. W., Stolley, P. D., Moye, N. E., and Kazandjian, V. A. (1993). Hysterectomy and race. Obstet. Gynecol. 82, 757-764.

172 30. Willett, W., Stampfer, M. J., Bain, C., Lipnick, R., Speizer, F. E., Rosner, B., Cramer, D., and Hennekens, C. H. (1983). Cigarette smoking, relative weight, and menopause. Am. J. Epidemiol. 117, 651-658. 31. Bromberger, J. T., Matthews, K. A., Kuller, L., Wing, R. R., Meilahn, E. N., and Plantinga, E (1997). Prospective study of the determinants of age at menopause. Am. J. Epidemiol. 145, 124-133. 32. Stanford, J. L., Hartge, E, Brinton, L. A., Hoover, R. N., and Brookmeyer, R. (1987). Factors influencing the age at natural menopause. J. Chronic. Dis. 409, 995-1002. 33. MacMahon, B., and Worcester, J. (1966). "Age at menopause: United States 1960-1962," Vital Health Stat. Ser. 11, No. 19, PHS Publ. no. 1000, 1-19. U.S. Govt. Printing Office, Washington, D.C. 34. Barrett-Connor, E. (1994). Heart disease in women. Fertil. Steril. 62, 127S-132S. 35. Cummings, S. R., Kelsey, J. L., Nevitt, M. C., and O'Dowd, K. J. (1985). Epidemiology of osteoporosis and osteoporotic fractures. Epidemiol. Rev. 7, 178-208. 36. Women's Health Initiative Study Group (1998). Design of the Women's Health Initiative clinical trial and observational study. Controlled Clin. Trials 19, 61 - 109. 37. Espeland, M. A., Bush, T. L., Mebane-Sims, I., Stefanick, M. L., Johnson, S., Sherwin, R., and Waclawiw, M. (1995). Rationale, design, and conduct of the PEPI trial. Controlled Clin. Trials 16, 3S-19S. 38. Greendale, G. A., Hogan, P., Kritz-Silverstein, D., Langer, R., Johnson, S. R., and Bush, T. (for the PEPI trial investigators) (1995). Age at menopause in women participating in the Postmenopausal Estrogen/Progestins Interventions (PEPI) Trial: An example of bias introduced by selection criteria. Menopause 2, 27-34. 39. Johannes, C. B., and Avis, N. E. (1996). The short-term health consequences of hysterectomy. J. Women's Health 5, 278. 40. Bush, T. L. (1990). The epidemiology of cardiovascular disease in postmenopausal women. Ann. N. Y. Acad. Sci. 592, 263-271. 41. Barrett-Connor, E. (1991). Postmenopausal estrogen and prevention bias. Ann. Intern. Med. 115, 455-456. 42. Barrett-Connor, E., and Bush, T. L. (1991). Estrogen and coronary heart disease in women. J. Am. Med. Assoc. 265, 1861-1867. 43. Barrett-Connor, E. (1996). The menopause, hormone replacement, and cardiovascular disease: The epidemiologic evidence. Maturitas 23, 227-234. 44. Hemminki, E., Malin, M., and Topo, P. (1993). Selection to postmenopausal therapy by women's characteristics. J. Clin. Epidemiol. 46, 211-219. 45. Matthews, K. A., Kuller, L. H., Wing, R. R., Meilahn, E. N., and Plantinga, P. (1996). Prior to use of estrogen replacement therapy, are users healthier than nonusers? Am. J. Epidemiol. 143, 971-978. 46. Vandenbroucke, J. P. (1995). How much of the cardioprotective effect of postmenopausal estrogens is real? Epidemiology 6, 207-208. 47. Posthuma, W. E M., Westendorp, R. G. J., and Vandenbroucke, J. P. (1994). Cardioprotective effect of hormone replacement therapy in postmenopausal women: Is the evidence biased? Br. Med. J. 308, 1268-1269. 48. Derby, C. A., Hume, A. L., Barbour, M. M., McPhillips, J. B., Lasater, T. M., and Carleton, R. A. (1993). Correlates ofpostmenopausal estrogen use and trends through the 1980s in two southeastern New England communities. Am. J. Epidemiol. 137, 1125-1135. 49. Johannes, C. B., Crawford, S. L., Posner, J. G., and McKinlay, S. M. (1994). Longitudinal patterns and correlates of hormone replacement therapy use in middle-aged women. Am. J. Epidemiol. 140, 439-452. 50. Rosenberg, L. (1993). Hormone replacement therapy; The need for reconsideration. Am. J. Public Health 83, 1670-1673. 51. Nachtigall, L. E., Nachtigall, R. H., Nachtigall, R. D., and Beckman, E. M. (1979). Estrogen replacement therapy I: A 10-year prospective study in the relationship to osteoporosis. J. Am. Coil. Obstet. Gynecol. 53, 277-281. 52. Criqui, M. H., Suarez, L. Barrett-Connor, E., McPhillips, J., Wingard,

SYBIL L. CRAWFORD

53.

54.

55.

56. 57.

58.

59.

60.

61.

62.

63.

64.

65. 66. 67.

D. L., and Garland, C. (1988). Postmenopausal estrogen use and mortality: Results from a prospective study in a defined, homogeneous community. Am. J. Epidemiol. 182, 606-614. Sturgeon, S. R., Schairer, C., Brinton, L. A., Pearson, T., and Hoover, R. N. (1995). Evidence of a healthy estrogen user survivor effect. Epidemiology 6, 227-231. Petitti, D. B. (1994). Coronary heart disease and estrogen replacement therapy: Can compliance bias explain the results of observational studies? Ann. Epidemiol. 4, 115-118. Gallagher, E. J., Viscoli, C. M., and Horwitz, R. I. (1993). The relationship of treatment adherence to the risk of death after myocardial infarction in women. J. Am. Med. Assoc. 270, 742-744. Vandenbroucke, J. E (1991). Postmenopausal oestrogen and cardioprotection. Lancet 337, 833- 834. Grady, D., Applegate, W., Bush, T., Furberg, C., Riggs, B., and Hulley, S. B. (1998). Heart and Estrogen/progestin Replacement Study (HERS): Design, methods, and baseline characteristics. Controlled Clin. Trials 19, 314-335. Johnson, S., Mebane-Sims, I., Hogan, P. E., and Stoy, D. B. (1995). Recruitment of postmenopausal women in the PEPI trial. Controlled Clin. Trials 16, 20S-35S. Bastian, L. A., Couchman, G. M., Rimer, B. K., McBride, C. M., Feaganes, J. R., and Siegler, I. C. (1997). Perceptions of menopausal stage and patterns of hormone replacement therapy use. J. Women's Health 6, 467-475. Brett, K. M. and Madans, J. H. (1997). Use of postmenopausal hormone replacement therapy: Estimates from a nationally representative cohort study. Am. J. Epidemiol. 145, 536-545. Hernandez-Avila, M., Stampfer, M. J., Ravnikar, V. A., Willett, W. C., Schiff, I., Francis, M., Longcope, C., and McKinlay, S. M. (1993). Caffeine and other predictors of bone density among pre- and perimenopausal women. Epidemiology 4, 128-134. Michelson, D., Stratakis, C., Hill, L., Reynolds, J., Galliven, E., Chrousos, G., and Gold, P. (1996). Bone mineral density in women with depression. N. Engl. J. Med. 335, 1176-1181. Zhang, Y., Kiel, D. P., Kreger, B. E., Cupples, L. A., Ellison, R. C., Dorgan, J. F., Schatzkin, A., Levy, D., and Felson, D. T. (1997). Bone mass and the risk of breast cancer among postmenopausal women. N. Engl. J. Med. 336, 611 - 617. Brambilla, D. J., and McKinlay, S. M. (1987). A comparison of responses to mailed questionnaires and telephone interviews in a mixed mode health survey. Am. J. Epidemiol. 126, 962-971. McKinlay, S. M., and McKinlay, J. B. (1986). Aging in a "healthy" population. Soc. Sci. Med. 236, 531-535. Snowden, R. (1977). The statistical analysis of menstrual bleeding patterns. J. Biosoc. Sci. 9, 107-120. Belsey, E. M., and Farley, T. M. M. (1987). The analysis of menstrual bleeding patterns: A review. Appl. Stochastic Models Data Analy. 3, 125-150.

68. Bean, J. A., Leper, J. D., Wallace, R. B., Sherman, B. M., and Jagger, H. (1979). Variation in the reporting of menstrual histories. Am. J. Epidemiol. 109, 181-185. 69. Kaufert, P. A., Gilbert, P., and Hassard, T. (1988). Researching the symptoms of menopause: An exercise in methodology. Maturitas 10, 117-131. 70. Johannes, C. B., Crawford, S. L., Woods, J., Goldstein, R. B., Tran, D., Mehrotra, S., Johnson, K. B., and Santoro, N. (2000). An electronic menstrual cycle calendar: Comparison of data quality with a paper version. Menopause, in press. 71. Burger, H. C. (1994). The menopause: When is it all over oris it? Aust. N.Z. J. Obstet. Gynaecol. 34, 293-295. 72. Burger, H. G. (1994). Diagnostic role of follicle-stimulating hormone (FSH) measurements during the menopausal transitionmAn analysis of FSH, oestradiol and inhibin. Eur. J. Endocrinol. 130, 38-42. 73. Hee, J., MacNaughton, J., Banagh, M., and Burger, H. G. (1993).

173

CHAPTER 10 M e t h o d o l o g i c Challenges in M e n o p a u s e Studies

74.

75. 76.

77.

78.

79. 80.

81.

82.

83.

84.

85.

86.

87. 88.

89.

90. 91.

92.

93.

94.

Perimenopausal patterns of gonadotrophins, immunoreactive inhibin, oestradiol and progesterone. Maturitas 18, 9-20. Metcalf, M. G., Donald, R. A., and Livesey, J. H. (1982). Pituitaryovarian function before, during, and after the menopause: A longitudinal study. Clin. Endocrinol. 17, 489-494. Metcalf, M. G. (1988). The approach of menopause: A New Zealand study. N.Z. Med. J. 101, 103-106. Reame, N. E., Kelche, R. P., Beitins, I. Z., Yu, M. Y., Zawacki, C. M., and Padmanabhan, V. (1996). Age effects of follicle-stimulating hormone and pulsatile luteinizing hormone secretion across the menstrual cycle of women. J. Clin. Endocrinol. Metab. 81, 1512-1518. Bungay, G. T., Vessey, M. P., and McPherson, C. K. (1980). Study of symptoms in middle life with special reference to the menopause. Br. Med. J. 281, 181-183. Morabia, A., and Flandre, P. (1992). Misclassification bias related to definition of menopausal status in case-control studies of breast cancer. Int. J. Epidemiol. 21,222-228. World Health Organization Scientific Group (1981). "Research on the Menopause," WHO Tech. Serv. Rep. Ser. 670. WHO, Geneva. World Health Organization Scientific Group (1996). "Research on the Menopause in the 1990s," WHO Tech. Serv. Rep. Ser. No. 866. WHO, Geneva. Kaufert, P., Lock, M., McKinlay, S., Beyenne, Y., Coope, J., Davis, D., Eliasson, M., Gognalons-Nicolet, M., Goodman, M., and Holte, A. (1986). Menopause research: The Korpilampi workshop. Soc. Sci. Med. 22, 1285-1289. Brambilla, D. J., McKinlay, S. M., and Johannes, C. B. (1994). Defining the perimenopause for application in epidemiologic investigations. Am. J. Epidemiol. 140, 1091-1095. Dudley, E. C., Hopper, J. L., Taffe, J., Guthrie, J. R., Burger, H. G., and Dennerstein, L. (1998). Using longitudinal data to define the perimenopause by menstrual cycle characteristics. Climacteric 1, 18-25. Cooper, G. S., and Baird, D. D. (1995). The use of questionnaire data to classify peri- and premenopausal status. Epidemiology 6, 6 2 5 628. Wing, R. R., Matthews, K. A., Kuller, L. H., Meilahn, E. N., and Plantinga, P. (1991). Weight gain at the time of menopause. Arch. Intern. Med. 151, 97-102. Shideler, S. E., DeVane, G. W., Kalra, P. S., Benirschke, K., and Lasley, B. L. (1989). Ovarian-pituitary hormone interactions during the perimenopause. Maturitas 11, 331-339. Burger, H. G. (1996). The endocrinology of the menopause. Maturitas 23, 129-136. Wide, L., Nillis, S. J., Gemzell, C., and Roos, P. (1973). Radioimmunosorbent assay of follicle-stimulating hormone and luteinizing hormone in serum and urine from men and women. Acta Endocrinol. (Copenhagen) 174, 3-58. Chakravarti, S., Collins, W. P., Forecast, J. D., Newton, J. R., Oram, D. H., and Studd, J. W. (1976). Hormonal profiles after the menopause. Br. Med. J. 2, 784-786. Wilson, J. D., and Foster, D. W., eds. (1992). "William's Textbook of Endocrinology, 8th Ed." Saunders, Philadelphia. Goldenberg, R. L., Grodin, J. M., Rodbard, D., and Ross, G. T. (1973). Gonadotropins in women with amenorrhea. Am. J. Obstet. Gynecol. 116, 1003-1012. Nordin, B. E. C., Morris, H. A., Need, A. G., Horowitz, M., and Robertson, W. G. (1990). Relationship between plasma calcium fractions, other bone-related variables, and serum follicle-stimulating hormone levels in premenopausal, perimenopausal, and postmenopausal women. Am. J. Obstet. Gynecol. 163, 140-145. Lenton, E. A., Sexton, L., Lee, S., and Cooke, I. D. (1988). Progressive changes in LH and FSH and LH:FSH ratio in women throughout reproductive life. Maturitas 10, 35-43. MacNaughton, J., Banah, M., McCloud, P., Hee, J., and Burger, H.

95.

96.

97.

98.

99. 100.

101. 102.

103. 104. 105. 106.

107.

108.

109.

110.

(1992). Age related changes in follicle stimulating hormone, luteinizing hormone, oestradiol and immunoreactive inhibin in women of reproductive age. Clin. Endocrinol. 36, 339-345. Reyes, F. I., Winter, J. S. D., and Faiman, C. (1977). Pituitary-ovarian relationship preceding the menopause. Am. J. Obstet. Gynecol. 129, 557-564. Metcalf, M. G., and Livesey, J. H. (1985). Gonadotrophic excretion in fertile women: Effect of age and the onset of the menopause. J. Endocrinol. 105, 357-362. Sherman, B. M., and Korenman, S. G. (1975). Hormonal characteristics of the human menstrual cycle throughout reproductive life. J. Clin. Invest. 55, 699-706. Stellato, R., Crawford, S., McKinlay, S., and Longcope, C. (1998). Can follicular stimulating hormone be used to define menopause? Endocr. Pract. 4, 137-141. Taffe, J., Garamszegi, C., Dudley, E., and Dennerstein, L. (1997). Determinants of self rated menopause status. Maturitas 27, 223-229. Paganini-Hill, A., Krailo, M. D., and Pike, M. C. (1984). Age at natural menopause and breast cancer: The effect of errors in recall. Am. J. Epidemiol. 119, 81-85. McKinlay, S. M., Jefferys, M., and Thompson, B. (1972). An investigation of the age at menopause. J. Biosoc. Sci. 4, 161-173. Benjamin, F. (1960). The age of the menarche and of the menopause in white South African women and certain factors influencing these times. S. Afr. Med. J. 34, 316-320. Treloar, A. E. (1974). Menarche, menopause and intervening fecundability. Hum. Biol. 46, 89-107. Treloar, A. E. (1981). Menstrual cyclicity and the pre-menopause. Maturitas 3, 249-264. Kaufert, P. A., Gilbert, P., and Tate, R. (1987). Defining menopausal status: The impact of longitudinal data. Maturitas 9, 217-226. Johannes, C. B., Crawford, S. L., Longcope, C., and McKinlay, S. M. (1996). Bleeding patterns and changes in the perimenopause: A longitudinal characterization of menstrual cycles. Clin. Consult. Obset. Gynecol. 8, 9-20. Kuller, L. H., Meilahn, E. N., Cauley, J. A., Gutai, J. P., and Matthews, K. A. (1994). Epidemiologic studies of menopause: Changes in risk factors and disease. Exp. Gerontol. 29, 495-509. den Tonkelaar, I. (1997). Validity and reproducibility of self-reported age at menopause in women participating in the DOM-project. Maturitas 27, 117-123. Colditz, G. A., Stampfer, M. J., Willett, W. C., Stason, W. B., Rosner, B., Hennekens, C. H., and Speizer, F. E. (1987). Reproducibility and validity of self-reported menopausal status in a prospective cohort study. Am. J. Epidemiol. 126, 319-325. Paganini-Hill, A., and Ross, R. K. (1982). Reliability of recall of drug usage and other health-related information. Am. J. Epidemiol. 116, 114-122.

111. Hahn, R. A., Eaker, E., and Rolka, H. (1997). Reliability of reported age at menopause. Am. J. Epidemiol. 146, 771-775. 112. Horwitz, R. I., and Yu, E. C. (1985). Problems and proposals for interview data in epidemiological research. Int. J. Epidemiol. 14, 4 6 3 467. 113. Greendale, G. A., James, M. K., Espeland, M. A., and Barrett-Connor, E. (1997). Can we measure prior postmenopausal estrogen/progestin use? Am. J. Epidemiol. 146, 763-770. 114. Goodman, M. T., Nomura, A. M. Y., Wilkens, L. R., and Kolonel, L. N. (1990). Agreement between interview information and physician records on history of menopausal estrogen use. Am. J. Epidemiol. 131, 815-825. 115. Jannausch, M. L. and Sowers, M. R. (1992). Consistency ofperimenopausal estrogen use reporting by women in a population-based prospective study. Maturitas 14, 161-169. 116. Creinin, M. D. (1996). Laboratory criteria for menopause in women using oral contraceptives. Fertil. Steril. 66, 101-104.

174 117. Schiff, I. (1980). The effects of conjugated estrogens on gonadotropins. Fertil. Steril. 33, 333-334. 118. Mathur, R. S., Landgrebe, S. C., Moody, L. O., Semmens, J. E, and Williamson, H. O. (1985). The effect of estrogen treatment on plasma concentrations of steroid hormones, gonadotropins, prolactin and sex hormone-binding globulin in post-menopausal women. Maturitas 7, 129-133. 119. Powers, M. S., Schenkel, L., Darley, E E., Good, W. R., Balestra, J. C., and Place, V. A. (1985). Pharmacokinetics and pharmacodynamics of transdermal dosage forms of 17fi-estradiol: Comparison with conventional oral estrogens used for estrogen replacement. Am. J. Obstet. Gynecol. 152, 1099-1106. 120. Siddle, N., Sarrel, E, and Whitehead, M. (1987). The effect of hysterectomy on the age at ovarian failure: Identification of a subgroup of women with premature loss of ovarian function and literature review. Fertil. Steril. 47, 94-100. 121. Greene, J. G. (1998). Constructing a standard climacteric scale. Maturitas 29, 25-31. 122. Alder, E. (1998). The Blatt-Kupperman menopausal index: A critique. Maturitas 29, 19-24. 123. Martin, M. C., Block, J. E., Sanchez, S. D., Arnaud, C. D., and Beyene, Y. (1993). Menopause without symptoms: The endocrinology of menopause among rural Mayan Indians. Am. J. Obstet. Gynecol. 168, 1839-1845. 124. Beyene, Y. (1986). The cultural significance and physiological manifestation of menopause: A biocultural analysis. Cult. Med. Psychiatry 10, 47-71. 125. Lock, M. (1998). Menopause: Lessons from anthropology. Psychosom. Med. 60, 410-419. 126. Lock, M., Kaufert, E, and Gilbert, E (1988). Cultural construction of the menopausal syndrome: The Japanese case. Maturitas 10, 317-332. 127. Morgan, B. J. T. (1992). "Analysis of Quantal Response Data." Chapman & Hall, London. 128. Krailo, M. D., and Pike, M. C. (1983). Estimation of the distribution of age at natural menopause from prevalence data. Am. J. Epidemiol. 117,356-361. 129. Shinberg, D. S. (1998). An event history analysis of age at least menstrual period: Correlates of natural and surgical menopause among midlife Wisconsin women. Soc. Sci. Med. 46, 1381-1396. 130. Harlow, S. D., and Zeger, S. L. (1991). An application of longitudinal methods to the analysis of menstrual diary data. J. Clin. Epidemiol. 44, 1015-1025. 131. Harlow, S. D., and Matanoski, G. M. (1991 ). The association between weight, physical activity, and stress and variation in the length of the menstrual cycle. Am. J. Epidemiol. 133, 38-49. 132. Harlow, S. D., and Zeger, S. L. (1991). An application of longitudinal methods to the analysis of menstrual diary data. J. Clin. Epidemiol. 44, 1015-1025. 133. Murphy, S. A., Bentley, G. R., and O'Hanesian, M. A. (1995). An analysis for time-varying covariates. Stat. Med. 14, 1843-1857.

SYBIL L. CRAWFORD 134. Harlow, S. D., and Campbell, B. (1996). Ethnic differences in the duration and amount of menstrual bleeding during the postmenarcheal period. Am. J. Epidemiol. 144, 980-989. 135. Harlow, S. D., and Campbell, B. C. (1994). Host factors that influence the duration of menstrual bleeding. Epidemiology 5, 352-355. 136. Collett, D., and Weerasooriya, N. (1993). A modelling approach to the analysis of menstrual diary data. Stat. Med. 12, 955-965. 137. Treloar, A. E., Boynton, R. E., Behn, B. G., and Brown, B. W. (1967). Variation of the human menstrual cycle through reproductive life. Int. J. Fertil. 12, 77-126. 138. Rodriguez, G., Faundes-Latham, A., and Atkinson, L. E. (1976). An approach to the analysis of menstrual patterns in the critical evaluation of contraceptives. Stud. Fam. Plann. 7, 42-51. 139. van Beresteijn, E. C. H., Korevaar, J. C., Huijbregts, P. C. W., Schouten, E. G., Burema, J., and Kok, F. J. (1993). Perimenopausal increase in serum cholesterol: A 10-year longitudinal study. Am. J. Epidemiol. 137, 383-392. 140. Jensen, J., Nilas, L., and Christiansen, C. (1990). Influence of menopause on serum lipids and lipoproteins. Maturitas 12, 321-331. 141. Falch, J. A., and Sandvik, L. (1990). Perimenopausal appendicular bone loss: A 10-year prospective study. Bone 11,425-428. 142. Crawford, S. L., Casey, V. A., Avis, N. E., and McKinlay, S. M. (2000). A longitudinal study of weight and the menopause transition: Results from the Massachusetts Women's Health Study. Menopause, in press. 143. Longcope, C., Herbert, E N., McKinlay, S. M., and Goldfield, S. R. (1990). The relationship of total and free estrogens and sex hormonebinding globulin with lipoproteins in women. J. Clin. Endocrinol. Metab. 71, 76-72. 144. Little, R. J. A., and Rubin, D. B. (1987). "Statistical Analysis with Missing Data." Wiley, New York. 145. Rubin, D. B. (1987). "Multiple Imputation for Nonresponse in Surveys." Wiley, New York. 146. National Center for Health Statistics, Pokras, R., and Hufnagel, V. G. (1987). "Hysterectomies in the United States, 1965-84," Vital Health Stat., Ser. 13, No. 92, DHHS Publ. no. (PHS) 88-1753. U.S. Gov. Printing Office, Washington, DC. 147. Pike, M. C., Ross, R. K., and Spicer, D. V. (1998). Problems involved in including women with simple hysterectomy in epidemiologic studies measuring the effects of hormone replacement therapy on breast cancer risk. Am. J. Epidemiol. 147, 718-721. 148. Schmidt, E J., and Rubinow, D. R. (1991). Menopause-related affective disorders: A justification for further study. Am. J. Psychiatry 148, 844-852. 149. Brincat, M., Magos, A., Studd, J. W., Cardozo, L. D., O'Dowd, T., Wardle, P. J., and Cooper, D. (1984). Subcutaneous hormone implants for the control of climacteric symptoms: A prospective study. Lancet 1, 16-18. 150. Stampfer, M. J., Colditz, G. A., and Willett, W. C. (1990). Menopause and heart disease: A review. Ann. N. Y Acad. Sci. 592, 286-294.

~HAPTER 1

SWAN: A Multi c enter,

Multiethnic, CommunityBased Cohort Study

of Women and the Menopausal Transition MARYFRAN SOWERS,* SYBIL L. CRAWFORD, t BARBARA STERNFELD,; DAVID M O R G A N S T E I N , wE L L E N B . G O L D , G A I L A . G R E E N D A L E , # D E N I S E V A N S , * * R O B E R T N E E R , tt K A R E N M A T T H E W S , ~ S H E R R Y S H E R M A N , w167 ANNIE LO, wGERSON WEISS, A N D J E N N I F E R K E L S E Y ## *Department of Epidemiology, School of Public Health, University of Michigan, Ann Arbor, Michigan 48109; tNew England Research Institutes, Watertown, Massachusetts 02472; *Department of Epidemiology and Biostatistics, Division of Research, Kaiser Permanente, Oakland, California 94611; ~Westat, Inc., Rockville, Maryland 20850; IIDepartmentof Epidemiology and Preventive Medicine, School of Medicine, University of California, Davis, Davis, California 95616; #Departments of Medicine and Obstetrics and Gynecology, University of California, Los Angeles, School of Medicine, Los Angeles, California 90024; **Rush Institute on Aging, Chicago, Illinois 60612; ttDivision of Endocrinology, Department of Medicine, Massachusetts General Hospital, Boston, Massachusetts 02114; **Departmentof Psychiatry, University of Pittsburgh, Pittsburgh, Pennsylvania 15261; ~NIH/NIA, Bethesda, Maryland 20892; IIIIDepartmentof Obstetrics and Gynecology, University of Medicine and Dentistry of New Jersey, Newark, New Jersey 07103; and ##Division of Epidemiology, Department of Health Research & Policy, Stanford University, School of Medicine, Stanford, California 94305

I. II. III. IV. V. VI.

Appendix A. SWAN Investigators Appendix B. Specific Sampling and Recruiting Strategies by Sites with List-Based Primary Sampling Frames Appendix C. Specific Sampling and Recruiting Strategies by Sites with RDD-Based Primary Sampling Frames References

Introduction Overview of the Study Design Data Collection Sampling and Recruitment Strengths and Limitations of SWAN Summary


completely understood [1,2]. Furthermore, much of what is known is based on data from Caucasian women, from women who are self-referred to menopause clinics, or from convenience samples of women seen in the clinical setting

Menopause is a universal phenomenon of women, yet, as discussed in other chapters in this book, it is inMENOPAUSE:


175


176 for other health problems. In the next two decades, approximately 40 million American women will experience the menopause [3] and by the year 2005, it is estimated that $3-5 billion will be spent annually for hormone replacement therapy (HRT) and the physician monitoring of that HRT use [4]. Additionally, study of the menopause poses special methodological challenges because of its transitional nature, the potential for involving multiple organ systems (i.e., bone, lipids, mental health), and the potential contribution of varied social, behavioral, and cultural factors (see Chapter 10). Thus, study of the menopausal transition is both important and complex. To address many of the knowledge deficits about the menopausal transition identified in chapters of this book, the Study of Women's Health Across the Nation (SWAN), a multisite, longitudinal study of the natural history of menopause, was funded by the National Institute on Aging, the National Institute of Nursing Research, and the Office of Research on Women's Health. The overall goal of SWAN is to describe the chronology of the biological and psychosocial characteristics of the menopausal transition and the effect of this transition on subsequent health and risk factors for agerelated chronic diseases. Because investigation of the menopausal experience in minority women has been neglected, SWAN placed special emphasis on including minority populations. This would allow SWAN to describe the sociocultural, lifestyle, psychological, and biological characteristics of these groups in relation to the menopausal transition [5]. In addition, emphasis was placed on recruiting a sample of women that was community or population based, rather than volunteer or clinically based, so that the sample would be representative of women from the full spectrum of socioeconomic status and cultural experiences. The specific aims of SWAN, shown below, are being addressed in a representative cohort of initially premenopausal women who are socially and culturally diverse. The aims are as follows: 1. To characterize the symptomatology, hormonal, and bleeding pattern characteristics related to the menopausal transition. 2. To investigate the hormonal and menstrual bleeding pattern characteristics related to change in bone mineral density, cardiovascular status markers, measures of carbohydrate metabolism, and body composition during the menopausal transition. 3. To examine the relations of psychosocial factors, personality characteristics, and behaviors, including lifestyle behaviors, as they may relate to age at onset, symptoms, and physiological changes of the menopausal transition. 4. To discern what changes observed over time are related to the menopausal transition as compared to age-related changes, including those changes that appear to accelerate the aging process.

SOWERS ET AL.

5. to describe and quantify cultural and ethnic differences among women with respect to midlife aging and the menopausal transition among the four race/ethnic groups of the cohort, in addition to non-Hispanic Caucasians. This chapter is an overview of the SWAN study design and includes a brief description of SWAN's comprehensive data collection. The data being collected mirrors the specific aims, reflecting the belief that the biologic process of menopause occurs within the context of diverse personality characteristics, psychosocial factors, and behavioral attributes as well as an ethnic and cultural context. Consequently, the methods used to recruit this important sample of multiethnic women are described and the strengths and limitations of those methods are discussed.

II. O V E R V I E W

OF THE STUDY DESIGN

SWAN is organized as a prospective, multicenter, multiethnic, multidisciplinary study of the natural history of the menopausal transition, under the auspices of a cooperative agreement between the National Institutes of Health and seven sites with clinical examination facilities, a data coordinating center, and two laboratories. Appendix 1 shows the locations, investigators, and roles of those investigators. SWAN includes a large and representative sample of African-American, non-Hispanic Caucasian, Chinese, Hispanic, and Japanese women. The study design, developed in a collaborative process, consists of a cross-sectional study and a longitudinal cohort study, both of which employ common protocols across the seven sites with clinical examination facilities. Focus groups were conducted to inform the development of the study design and the protocols and to ensure the relevance and the appropriateness of the protocols to the multiethnic cohort. The SWAN Cross-sectional Study consisted of a 15- to a 20-minute telephone interview (or face-to face interview in those instances in which no telephone number could be associated with the sampled respondent). The interview was administered to 16,065 women aged 40-55 years who were randomly selected from sampling frames established at each site with clinical examination facilities (described more fully in Section IV). The two purposes of the SWAN Cross-sectional Study were to identify women eligible for study longitudinally and assess, cross-sectionally, those factors associated with the age at natural menopause, the prevalence of surgical menopause, symptoms of menopause, health status, and health care use. Additional information about the eligibility criteria, sampling frames, and characteristics of participants are discussed in Section IV. On completion of the interview, eligibility for the longitudinal study was determined and women meeting the eligibility criteria were invited to join that cohort. The annual examinations of the SWAN Longitudinal Study include

CHAPTER 11 SWAN Cohort Study TABLE I

177

The Breadth of Measures and Their Frequency of Ascertainment in the SWAN Longitudinal Study

Type of measurement

Frequency a

Questionnaire

Type of measurement

Frequency

a

Specimen data b

Socioeconomic status Medical history Psychosocial environment Lifestyle behaviors Menstrual status Natural/surgical menopause Symptoms Use of medical services Use of medications Quality of life Sexual activity Food frequency

Annual Annual Annual Annual Annual Annual Annual Annual Annual Annual Annual Base line, F/U-04

Clinic measurements

Blood (serum) E 2, FSH, SHBG, DHEAS, testosterone TSH Glucose and insulin Fibrinogen, factor VII, PAI-1, TPA antigen Lipid profile, HDL subfractions, lipoprotein (a) Biochemical bone turnover markers (at five sites) Serum repository specimens

Annual Base line Annual Base line Base line, F/U-01, 02, 03, 05 Base line, F/U-01, 02, 03 Annual

Urine N-Teleopeptides (at five sites) Urine repository specimens

Annual Annual

Other data collection beyond annual evaluation

Anthropometry Blood pressure Bone density (at five sites)

Annual Annual Annual

Abstract medical records for hysterectomy Menstrual calendars Daily Hormone Study (subsample of 900)

Monthly One cycle, annually

a Note: F/U denotes a follow-up examination; the number denotes which follow-up examination. bAbbreviations: E2, estradiol; FSH, follicle-stimulating hormone; SHBG, sex hormone binding globulin; DHEAS, dehydroepiandrosterone sulfate; TSH, thyroid-stimulating hormone; PAI-1, plasminogen activator inhibitor-1; TPA, tissue plasminogen activator; HDL, high-density lipoprotein.

questionnaires, blood and urine specimen collection, and physical measures (Table I). Because the SWAN Longitudinal Study is focused on the menopausal transition, unique data collection activities are required. For example, the annual examinations are scheduled for days 2 - 5 after bleeding c o m m e n c e s to standardize serum h o r m o n e measures to the early phase of the menstrual cycle. In addition, the cohort is followed with monthly menstrual cycle calendars and a subset of the cohort participates in daily urine collection as well as maintaining a daily symptom diary for one complete menstrual cycle on an annual basis. The following section describes the data collection more fully.

III.

DATA A.

COLLECTION Theoretical

Approaches

As a multidisciplinary study, the SWAN data collection instruments and approaches were developed to address the potential contribution of the multiple theories surrounding the study of the menopausal transition [6]. For example, the biological approach ascribes the experience of the menopause particularly within the framework of alterations in metabolism and endocrine status. The psychological~psychosocial approach maintains the importance of stressors and losses as catalysts for symptoms. The sociocultural/environmental approach indicates that cultural constructs and lifestyle factors define our responses toward the menopause and the presentation of potential symptoms. Finally, the feminist

theory views the menopause as a normal developmental stage with its own unique challenges. The instruments and data collection activities of SWAN have reflected an inclusive approach that acknowledges the need for and value of each of these perspectives, while minimizing the reductionist approach to studying and interpreting the characteristics of the menopause transition.

B. Types of Data The types of data collected from SWAN participants in the annual examinations are shown below in examples that include the type of construct and contributing variables. Construct

Variable

Acculturation

Language used, cultural and religious practices, dietary practices Weight changes associated with each pregnancy, weight cycling Use of contraceptive methodologies Use of hormone preparations, past use of oral contraceptives, and current contraception methodology Smoking history and current passive smoke exposure; current caffeine and alcohol consumption; diet and dietary practices, including use of supplements; amount and frequency of physical activity practices, including planned exercise

Body size history Contraception Hormone use practices

Lifestyle behaviors

178

SOWERS ET AL.

Construct

Variable

Construct

Variable

Menstrual status

Current menstrual bleeding characteristics and their variation according to timing, duration, and intensity; usual premenstrual symptoms, if any

Bone status and its turnover

Psychological status

Depression, hostility, and stress

Bone mineral density of the femoral head, lumbar spine, and total body (from five sites with bone densitometry facilities); biochemical measures of bone formation and resorption

Recent medical care utilization

Frequency of prevention behaviors, including Pap smear, physical breast exam, and physician visit for health problem or routine check-up; use of complementary and alternative health approaches; health insurance

Carbohydrate and energy metabolism

Glucose, insulin and thyroid-stimulating hormone concentrations (the latter at base line)

Clotting factors

Fibrinogen, factor VIIc, plasminogen activator inhibitor-1, tissue plasminogen activator antigen

Relationships

Number, nature, and satisfaction with relationships; life satisfaction

Lipid metabolism

Reproductive history

Age at menarche, gravidity, parity, pregnancy losses, infertility, lactation practices

Total cholesterol, triglycerides, highand low-density lipoprotein cholesterol, high-density lipoprotein cholesterol subfractions, lipoprotein (a)

Reproductive hormones

Self-perception

Quality of life, health status, degree of physical activity

Estradiol, follicle-stimulating hormone, sex hormone binding globulin, progesterone, and testosterone

Sexuality

Types of practices, satisfaction, and attitudes toward sex

Significant life events

Marriage, divorce, death or birth in family, change in or loss of job, illness, social support, occupational stress (autonomy)

Significant medical history

Diagnosis by a physician of hypertension, cardiovascular disease, malignancies, or thyroid disease; fractures, pelvic surgery, urinary incontinence, current medications, family history of health events

Sociodemographic status

Age, birth date, birthplace, marital status, level of education, income of household, occupation and the physicality of one's work, household composition

Social environment

Discrimination, religiosity, and spirituality

The interview data will be linked with other information being collected annually that describes the physical and hormone status of enrollees. The general areas of interest and the variables that contribute to the constructs are shown below.

Construct

Variable

Blood pressure

Resting systolic and diastolic blood pressure, resting heart rate

Body composition and body topology

Weight, height, waist and hip circumference, body composition (the latter from five sites with bone densitometry facilities)

Two additional data collection elements, monthly menstrual cycle calendars and daily specimen/diary collection, are important in more precisely characterizing the transitional process. The monthly menstrual calendars provide a record of the changing characteristics of menstrual bleeding from month to month. These monthly calendars also include a record of the use of oral contraceptives or other hormones, symptoms, and the occurrence of any gynecological surgery or procedures. A more extensive calendar is in use at three clinical sites to ascertain lifestyle factors including dieting, shift work, exercise practice, and smoking behavior, as well as stress. A Daily H o r m o n e Study (DHS) contributes to the SWAN specific aims by providing a more complete understanding of the variation in hormone concentrations throughout menstrual cycles (or equivalent time periods) of the perimenopausal transition and characterizing changes in the nature and frequency of within-cycle events, such as ovulation. Blood and urine specimens are being collected from a subsample of 900 women, with participation at each of the seven sites and from each of the race/ethnic groups as well as the non-Hispanic Caucasian women. Participants collect daily urine specimens for one complete menstrual cycle each year. These urine specimens are assayed for excretion products of the pituitary (the gonadotropins, luteinizing hormone, and follicle-stimulating hormone) and the ovary (estrone conjugate and pregnanediol glucuronide). The goal is to describe the changes in the hormone concentrations at important points during the menopausal transition and prior to the final menstrual period. The DHS also includes a daily diary to characterize symptoms and social dimensions of each day during the cycle of the daily urine collection.

CHAPTER 11 SWAN Cohort Study

179

TABLE II The Site-Specific Recruiting Goals for Race/Ethnicity Percentage in the SWAN Longitudinal Study of the Menopausal Transition in Seven Geographic Locales a Primary race/ethnic self-identification (%) AfricanAmerican

Geographic locale Detroit, Michigan (Ypsilanti/Inkster) Chicago, Illinois (Morgan Park/Beverly) Boston/Cambridge, Massachusetts Pittsburgh, Pennsylvania (Allegheny County) Oakland, California (plus Hayward and Richmond) Newark, New Jersey (Hudson County) Los Angeles, California (South Bay/Sawtelle)

Chinese Hispanic Japanese Caucasian

66 55 45 33

33 45 55 66 45 33 45

55 66 55

a Each site was to recruit at least 450 women to the SWAN Longitudinal Study with the proportion of primary race/ ethnicity among women described in the table.

Collectively, the monthly menstrual calendars and the Daily Hormone Study help to refine the definition of the menopause by providing more frequent and supplemental data during the transitional period. It is anticipated that an outgrowth will be the provision of more comprehensive understanding of the bleeding and hormone markers of the onset of perimenopause and the stages within the transition process.

recruitment strategy that they considered optimal for the Study's scientific questions, the characteristics of the local site (including access to clinical facilities), and the specific minority population to be evaluated. The result was the use of multiple sampling frames and multiple sampling approaches implemented in a coordinated manner. SWAN thus also provides the opportunity to describe and evaluate the various sampling frames, the sampling approaches to recruiting women from those frames, and the relative impact of using the various sampling frames and approaches.

IV. SAMPLING AND RECRUITMENT A. Overview

B.

The SWAN sampling and recruiting was implemented in seven locations in the United States: Boston, Chicago, the Detroit area, Los Angeles, Newark, Pittsburgh, and Oakland, California. The recruitment goal for each of the seven sites was to obtain representative samples of at least 450 women [non-Hispanic Caucasian women and one designated minority group (African-American, Chinese, Hispanic, and Japanese)] in a proportion specific for each site (see Table II). To achieve this goal, each site developed a sampling and

SWAN Recruitment

As indicated previously, recruitment for SWAN was undertaken as a two-step process (Table III). The first recruitment step involved a cross-sectional study to act as a sampiing frame for the SWAN Longitudinal Study. The second recruitment step was the development of a longitudinal study cohort from among the SWAN Cross-sectional Study enrollees. To be eligible for participation in SWAN Cross-sectional Study, women had to meet the following criteria:

TABLE III Summary of Sampling Units Contacted to Determine Eligibility in the Two-Step Process to Identify SWAN Longitudinal Study Enrollees Recruitment step Cross-sectional study recruitment, sampling units contacted Longitudinal study recruitment, units from the cross-sectional study

Sampling units

No. eligible

No. recruited

Response rate (%)

202,985

34,985

16,065

46.6

3,306

50.7

16,065

6,521 a

aThere were 36 Caucasian women included in this table who were "filtered out" (i.e., eligible to enter the cohort, but not recruited because target recruitment had been met).

180

S O W E R S ET AL.

1. Primary residence in designated geographic area 2. Ablility to speak English or designated other language m Spanish, Cantonese, or Japanese 3. Age 4 0 - 5 5 years at time of contact 4. Cognitive ability to provide verbal informed consent 5. Membership in a specific site's targeted ethnic groups To identify women eligible for the cross-sectional study, sites screened the constituent sampling units from the sampiing frames. Depending on the site, the sampling units were the households, telephone numbers, or individual names of women; the sampling frames were the listings of the sampiing units. Study-wide, 202,485 sampling units from sampiing frames were evaluated, leading to the identification of 34,446 eligible women. Of these, 16,065 women completed the SWAN Cross-sectional Study. The eligibility criteria for the SWAN Longitudinal Study wereas follows: 1. Aged 4 2 - 5 2 years 2. No surgical removal of the uterus and/or both ovaries 3. Not currently using exogenous hormone preparations affecting ovarian function 4. At least one menstrual period in the previous 3 months 5. Self-identification with one of each site's designated race/ ethnic group From the SWAN Cross-sectional Study, 6557 women were identified as eligible for longitudinal study. Of these women, 36 Caucasian women were "filtered out," i.e., they were not asked to participate in the longitudinal study because the site had met its target sample size. Of the remaining 6521 women, 3306 were recruited for the SWAN Longitudinal Study (see Table IV). This is the cohort currently being followed.

TABLE IV

Geographic Primary locale frame type" Boston Chicago Detroit

List List List

Frames

To identify the cohort for the longitudinal study, sites had to address successfully three competing requirements. These requirements were to (1) identify populations representative of a defined and diverse community, (2) recruit women from a specified race/ethnic minority group in a proportion significantly greater than the groups' proportion in the general United States population, and (3) implement the recruitment in a defined and circumscribed geographic locale so that relatively intense longitudinal clinical studies could be sustained. To meet these requirements and to be cost-efficient, the seven sites selected study communities that had a relatively higher density of the particular racial/ethnic minority group designated for recruitment. Then, individual sites utilized a variety of sampling frames from which the sample(s) would be drawn. In general, these sampling frames included telephone numbers randomly generated from random digit dialing (RDD)-based and list-based frames (Table IV). The following sections describe both types of frames in the context of the SWAN geographic locations and racial/ethnic group requirements. Appendices 2 and 3 provide specific information about the sampling approach at each clinical site. 1. RANDOM DIGIT DIALING-BASED FRAMES

The sampling unit for RDD frames was a telephone number and the only eligibility information available from an RDD frame consisted of the telephone number itself, i.e., the geographic location associated with the telephone number's exchange. Three sites [Newark area, Pittsburgh area, and Los Angeles (Table IV)] use an RDD-based sampling frame as the major frame. Two of these sites (Newark and Los Angeles) used list-assisted RDD-based sampling, and the Pitts-

The Primary Sampling Frame, Supplemental Frames, and Type of Supplemental Information Provided to the Primary Frame According to Geographic Locales Primary frameb

Oakland List Los Angeles RDD

City census listing Enrollmentlist from earlier study Electricalutility company customer listing for communitycensus HMO enrollment list RDD 3+ approach

Pittsburgh Newark

RDD RDD 3+ approach

RDD RDD

C. T h e S a m p l i n g

Supplemental frames" None None None None VRL, telephone directory list, ethnic organization membershiplists, snowball VRL Snowball

Supplemental information added to frames Telephone numbers, face-to-facecontact None Telephone directory, race from organization lists and VRL, face-to-facecontact None

Telephone directory None

RDD, Random digit dialing. is a variation in random digit dialing that increases the likelihood that telephone numbers are households and not commercialfirms. cVRL, Voter registration list.

a

b 3+

CHAPTER 11 SWAN Cohort Study burgh site used voter registration lists as their important secondary frame. Sites with a primary RDD sampling frame implemented the following steps: 1. Each telephone number was screened to determine if it represented a household. 2. The household was then screened to verify that the household was in the target geographic area and to determine if any woman age-eligible for the cross-sectional study was in residence. 3. Personnel then determined whether the household included at least one age-eligible woman who was Caucasian or was from the site's designated racial/ethnic minority group. All three of the sites that used the RDD sampling frame supplemented the RDD frame with list-based or "snowball" (referral by other participants) sampling frames. 2. LIST-BASED FRAMES At four sites, lists representing households (Detroit area) or individual women (Boston, Chicago, and Oakland areas) comprised the primary sampling frames. The list-based frames were varied and included a state-mandated census in Boston, an electrical utility customer listing in the Detroit area, a census from a previous study in the Chicago area, and a health maintenance organization (HMO) enrollment list in the Oakland, California area. Although each of these sites recruited its entire sample using a list sampling frame, only one site had a single frame that included all the information necessary to determine eligibility a priori (age, address, telephone, geographic area, race/ethnicity, gender) for recruiting to the SWAN Cross-sectional Study. That single list-based frame had been developed in a previous research study.

D.

Sampling Strategies

Specific sampling procedures varied across sites and were a function of the sampling frame(s) used and the level of information available with the frame(s). Sampling procedures included conducting a census, implementing an area probability sampling, and identifying acquaintance networks with snowballing. For example, the Detroit site conducted a census in which every household in the selected geographic area was enumerated and contacted, with the probability of selection being 1. Area probability samples were developed and implemented in the Chicago, Oakland, and Pittsburgh areas, where women were sampled with a known probability of selection that was , 36.78 4

~

2

36.74

"'"

-30 A

o o

E _e .c_ era (D

=E

. . . . . . .

!

. . . . . . . . .

-20

I

. . . . . . . . .

-10

I

. . . . . . . . .

0

! . . . . . . . . .

I

10

20

10

20

....................

...................

-~

-~

, .........

-10

w. . . . . . . . .

0

35.4 35.3 35.2 J

35.1

o

%

35.0 34.9 34.8 -30

-20

-10

0

FIGURE 1 Peripheral physiological events of the hot flash. Data from Freedman [12]. Drawing by Jeri Pajor.

, ........

10

,,

20

CHAPTER 14 Menopausal Hot Flashes

217

through the chamber. Skin conductance level was also recorded from the sternum using a 0.5-V constant voltage circuit and disposable Ag/AgC1 electrodes. Both measures increased significantly during 29 hot flashes recorded in 14 women (Fig. 2). Measurable sweating occurred during 90% of the flashes and there was a close time correspondence between both measures. 3. CORE BODY TEMPERATURE

Homeotherms regulate core body temperature between upper thresholds, where sweating and peripheral vasodilation occur, and a lower threshold, where shivering occurs. If core body temperature were elevated in women with hot flashes, their symptoms of sweating and peripheral vasodilation could be explained. However, measurements of esophageal [14], rectal [13], and tympanic [15] temperatures were not elevated prior to hot flashes. These studies all found declines of about 0.3~ following hot flashes, probably due to increased heat loss (peripheral vasodilation) and evaporative cooling (sweating). However, esophageal and rectal temperatures have long thermal lag times and might respond too

14

A

0

r-

E "-I

12

m G,1

4. METABOLIC RATE

>' _1

10

o r-

8

Elevations in core body temperature can be caused by increased metabolic rate (heat production) and by peripheral vasoconstriction (decreased heat loss). In the last study [ 12] we sought to determine if either of these factors accounted for the core body temperature elevations preceding hot flashes. Twenty-nine flashes were recorded in 14 postmenopausal women. Significant elevations in metabolic rate (about 15%) occurred but were simultaneous with sweating and peripheral

oe~-o

.i,,a O

"0 tO 0

6 4

C ..,.

o0

slowly to appear along with the rapid peripheral events of the hot flash [ 18]. Additionally, it has been shown that tympanic temperature does not reliably measure core body temperature because it is affected by peripheral vasodilation and sweating [ 19]. We therefore conducted several studies in which we measured core body temperature using an ingested radiotelemetry pill, which has a faster response time than the esophageal and rectal methods (Fig. 3). The pill is swallowed 90 min before an experiment, to allow its egress from the stomach, and the signals are detected by a wire antenna and stored in a small digital recorder. The typical transit time through the gut is about 24 hr, during which the recorder samples the data every 30 sec. Hot flashes are recorded on a separate device, using the sternal skin conductance level as the marker. In the first study, 10 symptomatic women were recorded using ambulatory monitoring for 24 hr [20]. Of 77 hot flashes recorded, 46 (60%) were preceded by small but significant increases in core body temperature. In a second study, conducted during sleep in a temperature-controlled laboratory, 37 hot flashes occurred in 8 postmenopausal women [21]. Significant core temperature elevations preceded 24 of the flashes (65%), whereas rectal temperature had not significantly changed (Fig. 4). These results were replicated during a daytime study in the laboratory [ 12].

2 -'

0

1

2

0

1

2

3

4

5

6

7

8

0.14

A

.C

0.12

E r

0.10

0.08 E Q,t t~

0.06

n,,

0.04 Or)

0.02 0.00 ,,,o ~ -'

3

4

5

6

7

8

Minute

FIGURE 2 Time course of skin conductance and sweating in 29 hot flashes. From Freedman [12]. Reprinted by permission from the American Society for Reproductive Medicine (Fertility and Sterility, 1998, Vol. 70, 1-6).

FIGURE 3 Radiotelemetry pill.

218

ROBERT R. FREEDMAN

5. HEART RATE

Modest increases in heart rate, about 7-15 beats/min [ 13,22,23], occur at approximately the same time as the peripheral vasodilation and sweating.

36.50 36.49

36.48 36.47 (.9 u) (1)

36.46

III. OBJECTIVE

a~ 36.45 i_

MEASUREMENT

OF

HOT FLASHES

D 36.44 36.43

A. In the L a b o r a t o r y

36.42 36.41 36.40

' -30

. . . . . . . . .

t

. . . . . . . . .

-20

)

.........

i .........

-10

0

i .........

10

i .........

i

20

Temperature from the dorsum of one finger was proposed as the first physiological marker for menopausal hot flashes [24]. In 7 symptomatic women, 41 skin temperature elevations > 1~ occurred within approximately 1 min of the subjective hot flash. However, the duration of the temperature elevations averaged 31 min, whereas the duration of subjective flushing was 2.3 min. Also, precise definitions of the onset and offset of the temperature elevations were not reported.

Minutes

36.38 36.37 36.36 36.35 36.34

36.33 ~

1. F I N G E R TEMPERATURE

30

36.32

2. SKIN CONDUCTANCE

36.31 36.30 36.29 36.28

'. . . . . . . . . -30

i .........

-20

i .........

) .........

-10

0

i

. . . . . . . . .

10

i

.........

i

20

30

Minutes

8 o 7 E

e--

_A

r

6

4

. . . . . . . . .

-30

, . . . . . . . . .

-20

, . . . . . . . . .

-10

i . . . . . . . . .

0

i . . . . . . . . .

10

, . . . . . . . . .

20

i

30

Minutes

FIGURE 4 (A) Core body temperature (means) recorded from ingested telemetry pills during menopausal hot flashes. Time zero is the beginning of the sternal skin conductance response (in C). (B) Rectal temperature (means) during menopausal hot flashes. (C) Sternal skin conductance (means) during menopausal hot flashes. From Freedman and Woodward [21 ]. Reprinted by permission from the American Society for Reproductive Medicine (Fertility and Sterility, 1996, Vol. 65, 1141-1144).

vasodilation and did not precede the core temperature elevations (Fig. 1). Peripheral vasoconstriction did not occur. Thus, increased metabolic rate and peripheral vasoconstriction did not account for the core body temperature elevations in these women.

Subsequently, skin conductance recorded from the sternum was investigated as a hot flash marker. Tataryn et al. [25] found that 98% of 128 subjective flashes in 8 postmenopausal women were accompanied by elevations in sternal skin conductance compared to 82% for finger temperature and 81% for decreased tympanic temperature. All of these changes were significantly reduced by estrogen administration in 4 of the women. However, the precise characteristics of the skin conductance responses were not defined. Our laboratory subsequently sought to determine these characteristics [23]. Sternal skin conductance level, finger temperature, and heart rate were recorded for 4 hr in 11 postmenopausal and 8 premenopausal women. Twenty-nine subjective hot flashes were indicated by pushbutton in the first group. All of these were accompanied by an increase in sternal skin conductance ---2/xmho/30 sec. One skin conductance elevation occurred without a button press. All skin conductance elevations occurred within 66 sec of the button press. No skin conductance elevations occurred in the premenopausal women. Thus, there was a concordance of 95% between the skin conductance criterion and the reports of the subjects. Significant elevations in skin temperature and heart rate occurred during the flashes but were not as sensitive or specific as the skin conductance elevations. We replicated these findings in 18 symptomatic and 8 asymptomatic postmenopausal women [26]. There was a concordance of 80% between the sternal skin conductance criterion (2/xmho/30 sec) and the subjective reports (button press) in 15 flashes recorded in the symptomatic women. No events occurred in the asymptomatic women. Our findings were then independently replicated by an-

CHAPTER 14 Menopausal Hot Flashes other laboratory [27]. In two separate studies of 20 symptomatic women, a concordance of 90% was obtained between the sternal skin conductance criterion and subjective reports. Measurements of finger temperature and blood flow were less predictive and did not improve the concordance rate when added to the skin conductance measure.

B. A m b u l a t o r y M o n i t o r i n g To evaluate treatment studies it would be useful to have a method that could be used outside the laboratory over longer periods of time. We therefore developed methods for recording sternal skin conductance on ambulatory monitors for 24 hr. Using the same basic circuit and electrodes we found a concordance of 86% between the skin conductance criterion and button presses in 43 flashes recorded in 7 symptomatic women [23]. No such changes occurred in the 8 premenopausal women. We replicated these findings in a second study [26]. A concordance of 77% was obtained in 149 flashes recorded in 10 symptomatic women. Twelve skin conductance responses occurred in 8 putative asymptomatic women, representing a false response rate of about 8%. These ambulatory monitoring procedures were then successfully used to demonstrate the efficacy of a behavioral treatment for hot flashes in two subsequent studies [28,29].

C. P r o v o c a t i o n T e c h n i q u e s For laboratory investigations, it would be useful to provoke hot flashes reliably as opposed to waiting for them to occur during extended recording periods. Sturdee [22] observed that peripheral warming provoked objective and subjective hot flashes in 7 of 8 symptomatic women. We therefore sought to define this procedure operationally. Two 40 x 60-cm circulating water pads maintained at 42~ were placed on the torso of 11 supine symptomatic women in a 23~ room [23]. Eight hot flashes occurred within 30 min. A concordance of 73% was obtained between the skin conductance criterion (2/xmho/30 sec) and subjective report (button press). These findings were replicated in a subsequent study in 14 symptomatic women with a concordance of 84%. In this study, 25 hot flashes occurred during a 45-minute heating period. No objective or subjective responses occurred in 8 asymptomatic women.

IV. E N D O C R I N O L O G Y A. E s t r o g e n s Because hot flashes accompany the decline of estrogens in the vast majority of naturally and surgically menopausal

219 women, there is little doubt that estrogens play a role in the genesis of hot flashes. However, estrogens alone do not appear responsible for hot flashes because there is no correlation between the presence of this symptom and plasma [30], urinary [31 ], or vaginal [31 ] concentrations. No differences in unconjugated plasma estrogen concentrations were found in symptomatic versus asymptomatic women [32]. Additionally, clonidine significantly reduces hot flash frequency without altering circulating estrogen values [33]. Prepubertal girls have low estrogen production without hot flashes and hot flashes occur in the last trimester of pregnancy when estrogen production is high. Nevertheless, estrogen administration in hormone replacement therapy virtually eliminates hot flashes [34,35].

B. G o n a d o t r o p i n s Because gonadotropins become elevated at menopause, their possible role in the initiation of hot flashes has been investigated. Although no differences in luteinizing hormone (LH) concentrations were found between women with and without hot flashes [36], a temporal association was found between LH pulses and hot flash occurrence [37,38]. However, subsequent investigation revealed that women with a defect of gonadotropin-releasing hormone (GnRH) secretion (isolated gonadotropin deficiency) had hot flashes but no LH pulses and women with abnormal input to GnRH neurons (hypothalamic amenorrhea) had some LH pulses but no hot flashes [39]. Additionally, hot flashes occur in hypophysectomized women, who have no LH release [40], in women with pituitary insufficiency and hypoestrogenism [41], and in women with LH release suppressed by GnRH analog treatment [42,43]. Thus, LH cannot be the basis for hot flashes.

C. O p i a t e s It was observed that alcohol-induced flushing in subjects taking chlorpropamide, a drug that stimulates insulin release and lowers blood glucose, was related to opiate receptor activation [44]. Lightman et al. [45] subsequently found that naloxone infusion significantly reduced hot flash and LH pulse frequencies in six postmenopausal women. However, DeFazio et al. [46] attempted to replicate this study and found no effects. Tepper et al. [47] found that plasma fl-endorphin concentrations decreased significantly before occurrence of menopausal hot flashes whereas Genazzani et al. [48] found significantly increased values preceding hot flashes. Thus, there is no consistent evidence of the involvement of an opioidergic system in menopausal hot flashes.

220

ROBERTR. FREEDMAN D. C a t e c h o l a m i n e s

There is considerable evidence that norepinephrine plays an important role in thermoregulation mediated, in part, through ce2-adrenergic receptors [49]. Injection of norepinephrine into the preoptic hypothalamus causes peripheral vasodilation, heat loss, and a subsequent decline in core body temperature [49]. Additionally, there is considerable evidence that gonadal steroids modulate central noradrenergic activity [50]. Studies of plasma norepinephrine have not found increased concentrations prior to or during hot flashes [ 14,37]. However, brain norepinephrine content cannot be measured in plasma, due to the large amounts derived from peripheral organs [51 ]. We therefore measured plasma 3-methoxy-4-hydroxyphenylglycol (MHPG), the main metabolite of brain norepinephrine, to determine if central norepinephrine concentrations were elevated during hot flashes [52]. We studied 13 symptomatic and 6 asymptomatic postmenopausal women who were supine with an intravenous line in a 23~ room. Blood samples were drawn at the beginning and end of a 60-min period and during a hot flash, if one occurred. The same procedures were followed during a 45-min heating period. Basal MHPG levels were significantly higher in the symptomatic women (p < 0.0001, Table I) and increased significantly during resting and heat-induced flashes. There were no hot flashes or significant MHPG changes in the asymptomatic women, whose blood drawing times were yoked to those of 6 symptomatic women. However, approximately 50% of the free MHPG that enters the blood is metabolized peripherally to vanillylmandelic acid (VMA), and VMA formation can compete with MHPG production [53]. Thus, fluctuations in peripheral VMA formation could potentially distort measurements of plasma MHPG. Therefore, we measured both compounds simultaneously before and after hot flashes in 14 symptomatic women [12]. Plasma MHPG concentrations increased significantly (p < 0.02) between the preflash (3.7 ___ 1.4 ng/ml) and postflash (5.1 ___ 2.3 ng/ml) blood samples whereas

VMA levels did not significantly change (6.2 ___ 1.8 ng/ml vs. 6.1 ___ 2.5 ng/ml). Thus, there is evidence of increased brain norepinephrine content before hot flashes and these increase significantly when a flash occurs. Clonidine, an ce2-adrenergic agonist, reduces central noradrenergic activation and hot flashes [54-56]. Yohimbine, an ce2-adrenergic antagonist, increases central noradrenergic activation. We sought to determine if clonidine would ameliorate hot flashes and if yohimbine would provoke them in controlled laboratory conditions [57]. Nine symptomatic postmenopausal women, aged 4 3 - 6 3 years, served as subjects. Six asymptomatic women, aged 4 6 - 6 1 years, served as a comparison group. All women were in good health and had been amenorrheic for -- 2 years. In two blind laboratory sessions, subjects received either intravenous clonidine HC1 (1/xg/kg) or placebo followed by a 60-min waiting period and then by 45 min of peripheral heating. In two additional blind sessions, subjects received yohimbine HC1 (0.032-0.128 mg/kg intravenously)or placebo. Clonidine significantly (p = 0.01) increased the length of heating time needed to provoke a hot flash compared to placebo (40.6 ___ 3.0 min vs. 33.6 __+ 3.6 min) and reduced the number of hot flashes that did occur (2 vs. 8) (Fig. 5). In the symptomatic women, six hot flashes occurred during the yohimbine sessions and none during the corresponding placebo sessions, a statistically significant difference (p < 0.015). No hot flashes occurred in the asymptomatic women during either session (Fig. 6). These data support the hypothesis that a2-adrenergic receptors within the central noradrenergic system are involved in the initiation of hot flashes and are consistent with the idea that brain norepinephrine is elevated in this process. Animal studies have shown that yohimbine increases norepinephrine release by blocking inhibitory presynatic a2-adrenergic re-

"

Symptomatic

:..

/ "'.. ".

8

Plasma MHPG during Resting and Heating Periods in Symptomatic and Asymptomatic Women

TABLE

I

"

::

o

cE

i

6

_

j :

_

~

~ . _ . . . .

.

[.

.

. ' .......

.

.

.

Stage Resting Basal Hot flash Post Heating Basal Hot flash Post

MHPG (ng/ml) 3.5 __+0.2 4.3 +__0.3 3.8 _+0.2 3.4 ___0.2 3.9 ___0.3 3.4 ___0.2

Resting Basal Hot flash Post Heating Basal Hot flash Post

MHPG (ng/ml)

2

2.3 ___0.2 2.1 ___0.1 2.2 ___0.2

..

:~.....~......:~.....-.-....-...~-.........~..-.......~-..~..~-....:.-..-..~...`.`...-.....-.-..~ ...... "-"-"-'- ..................... .----

o

2.6 ___0.1 2.7 __+0.2 2.5 ___0.2

....' ........ -

Asymptomatic . . . . Clonidine ....... Placebo

Asymptomatic women Stage

.

9. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4 ..................... "

Symptomatic women

Clonidine Placebo

.

....,

J

.: .. ..

-.

0

I

I

I

I

I

t

I

I

t

5

10

15

20

25

30

,35

40

45

MINUTES

FIGURE 5 The occurrence of a hot flash during body heating was delayed after 1/xg/kg clonidine, comparedto placebo. No hot flashesoccurred in an asymptomaticwoman. From Freedman et al. [57]. Reprintedwith permission from the American College of Obstetricians and Gynecologists (Obstetrics and Gynecology, 1990, Vol. 76, 573-578).

221

CHAPTER 14 Menopausal Hot Flashes 10 -

gens modulate adrenergic receptors in many tissues [50]. It is possible, therefore, that hypothalamic ce2-adrenergic receptors are affected by the estrogen withdrawal associated with the menopause. As noted above, a decline in inhibitory presynatic ce2 receptors would lead to increased central norepinephrine concentrations and this is consistent with evi, dence from animal studies.

Symptomatic --Yohimbine ....... P l a c e b o o _c

Asymptomatic

E

-- Yohimbine ....... P l a c e b o

-:. .....................................

.................

0.052 mg/ikg 0

0.064

10

20

.-....-...:. ............

mg/kg

.....................................

0.128

30

40

mg/~]g 50

.-.

I 60

MINUTES

FIGURE 6 A hot flash, indicated by a sternal skin conductance response, occurred after intravenous infusion of 0.032 mg/kg yohimbine in a menopausal woman with hot flashes. No responses occurred in the matched placebo session or in an asymptomatic women givenhigher doses. From Freedman et al. [57]. Reprinted with permission from the American College of Obstetricians and Gynecologists (Obstetrics and Gynecology, 1990, Vol. 76,573-578).

ceptors [58]. These autoreceptors mediate the turnover of norepinephrine through a feedback mechanism, and a reduction in their number and/or sensitivity would result in increased norepinephrine release [59]. This mechanism is consistent with human studies showing that yohimbine elevates and clonidine reduces plasma levels of M H P G [60]. Therefore, the yohimbine provocation and the clonidine inhibition of hot flashes in symptomatic women may reflect a deficit in inhibitory ce2-adrenergic receptors not seen in asymptomatic women. Additionally, the injection of clonidine into the hypothalamus reduces body temperature and activates heat conservation mechanisms, effects that are blocked by yohimbine [61]. Thus, ce2-adrenoceptors in the hypothalamus may be responsible for the events of the hot flash that are characteristic of a heat dissipation response. There is considerable evidence demonstrating that estro-

V. THERMOREGULATION AND HOT FLASHES Increased thermosensitivity at menopause has been noted in the literature for many years and is reflected in reports of increased hot flash frequency and duration during warm weather [62,63]. Peripheral heating has been demonstrated to provoke hot flashes in most of our symptomatic subjects [23], and this has been found by others as well [22]. As noted earlier, core body temperature (Tc) in homeotherms is regulated by hypothalamic centers between the thresholds of Tc for sweating and peripheral vasodilation and shivering (Fig. 7). According to this mechanism, the heat dissipation responses of hot flashes (sweating, peripheral vasodilation) would be triggered if body temperature were elevated or the sweating threshold lowered. We previously demonstrated that peripheral heating induced hot flashes in symptomatic but not in asymptomatic postmenopausal women nor in premenopausal women [23, 26]. These data suggested that the sweating threshold was reduced in symptomatic postmenopausal women. Considerable research in humans and animals has shown that conditions that alter the sweating threshold tend to alter the shivering threshold in the same direction [49]. We therefore tested to see if the T~ shivering threshold was reduced in symptomatic women, similar to their reduction in sweating threshold. We found that the shivering threshold was elevated rather than reduced in symptomatic compared to asymptomatic women [64]. This result implies that the thermoneutral zone

T

IAI

Zone

HF

Y

NON HF

BRAIN NE CLON

t

MHPG?

t

FIGURE 7 We have shown that the thermoneutral zone is narrowed in symptomatic women. Elevated brain norepinephrine (NE) in animals reduces this zone. Yohimbine (YOH) elevates brain norepinephrine and should reduce this zone. Conversely, clonidine (CLON) should widen it. HF, Symptomatic women; Non-HF, asymptomatic women; MHPG, 3-methoxy-4-hydroxyphenylglycol (the primary brain NE metabolite).

222

ROBERT R. FREEDMAN

is narrowed in postmenopausal women with hot flashes. This hypothesis would explain the ability of small Tc elevations, as we found with the telemetry pill, to trigger the heat loss mechanisms of the hot flash (sweating, cutaneous vasodilation) and would also explain the shivering observed following many of the episodes. We therefore measured the thermoneutral zone in symptomatic and asymptomatic postmenopausal women, hypothesizing a reduction in the former group. We studied 12 symptomatic and 8 asymptomatic postmenopausal women [65]. We measured body temperature using a rectal probe, the ingested telemetry pill, and a weighted average of rectal and skin temperatures and determined the sweating and shivering thresholds for each. In a subsequent session, we raised body temperature to the sweating threshold using exercise. The symptomatic women had significantly smaller interthreshold zones compared to the asymptomatic women on all three measures of body temperature (Table II). Sweat rates were significantly higher in the former group. During exercise, all of the symptomatic and none of the asymptomatic women demonstrated hot flashes. Animal studies have shown that increased brain norepinephrine narrows the width of the interthreshold zone [49]. Conversely, clonidine reduces norepinephrine release, raises the sweating threshold, and lowers the shivering threshold in human studies [66]. Thus, we suggest that elevated brain norepinephrine narrows the thermoregulatory interthreshold zone in symptomatic postmenopausal women. This zone was so small as to be virtually zero using our methods. We propose that small elevations in core body temperature trigger hot flashes when the sweating threshold is crossed. Core

TABLE II Sweating Thresholds, Shivering Thresholds, and Interthreshold Zones for Rectal Temperature, Telemetry Pill Temperature, and Mean Body Temperature a

body temperature falls following hot flashes and patients often report shivering at this time. This likely represents the point where the shivering threshold is crossed, although this has not been directly measured.

VI. CIRCADIAN RHYTHMS The circadian rhythm of Tc is well known, and similar variations in other thermoregulatory parameters, such as heat conductance and sweating, have also been demonstrated. These patterns suggest that the thermoregulatory effector responses of hot flashes might also demonstrate temporal variations. A previous study showed circadian rhythmicity of self-reported hot flashes in some menopausal women, but no physiological data were collected [67]. We recruited and screened 10 symptomatic and 6 asymptomatic postmenopausal women [20]. Each received 24-hr ambulatory monitoring of sternal skin conductance level to detect hot flashes as well as ambient temperature, skin temperature, and Tc. The last measure was recorded using the ingested radiotelemetry pill. Cosinor analysis demonstrated a circadian rhythm (p < 0.02) of hot flashes with a peak around 1825 hr (Fig. 8). This rhythm lagged the circadian rhythm of Tc in symptomatic women by about 3 hr. Tc values of the symptomatic women were lower than those of the asymptomatic women (p < 0.05) from 0000-0400, and at 1500 and 2200 hr. The majority of hot flashes were preceded by elevations in Tc, a statistically significant effect (p < 0.05). Hot flashes began at significantly (p < 0.02) higher levels of Tc (36.82 ___0.04~ compared to all nonflash periods (36.70 ___ 0.005~ These data are consistent with the hypothesis that elevated Tc serves as part of the hot flash triggering mechanism.

VII. SLEEP A. T h e r m o r e g u l a t i o n a n d S l e e p

Temperature measurement (~ Group

Sweating

Shivering

Interthreshold

Symptomatic Asymptomatic P value

37.4 _+ 0.06 37.7 _ 0.05 0.001

Rectal 37.4 ___ 0.06 37.3 _+ 0.16 NS

0.0 _+ 0.06 0.4 ___ 0.18 0.005


37.2 ___ 0.09 37.5 +__0.14 0.008

T e l e m e t r y pill 37.2 ___ 0.15 37.1 ___ 0.09 NS

0.0 ___ 0.11 0.4 + 0.18 0.005


37.2 ___ 0.07 37.6 ___ 0.04 0.0003

Mean body 36.4 ___ 0.06 36.1 ___ 0.18 0.02

0.8 ___ 0.09 1.5 ___ 0.20 0.0006

a Values are means __+ S.E. P values for group differences, unpaired T-tests; NS, not significant.

Our observations of Tc elevations preceding hot flashes are supported by several other findings of our current work. Previous research had shown that external body heating applied 2 hr before sleep induced significantly higher amounts of subsequent stage 4 (slow wave) sleep [68]. We then observed [69] that postmenopausal women with hot flashes had significantly more stage 4 sleep than did asymptomatic women and that the number of hot flashes occurring 2 hr before sleep was significantly and positively correlated with the amount of subsequent stage 4 sleep. In our most recent work we used ambient cooling to reduce the frequency of hot flashes in the 2-hr period before sleep, which significantly reduced the amount of subsequent stage 4 sleep. Three consecutive nights of sleep recording were conducted on 11 symptomatic and 7 asymptomatic


12-

I

223

I HF Frequency Core Temperature, Symptomatic

10

D

- 37.2

..q

~::; .......................... 9

.....::"

9 ""

....[] Core Temperature, Asymptomatic

,

".

"

~: ....

..

:.~..

.D. . . . . "

- 37.1

9.'-.. .

9""

"

"

"O.-i:::q

..." :'""

"':;(.... "..

-1-

~

q

'tr

37.0 36.9

~"

36.8

~

36.7 R 9-. -."a-

"6

...2

It ." ........ r

-III [I II II II If II I1\"

~\ -a..........~ ..... o... 'y""/ 4

....... :"..

//

36.6

E

36.5

m ~

36.4

z

36.3 36.2 0 0

2

4

6

8

10

12

l

i

I

i

i

14

16

18

20

22

36.1

Hour F I G U R E 8 Hot flash (HF) frequency and core body temperature over 24 hr. Hot flash frequency in 10 symptomatic women shown as bars. Curves: best-fit cosine curve for hot flash frequency ( . . . . . ); 24-hr core temperature data for 10 symptomatic women ( o - - o ) with best-fit cosine curve (--); 24-hr core temperature data in 6 asymptomatic women (D ..... n) with best-fit cosine curve ( ..... ). From R. R. Freedman, D., Norton, S., Woodward, and G. Cornelissen (1995). Core body temperature and circadian rhythm of hot flashes in menopausal women. J. Clin. E n d o c r i n o l . M e t a b . 80(8), 2 3 5 4 - 2 3 5 8 . 9 The Endocrine Society.

menopausal women. Ambient temperature during the 2 hr before sleep was set at either neutral (23~ warm (30~ or cold (18~ Hot flash frequencies were lower in the cold condition (p < 0.02), and in the neutral condition (p < 0.01) relative to the warm condition. The percentage of time spent in stage 4 sleep was greater following both the neutral and warm conditions in the symptomatic group, compared to the asymptomatic group (p < 0.01). Within the symptomatic group, time spent in stage 4 sleep was significantly lower following the cold condition (p < 0.025), compared to the warm condition. These results demonstrate that the association between hot flashes and slow wave sleep can be selectively reversed by suppressing hot flashes with cooling prior to sleep onset.

flash frequency and in nighttime awakenings as well as increased rapid eye movement (REM) sleep. However, the causal relationships, if any, between hot flashes and sleep variables are not known.

VIII. T R E A T M E N T OF H O T F L A S H E S

A. Hormone Replacement Therapy Virtual elimination of hot flashes by hormone replacement therapy has been established [34,74]. Initially, estrogen alone was given, but to provide uterine protection women with a uterus are now given estrogen plus a progestin in combination or cyclically [hormone replacement therapy (HRT)].

B. Sleep Disturbance B. Laboratory sleep studies have demonstrated that symptomatic postmenopausal women wake up more frequently during the night than do asymptomatic women. Erlik et al. [70] first reported that objectively measured hot flashes were accompanied by EEG-defined waking episodes in 45/47 flashes recorded in 8 symptomatic women. This relationship did not occur in asymptomatic women. Three additional studies [71-73] showed that administration of estrogen to symptomatic women resulted in significant declines in hot

c~-AdrenergicAntagonists

As noted above, clonidine decreases central noradrenergic activation as well as thermally induced hot flashes [57]. In a double-blind cross-over study involving a small number of women, Clayden et al. [54] found that oral clonidine reduced hot flash frequency by 44% after 4 weeks and by 56% at 8 weeks, significantly better than the placebo group. In 11 symptomatic women, Schindler et al. [33] found that clonidine significantly reduced hot flash

224

ROBERTR. FREEDMAN

frequency and was well tolerated. Using objective measurements of hot flashes, Laufer et al. [55] found that clonidine significantly reduced hot flash frequency compared with baseline and with placebo. However, 4 of the subjects withdrew due to side effects. Using transdermal clonidine, Nagamani et al. [75] found an 80% decrease in hot flash frequency in the active drug group compared to 36% in the placebo group. Reported side effects were minimal. Thus clonidine, particularly in transdermal form, may be a useful treatment for hot flashes in women for whom HRT is contraindicated.

tance. The paced respiration group showed a significant decline in hot flash frequency (again about 50%) compared to no change in the control group. However, there were no significant changes in any biochemical measure for either group. Thus, the mechanism through which paced respiration reduces hot flash frequency remains to be determined. Two other small studies [78,79] have also found significant amelioration of hot flashes through behavioral relaxation procedures. Thus, behavioral treatments my be useful for women in whom HRT is contraindicated or who choose not to take medications.

C. B e h a v i o r a l T r e a t m e n t s

D. A l t e r n a t i v e T r e a t m e n t s

Although hormone replacement therapy offers effective control of hot flashes, it may be contraindicated for some women. We have presented evidence that central sympathetic activation is increased in women with hot flashes. Behavioral relaxation methods have been shown to reduce sympathetic activity in normal subjects and in some clinical populations [76]. Therefore, we treated seven menopausal women with hot flashes using a combination of progressive muscle relaxation exercises and slow deep breathing [77]. Seven additional women were assigned to receive a control procedure, alpha wave EEG biofeedback. The relaxation procedure significantly reduced objective symptoms recorded in the laboratory and diary-recorded hot flash frequency (by about 50%) compared to the control procedure. This investigation demonstrated that a combination of muscle relaxation exercises and slow deep breathing significantly reduced hot flash frequency in a small group of subjects. However, because two treatment procedures were combined, it was not possible to determine which component was responsible for the therapeutic effect. The physiological data showed that respiration rate was the only recorded variable that was significantly altered during training. Therefore, a second study was performed in which one group of subjects received slow deep breathing alone, another group received muscle relaxation exercises alone, and a third group received alpha EEG biofeedback [28]. Treatment outcome was assessed by ambulatory monitoring of sternal skin conductance responses, described earlier. Only the paced respiration group showed a significant decline in hot flash frequency (about 50%), decreased respiration rate, and increased tidal volume. There were no significant changes shown by the other two groups. We then sought to determine if reduced sympathetic activation was the mechanism by which paced respiration ameliorates hot flashes [29]. We therefore measured plasma MHPG, epinephrine, norepinephrine, and platelet ce2-receptors during paced respiration or alpha EEG biofeedback in 24 symptomatic women. Treatment outcome was again assessed by ambulatory monitoring of sternal skin conduc-

1. ACUPUNCTURE

One controlled study has thus far been published on the effects of acupuncture on hot flashes. Wyon et al. [80] randomized 24 symptomatic women to receive either active or placebo acupuncture for 8 weeks. Both groups showed significant declines in reported hot flash frequency at 4 weeks, 8 weeks, and 3 months posttreatment. 2. PHYTOESTROGENS

Phytoestrogens are plant compounds with estrogen-like biological activity [11 ]. Their possible beneficial effects on hot flashes are inferred from the low reported prevalence of symptoms in countries, such as Indonesia and Japan, with diets rich in these compounds (see discussion on epidemiology, Section I). A controlled study of 145 women in Israel compared a phytoestrogen-rich diet (tofu, soy drink, miso, flax seed) with the usual Israeli diet [81]. Both groups showed significant declines in hot flash scores but the decline in the active treatment group was significantly greater. Similar findings were obtained by Murkies et al. [82], who compared soy flour dietary supplementation with wheat flour supplementation (control group). The most recent investigation compared soy protein supplementation with casein (placebo) in 104 symptomatic women [83]. At week 12, the soy group had a significantly greater reduction in hot flash frequency (45%) compared to placebo (30%). Thus, dietary supplementation with phytoestrogens may be useful for treating hot flashes, although the effects are not dramatic when compared with placebo.

IX. S U M M A R Y

AND CONCLUSIONS

Hot flashes are the most common symptom associated with menopause, although prevalence estimates are lower in some rural and non-Western areas. The symptoms are characteristic of a heat-dissipation response and consist of sweating on the face, neck, and chest, as well as peripheral

225


vasodilation. Although hot flashes clearly accompany the estrogen withdrawal at menopause, estrogen alone is not responsible because levels do not differ in symptomatic and asymptomatic women. Until recently it was thought that hot flashes were triggered by a sudden, downward resetting of the hypothalamic setpoint, because there was no evidence of increased core body temperature. However, we obtained such evidence, using a rapidly responding ingested telemetry pill. We then found that the thermoneutral zone, within which sweating, peripheral vasodilation, and shivering do not occur, is virtually nonexistent in symptomatic women but is normal (about 0.4~ in asymptomatic women. Thus, we believe that small temperature elevations preceding hot flashes acting within a reduced thermoneutral zone constitute the triggering mechanism. We also demonstrated that central sympathetic activation is elevated in symptomatic women; in animal studies, this reduces the thermoneutral zone. Clonidine reduces central sympathetic activation, widens the thermoneutral zone, and ameliorates hot flashes. Estrogen virtually eliminates hot flashes but its mechanism of action is not known. Behavioral relaxation procedures reduce hot flash frequency to the same extent as clonidine (about 50%) but their mechanism of action is also not understood.

11. 12. 13. 14.

15.

16.

17. 18.

19.

20.

21. 22.

Acknowledgment Research conducted by the author was supported by NIH Merit Award, AG-05233, from NIA.

References 1. Neugarten, B. L., and Kraines, R. J. (1965). "Menopausal symptoms" in women of various ages. Psychosom. Med., 27, 266-273. 2. Feldman, B. M., Voda, A., and Gronseth, E. (1985). The prevalence of hot flash and associated variables among perimenopausal women. Res. Nurs. Health, 8, 261-268. 3. Hagstad, A., and Janson, P. O. (1986). The epidemiology of climacteric symptoms. Acta Obstet. Gynecol. Scand. Suppl. 134, 59-65. 4. Guthrie, J. R., Dennerstein, L., Hopper, J. L., and Burger, H. G. (1996). Hot flushes, menstrual status, and hormone levels in a population-based sample of midlife women. Obstet. Gynecol. 88(3), 437-442, 5. Kronenberg, E (1990). Hot flashes: Epidemiology and physiology. Ann. N. Y Acad. Sci. 592, 52-86. 6. Chakravarti, S., Collins, W. P., Newton, J. R., Oram, D. H., and Studd, J. W. W. (1977). Endocrine changes and symptomatology after oophorectomy in premenopausal women. Br. J. Obstet. Gynaecol. 84, 769775. 7. Schwingl, P. J., Hulka, B. S., and Harlow, S. D. (1994). Risk factors for menopausal hot flashes. Obstet. Gynecol. 84(1), 29-34. 8. Flint, M., and Samil, R. S. (1990). Cultural and subcultural meanings of the menopause.Ann. N. Y Acad. Sci. 592, 134-148. 9. Tang, G. (1994). Menopause: The situation in Hong Kong Chinese women. In "The Modem Management of the Menopause" (G. Berg and M. Hammar, eds.), vol. 8; pp. 47-55. Parthenon, New York. 10. Beyene, Y. (1986). Cultural significance and physiological manifesta-

23. 24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

tions of menopause: A biocultural analysis. Cult. Med. Psychiatry 10, 47-71. Murkies, A. L., Wilcox, G., and Davis, S. R. (1998). Phytoestrogens. J. Clin. Endocrinol. Metab. 83(2), 297-303. Freedman, R. R. (1998). Biochemical, metabolic, and vascular mechanisms in menopausal hot flushes. Fertil. Steril. 70(2), 1-6. Molnar, G. W. (1975). Body temperature during menopausal hot flashes. J. Appl. Physiol.: Respir. Environ. Exercise Physiol. 38, 499-503. Kronenberg, F., Cote, L. J., Linkie, D. M., Dyrenfurth, I., and Downey, J. A. (1984). Menopausal hot flashes: Thermoregulatory, cardiovascular, and circulating catecholamine and LH changes. Maturitas 6, 31- 43. Tataryn, I. V., Lomax, P., Bajorek, J. G., Chesarek, W., Meldrum, D. R., and Judd, H. L. (1980). Postmenopausal hot flushes: A disorder of thermoregulation. Maturitas 2, 101-107. Ginsburg, J., Swinhoe, J., and O'Reilly, B. (1981). Cardiovascular responses during the menopausal hot flush. Br. J. Obstet. Gynaecol. 88, 925-930. Sturdee, D. W., and Reece, B. L. (1979). Thermography of menopausal hot flushes. Maturitas 1,201-205. Molnar, G. W., and Read, R. C. (1974). Studies during open heart surgery on the special characteristics of rectal temperature. J. Appl. Physiol. 36, 333-336. Shiraki, K., Nobuhide, K., and Sagawa, S. (1986). Esophageal and tympanic temperature responses to core blood temperature changes during hyperthermia. J. Appl. Physiol. 61, 98-102. Freedman, R. R., Norton, D., Woodward, S., and Comelissen, G. (1995). Core body temperature and circadian rhythm of hot flashes in menopausal women. J. Clin. Endocrinol. Metab. 80(8): 2354-2358. Freedman, R. R., and Woodward, S. (1996). Core body temperature during menopausal hot flushes. Fertil. Steril. 65(6), 1141-1144. Sturdee, D. W., Wilson, K. A., Pipili, E., and Crocker, D. (1978). Physiological aspects of menopausal hot flush. Br. Med. J. 2, 79-80. Freedman, R. R. (1989). Laboratory and ambulatory monitoring of menopausal hot flashes. Psychophysiology 26(5), 573-579. Meldrum, D. R., Shamonki, I. M., Frumar, A. M., Tataryn, I. V., Chang, R. J., and Judd, H. L. (1979). Elevations in skin temperature of the finger as an objective index of postmenopausal hot flashes: Standardization of the technique. Am. J. Obstet. Gynecol. 135(6), 713-717. Tataryn, I. V., Lomax, P., Meldrum, D. R., Bajorek, J. G., Chesarek, W., and Judd, H. L. (1981). Objective techniques for the assessment of postmenopausal hot flashes. Obstet Gyneco157(3), 340-344. Freedman, R. R., Woodward, S., and Norton, D. (1992). Laboratory and ambulatory monitoring of menopausal hot flushes: Comparison of symptomatic and asymptomatic women. J. Psychophysiol. 6, 162-166. de Bakker, I. P. M., and Everaerd, W. (1996). Measurement of menopausal hot flushes: Validation and cross-validation. Maturitas 25, 8798. Freedman, R. R., and Woodward, S. (1992). Behavioral treatment of menopausal hot flushes: Evaluation by ambulatory monitoring. Am. J. Obstet. Gynecol. 167(2), 436-439. Freedman, R. R., Woodward, S., Brown, B., Javaid, J. I., and Pandey, G. N. (1995). Biochemical and thermoregulatory effects of behavioral treatment for menopausal hot flashes. Menopause 2(4), 211-218. Askel, S., Schomberg, D. W., Tyrey, L., and Hammond, C. B. (1976). Vasomotor symptoms, serum estrogens, and gonadotropin levels in surgical menopause. Am. J. Obstet. Gynecol. 126(2), 165-169. Stone, S. C., Mickal, A., Rye, E, and Rye, P. H. (1975). Postmenopausal symptomatology, maturation index, and plasma estrogen levels. Obstet. Gynecol. 45(6), 625-627. Hutton, J. D., Jacobs, H. S., Murray, M. A. E, and James, V. H. T. (1978). Relation between plasma esterone and estradiol and climacteric symptoms. Lancet 1, 671-681. Schindler, A. E., Muller, D., Keller, E., Goser, R., and Runkel, E (1979). Studies with clonidine (Dixarit) in menopausal women. Arch. Gynecol. 227, 341-347.

226 34. Johnson, S. (1998). Menopause and hormone replacement therapy. Med. Clin. North Am. 82(2), 297-320. 35. Kenemans, E, Barentsen, R., and van de Weijer, E (1996). Practical HRT. Med. Forum Int. 36. Campbell, S. (1976). Intensive steroid and protein hormonal profiles on postmenopausal women experiencing hot flashes and a group of controls. In "Management of the Menopause and Post-Menopausal Years" S. Campbell (ed.). MTP Press, London. 37. Casper, R. F., Yen, S. S. C., and Wilkes, M. M. (1979). Menopausal flushes: A neuroendocrine link with pulsatile luteinizing hormone secretion. Science 205, 823-825. 38. Tataryn, I. V., Meldrum, D. R., Lu, K. H., Frumar, A. M., and Judd, H. L. (1979). LH, FSH, and skin temperature during menopausal hot flush. J. Clin. Endocrinol. Metab. 49, 152-154. 39. Gambone, J., Meldrum, D. R., Laufer, L. Chang, R. J., Lu, J. K. H., and Judd, H. L. (1984). Further delineation of hypothalamic dysfunction responsible for menopausal hot flashes. J. Clin. Endocrinol. Metab. 59(6), 1092-1102. 40. Mulley, G., Mitchell, R. A., and Tattersall, R. B. (1977). Hot flushes after hypophysectomy. Br. Med. J. 2, 1062. 41. Meldrum, D. R., Erlik, Y., Lu, J. K. H., and Judd, H. L. (1981). Objectively recorded hot flushes in patients with pituitary insufficiency. J. Clin. Endocrinol. Metab. 52(4), 684-687. 42. Casper, R. E, and Yen, S. S. C. (1981). Menopausal flushes: Effect of pituitary gonadotropin desensitization by a potent luteinizing hormone releasing factor agonist. J. Clin. Endocrinol. Metab. 53, 1056-1058. 43. DeFazio, J., Meldrum, D. R., Laufer, L. Vale, W., Rivier, J., Lu, J. K., and Judd, H. L. (1983). Induction of hot flashes in premenopausal women treated with a long-acting GnRH agonist. J. Clin. Endocrinol. Metab. 56(3), 445-448. 44. Leslie, R. D. G., Pyke, D. A., and Stubbs, W. A. (1979). Sensitivity to enkephalin as a cause of non-insulin dependent diabetes. Lancet 1, 341-343. 45. Lightman, S. L., Jacobs, H. S., Maguire, A. K., McGarrick, G., and Jeffcoate, S. L. (1981). Climacteric flushing: Clinical and endocrine response to infusion of naloxone. Br. J. Obstet. Gynecol. 88, 919-924. 46. DeFazio, J., Vorheugen, C., Chetkowski, R., Nass, T., Judd, H. L., and Meldrum, D. R. (1984). The effects of naloxone on hot flashes and gonadotropin secretion in postmenopausal women. J. Clin. Endocrinol. Metab. 58(3), 578-581. 47. Tepper, R., Neri, A., Kaufman, H., Schoenfield, A., and Ovadia, J. (1987). Menopausal hot flushes and plasma fl-endorphins. Obstet. Gynecol. 70, 150-152. 48. Genazzani, A. R., Petraglia, F., Facchinetti, F., Facchini, V., Volpe, A., and Alessandrini, G. (1984). Increase of proopiomelanocortin-related peptides during subjective menopausal flushes. Am. J. Obstet. Gynecol. 149, 775-779. 49. Brtick, K., and Zeisberger, E. (1990). Adaptive changes in thermoregulation and their neuropharmacological basis. In "Thermoregulation: Physiology and Biochemistry" (E. Schtinbaum and E Lomax, eds.), pp. 255-307. Pergamon, New York. 50. Insel, E A., and Motulskey, H. J. (1987). Physiologic and pharmacologic regulation of adrenergic receptors. In "Adrenergic Receptors in Man" (E A. Insel, ed.), pp. 201-236. Dekker, New York. 51. Lambert, G. W., Kaye, D. M., Vaz, M., Cox, H. S., Turner, A. G., Jennings, G. L., and Elser, M. D. (1995). Regional origins of 3-methoxy-4-hydroxyphenylglycol in plasma: Effects of chronic sympathetic nervous activation and devervation, and acute reflex sympathetic stimulation. J. Autom. Nerv. Syst. 55, 169-178. 52. Freedman, R. R., and Woodward, S. (1992). Elevated az-adrenergic responsiveness in menopausal hot flushes: Pharmacologic and biochemical studies. In "Thermoregulation: The Pathophysiological Basis of Clinical Disorders" (Lomax and Sch6nbaum, eds.), pp. 6 - 9 . Karger, Basel.

ROBERT R. FREEDMAN 53. Kopin, I. J., Blombery, E, Ebert, M. H., Gordon, E. K., Jimerson, D. C., Markey, S. E, and Polinsky, R. J. (1984). Disposition and metabolism of M H P G - C D 3 in humans: Plasma MHPG as the principal pathway of norepinephrine metabolism and as an important determinant of CSF levels of MHPG. In "Frontiers in Biochemical and Pharmacological Research in Depression" (E. Usdin, et al., eds.), pp. 57-68. Raven Press, New York. 54. Clayden, J. R., Bell, J. W., and Pollard, E (1974). Menopausal flushing: Double blind trial of a non-hormonal medication. Br. Med. J. 1, 4 0 9 412. 55. Laufer, L. R., Erlik, Y., Meldrum, D. R., and Judd, H. L. (1982). Effect of clonidine on hot flushes in postmenopausal women. Obstet. Gynecol. 60, 583-589. 56. Schmitt, H. (1977). The pharmacology of clonidine and related products. Handb. Exp. Pharmacol. 39, 299-396. 57. Freedman, R. R., Woodward, S., and Sabharwal, S. C. (1990). ce2Adrenergic mechanism in menopausal hot flushes. Obstet. Gynecol. 76(4), 573-578. 58. Goldberg, M., and Robertson, D. (1983). Yohimbine: A pharmacological probe for study of the a 2- adrenoceptor. Pharmacol. Rev. 35, 143180. 59. Starke, K., Gothert, M., and Kilbringer, H. (1989). Modulation of neurotransmitter release by presynaptic autoreceptors. Physiol. Rev. 69, 864-989. 60. Charney, D. S., Heninger, G. R., and Sternberg, D. E. (1982). Assessment of a2-adrenergic autoreceptor function in humans: Effects of oral yohimbine. Life Sci. 30, 2033-2041. 61. Zacny, E. (1982). The role of az-adrenoceptors in the hypothermic effect of clonidine in the rat. J. Pharm. Pharmacol. 34, 455-456. 62. Molnar, G. W. (1981). Menopausal hot flashes: Their cycles and relation to air temperature. Obstet. Gynecol. 57(6)(Suppl.), 52-55. 63. Kronenberg, F., and Barnard, R. M. (1992). Modulation of menopausal hot flashes by ambient temperature. J. Therm. Biol. 17(1), 43-49. 64. Freedman, R. R., and Woodward, S. (1995). Altered shivering threshold in postmenopausal women with hot flashes. Menopause 2(3), 163-168. 65. Freedman, R. R., and Krell, W. (1999). Reduced thermoregulatory null zone in postmenopausal women with hot flashes. Am. J. Obstet. Gynecol. 181(1), 66-70. 66. Delaunay, L., Bonnet, E, Liu, N., Beydon, L., Catoire, E, and Sessler, D. I. (1993). Clonidine comparably decreases the thermoregulatory thresholds for vasoconstriction and shivering in humans. Anesthesiology 79, 470-474. 67. Albright, D. L., Voda, A. M., Smolensky, M. H., Hsi, B., and Decker, M. (1989). Circadian rhythms in hot flashes in natural and surgicallyinduced menopause. Chronobiol. Int. 6(3), 279-284. 68. Horne, J. A., and Reid, A. J. (1985). Night-time sleep EEG changes following body heating in a warm bath. Electoencephalogr. Clin. Neurophysiol. 60, 154-157. 69. Woodward, S., and Freedman, R. R. (1994). The thermoregulatory effects of menopausal hot flashes on sleep. Sleep 17(6), 497-501. 70. Erlik, Y., Tataryn, I. V., Meldrum, D. R., Lomax, P., Bajorek, J. G., and Judd, H. L. (1981). Association of waking episodes with menopausal hot flushes. JAMA, J. Am. Med. Assoc. 245(17), 1741-1744. 71. Thomson, J., and Oswald, I. (1977). Effect of oestrogen on the sleep, mood, and anxiety of menopausal women. Br. Med. J. 2, 1317-1319. 72. Schiff, I., Regestein, Q., Tulchinsky, D., and Ryan, K. J. (1979). Effects of estrogens on sleep and psychological state of hypogonadal women. JAMA, J. Am. Med. Assoc. 242(22), 2405-2414. 73. Scharf, M. B., McDannold, M. D., Stover, R., Zaretsky, N., and Berkowitz, D. V. (1997). Effects of estrogen replacement therapy on rates of cyclic alternating patterns and hot-flush events during sleep in postmenopausal women: A pilot study. Clin. Ther. 19(2), 304-311. 74. Ettinger, B. (1998). Overview of estrogen replacement therapy: A historical perspective. Proc. Soc. Exp. Biol. Med. 217 (1), 2-5.

CHAPTER 14 Menopausal Hot Flashes 75. Nagamani, M., Kelver, M. E., and Smith, E. R. (1987). Treatment of menopausal hot flashes with transdermal administration of clonidine. Am. J. Obstet. Gynecol. 156(3), 561-565. 76. Hoffman, J. W., Benson, H., Arns, E A., Stainbrook, G. L., and Langsberg, G. L. (1982). Reduced sympathetic nervous system responsivity associated with the relaxation response. Science 215, 190-192. 77. Germaine, L. M., and Freedman, R. R. (1984). Behavioral treatment of menopausal hot flashes: Evaluation by objective methods. J. Consult. Clin. Psychol. 52(6), 1072-1079. 78. Irvin, J. H., Domar, A. D., Clark, C., Zuttermeister, E C., and Friedman, R. (1996). The effects of relaxation response training on menopausal symptoms. J. Psychosom. Obstet. Gynecol. 17, 202-207. 79. Wijima, K., Melin, A., Nedstrand, E., and Hammar, M. (1997). Treatment of menopausal symptoms with applied relaxation: A pilot study. J. Behav. Ther. Exp. Psychiatry 28(4), 251-261.

227 80. Wyon, Y., Lindren, R., Lundeberg, T., and Hammar, M. (1995). Effects of acupuncture on climacteric vasomotor symptoms, quality of life, and urinary excretion of neuropeptides among postmenopausal women. Menopause 2(1), 3-12. 81. Brezezinski, A., Adlecreutz, H., Shaoul, R., R6sler, A., Shmueli, A., Tanos, V., and Schenker, J. G. (1997). Short-term effects of phytoestrogen-rich diet on postmenopausal women. Menopause 4(2), 8994. 82. Murkies, A. L., Lombard, C., Strauss, B. J., Wilcox, G., Burger, H. G., and Morton, M. S. (1995). Dietary flour supplementation decreases post-menopausal hot flushes: Effect of soy and wheat. Maturitas 21(3), 189-195. 83. Albertazzi, E, Pansini, E, Bonaccorsi, G., Zanotti, L., Forini, E., and De Aloysio, D. (1998). The effect of dietary soy supplementation on hot flushes. Obstet. Gynecol. 91, 6-11.

_~HAPTER | .

Cardiovascular Pathophysiology CAROL A.

DERBY

New England Research Institutes, Watertown, Massachusetts 02472

IV. Studies of Menopause and Cardiovascular Risk Factors V. Conclusions References

I. Introduction II. Sex-Specific Trends in Cardiovascular Risk with Age III. Epidemiologic Studies of Menopause and Cardiovascular Outcomes

Data regarding endogenous estrogen levels and cardiovascular risk are rare. One cross-sectional and one prospective study have examined circulating estrogen concentrations and cardiovascular disease in postmenopausal women, both with negative results [3,10]. Cauley et al. [10] found no cross-sectional association between estrone concentrations and coronary artery occlusion in a series of 87 postmenopausal women evaluated with coronary arteriography. The only prospective study to date found that age-adjusted endogenous estrogen concentrations did not predict cardiovascular death or fatal ischemic heart disease in a cohort of 651 postmenopausal women [3 ]. The remaining evidence for the hypothesis that menopause increases cardiovascular risk is based on comparisons of age-specific disease rates in men and women, observational studies of cardiovascular outcomes in women with menopause at an early age, and studies of risk factor changes around the time of menopause. This chapter reviews these lines of evidence based on vital statistics data, epidemiologic studies of menopause and cardiovascular outcomes, and studies of menopause in relation to blood pressure, lipids, and hemodynamic factors. Direct effects of estrogen on the arterial wall are also briefly summarized.

I. I N T R O D U C T I O N Cardiovascular diseases are the leading cause of death among women in the United States and in most developed countries [1]. In the United States, all cardiovascular diseases combined claim the lives of one-half million women annually, twice the number of deaths attributable to all forms of cancer [ 1,2]. Among women over the age of 50, these diseases account for over 50% of all deaths, and are among the major causes of morbidity and disability in postmenopausal women [2]. It has been hypothesized that the cessation of ovarian function with menopause is associated with increased cardiovascular risk, and that the increase is mediated by estrogen. However, whether menopause is independently associated with increased cardiovascular risk remains a topic of debate. The best evidence in support of this hypothesis is from studies of exogenous estrogen replacement [3-5]. However, these data are based predominantly on observational studies, and there is concern that results may be biased by selection factors that determine who uses hormone replacement therapy [6-9]. The cardiovascular consequences of hormone replacement therapy are discussed in Chapter 37.

MENOPAUSE:


229


230

CAROL A. DERBY

II. SEX-SPECIFIC CARDIOVASCULAR WITH

TRENDS

IN

RISK

AGE

Rates of coronary disease are very low among young women and increase sharply with age, with the largest apparent increase around the age of 50 years. Women have lower age-specific rates of fatal coronary heart disease (CHD) than do men at every age [5]. The sex ratio for fatal CHD is remarkably consistent across populations with varying rates of heart disease and lifestyles, with a 2.5- to 4.5-fold excess risk in men aged 4 5 - 6 9 years [11]. The sex differential in CHD deaths is greatest around age 50 years and diminishes thereafter [5,12-16]. The consistency of an apparent protective trait in women along with the decline in the sex ratio for CHD deaths around the age of menopause have been cited as evidence that natural menopause increases CHD risk [3,12-15,17]. However, closer examination of vital statistics data does not support this hypothesis. If natural menopause were responsible for increased CHD mortality, then the rate of increase in mortality rates with age would be expected to accelerate in the postmenopausal period. That is, when plotted on a log scale, the slope of the curve for age-specific death rates would be steeper in women over the age of 50 years compared with that for younger women. Examinations of vital statistics data for the United States [5,12,16], New York state [14], and England and Wales [13] have all failed to demonstrate an inflection point in the rate of increase in CHD mortality with age. This is in contrast to plots of the mortality rates for breast cancer, which show a clear shift around the age of menopause [5,16]. Similarly, close inspection of sex-specific rates of increase in CHD mortality with age suggest that menopause is not responsible for diminution of the male-to-female mortality ratio around age 50 years. Rather, the narrowing of the gender gap in CHD mortality is due to a slowing of the rate of increase in male death rates after age 50 years, and is not attributable to a postmenopausal shift in the rates of change for women [12,13,16]. Furthermore, although the average age of menopause has been reported to be fairly consistent across countries and over time [18], the age at which the male-to-female ratio of CHD deaths peaks has been shown to vary from 30 to 35 years in the Netherlands, to 55 to 60 years in Japan [16]. In summary, there is a widely held belief that estrogen protects premenopausal women from cardiovascular disease and that this protection is substantially reduced with the menopause [19]. However, analyses of age-specific disease rates in men and women have consistently refuted this assumption. Thus, vital statistics data do not support the hypothesis that menopause is accompanied by an increase in cardiovascular risk.

III. EPIDEMIOLOGIC OF MENOPAUSE

STUDIES

AND

CARDIOVASCULAR

OUTCOMES

Studies of menopause and cardiovascular risk have yielded inconsistent results, particularly studies of natural menopause. In part, this may be attributed to a range of methodologic issues. Many studies have not distinguished between natural and surgical menopause, or between women with hysterectomy alone and those with bilateral oophorectomy [15,20-22]. Definitions of menopause are not consistent across studies and often rely on self-report [15,22-29]. The definition of early menopause has ranged from 35 to 50 years, with age at menopause frequently based on recall. Inaccurate recall of age at menopause may be particularly problematic for studies of natural menopause. In some instances, specific age of menopause is not reported [23,28]. Risk estimates from many studies of coronary disease and early menopause are based on small numbers of cases, particularly in the premenopausal control group [24,29,30]. Several studies have included a large proportion of nonspecific, or "soft," cardiovascular end points such as angina, with very few more definitive cases such as myocardial infarction [21,29-33]. Because the majority of longitudinal studies begin with women at or very close to menopause, there are few longitudinal data on changes during the menopausal transition period preceding the actual cessation of menses. This perimenopausal period is characterized by wide fluctuations in circulating estrogen and may mark the beginning of estrogen-mediated changes in cardiovascular risk factors [18]. In general, prospective studies have been based on follow-up times that are relatively short given the lengthy natural history of cardiovascular disease. Many studies have not controlled for confounding by smoking, a major cardiovascular risk factor that is also associated with younger age at menopause [ 17]. Finally, it is very difficult to discriminate between changes related to aging and those attributable to menopause per se. Precise adjustment for age is crucial yet is often lacking [17]. A detailed discussion of methodologic issues in the study of menopause is included in Chapter 10.

A. S t u d i e s w i t h T y p e o f M e n o p a u s e U n s p e c i f i e d A few published studies relating early menopause to cardiovascular risk have failed to specify whether menopause was natural or surgical. Two of these [21,22] suggested that early menopause is associated with increased risk of cardiovascular disease, whereas a third [20] found no increased risk of fatal coronary heart disease among women with early menopause. Oliver [21] described a case-series of women with clinical coronary heart disease under the age of 45

CHAPTER 15 Cardiovascular Pathophysiology years, and reported that compared with women in the general population, a high proportion had experienced menopause (20% compared with 3 to 5%). Similarly, a population-based case-control study in Gothenberg, Sweden, showed that women with clinical heart disease were significantly more likely than healthy controls to have reached menopause prior to age 50 years [22]. In contrast, a case-control study reported by Mann and Inman [20] found no elevated risk of fatal coronary heart disease among women with menopause before age 50 years. Given that the type of menopause was not specified, and that a control group was either missing entirely [21 ] or was not well defined [20,22], inferences from these studies are limited.

B. S t u d i e s o f N a t u r a l M e n o p a u s e A number of studies have examined the relationship between natural menopause and cardiovascular risk, with inconsistent results. Sznajderman and Oliver [33] described a case-series of women evaluated for premature cessation of menstruation between the ages of 35 and 40 years, and determined the proportion that developed clinical features of ischemic heart disease 15 to 20 years later. When compared with a general population sample, the women with early menopause had a sevenfold increased risk. However, the interpretation of this study is limited by the lack of an appropriate control group and by the small number of cases. In a cross-sectional study, Witteman et al. [24] measured the extent of calcification in the abdominal aorta in a population-based sample of women aged 4 5 - 5 5 years. After adjustment for age and coronary risk factors, women with natural menopause were estimated to have a 3.4-fold elevated risk of calcifications compared with premenopausal women (95% confidence interval, 1.2-9.7). Among women with natural menopause, there was evidence for an increasing trend in risk of calcification with increasing time since menopause. However, estimates were based on small numbers of women with calcifications, and the temporal association between outcome and exposure could not be determined. Case-control studies have suggested either a positive association between early menopause and cardiovascular risk [34], or no association [23,25]. A study in Rochester, Minnesota compared incident cases of coronary heart disease prior to age 60 years and hospital-based controls on the proportion with menopause before age 50 years [34]. The estimated relative risk of myocardial infarction for women with early menopause was 1.3, and was not statistically significant. A hospital-based case-control study by LaVecchia et al. [23] also found no association between natural menopause and risk of myocardial infarction prior to age 55 years. This study did not specify age of menopause in cases and controls.

231 The largest reported case-control study of cardiovascular disease and menopause was based on data from the baseline interview of the Nurses Health Study [25]. Women under 56 years of age, with a history of myocardial infarction (N = 279), were age matched with over 5000 control women who had no history of infarction. The estimated relative risk of myocardial infarction for women with natural menopause compared with premenopausal women was 0.9 with a 95% confidence interval of 0.6-1.3. Multivariate adjustment for cardiovascular risk factors, including smoking, further attenuated this association. Myocardial infarction risk was elevated in women with natural menopause before age 35 years, although the relative risk for this age stratum was based on few cases and was not statistically significant. Prospective studies of natural menopause and cardiovascular risk have also failed to yield consistent results, with some suggesting increased risk in postmenopausal women [27,29,30] and others showing no evidence for elevated risk [26,28]. The Framingham study compared coronary disease event rates for 1934 women with natural menopause during follow-up and age-matched premenopausal women [29,30, 35]. Among women aged 4 5 - 5 4 years, the relative risk for coronary heart disease was 2.7, and was statistically significant. This risk estimate is based on a total of only 10 cases, the majority of which were angina. Risk estimates in other age strata were not significant [29]. An update from this study has reported a relative risk of 4.1 for natural menopause in women aged 5 0 - 5 9 years [35]. None of these analyses were adjusted for chronological age or smoking status. A study by van der Schouw et al. [27] has provided additional data from a large population-based study with 20 years of follow-up on 12,115 women initially aged 5 0 65 years. The age-adjusted hazard ratio for age at menopause was 0.982 per year (95% confidence interval, 0.968-0.996). Thus, for each year of delay in menopause (surgical and natural combined), cardiovascular mortality risk decreased by 2%. Adjustment for cardiovascular risk factors, including smoking, did not alter these results substantially. However, when natural menopause was analyzed separately, the hazard ratio did not reach statistical significance (0.98; 95% confidence interval, 0.97-1.00). These results may be limited by the fact that menopausal status and age of menopause were based on self-report. A population-based prospective study in Gothenburg, Sweden [26] examined 12-year incidence rates for various cardiovascular end points in relation to menopausal age in a cohort of 1462 women. Relative risks for myocardial infarction and angina were slightly elevated regardless of whether early menopause was defined at 40, 45, or 50 years. However, none of these estimates was statistically significant, and no adjustment was made for potential confounding by smoking, and menopausal age was determined by respondent recall.

232

CAROL A. DERBY

The largest prospective study of natural menopause and cardiovascular risk is the Nurses Health Study, which reported on 12 years of follow-up for 121,700 women aged 3 0 55 years at baseline [28]. After adjustment for age in 5-year increments, relative risk of coronary disease women with natural menopause was 1.7 (95% confidence interval, 1.12.8). However, more precise adjustment for age, in 1-year increments, showed no increased cardiovascular risk with natural menopause (relative risk 1.2, 95% confidence limits, 0.8-1.8). This study highlights the importance of adjusting closely for chronological age and smoking status [ 17]. Similarly, another study of women in Italy who were followed for 16 years found that although cardiovascular morbidity and mortality were higher among women with natural menopause as compared with premenopausal women, these differences were no longer apparent after age adjustment [36].

C. S t u d i e s o f S u r g i c a l M e n o p a u s e In contrast to the inconclusive findings for natural menopause, studies of bilateral oophorectomy have been more consistent, suggesting that early bilateral oophorectomy increases risk of cardiovascular disease. Nevertheless, it should be noted that studies of surgical menopause are limited by many of the methodologic issues outlined above. Furthermore, comparisons across studies are complicated by inconsistencies in the reference groups used. Oophorectomized women have been compared with women who had hysterectomy only, with premenopausal women, or with naturally menopausal women. Overall, autopsy studies suggest that bilateral oophorectomy is associated with an increased degree of coronary atherosclerosis. Autopsy studies provide clear definitions of ovarian status as well as anatomic definitions of disease end points. However, the data are cross-sectional, selection factors that determine inclusion in autopsy series are difficult to quantify, and control for confounding factors is limited by lack of information. Parrish et al. [37] studied the autopsy records of 80 women who had bilateral oophorectomy prior to age 50 years, comparing the extent of coronary atherosclerosis with that in age-matched women with intact ovaries. Bilateral oophorectomy was associated with more extensive atherosclerosis only among women with surgery prior to age 40 years. Wuest et al. [38] compared autopsy findings for oophorectomized women, a control series of women, and a series of men. Women with bilateral oophorectomy had more extensive coronary atherosclerosis than did female controls, and oophorectomized women had disease rates that approximated those of men in the same age groups. Similarly, Rivin and Dimitroff [39] compared the autopsy records of women with bilateral oophorectomy before age 50 years and those from a series of 600 control women. The oophorectomized

group had two to three times the rate of severe atherosclerosis compared to controls. The latter two studies are limited in that the composition of the female control group with respect to menopause status and the comparability of the ages for cases and controls are not clear. In contrast to these studies, Novak and Williams [40] found no significant difference in the extent of atherosclerosis in oophorectomized women compared with controls. Analyses by age at menopause also showed no effect, although the sample sizes in the age strata were quite small and there was a wide range of ages at death. A few cross-sectional studies have examined anatomically defined atherosclerosis in samples of women undergoing arteriography or noninvasive radiographic screening tests. Manchester et al. [41 ] compared the prevalence of angiographically documented coronary artery disease for 20 women with history of bilateral oophorectomy at least 5 years prior to angiography, and a control group of 65 women. All women were under the age of 50 years, and had been referred for angiography for the evaluation of angina. The majority of the control group (95%) was premenopausal. The relative odds of significant coronary artery disease (greater than a 50% occlusion) was 0.6 (p = 0.4) for oophorectomized women compared with controls. This estimate was based on a small sample of women, and only five in the oophorectomy group had significant coronary disease. In addition, arteriography studies are subject to a number of selection biases that may limit their generalizbility [42]. A much larger cross-sectional study of anatomically defined disease found that women with bilateral oophorectomy had 5.5 times the risk of calcifications in the abdominal aorta than did premenopausal women of the same age [24]. Women with hysterectomy and intact ovaries had no excess risk. These results support the hypothesis that increased cardiovascular risk in oophorectomized women is due to loss of ovarian function. Oliver and Boyd [43] have also reported results consistent with this hypothesis. Women with either bilateral or unilateral oophorectomy before age 35 were compared with respect to the incidence of clinical coronary heart disease in the 25 years following surgery. Those with bilateral oophorectomy appeared to have an elevated risk of coronary heart disease compared with either the unilateral oophorectomy group (odds ratio approximately 10) or with a general practice sample (odds ratio approximately 8). In contrast, the coronary heart disease rate for women with unilateral oophorectomy was similar to that in the general medical practice sample. In general, case-control studies of myocardial infarction and surgical menopause also suggest that surgical menopause increases risk. However, in several of these studies sample sizes were small and results were not statistically significant. At least one study has suggested that factors leading to gynecologic surgery may be related to increased cardiovascular risk independent of ovarian status [44]. Ritterband et al. [44] reviewed hospital records from 10 years prior

CHAPTER 15 Cardiovascular Pathophysiology to identify women with oophorectomy and a group with hysterectomy only before the age of 35. The women were evaluated for history or physical evidence of coronary heart disease. Coronary disease rates for women with hysterectomy only and those with bilateral oophorectomy were similar even after adjusting for age and estrogen replacement therapy. When compared with a control group of their sisters, both gynecologic surgery groups had twice the rate of heart disease, although the number of cases was small and differences were not statistically significant. Robinson et al. [31] also compared cardiovascular disease prevalence in women with history of oophorectomy or hysterectomy only prior to age 45. In contrast to the previous study, the estimated risk of clinical coronary heart disease was approximately 4.0 times higher for women with oophorectomy, and the difference was statistically significant. Winklestein et al. [15] studied 50 women with myocardial infarction and compared the prevalence of surgical menopause with prevalence in paired neighbor controls and with controls from a random probability population sample. Myocardial infarction patients had a higher probability of having had surgical menopause, although the result was only of borderline statistical significance. This study did not distinguish between surgical menopause with or without bilateral oophorectomy. In another hospital-based case-control study, La Vecchia et al. [23] studied women with myocardial infarction (MI) before age 55 years, and a group of hospital controls. They reported no association between surgical menopause and acute myocardial infarction (estimated relative risk 0.7, 95% confidence interval, 0.39-1.23). It was not clear whether the surgical menopause group excluded women with hysterectomy only and intact ovaries. In a similar study, Beard et al. [34] identified women with coronary heart disease prior to age 60 years, and compared the prevalence of surgical menopause with a group of age-matched controls from the Mayo Clinic. The relative risk for surgical menopause compared with natural menopause was 1.6, and was of borderline statistical significance (95% confidence interval, 1.0-2.5). Johansson et al. [32] identified a group of women who had undergone bilateral oophorectomy between 1904 and 1910, and attempted to obtain follow-up information regarding their cardiovascular experience. This was compared with the cardiovascular status of women with surgery for uterine prolapse. Among women under the age of 65 at follow-up, the estimated odds ratio for coronary heart disease was 3.0 but was not statistically significant. This study is limited in that follow-up information was obtained for less than half of the women with oophorectomy, and risk estimates were based on only 17 cases and 17 controls. The largest case-control study of surgical menopause and cardiovascular risk was from the Nurses Health Study [25]. Bilateral oophorectomy was associated with an overall estimated relative risk of 2.9 (95% confidence interval, 2.1-

233 4.0). Risk associated with oophorectomy increased with decreasing age at surgical menopause. Women with hysterectomy without bilateral oophorectomy had no increased risk of acute myocardial infarction (estimated relative risk 0.9, 95% confidence interval, 06-1.3). This finding is consistent with the cross-sectional results of Witteman et al. [24], and in contrast to the case-control study by Ritterband et al. [44]. Prospective data also suggest that bilateral oophorectomy is associated with increased cardiovascular risk. In the Framingham study, women with surgical menopause tended to have elevated risk of coronary heart disease, although, the association was significant only among women aged 4 0 - 4 4 years [29,30]. In addition, the number of CHD cases was small, and angina constituted the majority of cases. A more recent prospective study estimated the hazards ratio for age at menopause among women with hysterectomy only and for those with bilateral oophorectomy [27]. Age at menopause was not associated with cardiovascular mortality among those with hysterectomy only. Among women with oophorectomy, there was a significant inverse relation between age of menopause and cardiovascular mortality risk, with a 6% reduction in annual risk with each year increase in age of menopause. The Nurses Health Study found that after adjusting for age and smoking, women with hysterectomy and bilateral oophorectomy had a relative risk of 2.2 (95% confidence interval, 1.2-4.2) [28]. This study provides some of the strongest evidence to date that increased cardiovascular risk following surgical menopause is the result of decreased estrogen levels. Although women with surgical menopause had a twofold increased risk of cardiovascular disease, women who used hormone replacement therapy after bilateral oophorectomy had no greater risk than did premenopausal women.

D. S u m m a r y Epidemiological evidence that natural menopause increases cardiovascular risk is lacking. A number of studies ranging from case series to cohort studies have shown inconsistent results. This inconsistency may be the result of a number of methodologic issues, variations in study design, or definitional inconsistencies. Given that the natural history of cardiovascular disease can span decades and that ovarian function declines gradually for a period of 10 to 15 years prior to the cessation of menses, the lack of an abrupt increase in cardiovascular risk following natural menopause is not surprising [5,17,30]. Data regarding surgical menopause have been more consistent suggesting that bilateral oophorectomy prior to age 40 or 45 years is associated with increased cardiovascular risk. The lack of increased risk in women with bilateral oophorectomy who take hormone replacement therapy is the

234

CAROL A. DERBY

best evidence that estrogen is the mediating factor in this relationship. However, factors leading to surgical menopause may be related to cardiovascular risk and it is not possible to separate these factors from the effects of decreased postsurgical estrogen levels [44,45]. Thus, the role of menopause in the development of atherosclerotic disease remains controversial. Larger studies, using standard definitions and with long-term follow-up beginning prior to the perimenopausal period, are needed to clarify the relation of natural menopause to cardiovascular risk. The inferences from studies of surgical menopause would be strengthened by additional data regarding factors leading to surgery. Finally, studies regarding the role of endogenous estrogen levels in relation to cardiovascular outcomes are needed to confirm whether there is a causal link between menopausal declines in estrogen and cardiovascular risk.

IV. STUDIES OF MENOPAUSE AND CARDIOVASCULAR RISK FACTORS An alternative approach to evaluating the influence of menopause on cardiovascular risk is to examine the relationship between menopausal status and cardiovascular risk factors. In this section the epidemiological evidence for associations between menopause status and the traditional risk factors--smoking, blood pressure, and lipids--is discussed. Also discussed is the association between menopause and hemodynamic factors and evidence for direct effects of estrogen on arterial walls. The associations of menopause with carbohydrate metabolism and body composition are described in Chapter 16.

A. Smoking Although menopause status is not a determinant of smoking behavior, smoking is a potentially strong confounder in studies of cardiovascular risk and menopause [17,28,45]. Smoking is well established as a coronary risk factor, both independently [46] and via positive associations with blood pressure and lipids [47-49] and an inverse association with body mass index [50]. In addition, smoking has consistently been associated younger age of menopause. Women who smoke cigarettes experience menopause 1 to 2 years earlier than do nonsmoking women [51-53]. Proposed mechanisms include lower estrogen levels, more rapid estrogen metabolism, accelerated aging of ovarian follicles, and impaired estrogen receptor binding in smokers [54]. An understanding of this association is important to the interpretation of studies regarding menopausal changes in cardiovascular risk.

B. Blood Pressure Blood pressure increases gradually with age in both men and women. However, there is a cross-over in the relative levels of blood pressure after the age of menopause. Although women have lower levels of systolic and diastolic blood pressure prior to age 60 years, men tend to have lower levels at older ages [55]. Similarly, hypertension is less prevalent in women until middle age, when the sex difference disappears or reverses [2]. These patterns have led to the hypothesis that estrogens may protect premenopausal women from hypertension. Studies showing blood pressure fluctuations with the menstrual cycle have also suggested an association with female hormones [56-58]. However, there is no clear evidence linking menopause with adverse changes in blood pressure. Four decades ago, Taylor et al. [59] observed that arterial hypertension was no more common in a series of postmenopausal women than in the general population. The interpretation of this study was limited given that inferences were based on a case series and that the study lacked an adequate control group. Nevertheless, subsequent studies have failed to produce consistent evidence that would refute Taylor's early finding of no association between menopause and blood pressure. Cross-sectional studies of blood pressure in relation to menopause status have yielded discrepant results. These have shown no difference between premenopausal and postmenopausal women [56,60-62], or have found increased systolic [63], decreased systolic [64], or increased diastolic blood pressure [65] in postmenopausal women. Wu et al. [56] found in increased prevalence of both hypotension and hypertension in postmenopausal women, whereas Portaluppi et al. [61] reported no difference in the prevalence of hypertension after adjusting for differences in age and body mass index. In general, prospective studies have failed to show an association between menopause and blood pressure. Data from the Framingham study indicate no relationship between changes in menopausal status and changes in systolic or diastolic blood pressure [66]. These comparisons considered the effects of natural or surgical menopause with and without bilateral oophorectomy. Matthews et al. [67,68] followed a cohort of women through the menopause transition from pre- to peri- to postmenopause, and found no significant changes in either systolic or diastolic blood pressure. A longitudinal population study in Gothenberg, Sweden also observed no increase in blood pressure for women who became postmenopausal during follow-up [69]. In fact, this study reported that systolic blood pressure decreased with increasing time since menopause. Declining blood pressure with increasing time since menopause was also observed in a cohort of 168 perimenopausal women followed for 7 years in Sweden [70].

CHAPTER 15 Cardiovascular Pathophysiology In contrast, two recent prospective studies have shown increases in systolic blood pressure with menopause. Poehlman et al. [71] followed a small sample of 38 women for 6 years, and found that systolic blood pressure increased significantly more among women who became postmenopausal during follow-up. The observed blood pressure changes were not correlated with changes in body fatness. A larger study reported by Staessen et al. [72] followed 315 women for 5 years and showed an increase in systolic blood pressure of 3 - 4 mm Hg for postmenopausal women, but no change among premenopausal women. This difference persisted after adjustment for age, changes in body mass index, and antihypertensive medication use. A higher incidence of hypertension among postmenopausal women was also reported, although this was not statistically significant after adjustment for confounders. Increases in blood pressure around the time of menopause may be the consequence of aging, rather than changing hormonal status. Increasing blood pressure with age has been documented for women with unchanged menstrual status [67-69]. A 30-year follow-up study of over 4000 female atomic bomb survivors in Japan found an increasing trend in systolic blood pressure with age, but no change in the slope of this trend around the time of either natural or surgical menopause [73]. Similarly, van Beresteyn et al. [70] observed increasing systolic blood pressure with chronological aging, but no blood pressure effect of menopause. Weight gain around the time of menopause, rather than menopause per se, may also influence blood pressure levels. Excess body weight and excess abdominal fat are positively correlated with increased blood pressure [47], and an ageassociated weight gain has been observed among women around the time of menopause [67,74]. Both cross-sectional [56,61] and prospective studies [70,74] have shown a positive correlation between weight gain and blood pressure among women of menopausal age, but no independent blood pressure effect of menopause. In contrast, others have reported that adjustment for body mass index did not explain a positive relation between menopause and systolic blood pressure [63,71,72]. In summary, the preponderance of evidence from epidemiological studies does not support a causal association between menopause and adverse changes in blood pressure. Whether menopause exerts an independent effect on blood pressure is difficult to assess given the joint effects of age and blood pressure. If menopause does exert an adverse effect, the lack of consistency across numerous studies suggests that it may be weak, and limited to systolic blood pressure.

C. L i p i d s Perhaps the best evidence for an effect of menopause on CVD risk is that suggesting adverse changes in the lipid pro-

235 file around the time of menopause. This evidence is based on sex differences in age trends for lipoproteins and results from cross-sectional and longitudinal population studies and studies of postmenopausal hormone replacement therapy. Total cholesterol levels increase with age in both sexes. However, among men, the increase begins in the third decade of life, whereas in women the age-related increase is delayed until the fifth decade [75]. Similarly, although low-density lipoprotein (LDL) cholesterol increases with age for men and women, the timing of the increase differs by sex. Women experience a more gradual increase until middle age, when their rate of increase accelerates and that of men levels off [76,77]. In both sexes, increasing LDL with age is thought to be the result of an age-dependent decline in LDL receptors [78]. Animal experiments have shown that estrogens increase LDL receptors [79,80]. These observations have led to the hypothesis that age-related increases in LDL for women are modulated by the presence of estrogen, and that increasing LDL levels around the time of menopause may be the result of a reduction in LDL receptors triggered by declining estrogen levels [77,81 ]. In contrast, sex-specific trends in high-density lipoprotein (HDL) cholesterol with age suggest that differential levels of testosterone rather than estrogen determine sex differences [77]. During adolescence, HDL levels rise slowly in girls, and decline abruptly in boys in conjunction with sexual maturation [76,77,82]. HDL levels remain constant in adults, with levels in men remaining below those in women, even after menopause [54,76]. The majority of observational studies, conducted in various populations with differing definitions and study designs, have supported an association between menopause and increasing serum cholesterol levels. These include studies in the United States [30,65,66,71,83-85], Europe [60,64,69, 86,87], Britain [88], Japan [73,89], and China [56]. Most comparisons have been for women with natural menopause and premenopausal controls, with higher total cholesterol levels in the postmenopausal group. This has been observed in both cross-sectional [56,60,64,87,89] and longitudinal studies [30,66,69,86]. Similar findings have been reported when the postmenopausal group included both naturally and surgically menopausal women [30,62,63,65]. One crosssectional [88] and one prospective study [73] have compared total cholesterol levels in naturally menopausal women with those in men. Razay et al. [88] reported that age predicted increasing total cholesterol levels in women, and that women over 50 years of age had higher levels than did younger women. The lack of change in total cholesterol levels with age in men was interpreted as suggesting that the observations in women were due to menopause. It is important to note that this evidence is indirect, and the study is limited by the fact that postmenopausal status was based solely on age. Akahoshi et al. [73] compared longitudinal changes in lipids for women who became naturally menopausal and

236 age-matched male controls. A sharp increase in total cholesterol levels around the time of menopause was observed only in the women. Increased serum total cholesterol levels in postmenopausal women appear to be independent of age and are not explained by increased body mass index among postmenopausal women [56,60,63,65,66,73,85,87,89], although not in all studies [36,90]. A cross-sectional study by Campos et al. [90] and a prospective analysis reported by Casiglia et al. [36] each reported that differences in total cholesterol levels for pre- and postmenopausal women were no longer statistically significant after adjustment for age [36], or age and body mass index [90]. Total cholesterol levels also appear to be elevated among women with surgical menopause, although fewer studies have addressed this issue. A small, cross-sectional study reported by Notelevitz et al. [83] showed elevated total cholesterol levels in young women with bilateral oophorectomy compared to age-matched women with intact ovaries. Hjortland et al. [66] demonstrated prospectively that relative to premenopausal controls, the increase in total cholesterol levels for women with surgical menopause was similar to the increase observed for women with natural menopause. Similarly, the prospective study of Akahoshi et al. [73] showed that serum cholesterol levels increased abruptly with menopause, whether surgical or natural, and that no such change occurred in an age-matched group of male controls. There are few data regarding the relation of menopause to serum cholesterol levels in nonwhite women. The few studies to have examined this issue suggest that racial differences exist [84,85]. A report from the Evans County Cardiovascular Disease Study [85] showed no significant relationship between menopause status and total cholesterol levels in black women, and a significant relationship among whites that was independent of age, body mass index, and smoking. A more recent cross-sectional analysis of data from the Minnesota Heart Survey found similar results. Demirovic et al. [84] reported higher total cholesterol levels in white postmenopausal women compared with premenopausal controls. In contrast, among blacks, postmenopausal women did not have significantly higher total cholesterol levels than did premenopausal women. The lack of an association between menopause and serum cholesterol levels has also been reported among Pima Indian women in the southwestern United States [91]. More information is required regarding menopausal changes in minority women. Increases in total cholesterol with menopause appear to be attributable to increases in LDL [30,63,67,71,86,90]. This is consistent with the hypothesis that declining LDL receptors in response to decreasing estrogen levels mediate menopausal changes in lipids and lipoproteins. A relationship between menopause and LDL levels has been observed in cross-sectional [60,63,90] and prospective studies [67,

CAROL A. DERBY

71,86]. Studies have not consistently controlled for age, although those that did incorporate age controls have suggested an independent effect [60,63,90]. In contrast, Matthews et al. [68] have studied women as they transition from premenopause through perimenopause, and concluded that elevations in LDL are associated with aging and not with menopausal status. Cross-sectional results from the Framingham Offspring Study [90] have suggested that menopause may be related to changes in the quality as well as the quantity of LDL. Postmenopausal women in this study were shown to have significantly increased numbers of small, dense LDL particles compared to premenopausal women of the same age. These particles, rich in apoprotein-B, have been associated with risk of premature coronary artery disease [92,93]. Several studies have shown a relation between increased triglyceride levels and either natural [56,63,69,71,87] or surgical menopause [83]. However, this has not been shown consistently. Other studies have reported either no association or have shown that adjustment for age attenuated the association between menopause status and triglycerides [36,60,90]. Studies of HDL and menopause have also yielded inconsistent findings. Cross-sectional studies have found menopause to be associated with decreased [83,90] or increased [56] HDL, or have failed to show an association [30,60, 63,84,89,94]. Prospective studies have tended to show decreasing HDL with menopause [67,71,86]. Jensen et al. [86] longitudinally studied premenopausal women as they transitioned through menopause, and reported that HDL declined gradually beginning in the 2 years preceding the cessation of menses. Thus, the apparent discrepancy between crosssectional and prospective studies might be explained by the inability of cross-sectional analyses to detect small, gradual changes in HDL [86]. In contrast, Matthews et al. [68] have suggested that changes in HDL may be the effect of aging rather than menopause status. The failure of observational studies to demonstrate a consistent relation between menopause and HDL seem to conflict with studies of exogenous hormone therapy, which have consistently shown increased HDL with treatment [95-97]. This apparent discrepancy may be due to the direct hepatic effects of pharmacological doses of oral estrogen on lipid metabolism [77]. The HDL effects of hormone replacement therapy appear to be attenuated with transdermal hormone administration, which attains hormone levels more closely resembling physiologic levels and does not stimulate the liver on absorption [98]. Lipoprotein(a) [Lp(a)] has been established as a coronary risk factor [99,100], yet data regarding the relation of menopause to Lp(a) are limited. Jenner et al. [101] reported no differences in Lp(a) levels for postmenopausal and premenopausal women, after adjusting for age. This is in contrast to

CHAPTER 15 Cardiovascular Pathophysiology the cross-sectional study reported by Heinrich et al. [102] that showed significantly higher Lp(a) levels in postmenopausal women. This study also showed that concentrations of Lp(a) levels changed slightly in both men and women prior to age 45 years, whereas after age 45, levels in women increased steadily. Further studies are needed to determine the effects of menopause on Lp(a). The best evidence that estrogens mediate changes in lipoproteins with menopause is derived from studies of women taking exogenous hormone replacement therapy (Chapter 37). Although these findings have been generally consistent, the majority are based on observational data, and concerns have been raised regarding the influences of selection biases on the characteristics of women who take hormone replacement therapy [6,7,9]. However, in the largest randomized placebo controlled trial to date, the Postmenopausal Estrogen/Progestin Interventions (PEPI) Trial, estrogen replacement therapy, either opposed or unopposed, was associated with significant increases in HDL and decreases in LDL [95]. Data directly linking endogenous estrogen levels with lipid and lipoprotein changes are limited and inconsistent. Wu et al. [56] showed that premenopausal women had increased estradiol levels relative to postmenopausal women, and that estradiol levels were inversely associated with total cholesterol. In contrast, a cross-sectional study of women who had been postmenopausal for an average of 9 years found that endogenous estrogens were not related to lipids [103]. Similarly, a prospective study reported by BarrettConnor and Goodman-Gruen [3] found that estrone concentrations were only weakly related to total cholesterol levels. A few studies have examined the time course of lipid and lipoprotein changes in relation to menopause, and have provided indirect evidence that these changes are mediated by declining estrogen levels. Declining estradiol concentrations have been shown in women approaching menopause [ 104,105], and several studies have suggested a temporal relationship between lipid and lipoprotein changes and this perimenopausal decline. Jensen et al. [86] showed that lipid changes were correlated with decreasing estrogen and increasing gonadotrophins during the perimenopausal period. Akahoshi et al. [73] demonstrated prospectively that increased total cholesterol preceded natural menopause by approximately 3 years. Hjortland et al. [66] also found higher premenopausal serum total cholesterol levels in women approaching menopause compared with those who remained premenopausal during follow-up, although the timing of evaluations in relation to menopause may have caused these differences. Lindquist and Bengtsson [64] found that premenopausal increases in total cholesterol were related to the time remaining until menopause. Finally, a cross-sectional study of total and HDL cholesterol levels in premenarchal girls and menstruating, pregnant, and post-

237 menopausal women also suggested that ovarian hormones influence lipid levels [89]. The timing of lipid changes in relation to surgical menopause also suggests a temporal association with the loss of ovarian function. Serum cholesterol levels have been shown to increase abruptly following bilateral oophorectomy [73]. In addition, the Framingham study reported that total cholesterol levels increased significantly in women with bilateral oophorectomy, but not among women with hysterectomy without oophorectomy or with unilateral oophorectomy [66]. In summary, epidemiologic data have consistently shown a relation between menopause and increased serum total cholesterol levels. This increase may be attributable to increased levels of LDL cholesterol, possibly as the result of decreased LDL receptors in response to declining estrogen levels. Increased triglyceride levels may also occur with menopause, although these data have been less consistent. There is no consistent evidence that menopause per se has a deleterious effect on HDL cholesterol levels. Important issues must be considered when weighing the available evidence relating menopause to adverse changes in the lipid profle. Both menopause and lipid levels are strongly related to age, as suggested by the gradual increases observed in both men and women with chronologic age. Thus, the effects of menopause independent of age are difficult to assess. Second, many of the data supporting a causal role of estrogen in lipid changes with menopause are derived from studies of exogenous hormone replacement therapy. Few studies have examined the relation of endogenous hormones to lipid levels, and these have been inconsistent, or have provided only indirect evidence by establishing a temporal association. Additional studies are needed to understand the timing, duration, and extent of lipid changes throughout the menopause transition and to establish whether these changes are determined by changes in endogenous ovarian hormone levels.

D. H e m o s t a t i c C h a n g e s Elevations of plasma fibrinogen [ 106-108] and factor VII [ 106,109,110] have been associated with increased risk of coronary heart disease. There is evidence that these coagulation factors may be influenced by estrogen. Fibrinogen increases with age [ 111 ], and in women this increase begins in the fifth decade [112]. Studies of fibrinogen during pregnancy and throughout the menstrual cycle also suggest an influence of endogenous estrogen [ 113]. The use of hormone replacement therapy, either estrogen alone [95,114] or estrogen plus progestin [95,115], has been associated with reduced concentrations of fibrinogen. Factor VII activity increases with the use of estrogen alone, although these changes normalize with the addition of progestins [115].

238 Population-based data regarding the effect of menopause on hemostatic factors are sparse and are primarily crosssectional. Several cross-sectional studies have reported elevated fibrinogen concentrations in postmenopausal women compared with premenopausal women [ 111,113,116-121 ]. Other cross-sectional comparisons have shown no differences in fibrinogen concentration by menopause status [63, 112,122]. Lee et al. [120,121] reported that although levels were elevated in postmenopausal women, menopause status explained only a small portion of the total variation in fibrinogen. Lindoff et al. [118] compared premenopausal women with women who had been postmenopausal for less than 18 months, and with a second group of women who had been postmenopausal for longer than 18 months. These comparisons suggested that increases in fibrinogen levels were related to the time elapsed since menopause. Meade et al. [117] prospectively observed that circulating fibrinogen increased in women who became postmenopausal during follow-up. Fibrinogen concentrations have been correlated with BMI, lipoproteins, and smoking [ 121,123], which are also related to menopause status. Although some studies have adjusted comparisons for age and these other factors [ 111,113,116, 117,120,121], others have not [63,112,118,119,122]. Thus, the lack of consistency across studies may be due in part to confounding and differential levels of adjustment. Cross-sectional studies have also shown elevated values for factor VII in postmenopausal women compared with those who are premenopausal [111-113,117]. This association has also been observed prospectively [116]. Increased factor VII may be a consequence of increased triglycerides in postmenopausal women [111,124]. Very-low-density lipoprotein cholesterol concentrations may rise after menopause [76], and studies in men have shown that triglycerides may increase factor VII activity [ 125]. Enhanced coagulability due to elevations of fibrinogen and factor VII in postmenopausal women may be offset by a concurrent increase in antithrombin III activity in postmenopausal women [118,126]. Postmenopausal women have been shown to have higher levels of antithrombin III compared with premenopausal women [113,116,118] or men [116]. In summary, most information regarding the effect of menopause on hemostatic factors is derived from studies of hormone replacement therapy. Studies of menopausal changes in hemostatic factors in untreated women are primarily cross-sectional. Some, but not all, of these have shown elevated concentrations of fibrinogen and factor VII in postmenopausal women. Whether these changes are attributable to estrogen or to changes in other factors at menopause has not been clearly demonstrated. Furthermore, because increased antithrombin III levels have also been shown

CAROL A. DERBY

in postmenopausal women, the net impact on risk of thrombosis may be minimal.

E. Direct Effects of Estrogens on the Vascular Wall Estrogen receptors are present in the cardiovascular system [ 127,128], and the direct effects of estrogens on the vascular wall may influence both the development of atherosclerosis and the regulation of arterial blood flow [ 124,129,130]. It has been estimated that the majority of the cardiovascular benefit of hormone replacement is attributable to direct effects of estrogen on the arterial wall, whereas only 2 0 - 3 0 % is attributable to lipid effects [ 126,131,132]. The molecular mechanisms by which estrogen affects arterial wall functions are not fully understood. Proposed mechanisms for the vasodilatory effects of estrogen include effects on calcium channels, potentiation of endothelium-dependent vasodilation, and increased synthesis of prostacyclin [124,126,130,133,134]. Estrogens may inhibit atherosclerosis by acting on vascular connective tissue, impeding vascular smooth muscle cell proliferation, reducing LDL accumulation in the arterial wall, inhibiting platlet aggregation, inhibiting stress-induced endothelial injury, and inhibiting the formation of foam cells [124,126,130,135]. Evidende for these effects is based on in vivo and in vitro animal studies [127,136-142] and studies of the effects of estrogen administration in postmenopausal women [131, 143-145]. Hormone replacement has been shown to enhance vascular tone and to inhibit progression of atherosclerosis [ 146-152]. Human studies of the effects of estrogen on vascular tone and atherogenesis are limited due to the lack of noninvasive techniques. Evidence for menopausal effects in women not treated with hormones is limited to data from studies of agerelated changes in vascular reactivity. In men, flow-mediated vasodilation of the brachial artery declines after the age of 40, whereas in women the decline begins only after age 50 and is steep at the time of menopause [153,154]. Flow-mediated vasodilation has also been shown to vary throughout the menstrual cycle [ 155]. Taddei et al. [ 156] studied the relationship of age to endothelial function in premenopausal women less than 45 years old, postmenopausal women over the age of 45, and men. Among men, there was a constant age-related decline in the vasodilatory response to acetycholine, whereas in women there was only a gradual decline up to age 49, after which vascular response to acetylcholine decreased more quickly than in men. The sex difference in agerelated endothelial dysfunction was interpreted as evidence that menopause influences endothelium-dependent vasodilation. Finally, Gangar et al. [ 144] demonstrated that pulsatility index in the internal carotid arteries of postmenopausal

239

CHAPTER 15 Cardiovascular Pathophysiology w o m e n , an indicator of blood flow impedance, was correlated with length of time since menopause. In summary, estrogens m a y help maintain vascular tone and protect the arterial wall from atherosclerotic processes. These m e c h a n i s m s may potentially explain the major portion of m e n o p a u s a l effects on cardiovascular risk. However, the existing evidence is based on the effects of h o r m o n e rep l a c e m e n t or e x p e r i m e n t a l administration of estrogens. Additional studies are needed to determine the impact of m e n o pausal changes in e n d o g e n o u s h o r m o n e levels on arterial wall function. Elucidation of these processes m a y be key to understanding the influence of m e n o p a u s e on the developm e n t of atherosclerotic disease.

V. C O N C L U S I O N S Cardiovascular disease rates are higher in p o s t m e n o p a u sal than in p r e m e n o p a u s a l w o m e n , and the m a l e - t o - f e m a l e ratio of disease rates diminishes after the age of m e n o p a u s e . W h e t h e r these trends are due to declining ovarian function with m e n o p a u s e remains controversial. Vital statistics data describing sex-specific trends with aging do not support the c o m m o n l y held belief that cardiovascular disease rates in w o m e n accelerate as a result of menopause. E p i d e m i o l o g i c a l studies e x a m i n i n g cardiovascular risk in relation to natural m e n o p a u s e have also failed to demonstrate increased risk with m e n o p a u s e , with results inconsistent across studies. Studies of surgical m e n o p a u s e have been m o r e consistent, suggesting a relationship b e t w e e n bilateral o o p h o r e c t o m y at an early age, and increased cardiovascular risk. Given the lengthy natural history of atherosclerosis, an abrupt increase in risk of clinical cardiovascular disease m i g h t not be expected at the age of m e n o p a u s e . The influence of m e n o p a u s e on risk m i g h t be m o r e gradual, m e d i a t e d t h r o u g h changes in cardiovascular risk factors. E p i d e m i o l o g i c a l studies have not shown a consistent effect of m e n o p a u s e on blood pressure. However, m e n o p a u s e has been associated with adverse changes in lipids and lipoproteins. W h e t h e r m e n o p a u s e per se is associated with adverse changes in hemostatic factors remains unclear. B e c a u s e estrogens appear to have favorable effects directly on arterial wall function, declining estrogen levels with m e n o p a u s e m a y facilitate the developm e n t of atherosclerosis and impaired blood flow. N u m e r o u s m e t h o d o l o g i c a l issues have limited prior studies of m e n o pause and cardiovascular disease. A major limitation of the current body of k n o w l e d g e is the lack of data regarding the relationship of e n d o g e n o u s h o r m o n e s to cardiovascular disease and related risk factors. Also lacking are data regarding racial differences in changes with m e n o p a u s e , and information regarding changes that occur during the p e r i m e n o p a u s a l period. To address these issues and to clarify the influence of m e n o p a u s e on cardiovascular risk, future studies m u s t in-

clude large, multiracial populations, with l o n g - t e r m followup and the ability to correlate biological and clinical markers of cardiovascular disease with e n d o g e n o u s h o r m o n e concentrations.

References 1. Mosca, L., Manson, J. E., Sutherland, S. E., Langer, R. D., Manolio, T., and Barrett-Connor, E. (1997). Cardiovascular disease in women: A statement for healthcare professionals from the American Heart Association. Circulation 96, 2468-2482. 2. American Heart Association (1998). "Heart and Stroke Statistical Update." American Heart Association, Dallas, TX. 3. Barrett-Connor, E., and Goodman-Gruen, D. (1995). Prospective study of endogenous sex hormones and fatal cardiovascular disease in postmenopausal women. Br. Med. J. 311, 1193-1196. 4. Grodstein, F., and Stampfer, M. (1995). The epidemiology of coronary heart disease and estrogen replacement in postmenopausal women. Prog. Cardiovasc. Dis. 38, 199-210. 5. Bush, T. L. (1990). The epidemiology of cardiovascular disease in postmenopausal women. Ann. N.Y. Acad. Sci. 592, 263-271. 6. Derby, C. A., Hume, A. L., Barbour, M. M., McPhillips, J. B., Lasater, T. M., and Carleton, R. A. (1993). Correlates of postmenopausal estrogen use and trends through the 1980's in two southeastern New England communities. Am. J. Epidemiol. 137, 1125-1135. 7. Johannes, C. B., Crawford, S. L., Posner, J. G., and McKinlay, S. M. (1994). Longitudinal patterns and correlates of hormone replacement therapy use in middle-aged women. Am. J. Epidemiol. 140, 439-452. 8. Rosenberg, L., Shapiro, S., Kaufman, D. W., Slone, D., Miettenen, O. S., and Stolley, P. D. (1979). Patterns and determinants of conjugated estrogen use. Am. J. Epidemiol. 109, 676-686. 9. Barrett-Connor, E., Wingard, D. L., and Criqui, M. H. (1989). Postmenopausal estrogen use and heart disease risk factors in the 1980's. JAMA, J. Am. Med. Assoc. 261, 2095-2100. 10. Cauley, J. A., Gutai, J. P., Glenn, N. W., Paternoster-Bales, M., Cottington, E., and Kuller, L. H. (1994). Serum estrone concentrations and coronary artery disease in postmenopausal women. Arterioscler. Thromb. 14, 14-18. 11. Zumoff, B., Troxler, R. G., O'Connor, J., Rosenfeld, R. S., Keream, J., Levin, J., Hickman, J. R., Sloan, A. M., Walker, W., Cook, R. L., and Fukushima, D. K. (1982). Abnormal hormone levels in men with coronary artery disease. Arteriosclerosis 2, 58-67. 12. Furman, R. H. (1968). Are gonadal hormones (estrogens and androgens) of significance in the development of ischemic heart disease? Ann. N.Y. Acad. Sci. 149, 822-833. 13. Heller, R. E, and Jacobs, H. S. (1978). Coronary heart disease in relation to age, sex and the menopause. Br. Med. J. 1, 472-474. 14. Winklestein, W., and Rekate, A. C. (1964). Age trend of mortality from coronary artery disease in women and observations on the reproductive patterns of those affected. Am. Heart J. 67, 481-488. 15. Winklestein, W., Stenchever, M. A., and Lilienfeld, A. M. (1958). Occurrence of pregnancy, abortion, and artificial menopause among women with coronary artery disease: A preliminary study. J. Chronic Dis. 7, 273-286. 16. Tracy, R. E. (1966). Sex difference in coronary disease: Two opposing views. J. Chronic Dis. 19, 1245-1251. 17. Stampfer, M. J., Colditz, G. A., and Willett, W. C. (1990). Menopause and heart disease. Ann. N.Y. Acad. Sci. 592, 193-204. 18. WHO Scientific Group on Research on the Menopause in the 1990's (1996). "Research on the menopause: Report of a WHO Scientific Group." WHO Tech. Rep. Ser. 866. WHO, Geneva.

240 19. Tunstall-Pedoe, H. (1998). Myth and paradox of coronary risk and the menopause. Lancet 351, 1425-1427. 20. Mann, J. I., and Inman, W. H. W. (1975). Oral contraceptives and death from myocardial infarction. Br. Med. J. 2, 245-248. 21. Oliver, M. F. (1974). Ischaemic heart disease in young women. Br. Med. J. 4, 253-259. 22. Bengtsson, C., Ryobo, G., and Westerberg, H. (1979). Number of pregnancies, use of oral contraceptives, and menopausal age in women with ischaemic heart disease, compared to a population sample of women. Acta Med. Scand. 599, 75-81. 23. La Vecchia, C., Decarli, A., Francschi, S., Gentile, A., Negri, E., and Parazzini, F. (1987). Menstrual and reproductive factors and the risk of myocardial infarction in women under fifty-five years of age. Am. J. Obstet. Gynecol. 157, 1108-1112. 24. Witteman, J. C. M., Grobbee, D. F., Kok, F. J., Hofman, A., and Valkenberg, H. A. (1989). Increased risk of atherosclerosis in women after the menopause. Br. Med. J. 298, 642-644. 25. Rosenberg, L., Hennekens, C. H., Rosner, B., B61anger, C., Rothman, K. J., and Speizer, F. E. (1981). Early menopause and the risk of myocardial infarction. Am. J. Obstet. Gynecol. 139, 47-51. 26. Lapidus, L., Bengtsson, C., and Lindquist, O. (1985). Menopausal age and risk of cardiovascular disease and death. Acta Obstet. Gynecol. Scand., Suppl. 130, 37-41. 27. van der Schouw, Y. T., van der Graaf, Y., Steyerberg, E. W., Eijkemans, M. J. C., and Banga, J. D. (1996). Age at menopause as a risk factor for cardiovascular mortality. Lancet 347, 714-718. 28. Colditz, G. A., Willett, W. C., Stampfer, M. J., Rosner, B., Speizer, F. E., and Hennekens, C. H. (1987). Menopause and the risk of coronary heart disease in women. N. Engl. J. Med. 316, 1105-1110. 29. Gordon, T., Kannel, W. B., and Hjortland, M. C. (1978). Menopause and coronary heart disease: The Framingham Study. Ann. Intern. Med. 89, 157-161. 30. Kannel, W. B., Hjortland, M. C., McNamara, P. M., and Gordon, T. (1976). Menopause and risk of cardiovascular disease: The Framingham Study. Ann. Intern. Med. 85, 447-452. 31. Robinson, R. W., Higano, N., and Cohen, W. D. (1959). Increased incidence of coronary heart disease in women castrated prior to the menopause. Arch. Intern. Med. 104, 908-913. 32. Johansson, B. W., Kaij, L., Kullander, S., Lenner, H. C., Svanberg, L., and Astedt, B. (1975). On some late effects of bilateral oophorectomy in the age range 15-30 years. Acta Obstet. Gynecol. Scand. 54, 4 4 9 461. 33. Sznajderman, M., and Oliver, M.F. (1963). Spontaneous premature menopause, ischaemic heart disease, and serum-lipids. Lancet 1 , 9 6 2 965. 34. Beard, C. M., Fuster, V., and Annegers, J. F. (1984). Reproductive history in women with coronary heart disease: A case-control study. Am. J. Epidemiol. 120, 108-114. 35. Lerner, D. J., and Kannel, W. B. (1986). Patterns of coronary heart disease morbidity and mortality in the sexes: A 26-year follow-up of the Framingham population. Am. Heart J. 111, 383-390. 36. Casiglia, E., d' Este, D., Ginocchio, G., Colangeli, G., Onesto, C., Tramontin, P., Ambrosio, G. B., and Pessina, A. C. (1996). Lack of influence of menopause on blood pressure and cardiovascular risk profile: A 16-year longitudinal study concerning a cohort of 568 women. J. Hypertens. 14, 729-736. 37. Parrish, H. M., Carr, C. A., Hall, D. G., and King, T. M. (1967). Time interval from castration in premenopausal women to development of excessive coronary atherosclerosis. Am. J. Obstet. Gynecol. 99, 155162. 38. Wuest, J. H., Dry, T. J., and Edwards, J. E. (1953). The degree of coronary atherosclerosis in bilaterally oophorectomized women. Circulation 7, 801-808. 39. Rivin, A. U., and Dimitroff, S. P. (1954). The incidence and severity of atherosclerosis in estrogen-treated males, and in females with hy-

CAROL A. DERBY poestrogenic or a hyperestrogenic state. Circulation 9, 533-539. 40. Novak, E. R., and Williams, T. J. (1960). Autopsy comparison of cardiovascular changes in castrated and normal women. Am. J. Obstet. Gynecol. 80, 863-872. 41. Manchester, J. H., Herman, M. V., and Gorlin, R. (1971). Premenopausal castration and documented coronary atherosclerosis. Am. J. Cardiol. 28, 33-37. 42. Pearson, T. A., and Derby, C. A. (1991). Should arteriographic casecontrol studies be used to identify causes of atherosclerotic coronary artery disease? (Invited commentary). Am. J. Epidemiol. 134, 123128. 43. Oliver, M. E, and Boyd, G. S. (1959). Effect of bilateral ovariectomy on coronary-artery disease and serum-lipid levels. Lancet 1,690-694. 44. Ritterband, A. B., Jaffe, I. A., Densen, P. M., Magagna, J. F., and Reed, E. (1963). Gonadal function and the development of coronary heart disease. Circulation 27, 237-251. 45. Barrett-Connor, E. (1996). The menopause, hormone replacement, and cardiovascular disease: The epidemiologic evidence. Maturitas 23, 227-234. 46. Surgeon General's Advisory committee on Smoking and Health (1990). "Smoking and Health: Report of the Advisory Committee to the Surgeon General of the Public Health Service," PHS No. 1103. U.S. Department of Health and Human Services, Bethesda, MD. 47. Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (1997). "The sixth report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure," National High Blood Pressure Education Program. NIH Publ. No. 98-4080. National Institutes of Health, National Heart, Lung, and Blood Institute, Washington, DC. 48. Criqui, M., Wallace, R., Heiss, G., Mishkel, M., Schonfeld, G., and Jones, G. T. (1980). Cigarette smoking and plasma high-density lipoprotein cholesterol: The Lipid Research Clinics Program Prevalence Study. Circulation 62 (Suppl. IV), 70-76. 49. Fortmann, S. P., Haskell, W. L., and Williams, P. T. (1986). Changes in plasma high density lipoprotein cholesterol after changes in cigarette use. Am. J. Epidemiol. 124, 706-710. 50. Manson, J. E., Colditz, G. A., Stampfer, M. J., Willett, W. C., Posner, B., Monson, R. R., Speizer, E E., and Hennekens, C. H. (1990). A prospective study of obesity and risk of coronary heart disease in women. N. Engl. J. Med. 322, 882-889. 51. Willett, W., Stampfer, M. J., Bain, C., Lipnick, R., Speizer, F. E., Rosner, B., Cramer, D., and Hennekens, C. H. (1983). Cigarette smoking, relative weight, and menopause. Am. J. Epidemiol. 117, 651-658. 52. McKinlay, S. M., Bifano, N. L., and McKinlay, J. B. (1985). Smoking and age at menopause in women. Ann. Intern. Med. 103, 350-356. 53. Brambilla, D. J., and McKinlay, S. M. (1989). A prospective study of factors affecting age at menopause. J. Clin. Epidemiol. 42, 10311039. 54. Sowers, M. F., and La Pietra, M. T. (1995). Menopause: Its epidemiology and potential association with chronic diseases. Epidemiol. Rev. 17, 287-302. 55. Staessen, J. A., Amery, A., and Fagard, R. (1990). Editorial review. Isolated systolic hypertension in the elderly. J. Hypertens. 8, 393405. 56. Wu, Z., Wu, X., and Zhang, Y. (1990). Relationship of menopausal status and sex hormones to serum lipids and blood pressure. Int. J. Epidemiol. 19, 297-302. 57. Greenberg, G., Imeson, J. D., Thompson, S. G., and Meade, T. W. (1985). Blood pressure and the menstrual cycle. Br. J. Obstet. Gynaecol. 92, 1010-1014. 58. Casslen, B. (1986). Blood pressure alters during the normal menstrual cycle. Br. J. Obstet. Gynaecol. 93, 523-526. 59. Taylor, R. D., Corcoran, A. C., and Page, I. H. (1947). Menopausal hypertension: A critical study. Am. J. Med. Sci. 213, 475-476.

241

CHAPTER 15 Cardiovascular Pathophysiology 60. Davis, C. E., Pajak, A., Rywik, S., Williams, D., Broda, G., Pazucha, T., and Ephross, S. (1994). Natural menopause and cardiovascular disease risk factors: The Poland and US Collaborative Study on Cardiovascular Disease Epidemiology. Ann. Epidemiol. 4, 445-448. 61. Portaluppi, E, Pansini, E, Manfredini, R., and Mollica, G. (1997). Relative influence of menopausal status, age, and body mass index on blood pressure. Hypertension 29, 976-979. 62. Shibata, H., Matsuzaki, T., and Hatano, S. (1979). Relationship of relevant factors of atherosclerosis to menopause in Japanese women. Am. J. Epidemiol. 109, 420-424. 63. Bonithon-Kopp, C., Scarabin, E-Y., Dame, B., Malmejac, A., and Guize, L. (1990). Menopause-related changes in lipoproteins and some other cardiovascular risk factors. Int. J. Epidemiol. 19, 42-48. 64. Lindquist, O., and Bengtsson, C. (1980). Serum lipids, arterial blood pressure, and body weight in relation to the menopause: Results from a population study of women in Goteborg, Sweden. Scand. J. Clin. Lab. Invest. 40, 629-636. 65. Weiss, N. S. (1972). Relationship of menopause to serum cholesterol and arterial blood pressure: The United States' Health Examination Survey of Adults. Am. J. Epidemiol. 96, 237-241. 66. Hjortland, M. C., McNamara, E M., and Kannel, W. B. (1976). Some atherogenic concomitants of menopause: The Framingham Study. Am. J. Epidemiol. 103, 304-311. 67. Matthews, K. A., Meilahn, E. N., KuIler, L. H., Kelsey, S., Caggiula, A. W., and Wing, R. R. (1989). Menopause and risk factors for coronary heart disease. N. Engl. J. Med. 321, 641-646. 68. Matthews, K. A., Wing, R. R., Kuller, L. H., Meilahn, E. N., and Plantinga, E (1994). Influence of the perimenopause on cardiovascular risk factors and symptoms of middle-aged healthy women. Arch. Intern. Med. 154, 2349-2355. 69. Lindquist, O., Bengtsson, C., and Lapidus, L. (1985). Relationships between the menopause and risk factors for ischaemic heart disease. Acta Obstet. Gynecol. Scand., Suppl. 130, 43-47. 70. van Beresteyn, E. C. H., van Hof, M. A., and de Waard, H. (1989). Contributions of ovarian failure and aging to blood pressure in normotensive perimenopausal women: A mixed longitudinal study. Am. J. Epidemiol. 129, 947-955. 71. Poehlman, E. T., Toth, M. J., Ades, E A., and Rosen, C. J. (1997). Menopause-associated changes in plasma lipids, insulin-like growth factor I and blood pressure: A longitudinal study. Eur. J. Clin. Invest. 27, 322-326. 72. Staessen, J. A., Ginocchio, G., Thijs, L., and Fagard, R. (1997). Conventional and ambulatory blood pressure and menopause in a prospective population study. J. Hum. Hypertens. 11, 507-514. 73. Akahoshi, M., Soda, M., Nakashima, E., Shimaoka, K., Seto, S., and Yano, K. (1996). Effects of menopause on trends of serum cholesterol, blood pressure, and body mass index. Circulation 94, 61-66. 74. Wing, R., Matthews, K. A., Kuller, L. H., Meilahn, E. N., and Plantinga, E L. (1991). Weight gain at the time of menopause. Arch. Intern. Med. 151, 97-102. 75. Burke, G. L., Sprafka, M., Folsom, A. R., Hahn, L. E, Luepker, R. V., and Blackburn, H. (1991). Trends in serum cholesterol levels from 1980 to 1987: The Minnesota Heart Survey. N. Engl. J. Med. 324, 941- 946. 76. U.S. Department of Health and Human Services, Public Health Service, National Institutes of Health (1980). "The Lipid Research Clinics' Population Studies Data Book. Vol. I. The Prevalence Study," N.I.H. Publ. No. 80-1527. USDHHS, PHS, NIH, Bethesda, MD. 77. Sacks, E M., and Walsh, B. W. (1990). The effects of reproductive hormones on serum lipoproteins: Unresolved issues in biology and clinical practice. Ann. N.Y. Acad. Sci. 592, 272-285. 78. Miller, N. E. (1984). Why does plasma LDL concentration in adults increase with age? Lancet 1. 263-267. 79. Windler, E., Kovanen, E T., Chao, Y. S., Brown, M. S., Havel, R. J.,

80.

81.

82.

83.

84.

85.

86. 87.

88.

89.

90.

91.

92.

93.

94.

95.

96.

97.

and Goldstein, J. L. (1980). The estradiol stimulated lipoprotein receptor of rat liver. J. Biol. Chem. 255, 10464-10471. Veldhuis, J. D., Gwynne, J. T., Azimi, E, Garmey, J.,and Juchter, D. (1985). Estrogen regulates LDL metabolism by cultured swine granulosa cells. Endocrinology (Baltimore) 117, 1321-1327. Kuller, L. H., Meilahn, E. N., Gutai, J, et al. (1990). Lipoproteins, estrogens and the menopause. In "The Menopause: Biological and Clinical Consequences of Ovarian Failure: Evolution and Management" (S.G. Korenman, ed.), pp. 179-197. Serono Symposia USA, Norwell, MA. Tell, G. S., Mittlemark, M. B., and Vellar, O. D. (1985). Cholesterol, high density lipoprotein cholesterol and triglycerides during puberty: The Oslo Youth Study. Am. J. Epidemiol. 122, 750-761. Notelovitz, M., Gudat, J.C., Ware, M.D., and Dougherty, M.C. (1983). Lipids and lipoproteins in women after oophorectomy and the response to oestrogen therapy. Br. J. Obstet. Gynaecol. 90, 171-177. Demirovic, J., Sprafka, J. M., Folsom, A. R., Laitinen, D., and Blackburn, H. (1992). Menopause and serum cholesterol: Differences between blacks and whites: The Minnesota Heart Survey. Am. J. Epidemiol. 136, 155-164. Baird, D. D., Tyroler, H. A., Heiss, G., Chambless, L. E., and Hames, C. G. (1985). Menopausal change in serum cholesterol: Black/white differences in Evans County, Georgia. Am. J. Epidemiol. 122, 982993. Jensen, J., Nilas, L., and Christiansen, C. (1990). Influence of menopause on serum lipids and lipoproteins. Maturitas 12, 321-331. Hallberg, L., and Svanborg, A. (1967). Cholesterol, phospholipids, and triglycerides in plasma in 50-year-old women. Acta Med. Scand. 181, 185-194. Razay, G., Heaton, K. W., and Bolton, C. H. (1992). Coronary heart disease risk factors in relation to the menopause. Q. J. Med. [N. S.] 85, 889-896. Shibata, H., Haga, H., Suyama, Y., Kumagai, S., and Seino, T. (1987). Serum total and HDL cholesterol according to reproductive status in Japanese females. J. Chronic Dis. 40, 209-213. Campos, H., McNamara, J. R., Wilson, P. W. F., Ordovas, J. M., and Schaeffer, E. J. (1988). Differences in low density lipoprotein subfractions and apolipoproteins in premenopausal and postmenopausal women. Endocrinol. Metab. 67, 30-35. Hamman, R. F., Bennett, P. H., and Miller, M. (1975). The effect of menopause on serum cholesterol in American (Pima) Indian women. Am. J. Epidemiol. 102, 164-169. Sniderman, A., Shapiro, S., Marpole, D., Skinner, B., Teng, B.,and Kwiter0vich, P. O. (1980). Association of coronary atherosclerosis with hyperapobetalipoproteinemia (increased protein but normal cholesterol levels in human plasma low density (B) lipoproteins). Proc. Natl. Acad. Sci. U.S.A. 77, 604-608. Kwiterovich, P. O., Coresh, J., Smith, H. H., Bachorik, P. S., Derby, C. A., and Pearson, T. A. (1992). Comparison of the plasma levels of apolipoproteins B and A-1, and other risk factors in men and women with premature coronary artery disease. Am. J. Cardiol. 69, 10151021. Bush, T., Cowan, L., Heiss, G., Chambliss, L., and Wallace, R. (1984). Ovarian function and lipid/lipoprotein levels: Results from the Lipid Research Clinics Program. Am. J. Epidemiol. 120, 489. Writing Group for the PEPI Trial (1995). Effects of estrogen or estrogen/progestin regimens on heart disease risk factors in postmenopausal women: The postmenopausal estrogen/progestins (PEPI) trial. JAMA, J. Am. Med. Assoc. 273, 199-208. Bush, T. L., Barrett-Connor, E., Cowan, L. D., Criqui, M. H., Wallace, R. B., Suchindran, C. M., Tyroler, H. A., and Rifkind, B. M. (1987). Cardiovascular mortality and noncontraceptive use of estrogen in women: Results from the Lipid Research Clinics program follow-up study. Circulation 75, 1102-1109. Hazzard, W. R. (1989). Estrogen replacement and cardiovascular dis-

242

98.

99.

100.

101.

102.

103.

104.

105.

106.

107.

108.

109.

110.

111.

112.

113.

114.

CAROL A. DERBY

ease: Serum lipids and blood pressure effects. Am. J. Obstet. Gynecol. 161, 1847-1853. Tikkanen, M. J. (1996). The menopause and hormone replacement therapy: Lipids, lipoproteins, coagulation and fibrinolytic factors. Maturitas 23, 209-216. Dahlen, G. H., Guyton, J. R., Attar, M., Farmer, J. A., Kautz, A., and Gotto, A. M. (1986). Associations of levels of lipoprotein Lp(a), plasma lipids and other lipoproteins with coronary artery disease documented by angiography. Circulation 74, 758-765. Hoefler, G., Harnoncourt, F., Paschke, E., Mirtl, W., Pfeiffer, K. H., and Kostner, G. M. (1986). Lipoprotein Lp(a)-a risk factor for myocardial infarction. JAMA, J. Am. Med. Assoc. 256, 2540-2545. Jenner, J. L., Ordovas, J. M., Lamon-Fava, S., Schaefer, M. M., Wilson E W. F., Castelli, W. E, and Schaefer, E. J. (1993). Effects of age, sex, and menopausal status on plasma lipoprotein(a) levels: The Framingham Offspring Study. Circulation 87, 1135-1141. Heinrich, J., Sandkamp, M., Kokott, R., Schulte, H., and Assmann, G. (1991). Relationship of lipoprotein(a) to variables of coagulation and fibrinolysis in a healthy population. Clin. Chem. (Winston-Salem, N.C.) 37, 1950 - 1954. Cauley, J. A., Gutai, J. E, Kuller, L. H., and Powell, J. G. (1990). The relation of endogenous sex steroid hormone concentrations to serum lipid and lipoprotein levels in postmenopausal women. Am. J. Epidemiol. 132, 884-894. Adamopoulos, D. A., Loraine, J. A., and Dove, G. A. (1971). Endocrinological studies in women approaching the menopause. J. Obstet. Gynaecol. Br. Commonw. 78, 62-79. Sherman, B. M., West, J. H., and Korenman, S. G. (1976). The menopausal transition: Analysis of LH, FSH, estradiol and progesterone concentration during menstrual cycles of older women. J. Clin. Endocrinol. Metab. 42, 629-636. Meade, T. W., Mellows, S., Brozovic, M., Miller, G. J., Chakrabati, R. R., North, W. R., Haines, A. E, Stifling, Y., Imeson, J. D., and Thompson, S. G. (1986). Haemostatic function and ischemic heart disease: Principal results of the Northwick Park Heart Study. Lancet 2, 533-537. Kannel, W. B., Wolf, P. A., Castelli, W. E, and D'Agostino, R. B. (1987). Fibrinogen and risk of cardiovascular disease: The Framingham Study. JAMA, J. Am. Med. Assoc. 258, 1183-1186. Wilhelmsen, L., Svardsudd, K., Korsan-Bengsten, K., Larsson, B., Welin, L., and Tibblin, G. (1984). Fibrinogen as a risk factor for stroke and myocardial infarction. N. Engl. J. Med. 201, 501-505. Dalaker, K., Hjermann, I., and Prydz, H. (1985). A novel form of Factor VII in plasma from men at risk for cardiovascular disease. Br. J. Haematol. 61, 205-222. Hoffman, C. J., Miller, R. H., Lawson, W. E., and Hultin, M. B. (1988). Elevation of factor VII activity and mass in ischemic heart disease and in young subjects at high risk of ischemic heart diseases. Circulation 78(Suppl. II), 206. Folsom, A. R., Wu, K. K., Davis, C. E., Conlan, M. G., Sorlie, P. D., and Szklo, M. (1991). Population correlates of plasma fibrinogen and factor VII, putative cardiovascular risk factors. Atherosclerosis 91, 191-205. Balleisen, L., Bailey, J., Epping, E H., Schulte, H., and van de Loo, J. (1985). Epidemiological study on factor VII, Factor VIII and fibrinogen in an industrial population: I. Baseline data on the relation to age, gender, body-weight, smoking, alcohol, pill-using, and menopause. Thromb. Haemostasis 54, 4 7 5 - 479. Meilahn, E. N., Kuller, L. H., Matthews, K. A., and Kiss, J. E. (1992). Hemostatic factors according to menopausal status and use of hormone replacement therapy. Ann. Epidemiol. 2, 445-455. Manolio, T. A., Furberg, C. D., Shemanski, L., Psaty, B. M., O'Leary, D. H., Tracy, R. E, and Bush, T. L. (1993). Associations of postmenopausal estrogen use with cardiovascular disease and its risk factors in older women. Circulation 88 (Part 1), 2163 -2171.

115. Nabulsi, A. A., Folsom, A. R., White, A., Patsch, W., Heiss, G., Wu, K. K., and Szklo, M. (1993). Association of hormone replacement therapy with various cardiovascular risk factors in postmenopausal women. N. Engl. J. Med. 328, 1069-1075. 116. Meade, T. W., Dyer, S., Howarth, D. J., Imeson, J. D., and Stifling, Y. (1990). Antithrombin III and procoagulant activity: Sex differences and effects of the menopause. Br. J. Haematol. 74, 77-81. 117. Meade, T. W., Haines, A. E, Imeson, J. D., Stirling, Y., and Thompson, S. G. (1983). Menopausal status and haemostatic variables. Lancet 1, 22-24. 118. Lindoff, C., Petersson, F., Lecander, I., Martinsson, G., and Astedt, B. (1993). Passage of the menopause is followed by haemostatic changes. Maturitas 17, 17-22. 119. Iso, H., Folsom, A. R., Sato, S., Wu, K. K., Shimamoto, T., Koike, K., Iida, M., and Komachi, Y. (1993). Plasma fibrinogen and its correlates in Japanese and US population samples. Arteriosclerosis Thromb. 13, 783 -790. 120. Lee, A. J., Lowe, G. D. O., Smith, W. C. S., and Tunstall-Pedoe, H. (1993). Plasma fibrinogen in women: Relationships with oral contraception, the menopause and hormone replacement therapy. Br. J. Haematol. 83, 616-621. 121. Lee, A. J., Smith, W. C. S., Lowe, G. D. O., and Tunstall-Pedoe, H. (1990). Plasma fibrinogen and coronary risk factors: The Scottish Heart Health Study. J. Clin. Epidemiol. 43, 913-919. 122. Pinto, S., Rostagno, C., Coppo, M., Paniccia, R., Prisco, D., Bruni, V., Rosati, D., and Abbate, R. (1990). No signs of increased thrombin generation in menopause. Thromb. Res. 58, 645-651. 123. Stefanick, M. L., Legault, C., Tracy R. P., Howard, G., Kessler, C. M., Lucas, D. L., and Bush, T. L. (1995). Distribution and correlates of plasma fibrinogen in middle-aged women: Initial findings of the postmenopausal estrogen/progestin interventions (PEPI) study. Arterioscler. Thromb. Vasc. Biol. 15, 2085-2093. 124. Samaan, S. A., and Crawford, M. H. (1995). Estrogen and cardiovascular function after menopause. J. Am. Coll. Cardiol. 26, 1403-1410. 125. Skartlein, A. H., Lyberg-Beckman, S., Holme, I., Hjermann, I., and Prydz, H. (1989). Effect of alteration in triglyceride levels on factor VII-phospholipid complexes in plasma. Arteriosclerosis 9, 798-801. 126. Gupta, S., and Rymer, J. (1996). Hormone replacement therapy and cardiovascular disease. Int. J. Gynecol. Obstet. 52, 119-125. 127. Losordo, D. W., Kearney, M., Kim, E. A., Jekanowski, J., and Isner, J. M. (1994). Variable expression of the estrogen receptor in normal and atherosclerotic coronary arteries of premenopausal women. Circulation 89, 1501 - 1510. 128. Karas, R. N., Patterson, B. L., and Mendelsohn, M. E. (1994). Human vascular smooth muscle cells contain functional estrogen receptors. Circulation 89, 1943-1950. 129. Pines, A., Mijatovic, V., van der Mooren, M., and Kenemans, E (1997). Hormone replacement therapy and cardioprotection: Basic concepts and clinical considerations. Eur. J. Obstet., Gynecol. Reprod. Biol. 71, 193-197. 130. Rosano, G. M. C., Chierchia, S. L., Leonardo, F., Beale, C. M., and Collins, P. (1996). Cardioprotective effects of ovarian hormones. Eur. Heart J. 17(Suppl D), 15-19. 131. Baron, Y. M., Galea, R., and Brincat, M. (1998). Carotid artery wall changes in estrogen-treated and -untreated postmenopausal women. Obstet. Gynecol. 91,982-986. 132. Stevensen, J. (1995). The metabolic and cardiovascular consequences of HRT. Br. J. Clin. Pract. 49, 87-90. 133. Glasser, S. E, Selwyn, A. E, and Ganz, E (1996). Atherosclerosis: Risk factors and the vascular endothelium. Am. Heart J. 131, 379-384. 134. Collins, E, Rosano, G., Jiang, C., Lindsay, D., Sarrel, E M., and Poole-Wilson, E A. (1993). Cardiovascular protection by oestrogen-A calcium antagonist effect? Lancet 341, 1264-1265. 135. Wild, R. A. (1996). Estrogen: Effects on the cardiovascular tree. Obstet. Gynecol. 87, 27s-35s.

CHAPTER 15 Cardiovascular Pathophysiology 136. Williams, J. K., Adams, M. R., and Klopfenstein, H. S. (1990). Estrogen modulates responses of atherosclerotic coronary arteries. Circulation 81, 1680-1687. 137. Williams, J. K., Adams, M. R., Herrington, D. M., and Clarkson, T. B. (1992). Short-term administration of estrogen and vascular responses of atherosclerotic coronary arteries. J. Am. Coll. Cardiol. 20,452-457. 138. Williams, J. K., Shively, C. A., and Clarkson, T. B. (1994). Determinants of coronary artery reactivity in premenopausal female cynomolgus monkeys with diet-induced atherosclerosis. Circulation 90, 983-987. 139. Clarkson, T. B., Anthony, M. S., and Klein, K. P. (1994). Effects of estrogen treatment on arterial wall structure and function. Drugs 47(Suppl. 2), 42-51. 140. Clarkson, T. B., Hughes, C. L., and Klein, K. E (1995). The nonhuman primate model of the relationship between gonadal steroids and coronary heart disease. Prog. Cardiovasc. Dis. 38, 189-198. 141. Gisclard, V., Miller, V. M., and Vanhoutte, P. M. (1988). Effect of 17/3estradiol on endothelium-dependent responses in the rabbit. J. Pharmacol. Exp. Ther. 244, 19-22. 142. Hayashi, T., Fukuto, J. N., Ignarro, L. J., and Chaudhuri, G. (1992). Basal release of nitric oxide from aortic rings is greater in female rabbits than in male rabbits: Implications for atherosclerosis. Proc. Nat. Acad. Sci. U.S.A. 89, 11259-11263. 143. A1-Khalili, F., Eriksson, M., Landgren, B., and Schenck-Gustafsson, K. (1998). Effect of conjugated estrogen on peripheral flow-mediated vasodilation in postmenopausal women. Am. J. Cardiol. 82, 215"218. 144. Gangar, K. E, Vyas, S., Whitehead, M., Crook, D., Meire, H., and Campbell, S. (1991). Pulsatility index in internal carotid artery in relation to transdermal oestradiol and time since menopause. Lancet 338, 839-842. 145. Herrington, D. M., Braden, G. A., Williams, J. K., and Morgan, T. M. (1994). Endothelial-dependent coronary vasomotor responsiveness in postmenopausal women with and without estrogen replacement therapy. Am. J. Cardiol. 73, 951-952. 146. Gruchow, H. W., Anderson, A. J., Barbriak, J. J., and Sobocinski, K. A. (1988). Postmenopausal use of estrogen and occlusion of coronary arteries. Am. Heart J. 115, 954-963.

243 147. Sullivan, J. M., Vander Zwaag, R., Lemp, G. F., Hughes, J. R, Maddock, V., Koetz, F. W., Ramanathan, K. B., and Mirvis, D. M. (1988). Postmenopausal estrogen use and coronary atherosclerosis. Ann. Intern. Med. 108, 358-363. 148. Volterrani, M., Rosano, G. M. C., Coats, A., Beale, C., and Collins, E (1995). Estrogen acutely increases peripheral blood flow in postmenopausal women. Am. J. Med. 99, 119-122. 149. Reis, S. E., Gloth, S. T., Blumenthal, R. S., Resar, J. R., Zacur, H. A., Gerstinblith, G., and Brinker, J. A. (1994). Ethinyl oestradiol acutely attenuates abnormal coronary vasomotor responses to acetylcholine in postmenopausal women. Circulation 89, 52-60. 150. Collins, E, Rosano, G. M. C., Sarrel, E M., Ulrich, L., Adamopoulos, S., Beale, C. M., McNeill, J. G., and Poole-Wilson, E A. (1995). Oestradiol-17-fl attenuates acetylcholine-induced coronary arterial constriction in women but not men with coronary heart disease. Circulation 92, 24-30. 151. Gilligan, D. M., Badar, D. M., Panza, J. A., Quyyumi, A. A., and Cannon, R. O., III (1994). Acute vascular effects of estrogen in postmenopausal women. Circulation 90, 786-791. 152. Gilligan, D. M., Quyyumi, A. A., and Cannon, R. O., III (1994). Effects of physiological levels of oestrogen on coronary vasomotor function in postmenopausal women. Circulation 89, 2545-2551. 153. Celermajer, D. S., Sorensen, K. E., Bull, C., Robinson, J., and Deanfield, J. E. (1994). Endothelium-dependent dilation in the systemic arteries of asymptomatic subjects relates to coronary risk factors and their interaction. J. Am. Coll. Cardiol. 24, 1468-1474. 154. Celermajer, D. S., Sorensen, K. E., Spiegelhalter, D. J., Georgakopoulos, D., Robinson, J., and Deanfield, J. E. (1994). Aging is associated with endothelial dysfunction in healthy men years before age-related decline in women. J. Am. Coll. Cardiol. 24, 471-476. 155. Hashimoto, M., Akishita, M., Eto, M., Ishikawa, M., Kozaki, K., Toba, K., Taketani, Y., Orimo, H., and Ouchi, Y. (1995). Modulation of endothelium-dependent flow-mediated dilation of the brachial artery by sex and menstrual cycle. Circulation 92, 3431-3435. 156. Taddei, S., Virdis, A., Ghiadoni, L., Mattei, E, Sudano, I., Bernini, G., Pinto, S., and Salvetti, A. (1996). Menopause is associated with endothelial dysfunction in women. Hypertension 28, 576-582.

7HAPTER 1(

Insulin Bo dy 9 Resistance, " Weight Obesity, Body Co os t on, and the Menopausal Transition MARYFRAN SOWERS AND JENNIFER TISCH Department of Epidemiology, School of Public Health, University of Michigan, Ann Arbor, Michigan 48109

V. Linking Body Composition and Carbohydrate Metabolism with Hormone Status VI. Summary Appendix References

I. I n t r o d u c t i o n

II. Mechanisms for Sex Hormones, Body Composition/ Topology, and Insulin Regulation III. Ovarian Hormone Status and Body Composition/Body Topology IV. Carbohydrate Metabolism and Change in Ovarian Hormone Status


women. Differences in body composition have been linked to mortality, heart disease, gall bladder disease, certain cancers, osteoporosis, and arthritis in pre- and postmenopausal women [ 1-4]. Additionally, differences in body composition are also associated with a number of undesirable metabolic characteristics, including glucose intolerance, hyperinsulinemia, elevated triglycerides, low high-density lipoprotein (HDL) cholesterol concentrations, and potentially plasma leptin concentrations [5,6]. This suggests that if menopauserelated events influence body composition, the metabolic impact is particularly important for those diseases linked to disordered carbohydrate metabolism, including diabetes and heart disease in women.

Possible changes in body composition associated with events linked to a menopausal transition have not been well characterized, although a limited number of studies, with selected populations, have provided some preliminary information. Furthermore, although there is some evidence of differences in body composition according to race or ethnicity, the preponderance of existing data are limited to studies of Caucasian women or use methodological approaches that preclude comparability between race/ethnic groups. It is increasingly well appreciated that body composition, including topology, is related to disease conditions in

MENOPAUSE: BIOLOGY AND PATHOBIOLOGY

245


246

SOWERS AND TISCH

This chapter is organized into four topics. First, we review potential mechanisms by which the events of the menopausal transition, including change in sex steroid hormone concentrations, are related to body composition changes or to change in measures of carbohydrate metabolism. Second, studies of body size are related to three states closely associated with the menopausal transition. These states include not only the menopausal transition but also surgical menopause (with a particular focus on oophorectomy) and the use of hormone replacement therapy. Third, we relate studies of the menopausal transition, surgical menopause, and hormone replacement therapy to measures of carbohydrate metabolism. Finally, we relate potential body composition changes in the menopausal transition to changes in carbohydrate metabolism and insulin resistance. The Appendix encompasses a description of methods used to evaluate body sizemincluding weight, obesity indices, body composition assessment, and body distribution (topology)--for those readers unfamiliar with body composition methodology.

FOR SEX BODY COMPOSITION/ AND INSULIN REGULATION

Weight gain and body topology

Menopause

Impaired Carbohydrate Metabolism

FIGURE 1 Model 1: Hormone changes with menopause lead to weight gain and change in body topology and, in turn, this leads to impaired carbohydrate metabolism.

Rarely have investigators explored the multiple pathways through which sex steroids (or their change with the menopausal transition), body composition, and measures of carbohydrate metabolism interact to influence the health of women. This review addresses mechanisms and studies that relate sex steroid concentrations to body composition and to measures of carbohydrate metabolism. We summarize several suggested mechanisms for the interaction between sex steroid hormones and body composition. It should be noted that these proposed mechanisms tend to evoke increasing insulin resistance as an integral part of the process or designate insulin resistance as a result of the process (see Figs. 1 and 2)

II. M E C H A N I S M S HORMONES, TOPOLOGY,

It is hypothesized that sex steroid hormones influence tissue mass, both lean and fat mass, albeit this has limited support in the literature. Because the menopausal transition is associated with marked shifts in gonadotropin and ovarian hormone concentrations, it is further hypothesized that the menopausal transition has a major impact on total weight and body composition, promoting an increase in fat mass and a redistribution of that fat mass to the abdomen. It is further believed that this influence is independent of a weight accumulation with aging. Several mechanisms have been proposed as to how the change in sex steroid hormones associated with the menopausal transition influences body composition and its topology. It is difficult to disaggregate the mechanisms associated with the accumulation of fat tissues (obesity) as compared to the role of fat patterning, suggesting either an immature understanding of these mechanisms or that body composition and its patterning are intrinsically linked and should not be conceptually divorced from each other. It is believed that the change in sex steroid hormones associated with the menopausal transition also affects the activities of insulin. It is hypothesized that there is probably a direct and indirect role. The direct role is probably mediated through hepatic glycogen deposition or sensitivity of the muscle bed to insulin action. An indirect mechanism may be related to an increasing insulin resistance associated with a change in body composition, particularly an increase in fat mass and decrease in muscle mass.

A. E n e r g y B a l a n c e It has been suggested that any increase in weight with the menopausal transition (or with a change in sex steroid concentrations) is a direct function of direct energy intake coupled with resting energy expenditure (REE). Resting energy expenditure decreases with increasing age, a decline that is proposed to begin with the menopausal transition. Furthermore, if there were a decline in physical activity, the combined effect would be to further increase the likelihood of weight gain [7]. The hypothesis indicates the difficulty in separating an age effect from an menopausal transition effect. Recent studies have reported that with age, an increase in adiposity and a decline in physical activity are associated with a decline in the levels of insulin-like growth factor I (IGF-I). IGF-I has anabolic growth-promoting effects of

HormoneChange]

~,, odyTopology

Impaired Carbohydrate Metabolism

Weight Increase

FIGURE 2 Model 2: Hormone change with menopause (estradiol, progesterone, and testosterone) and increasing weightgain leadto modification of body topologyand impaired carbohydratemetabolism.

CHAPTER 16 Body Composition, Insulin, and the Menopause growth hormone and insulin-like activities. Poehlman et al. [8] have shown that decline in IGF-I was related to a decline in physical activity and they have proposed that this decline in IGF-I might be related to the change in body composition with menopause. Studies have inadequately addressed the role of sex steroids in the regulation of IGF-I particularly in relation to body composition.

B. L e p t i n Estrogen may also influence body composition though its interaction with leptin. Leptin may be the hormone produced by body stores to signal adequate energy stores (adipose tissue) for reproduction. In some [9,10], but not all [11] studies leptin concentrations are higher in premenopausal women than postmenopausal women. Furthermore, estrogen regulates leptin production in rats and human subjects in vivo [9].

C. L i p o p r o t e i n L i p a s e and F a t t y A c i d D e p o s i t i o n Another mechanism includes a direct mechanism of estrogen concentrations linked to the patterning of android and gynoid fat. Estrogen is postulated to regulate directly lipoprotein lipase (LPL) activity in the gluteofemoral adipocytes. LPL is a key enzyme in the regulation of fatty acid deposition. It is proposed that during the premenopausal period, lipid deposition in the gluteofemoral adipocytes is facilitated by LPL via estrogen pathways in order to assure adequate energy stores for reproduction. For example, during lactation there is mobilization of energy from the gluteofemoral adipocytes to the breast for milk, a physiological mechanism mediated by suppression of estrogen and controlled by prolactin. Furthermore, there is some evidence of enlargement of femoral subcutaneous adipocytes to support the demands of pregnancy and lactation and this enlargement is reported to disappear with menopause [ 12]. Concurrently, it has been suggested that progesterone competes with glucocorticoid receptors [13,14] and may protect adipocytes from glucocorticoid effects during the late luteal phase of the menstrual cycle, a protection that would be increasingly compromised with the menopausal transition. It is extrapolated that with the events of the menopausal transition, LPL levels are minimally or no longer under the influence of estrogen (and potentially progesterone) and the gluteofemoral adipocytes no longer function as the major source of energy storage. These mechanisms have been suggested as being reflected in the relative location of regional fat deposition. In premenopausal women, the gluteofemoral pool is associated with higher LPL activity and low lipolytic activity. If the abdominal pool is the source of high lipolytic activity, there is rapid turnover of nonesterified fatty acids. These fatty acids

247 drain into the portal system and may compete with glucose as fuel, leading to insulin resistance. In postmenopausal women, the femoral pool is associated with lower lipoprotein lipase activity while the abdomen is associated with increasing visceral fat disposition and with higher lipolytic activity. In women using hormone replacement therapy (HRT), the femoral pool would again reflect stimulated lipoprotein lipase activity while the abdominal pool would have lower lipoprotein lipase activity. Thus, in premenopausal women, there is greater activity in the gluteofemoral region. With the menopausal transition, the abdomen becomes the "default" region, in the absence of estrogen replacement [ 15].

D. Stress and the C R F - A C T H - A d r e n a l

Axis

An extension of these mechanisms is based in a belief that increased glucocorticoid stimulation [ 16] is manifest in the increased size of the abdominal visceral adipose pool. In the states of depression, dysphoria, maladaption, and chronic stress, as well as in smoking behavior and alcohol use, chronic hypothalamic stimulation is reflected as increased activity in the CRF-ACTH-adrenal axis. This stress-based condition is proposed to occur with the menopausal transition [ 1].

E. W e i g h t and A n d r o g e n i s m A modification of these mechanisms suggests that with an increase in weight, there is an increase in insulin resistance (reflected as increased insulin concentrations). This is further associated with an inhibition of sex hormone binding globulin which, in turn, is associated with increased free testosterone concentrations and greater androgenicity, reflected by more centroid obesity [ 1,17-19]. Investigators have suggested a number of mechanisms whereby estrogen status is related to insulin action [17-19]. These mechanisms typically relate to glycogen status in the liver or muscle uptake of glucose for energy. The mechanisms are as follows: 1. Estrogens may have a direct effect on glycogen deposition in the liver. The impact of a menopausally related decline in estradiol concentrations and a relative increase in estrone has not been investigated relative to glycogen deposition. 2. Alternatively, estrogens may enhance the permissive effect of corticosteroids on hepatic glycogen deposition. 3. Estrogen decreases glucagon secretion; however, again the secretion of glucagon relative to shift in the estrogen levels with the menopausal transition has not been investigated. 4. Estrogen may enhance sensitivity of glucose uptake in muscle. The change in sex steroid concentrations may have a dual impact with respect to muscle mass. If the loss of estrogen minimizes glucose uptake in muscle, and if there is a loss

248

SOWERS AND TISCH

of muscle mass (in the presence of an increase in fat mass), the menopausal transition could pose a substantial impact both directly and indirectly on carbohydrate metabolism.

diminution of ovarian estrogen and progesterone achieved over a period of years), and hormone replacement therapy (with a rapid reintroduction of estrogen and progestins).

A. W e i g h t as a M e a s u r e o f B o d y C o m p o s i t i o n E

Summary 1. WEIGHT GAIN AND MENOPAUSE

The mechanisms by which changes in sex steroid hormone concentrations may affect body composition, body topology, or insulin activity are incompletely understood. A major hypothesis suggests that changes in body composition and insulin activity with the menopausal transition may be a function of change in energy expenditure and the decline in concentrations of growth factors. These changes would include an increase in fat mass relative to lean and a greater likelihood of the deposition of that fat mass in the abdomen as visceral fat. A second hypothesis links estrogen, and potentially progesterone, with lipoprotein lipase activity, particularly in the gluteofemoral region to optimize energy stores for reproduction, a mechanism that is minimized with the diminution of estradiol and progesterone production during and after the menopausal transition. A third mechanism, with a focus on the stress-related responses of the adrenal cortex, may suggest an approach to increasing the size of the abdominal visceral pool, but is less closely linked with the change in ovarian function. Finally, there is the mechanism that with loss of ovarian function with either natural or surgical menopause, the ratio of estrogen to androgen (testosterone) shifts and that a more androgenic state is more likely to be associated with the deposition of fat viscerally. Changes in sex steroid hormone concentrations may directly influence glycogen deposition and/or secretion, or through glucose uptake in muscle. It is quite possible that each of these mechanisms contribute interactively during and after the menopausal transition in a dynamic process.

III. OVARIAN HORMONE BODY COMPOSITION/BODY

STATUS AND TOPOLOGY

Because most studies have not simultaneously considered the interaction of the menopausal transition (or change in sex steroid concentrations), body composition, and measures of carbohydrate metabolism, we are forced to review these components independently. In this section, we review findings from studies that have reported measures of weight and body mass index (BMI). We will also review the more direct measures of body composition, including dual energy X-ray absorptiometry (DXA), underwater weighing and bioelectrical impedance (BIA), and body topology (waist-to-hip ratio) in three conditions in which there are changes in ovarian hormone status. These three states include oophorectomy (with a rapid and typically complete loss of ovarian estrogen and progesterone), natural menopause (with a more erratic

Weight gain, particularly after the menopausal transition, is thought to be a common occurrence [20]; however, remarkably few studies have simultaneously considered both age and menopausal status in relation to weight gain. For example, Kirchengast [21] reported an increase in weight across three age groups of Viennese women recruited from groups defined in clinical settings. The "older" premenopausal group (aged 32-41 years) weighed 64.3 kg; the perimenopausal group (aged 3 8 - 4 9 years) weighed 67.6 kg and the postmenopausal group (47-64 years) weighed 72.4 kg. The investigators excluded all potential enrollees with a BMI greater than 30 kg/m 2. They reported this weight increase as a function of menopausal status without adjusting for an age contribution. The longitudinal studies of weight suggest no association of change in weight with menopause after adjusting for the contribution of age. For example, in a 3-year follow-up of Pittsburgh women, Wing et al. [20] reported that similar weight gain was seen across time in both pre- and postmenopausal women. Similarly, in a 6-year follow-up of women in G6teborg, Sweden, Lindquist [22,23] concluded that the increase in weight appeared to be age, not menopause, related. These investigators excluded HRT users (7.5% of their population) and women with surgical menopause (9% of their population) from their analyses. The patterns of agerelated weight change may not be simple linear increases in the peri- and postmenopausal periods. In the 1996 report from G6teborg, investigators reported that women who remained premenopausal at two examinations gained weight; women who were pre- and then postmenopausal likewise gained weight, and yet women who were postmenopausal at both examinations lost weight in the interim time between evaluations. Thus, there are reports from a limited number of crosssectional studies of an association between the menopausal state and weight increase. However, rarely do these studies consider whether this association reflects an additional contribution of menopausal status to age or whether the weight gain is really just an age-related phenomenon that happens to coincide with a menopausal transition. The longitudinal studies suggest that when the contribution of age is considered, menopausal status per se is not a statistically important contributor to weight gain. Furthermore, although weight gain reflects an increase in total mass, it does not reflect changes in the constituents of the mass (lean vs. fat tissue) nor does it reflect any changes in the regional distribution or topology of that mass.

CHAPTER 16 Body Composition, Insulin, and the Menopause 2. WEIGHT GAIN WITH OOPHORECTOMY In considering oophorectomy as a model for evaluating the role of weight gain and the hormone change with menopause, it is important to distinguish the difference between oophorectomy and hysterectomy with and without ovarian conservation. Obviously, women who have hysterectomy with ovarian conservation do not reflect the same endogenous hormonal environment as those women without ovaries. These studies are not reporting data on weight alone, but instead use BMI, which takes into consideration weight per unit of height and show no association. 3. WEIGHT GAIN AND HRT USE

Studies of weight in women using hormone replacement therapy do not particularly clarify the nature of the relation. Short-term studies may show an increase in weight. For example, Gambacciani et al. [24] reported that a group of 12 control women not using HRT increased their body weight by 1.9 kg in 12 months whereas those using HRT, on average, had no weight increase. Yet, in a study that spanned 15 years, Kritz-Silverstein and Barrett-Connor [25] reported that there was little difference in the weight change between those women who continuously used estrogen replacement therapy (ERT) as compared to those without such use. Finally, the Postmenopausal Estrogen/Progestin Intervention (PEPI) Trial reported a very modest increase in weight in the placebo group during the 3 years of the trial as compared to the four treatment groups, but none of the differences was statistically different [26]. 4. SUMMARY

249 the average difference is approximately 1 BMI unit (killigrams per meter squared), translatable to a difference of approximately 10 pounds [27-29]. Longitudinal studies of BMI and the menopausal transition suggest a different relation than do the cross-sectional studies. In the longitudinal studies, the increase in BMI during the menopausal transition is not independent of age. This is observed across a number of populations including United States women in Pittsburgh [20] and Framingham [30], Japanese women in Nagasaki [31 ], and Swedish women in G6teborg [22,23]. These reports also span several age cohorts. The Framingham women were evaluated during the 1960s and 1970s whereas the Pittsburgh women were observed during the 1980s and 1990s, suggesting these observations are not limited to unique cohorts of women. The differences in findings of the cross-sectional studies as compared to the longitudinal studies suggest the need to determine whether the cross-sectional studies reflect clinic populations, who will have a different experience related to the age at which they will have their menopausal transition. Are premenopausal women in cross-sectional studies chronologically aged 35, 40, or 45 ? How long will it be before they experience the year of amenorrhea associated with menop a u s a l - 5 , 7, 11, or 13 years? Cross-sectional studies are potentially forced to misclassify women according to their menopausal status based on age and information from a single point of time. Typically, the cross-sectional studies frequently enroll many fewer women than do longitudinal studies. These characteristics may explain the discrepancies in findings between the cross-sectional and longitudinal studies. 2. B M I AND OOPHORECTOMY

The evidence is highly inconsistent as to the role of weight gain accompanying transitions associated with the menopausal period. Weight gain does occur, but there is substantial difficulty in linking this weight gain just to the events of a transition. Most of the studies, which are typically crosssectional in design, do not try to separate the independent effects of age and menopause status. Additionally, readers should exercise caution in generalizing the findings of the studies to external populations, because there are substantial opportunities for selection bias in using or observing clinical populations.

B. B o d y M a s s I n d e x as a M e a s u r e of Body Composition 1. BMI AND THE MENOPAUSAL TRANSITION Measures of weight alone do not consider the amount of body tissue per unit of height; thus, the use of BMI (weight/ height 2) is a better measure of body habitus. There have been a number of cross-sectional studies that have reported a greater BMI among postmenopausal women as compared to premenopausal women, after adjusting for age. Typically,

A Finnish study [32] indicates that women with hysterectomy and preservation of at least one ovary had a greater BMI than did those who had not undergone hysterectomy. 3. BMI ANn HORMONE REPLACEMENT In a number of prospective studies and clinical trials, there have been conflicting findings as to whether estrogen or hormone replacement therapy is associated with an increase in B MI. For example, the results from the Nurses Health Study (wherein BMI was self-reported) indicated that estrogen use was associated with lower BMI [33]. In a study among Pittsburgh women, Matthews et al. [34] reported greater weight gain among estrogen users. Notelovitz [35], Nachtigall [36], and Utian [37] have each reported no influence of estrogen replacement on weight following menopause. The Pittsburgh investigators have argued that estrogen users are a highly selective group and that there are differential factors, including baseline weight, that affect their weight change patterns as compared to women who do not use HRT [38]. These contradictory findings indicate the nature of problems associated with making definitive statements about weight or BMI and the menopause. Prospective studies of

250

SOWERS AND TISCH

women transitioning the menopause are uncommon. Rarely do prospective studies report the role of initial or baseline weight in explaining the amount of weight or BMI change. Furthermore, information about factors that influence weight, such as smoking, diet, and exercise habits, are frequently unavailable or not considered in the statistical adjustment. Increasingly, there is an appreciation of the kinds of bias that can lead to an inappropriate interpretation of the true association between menopause, HRT use, and weight. Many studies may not consider the age of menopause and whether the menopause was natural or surgical. The latter could introduce significant bias if there is an important association of socioeconomic status and weight, and those with lower socioeconomic status are more likely to have a surgical menopause. Among those studies of hormone replacement therapy, the sample sizes are frequently small and limit the ability to generalize. Formulations of hormone replacement are mixed. The preparations are strikingly different and include conjugated equine estrogens, alkalyted estrogens, or estrogen/progestin combinations. Doses are wide ranging. The evaluation of hormone replacement therapy also rarely accounts for why women were taking the therapy and how this might influence weight change pattern. Thus, the change in BMI could readily be different for the 40-year-old woman with a hysterectomy/oophorectomy associated with dysfunctional uterine bleeding as compared to the 55-year-old woman with postmenopausal hot flashes. 4. SUMMARY Currently, there is evidence that weight gain (as an absolute measure of weight or BMI) around the menopausal transition is likely to be linked to age, and is not unique to the menopausal transition. This has been most clearly demonstrated in prospective studies whose populations are not specifically selected from clinical patients. This is in direct contradiction to the commonly held perceptions of weight gain during the menopausal transition. Although there is little to support the concept of a major increase in weight with the menopause, supplementary data are needed to address whether there is a shift in the proportion of fat relative to the amount of lean tissue. Current data also do not address the potential shift in the location of large amounts of fat tissue to visceral stores in the abdomen.

C. Direct Measures of Body Compartment (Fat and Lean) Only a few studies have involved more direct measures of the lean and fat compartments in relation to menopause. Kirchengast [21] reported an increase in percentage body fat as measured by DXA across three age groups of Viennese women recruited from clinical populations. The older premenopausal group (aged 32-41 years) was 32.4% fat, the

perimenopausal group (aged 3 8 - 4 9 years) was 36.8%, and the postmenopausal group (47-64 years) was 38.1% fat. Two other studies using DXA [39,40] reported an 8-9% greater fat mass in postmenopausal women than in premenopausal women. Other studies, using more direct measures of body composition, have found little body composition change in relation to menopausal status [41-44]. Thus, approximately half of the studies suggest an increase in percentage body fat whereas the other half do not. Limited evidence from the very few longitudinal studies suggests that the sizes of body compartments (fat and lean mass) change with both ovarian aging and chronological aging. Sowers has shown that, on average, the 0.9 kg total weight gain annually was associated with a 1.4 kg increase in the fat compartment and a 0.5 kg decrease in the lean compartment, after adjusting for age [2]. The population includes more than 400 pre- and perimenopausal women, aged 2 5 45 years, and those women with hysterectomy were more likely to have an increase in the size of the fat compartment. In a widely cited study incorporating data from just 38 women, aged 4 4 - 4 8 years, who were assessed with underwater weighing, Poehlman [8] described a "menopause" effect in size of the fat compartment in a 6-year follow-up. Essentially, the definitive studies have not been undertaken to characterize the change in the size of the fat and lean body compartments with menopausal transition, after accounting for age effects. 1. BODY COMPOSITION AND EXOGENOUS HORMONE USE With estrogen/progestin preparations, Haarbo et al. [45] reported no impact on fat mass in 62 early postmenopausal women, consistent with the work of Hassager and Christiansen [46], who reported that estrogen replacement prevented an increase in fat mass. 2. SUMMARY Although suggestive of an increase in fat mass, the limited number of studies considering the sizes of the fat and lean compartments are insufficient to determine if there are actual shifts in the proportion of fat and lean with the menopausal transition. Furthermore, it is uncertain that these changes might occur apart from that which would be observed with age alone. Likewise, there is insufficient information about the differential impact of exogenous hormone replacements that include progestins as compared to those based on conjugated equine estrogens. Although speculation, conjecture, and hypotheses abound, evidence is minimal.

D. Body Topology (Patterning) and the Menopause There is extensive evidence of the influence of ovarian aging with respect to body composition topology [44]. Usually,

CHAPTER 16 Body Composition, Insulin, and the Menopause this association is described with waist-to-hip ratio (WHR). The abundance of evidence of the influence of ovarian aging on body topology may be a function of the relative ease of measuring WHR in larger populations. It is more costly and more difficult to use computerized tomography to measure body topology. Alternatively, total body DXA scans indicate differences in the estimate of android (upper body segment, trunk) and gynoid (hip and thigh region) distributions; however, this measurement does not differentiate between subcutaneous fat and visceral adipose tissue and whether the visceral fat tissue is saturated adipose tissue. These latter depots are thought to be most deleterious to health. Current data from the dozen or so cross-sectional studies of body fat distributions are contradictory in their findings. Approximately half of the studies suggest a shift to a more negative topological profile with menopause (following adjustment for age and/or BMI). The remaining studies suggest retaining a positive profile, reflecting a gynoid, not android, patterning [8,21,27-29,40,42-44,47-53]. In two longitudinal studies, there was apparently a menopause-based change in body topology. Poehlman [8] reported a menopause-related change in 38 women followed for 6 years. Bjorkelund [54] reported 1237 women from G6teborg, aged 38-60, to have an increase in waist and a decrease in the hip circumference. They suggested the change in body composition was a gradual process that started well before the cessation of menstrual bleeding and the process continued at a slower rate in the postmenopause. In a longitudinal study comparing pre- and perimenopausal women, Sowers reported that perimenopausal women had a greater waist girth relative to hip girth than the younger premenopausal women, after adjusting for age and other important variables such as smoking, physical activity, and parity [2]. 1. BODY TOPOLOGY AND EXOGENOUS HORMONE USE

Following menopause, it appears that there is an increase in visceral fat mass that is preventable, at least in part, by hormonal replacement therapy [24,45,55]. For example, Gambacciani et al. [24] reported that a group of 12 women using HRT, on average, had no weight increase whereas those not using HRT increased their body weight by 1.9 kg in 12 months. Simultaneously, in women not using HRT, the fat was patterned as an increase in total body fat mass to the trunk and arms. In the HRT-treated group (n = 15), there were no pattern modifications to the trunk. Studies by Haarbo [45], Kaye [56], den Tonkelaar [29], and Wing [20] each have reported that postmenopausal hormone use was associated with lower waist-to-hip ratio, albeit den Tonkelaar et al. [47] indicated that the association was no longer present following adjustment for age and obesity in their crosssectional study. In contrast, the PEPI Trial found no difference in the waist-to-hip ratio of those treated for 3 years with four different hormone preparations versus those in the placebo-control group [26].

251 2. SUMMARY

It appears that a redistribution of adipose tissue may occur with menopause based on longitudinal studies of the menopausal transition and the use of exogenous hormones. More studies are needed, however, that incorporate factors infrequently considered when evaluating waist-to-hip ratio, including ethnicity, the baseline weight, the type of preparation or formulation of replacement therapy used, smoking behavior, and socioeconomic status. Furthermore, more viable measures of both visceral adiposity and saturation of adipocytes are needed for implementation in both clinical and epidemiological studies. For example, both published and unpublished studies have shown that waist circumference explained more variation in HDL cholesterol and triglyceride concentrations than did WHR [4].

IV. C A R B O H Y D R A T E AND CHANGE HORMONE

METABOLISM

IN OVARIAN

STATUS

Carbohydrate metabolism and insulin resistance are believed to be important for two reasons. First, as carbohydrate metabolism becomes increasingly disordered and insulin resistance increases, greater numbers of women will have both diagnosed and undiagnosed diabetes. Second, insulin resistance is an important risk factor for coronary heart disease (CHD) in both men and women. Fasting plasma insulin concentrations correlate well with measures of insulin resistance using clamp studies, albeit not as glucose tolerance becomes abnormal [57]. Fasting insulin concentrations are correlated with other risk factors for coronary disease, including increases in blood pressure, circulating cholesterol and triglycerides, and decreases in HDL cholesterol in both Caucasians and African-Americans [58,59]. Some have postulated that changes in insulin and insulin sensitivity at the menopause may play a central role in women's increased risk for coronary heart disease at older ages, but findings have been conflicting. Fasting insulin concentrations are related to body fat distribution and weight gain [60,61 ].

A. A n i m a l S t u d i e s More than 30 years ago, Foglia [62] demonstrated that in castrated, pancreatectomized rats, estrogen replacement was associated with an initial deterioration in glycemic control followed by subsequent protection as pancreatic islet cells were regenerated. More recently, it was shown that castration increased the incidence of diabetes in female rats and this could be reversed with the administration of estrogen or testosterone [63]. In general, the work in animals suggests a beneficial effect for estrogen on insulin secretion in castrated and

252

SOWERS AND TISCH

intact rats. Investigators have generally observed an increased weight gain in ovariectomized rats as well as an increased insulin responsiveness in treated animals as compared to untreated animals [ 19]. Animal studies have also suggested other ovarian hormones, apart from estrogen, need to be considered. For example, there is some evidence from animal studies of a direct effect of progesterone and progestins on islet cells. Progesterone enters the islet cells, binds to cytosolic receptors, and enters the nucleus [64,65]. Progesterone and progestins can reduce glucose uptake in skeletal muscle and the uptake of glucose into lipid [19], resulting in a progesterone-induced resistance to insulin action in peripheral tissues.

B. H u m a n S t u d i e s There are relatively few studies of the effects of human menopause, either natural or surgical, on carbohydrate metabolism. Several studies have included both pre- and postmenopausal groups, although it is frequently unclear as to the perimenopausal status of the premenopausal women. 1. NATURAL AND SURGICAL MENOPAUSE

The role for menopause, surgical or natural, in carbohydrate metabolism is ill defined. One cross-sectional study showed a fall in insulin concentrations when comparing 426 pre- versus postmenopausal Italian women [42,43]. In contrast, another cross-sectional study of Turkish women [66] identified no difference in the insulin:glucose ratio among women with premature ovarian failure and natural menopause. Other studies have described increased fasting insulin concentration with menopause and suggested that the loss of ovarian function may explain this increase [58,67,68]. Two longitudinal studies have examined measures of carbohydrate metabolism and menopausal status. Natural menopause was not associated with increased glucose concentrations [30] in the Framingham Study. Likewise, natural menopause was not associated with the response to oral glucose tolerance tests in the participants of the Healthy Women's Study [34]. In the Framingham Study, ovariectomy was not associated with increased glucose levels [30]. These inconsistencies in study design and the measures used to evaluate different aspects of carbohydrate metabolism suggest the importance of determining whether the decline of estrogen and progesterone hormones at the menopause initially reduces both insulin secretion and elimination. It is also important to determine whether increasing insulin resistance at the cellular level induces an increase in circulating insulin concentrations, what causes the insulin resistance, and when the insulin resistance is likely to be sufficiently great as to generate a clinical diagnosis of disease. Godsland [ 19] suggested that postmenopausal women have a reduction in glucose-induced insulin secretion with a compensatory reduc-

tion in insulin elimination [69]. In this situation, detection of a menopausal difference will be difficult because a deficiency in insulin secretion but a compensatory reduction in insulin elimination could result in no net change in glucose tolerance or circulating insulin concentrations.

2. H o r m o n e R e p l a c e m e n t and Carbohydrate Metabolism The literature on hormone replacement and carbohydrate metabolism is much more extensive than the literature associated with natural or surgical menopause without replacement. Although there is an advantage in terms of frequency of a greater number of studies, the findings may require interpretation within the type and amount of the estrogen and/ or progestin used or within the degree to which carbohydrate metabolism was normal at the time treatment was initiated. For example, studies have shown that 2-hr insulin secretion in postmenopausal women did not differ between those who received hormone replacement therapies and those who did not receive therapy [26]. However, in 25 postmenopausal women with non-insulin-dependent diabetes mellitus (NIDDM), treatment with 17fl-estradiol for 3 months resulted in the diminution of hyperandrogenic characteristics and improved glucose homeostasis as well as plasma lipids [70]. In an excellent review, Godsland [19] concluded that estradiol replacement in postmenopausal women is associated with improved insulin sensitivity. When the replacement therapy was conjugated equine estrogens (Premarin) at the 0.625-mg/day dose, there was either an improvement or no change in insulin sensitivity, whereas administration of the 1.25-mg/day dose was associated with deterioration in insulin sensitivity. The deterioration in insulin sensitivity with the higher doses may reflect a secondary effect of increasing glucocorticoid activity. When the therapeutic regimen was alkylated estrogen, such as ethinyl estradiol and mestranol, there was deterioration in glucose tolerance and insulin resistance. Transdermal methods of delivery, in which exposure to the liver is averted, were not associated with changes in carbohydrate metabolism. Formulations incorporating progestins probably did not have the same effect as those based on estrogens [71]. A number of mechanisms have been proposed to explain the potential glucose tolerance and insulin resistance seen in formulations that are estrogen/progestin combinations and these mechanisms parallel those associated with mechanisms described in relation to body composition. These include increased circulating fatty acid levels, impact on growth hormone concentration, increased glucocorticoid activity in the obese via increased synthesis and decreased corticosteroid half-life, and reduced insulin receptor concentrations [ 19].

CHAPTER 16 Body Composition, Insulin, and the Menopause 3. SUMMARY

In animal models, estrogen deficiency is associated with impaired carbohydrate metabolism and increased risk of diabetes. This was reversible with the administration of an estrogen replacement. In humans, there is relatively little information about the process of the menopausal transition and measures of carbohydrate metabolism. There is substantially more information in women generated through studies of both estrogen replacement therapy and hormone replacement therapy, suggesting that the ovarian hormones estrogen and progesterone can have an impact on carbohydrate metabolism. However, the impact of a hormone replacement regimen depends on the dose of the estrogenic material, the formulation (i.e., alkylated vs. conjugated equine estrogens), the delivery system (tablet vs. transdermal), and the presence or absence of progestins. Estrogen replacement could result in improvement of markers of carbohydrate metabolism and insulin status. However, higher estrogen concentrations have resulted in deterioration of glucose tolerance, potentially through increased corticosteroid action and tryptophan metabolism. Understanding these mechanisms may be important linkage of body composition to subsequent risk of heart disease and/ or diabetes. There has been much focus on the role of estrogen status in the link between body composition and glucose tolerance, whereas the importance of progesterone and progestins is probably inadequately appreciated. In the menopause, ultimately, there is not only a diminution of estradiol production, there is also a loss of progesterone activity. Progesterone and some progestagens induce insulin resistance and this element must be considered in future studies that attempt to describe the impact of the menopausal transition on carbohydrate metabolism.

V. LINKING BODY COMPOSITION AND CARBOHYDRATE METABOLISM WITH H O R M O N E STATUS Cross-sectional studies have indicated that visceral adiposity is associated with alterations in glucose/insulin homeostasis. Furthermore, there is an association of age and menopausal status with the visceral adiposity, most frequently expressed as the ratio of size of the waist circumference to the hip circumference. Based on these relationships, it has been speculated that those experiencing the menopausal transition will be more likely to present with disruptions in glucose/insulin homeostasis, even to the point of having greater incidence of diabetes. In one of the few studies of carbohydrate metabolism and measured visceral adipose tissue (AT), Lemieux et al. [72,73] suggested that the group with the greatest gain in visceral AT had the greatest

253 deterioration in glucose/insulin homeostasis. Unfortunately, sample sizes were inadequate to evaluate the role of the menopausal transition of some study enrollees through the menopause. Other studies have linked body composition and carbohydrate metabolism with menopausal status. In the Healthy Women's Study, women transitioning the menopause had increases in fasting glucose and insulin concentrations, but these increases were simultaneously seen in premenopausal women, and appeared to be independent of measures of body composition. This is interpreted as indicating that changes in carbohydrate metabolism markers were reflecting aging [34]. In a study of pre- and postmenopausal women, postmenopausal women had a reduced pancreatic insulin secretion and increased insulin as compared to the premenopausal women [60,69]. The investigators suggest that additionally, menopausally associated changes are then accompanied by increasing insulin resistance with aging.

VI. SUMMARY Some investigators have postulated that changes in insulin and insulin sensitivity at the menopause may play a role in women's increased risk for coronary heart disease at older ages; however, the findings have been conflicting. Fasting insulin concentrations are correlated with other risk factors for coronary disease, including increases in blood pressure, total cholesterol and triglycerides, and decreases in HDL cholesterol [58,59]. In spite of the importance in understanding the role of body composition and the hormone changes of the menopausal transition in relation to carbohydrate metabolism, there is much to be learned of the temporal sequence of the relationship and the magnitude of the contribution. Most current studies are not longitudinal, thus making difficult the establishment of the temporal sequence of events. The use of weight and BMI rather than the more direct markers of body composition does not allow for the accounting of changes in the lean and fat compartments or their topological redistribution. Likewise, markers of body topology or patterning have primarily been based on circumference measures that become less precise as obesity increases. The markers of carbohydrate metabolism are limited in their capacity to not only describe trends, but also supply adequate information to inform potential understandings of mechanisms. In spite of the aforementioned limitations, emerging methodologies for the more direct measure of body composition and the understanding that waist circumferences are probably adequate to approximate visceral adipose tissue will facilitate the development of this area. Furthermore, a number of studies of the menopausal transition, including the Study of Women's Health Across the Nation (SWAN) in the United States, will allow the longitudinal and simultaneous measure

254

SOWERS AND TISCH

of ovarian hormone levels, body composition, glucose, and insulin. It is anticipated that the product will be a more complete understanding of the subtle changes that occur across a relatively long time span (estimated to be 8-10 years). This understanding is important because several of the major chronic diseases arising in the last third of the life span are thought to be affected by body composition, obesity, and impaired glucose metabolism.

APPENDIX

Measuring Body Composition Body composition technically is the amount and type of tissue represented in a living organism. Within humans, this has been largely restricted to understanding the size of three compartments: (1) the amount of fat, (2) the amount of lean, and (3) the amount of bone mineral content, which is frequently included as a subcompartment of the lean tissue mass. The size of these compartments is evaluated typically for the sake of describing obesity (the amount of excess fat in relation to the amount of lean mass), starvation (the loss of both fat mass and lean mass), or the loss of bone mineral density as a risk for osteoporotic fracture. More recently, it has come to be understood that not only is the amount of fat and lean mass important, but among the more corpulent, the distribution of that fat mass is also important. Because measurement of body composition and its distribution is technically difficult in humans, particularly in large numbers of humans in epidemiology studies, a number of measures have been developed that act (with more or less success) as surrogates for the measures of body composition and its distribution. In this section, we describe how well weight and body mass index approximate body composition and then describe three techniques that more directly estimate body composition, including underwater weighing, dual X-ray densitometry, and bioelectrical impedance (BIA). To describe distributions of adipose tissue, only one measure will be described (waist-to-hip ratio), because much of the clinical and epidemiological work uses this approach.

Weight Body weight is representative of the sum of protein, fat, water, and bone mineral mass, and does not provide any information on these four chemical components. In normal adults, there is a tendency to have increased fat deposition with age, concurrent with a reduction in muscle protein. Such changes are not evident in body weight measurements and can only be evaluated by determining body fat and/or fat-free mass. Kvist et al. [74] measured whole-body adipose tissue volume by computed tomography (CT), in a group of men and women differing in adiposity. One of the best pre-

dictors was weight. This observation is similar to those by Ross et al. [75] for obese men and women. They also reported that whole-body lean tissue from magnetic resonance imaging (MRI) in obese men and women can be predicted from weight.

Body Mass Index (Quetelet Index) The body mass index was originally described by the French biostatistician, Quetelet, and named after him. It represents weight relative to stature [weight (kg)/height (m2)] and has been associated with mortality and morbidity in a J-shaped curve. The very low values have been associated with increased mortality from digestive and pulmonary diseases, and higher values associated with increases in cardiovascular diseases, diabetes, and other chronic diseases [76]. Values of B MI less than 20 are regarded as indicative of underweight, values greater than 25 are indicative of overweight, and values greater than 30 indicate obesity [77]. A BMI of 20-25 is the "ideal" index range associated with the lowest risk of illness for most people. Measures of height and weight are generally reliable and have small technical errors of measurement. Using BMI for field studies is beneficial because the measure is highly reproducible. It provides important information apart from the independent measures of weight and height in that it controls for differences in weight normally accounted for by age, gender, ethnicity, and height, thereby facilitating the assessment of obesity in different populations. Although weight adjusted for height is highly correlated with the amount of fat mass (correlations of 0.80 to 0.90), this is a less appropriate approach to the measurement of muscle mass, protein status, or lean tissues (correlations of 0.4 to 0.6). The lack of correlation between BMI and lean mass is relevant. It is in the lean compartment that metabolism of glucose takes place and where insulin resistance is manifest. Furthermore, both fat and lean compartments independently predict menstrual cycle length [78], suggesting that the lean mass may be related to sex hormone relationships. For example, the lean compartment has now been shown to be a better predictor of bone mass and its change in white [79] and Japanese women [80].

Anthropometry Anthropometry is the process of measuring various dimensions of the human body. It is the science that deals with measurement of the size, weight, and proportional dimensions. Anthropometric measurements are of two types: growth and body composition measurements.

Skinfold Thickness The use of skinfold thickness to predict body fat is one of the most common field anthropometric techniques in body composition assessment. Skinfold thickness has been used

CHAPTER 16 Body Composition, Insulin, and the Menopause extensively as a means for estimating body density and fatness. Administered correctly, skinfolds can give accurate results. These techniques are widely used in clinical and field studies because they are relatively easy to administer to large groups of individuals and the equipment is inexpensive (skinfold calipers, tape measure, and anthropometer). The main purpose of skinfold measurements is to estimate general fatness and the distribution of subcutaneous adipose tissue. Fat is pinched between a two-pronged caliper on designated body sites such as triceps, biceps, abdomen, iliac crest, just below the scapula, the thigh and the chest. A skinfold caliper is designed specifically for simple accurate measurement of subcutaneous tissue. The skinfold measurements are used in multiple equations to predict percentage body density; this has been reviewed extensively by Brodie et al. [81 ]. The equations make several assumptions about the relationships between skinfold thickness and adiposity. The first assumption is that a fixed relationship exists between subcutaneous and deep adipose tissue. Second, subcutaneous adipose tissue is representative of the total body fat. Third, selected skinfold thicknesses and adiposity change in proportion to each other. Fourth, fat-free mass is relatively constant for a given body size and skinfold thickness. In addition, there are individual differences that can invalidate these established equations. For specific populations, the validity of the equations rests on the assumption that, for each population of interest, the composition of the fat-free mass is similar. For purposes of generalizability, these equations are not appropriate because many of them are specific to a particular population, and are dependent on the age, sex, nutritional status, and genetic background [81]. Therefore, care must be used in choosing appropriate equations for the population of interest. In addition to these limitations in the application of the equations, there are logistical limitations associated with the skinfold measurement process that can result in an inappropriate estimation of the subcutaneous fat thickness and, consequently, total body fat. Some of the problems associated with this method are inability to palpate the fat/muscle interface, difficulty in obtaining interpretable measurements in obese subjects, use of calipers that do not exert constant pressure at the skinfold site, and incorrect site location. A critical disadvantage in the use of skinfolds is that it is limited to the assessment of subcutaneous fat only, and cannot accurately distinguish between abdominal visceral adipose tissue and saturated adipose tissue. A ratio of triceps and subscapular skinfolds has been used to characterize the relative amount of appendicular to axial fat. Skinfold thicknesses have low correlations with fat-free mass (approximately 0.2) but they are highly correlated with percentage body fat (r = 0.7-0.9), and these correlations do not differ significantly among the common measured sites [82-84]. Although there are relatively high correlations between skinfold thicknesses at single sites and percentage

255 body fat, no one skinfold thickness is an accurate predictor of percentage body fat, albeit triceps skinfolds appear to have the best predictability in population studies [83].

Measures of B o d y C o m p o s i t i o n In field-based studies, measures of body composition have historically been limited to simple measures using weight related to height (power indices) as measures of adiposity. The use of more complex measures of body composition have typically been precluded by the logistical demands of implementation, including number of persons to be evaluated, the lack of available facilities, and time required for implementation.

Underwater Weighing Hydrodensitometry, or underwater weighing, is the classic approach to determining body composition. Based on principles promulgated by Archimedes, the technique generates knowledge of two compartments, the fat mass and the fat-free mass. When a body is submerged in water, there is a buoyant counterforce equal to the weight of the water that is displaced. Because bone and muscle have greater density than water, a person with a larger percentage of fat-free mass will weigh more in the water. Conversely, a larger amount of fat mass will make the body lighter in the water. The individual is measured for the amount of water displacement by submerging in water while sustaining a 30-sec forced expiration. This step is required because air trapped in the lungs also contributes to the amount of water displaced by the subject. The underwater weight is recorded at the end of the forced expiration. This is then compared to the subject's weight in air to obtain body density. Estimates of the fat body and the fat-free body densities are used to calculate the size of these two body composition compartments. The fat-free mass is a heterogeneous compartment that could be further subdivided according to its primary constituents: water (73.8%), protein (19.4%), and mineral (7.8%). Although not feasible for implementation in field studies, the hydrodensitometry approach is used as the gold standard for validating other methods [85-87]. This methodology is compromised because densitometry equations were developed from direct analysis of white cadavers [85] and will lead to the systematic underestimation of relative fatness in American Indian women, black women, and Hispanic women. The fat-free body density in these race/ethnic groups exceeds the assumed value of 1.1 g/ ml [88].

Dual X-Ray Densitometry Assessment using dual X-ray densitometry was developed in the late 1980s and early 1990s as an alternative

256 approach to underwater weighing [89-91]. Currently, the measurement of body composition by dual X-ray densitometry is rapidly replacing underwater weighing as the gold standard for body composition and is logistically more viable than is underwater weighing for studies of clinical and epidemiological populations [89,90]. All DXA devices consist of three components. First, there is an energy source capable of emitting low and high energy levels. Then, a series of detectors capable of discerning amounts of energy attenuation is linked to electrical signals, and these signals provide information used in solving simultaneous equations. Finally, a mechanical system is required to allow the integrated movement of energy source and detectors to determine changes in the degree of attenuation throughout different parts of the body. The fundamental premise is that the amount of density of tissue can be estimated by measuring the amount of ionizing energy transferred through material from a source transmitting at higher and lower energy levels. The estimated fat content in bonefree lean tissue is derived by the constant attenuation of pure fat (Rf = 1.18-1.21) and the attenuation of bone-free lean tissue (R l = 1.399), where R is the attenuation coefficient. The ratio of the attenuation at the lower energy relative to the higher energy in soft tissue [for the low- and high-energy Xrays (40 and 70 keV)] is a function of the proportion of the R values for fat and lean in each pixel (area unit). Traditionally, body composition has been expressed as a two-compartment system of fat tissue and lean tissue (that included muscle, bone, and water). Measurement using DXA allows for the description of a three-compartment system composed of adipose, muscle/water, and bone mineral content. The accuracy of body composition by dual X-ray densitometry is highly correlated with that of hydrodensitometry and the precision error with scans repeated a week apart is less than 2%. There are limitations in DXA measures of body composition. DXA cannot be undertaken in the morbidly obese for two reasons. First, the table is only mechanically stable to weights between 260 and 290 pounds (depending upon the manufacturer), but deforms with loads beyond those weights, jeopardizing the entire system. Second, an assumption underlying the use of DXA is that measurements are not affected by the anteroposterior thickness of the body. However, studies have consistently shown that thickness greater than 25 cm does have an impact on evaluating the energy signal and typically overestimates the fat mass [92]. The sensitivity to change in hydration status has the potential to affect the bone-free lean tissue; however, studies indicate that this is a relatively minor source of error [93]. Last, the estimation of body composition is a function of the 40-50% of the pixels that do not contain bone. Thus, measurements of regions of the body (including the thorax and the arm, which may have relatively fewer pixels) without bone are more prone to measurement error.

SOWERS AND TIscI4

Bioelectrical Impedance A third methodology, bioelectrical impedance, is lower in cost than DXA and is logistically easier to implement in field settings. All BIA devices consist of essentially an alternating electrical current source (usually less than 0.25 V), cables, electrodes for inducing the current into the body, and a system for sensing the voltage drop due to impedance from body tissues. This approach operationalizes the assumption that the electronic conduction in biological tissues is mainly ionic, that is, electrical charges are transferred by ionized salts, bases and acids dissolved in body fluids. Thus, simplistically conceived, the body has highly conductive intracellular and extracellular materials that can be measured, separated by insulating layers of materials such as lipids. The measures generated by the technique, which are resistance and reactance, can be used to derive estimates of total body water and, by extension, lean tissue and fat mass [94]. Because fat-free mass is composed of water, proteins, and electrolytes, conductivity is greater in fat-free mass than in fat [95]. Hence, conductivity is greater in lean than in fat tissue. Thus, this technology establishes a twocompartment model of body composition (fat tissue and lean tissue, which includes bone) based on the transmission speed of low-level, alternating current. Furthermore, specific equations have been validated for blacks [96], Hispanics [97], and Asians [88].

M e a s u r e s o f B o d y T o p o l o g y (Patterning) The terms topology and distribution refer to the relative amount of a tissue in different physical regions of the body. The term patterning is also used in discussions of regional body composition. It is ordinarily used to characterize a specific pattern of tissue distribution. Terms associated with body topology include gynoid obesity, android obesity, and central adiposity. Excess fat concentrated mostly in the abdomen is described as android obesity, whereas fat mostly below the waist and particularly in the hips is described as gynoid obesity. Central adiposity is located in the axial skeleton in amounts disproportionately greater than adiposity observed in the limbs or appendicular skeleton. The ultimate goal for many of the measures of topology is to act as a surrogate marker for the amount of visceral adiposity rather than a measure of subcutaneous adiposity.

Waist-to-Hip Ratio The ratio of hip and waist circumferences (WHR) represents the most commonly used measure of body topology. Recent reports have characterized individuals whose fat is concentrated mostly in the abdomen (android obesity) as more likely to develop many of the health risks associated with obesity, compared to those with gynoid obesity. A

257

CHAPTER 16 Body Composition, Insulin, and the Menopause W H R o f 0.8 or greater in w o m e n is indicative of android obesity. Percentiles are available for the W H R from a large French sample [98] (8646 men; 9747 women) 1 7 - 6 0 years of age and a large Danish sample [99] (1527 men; 1467 women) 3 5 - 6 5 years of age. Conceptually, android obesity is associated with excess visceral adipose deposits. However, the accuracy of the W H R in distinguishing abdominal visceral adipose tissue from subcutaneous adipose tissue is not defined. Estimates of abdominal visceral adipose tissue as well as subcutaneous tissue increase with increasingly greater body mass index, although the proportions of subcutaneous and visceral fat may begin to shift in w o m e n at around 60 years of age [ 100]. In obese individuals, changes in visceral adipose tissue after weight loss are not well related to changes in the W H R [ 101 ]. Thus, conceptually, measures of body topology cannot be considered independently of measures of composition. Circumferences contributing to the waist-to-hip ratio are relatively easy to measure with measuring tapes; however, they can be difficult to measure in persons who are markedly overweight. Standardized protocols for m e a s u r e m e n t of waist-to-hip ratio have not been developed. For example, protocols have not been developed to encourage the measurement of the waist on inspiration or expiration or to define how the hip circumference should be taken if there is sizable abdominal girth. The location of the waist moves up and down with changes in weight and muscle tone, so typically long-term reproducibility is problematic. It is more difficult to interpret trunk circumferences than limb circumferences because the trunk includes organs in addition to various tissues. Interpretation of hip circumference is uncertain because the hips include large amounts of adipose tissue and muscle and are affected by pelvic size and shape.

Summary M e a s u r e m e n t of body composition typically characterizes three compartments: (1) the amount of fat mass, (2) the amount of lean mass, and (3) the amount of bone mineral content (which, in some methodologies, is a subcompartment of the lean tissue mass compartment). It is increasingly well appreciated that body composition is related to chronic disease conditions in women, including diabetes. It is now understood that not only is the amount of fat and lean mass important, but among the more corpulent, the distribution of the fat mass is also important. It is also believed that sex h o r m o n e s may have a regulatory role in body composition and topology. Because m e a s u r e m e n t of body composition and component distribution is technically difficult in humans, particularly in epidemiological studies, a n u m b e r of measures have been used (with more or less success) as surrogates for the m e a s u r e m e n t of body composition and c o m p o n e n t distribu-

tion. These include weight and w e i g h t / h e i g h t indices as surrogate measures of body composition and the waist-to-hip circumference ratio as a measure of body topology. Increasingly, sophisticated measures of body composition are being applied in clinical and epidemiologic studies. These include the use of dual X-ray densitometry and bioelectrical impedance. Although these are likely to give more robust measures of both the fat and the lean compartments, a more technical understanding of the underlying assumptions used to generate information is required.

References 1. Bj6rntorp, R (1992). Abdominal'fat distribution and disease: An overview of epidemiological data. Ann. Med. 24, 15-18. 2. Sowers, M. F., Crutchfield, M., Jannausch, M. L., and RussellAulet, M. (1996). Longitudinal changes in body composition in women approaching the midlife. Ann. Hum. Biol. 23, 253-265. 3. Wing, R. R., Bunker, C. H., Kuller, L. H., and Matthews, K. A. (1989). Insulin, body mass index, and cardiovascular risk factors in premenopausal women. Arteriosclerosis 9, 479-484. 4. Sowers, M. E, and Sigler, C. (1999). Complex relation between increasing fat mass and decreasing high density lipoprotein cholesterol levels: Evidence from a population-based study of premenopausal women. Am. J. Epidemiol. 149, 47-54. 5. Larsson, H., Elmstahl, S., and Ahren, B. (1996). Plasma leptin levels correlate to islet function independently of body fat in postmenopausal women. Diabetes 45, 1580-1584. 6. Nicklas, B. J., Toth, M. J., Goldberg, A. R, and Poehlman, E. T. (1997). Racial differences in plasma leptin concentrations in obese postmenopausal women. J. Clin. Endocrinol. Metab. 82, 315-317. 7. Wade, G. N., and Gray, J. M. (1979). Gonadal effects on food intake and adiposity: A metabolic hypothesis. Physiol. Behav. 22, 583-593. 8. Poehlman, E. T., Toth, M. J., Ades, P. A., and Rosen, C. J. (1997). Menopause-associated changes in plasma lipids, insulin-like growth factor I and blood pressure: A longitudinal study. Eur. J. Clin. Invest. 27, 322-326. 9. Shimizu, H., Shimomura, Y., Nakanishi, Y., Futawatari, T., Ohtani, K., Sato, N., and Mori, M. (1997). Estrogen increases in vivo leptin production in rats and human subjects. J. Endocrinol. 154, 285-292. 10. Rosenbaum, M., Nicolson, M., Hirsch, J., Heymsfield, S. B., Gallagher, D., Chu, E, and Leibel, R. L. (1996). Effects of gender, body composition, and menopause on plasma concentrations of leptin. J. Clin. Endocrinol. Metab. 81, 3424-3427. 11. Sumner, A. E., Falkner, B., Kushner, H., and Considine, R. V. (1998). Relationship of leptin concentration to gender, menopause, age, diabetes, and fat mass in African Americans. Obes. Res. 6, 128-133. 12. Rebuff6-Scrive,M., Enk, L., Crona, N., Lonnroth, R, Abrahamsson, L., Smith, U., and Bj6rntorp, P. (1985). Fat cell metabolism in different regions in women: Effect of menstrual cycle, pregnancy, and lactation. J. Clin. Invest. 75, 1973-1976. 13. Rebuff6-Scrive, M., Lundholm, K., and Bj6rntorp, P. (1985). Glucocorticoid hormone binding to human adipose tissue. Eur. J. Clin. Invest. 15, 267-272. 14. Xu, X., Hoebeke, J., and Bj6rntorp, E (1990). Progestin binds to the glucocorticoid receptor and mediates antiglucocorticoid effect in rat adipose precursor cells. J. Steroid Biochem. 36, 465-471. 15. Samaras, K., Spector, T. D., Nguyen, T. V., Baan, K., Campbell, L. V., and Kelly, R J. (1997). Independent genetic factors determine the amount and distribution of fat in women after the menopause. J. Clin. Endocrinol. Metab. 82, 781-785.

258 16. Bjorntorp, E (1996). Growth hormone, insulin-like growth factor-I and lipid metabolism: Interactions with sex steroids. Horm. Res. 46, 188-191. 17. Bullough, W. S. (1953). Sex hormones, pregnancy and carbohydrate metabolism: Oestrogens, carbohydrate metabolism and mitosis. CI/ BA Found. Colloq. Endocrinol. 6, 278-194. 18. McKerns, K. W., Coulomb, B., Kaleita, E., and De Renzo, E. C. (1958). Some effects of in vivo administered estrogens on glucose metabolism and adrenal cortical secretion in vitro. Endocrinology (Baltimore) 63, 709-722. 19. Godsland, I. F. (1996). The influence of female sex steroids on glucose metabolism and insulin action. J. Intern. Med. 240, 1-60. 20. Wing, R. R., Matthews, K. A., Kuller, L. H., Meilahn, E. N., and Plantinga, E L. (1991). Weight gain at the time of menopause. Arch. Intern. Med. 151, 97-102. 21. Kirchengast, S. (1993). Anthropometric-hormonal correlation patterns in fertile and postmenopausal women from Austria. Ann. Hum. Biol. 20, 47-65. 22. Lindquist, O. (1982). Influence of the menopause on ischaemic heart disease and its risk factors and on bone mineral content. Acta Obstet. Gynecol. Scand., Suppl. 110,1-28. 23. Lindquist, O. (1982). Intraindividual changes of blood pressure, serum lipids, and body weight in relation to menstrual status: Results from a prospective population study of women in Goteborg, Sweden. Prev. Med. 11, 162-172. 24. Gambacciani, M., Ciaponi, M., Cappagli, B., Piaggesi, L., De Simone, L., Orlandi, R., and Genazzani, A. R. (1997). Body weight, body fat distribution, and hormonal replacement therapy in early postmenopausal women. J. Clin. Endocrinol. Metab. 82, 4 1 4 - 417. 25. Kritz-Silverstein, D., and Barrett-Connor, E. (1996). Long-term postmenopausal hormone use, obesity and fat distribution in older women. JAMA, J. Am. Med. Assoc. 275, 4 6 - 4 9 . 26. Anonymous (1995). Effects of estrogen or estrogen/progestin regimens on heart disease risk factors in postmenopausal women. The Postmenopausal Estrogen/Progestin Interventions (PEPI) Trial. The Writing Group for the PEPI Trial. JAMA, J. Am. Med. Assoc. 275, 370-375. 27. Ley, C. J., Lees, B., and Stevenson, J. C. (1992). Sex- and menopauseassociated changes in body fat distribution. Am. J. Clin. Nutr. 55, 950-954. 28. Pasquali, R., Casimirri, E, Labate, A. M., Tortelli, O., Pascal, G., Anconetani, B., Gatto, M. R., Flamia, R., Capelli, M., and Barbara, L. (1994). Body weight, fat distribution and the menopausal status in women. Int. J. Obes. 18, 6 1 4 - 621. 29. den Tonkelaar, I., Blankenstein, M. A., Collette, H. J., de Waard, E, and Thijssen, J. H. (1989). A prospective study on corpus luteum function and breast cancer risk. Gynecol. Endocrinol. 3, 11-19. 30. Hjortland, M. C., McNamara, P. M., and Kannel, W. B. (1976). Some atherogenic concomitants of menopause: The Framingham Study. Am. J. Epidemiol. 103, 304-311. 31. Akahoshi, M., Soda, M., Nakashima, E., Shimaoka, K., Seto, S., and Yano, K. (1996). Effects of menopause on trends of serum cholesterol, blood pressure, and body mass index. Circulation 94, 61-66. 32. Luoto, R., Kaprio, J., Reunanen, A., and Rutanen, E. M. (1995). Cardiovascular morbidity in relation to ovarian function after hysterectomy. Obstet. Gynecol. 85, 515-522. 33. Stampfer, M. J., Willett, W. C., Colditz, G. A., Rosner, B., Speizer, E E., and Hennekens, C. H. (1985). A prospective study of postmenopausal estrogen therapy and coronary heart disease. N. Engl. J. Med. 313, 1044-1049. 34. Matthews, K. A., Meilahn, E., Kuller, L. H., Kelsey, S. E, Caggiula, A. W., and Wing, R. R. (1989). Menopause and risk factors for coronary heart disease. N. Engl. J. Med. 321, 641-646. 35. Notelovitz, M., Kitchens, C., Rappaport, V., Coone, L., and Dough-

SOWERS AND TISCH

36.

37. 38.

39.

40.

41.

42.

erty, M. (1981). Menopausal status associated with increased inhibition of blood coagulation. Am. J. Obstet. Gynecol. 141, 149-152. Nachtigall, L. E., Nachtigall, R. H., Nachtigall, R. D., and Beckman, E. M. (1979). Estrogen replacement therapy II: A prospective study in the relationship to carcinoma and cardiovascular and metabolic problems. Obstet. Gynecol. 54, 74-79. Utian, W. H. (1978). Effect of postmenopausal estrogen therapy on diastolic blood pressure and body weight. Maturitas 1, 3-8. Matthews, K. A., Kuller, L. H., Wing, R. R., Meilahn, E. N., and Plantinga, E (1996). Prior to use of estrogen replacement therapy, are users healthier than nonusers? Am. J. Epidemiol. 143, 971-978. Rico, H., Revilla, M., Villa, L. F., Ruiz-Contreras, D., Hernandez, E. R., and Alvarez de Buergo, M. (1994). The four-compartment models in body composition: Data from a study with dual-energy x-ray absorptiometry and near-infrared interactance on 815 normal subjects. Metab., Clin. Exp. 43, 417-422. Svendsen, O. L., Hassager, C., and Christiansen, C. (1995). Age- and menopause-associated variations in body composition and fat distribution in healthy women as measured by dual-energy x-ray absorptiometry. Metab., Clin. Exp. 44, 369-373. Panotopoulos, G., Ruiz, J. C., Raison, J., Guy-Grand, B., and Basdevant, A. (1996). Menopause, fat and lean distribution in obese women. Maturitas 25, 11-19. Pasquali, R., Vicennati, V., Bertazzo, D., Casimirri, E, Pascal, G., Tortelli, O., and Labate, A. M. (1997). Determinants of sex hormonebinding globulin blood concentrations in premenopausal and postmenopausal women with different estrogen status. Metab., Clin. Exp.

46, 5-9. 43. Pasquali, R., Casimirri, F., Pascal, G., Tortelli, O., Morselli Labate, O., Bertazzo, D., Vicennati, V., and Gaddi, A. (1997). Influence of menopause on blood cholesterol levels in women: The role of body composition, fat distribution and hormonal milieu. The Virgilio Menopause Health Group. J. Intern. Med. 241, 195-203 44. Tremollieres, E A., Pouilles, J.-M., and Ribot, C. A. (1996). Relative influence of age and menopause on total and regional body composition changes in postmenopausal women. Am. J. Obstet. Gynecol. 175, 1594-1600. 45. Haarbo, J., Marskew, U., Gottfredsen, A., and Christiansen, C. (1991). Postmenopausal hormone replacement therapy prevents central distribution of body fat after menopause. Metab., Clin. Exp. 40, 323-326. 46. Hassager, C., and Christiansen, C. (1989). Estrogen/gestagen therapy changes soft tissue body composition in postmenopausal women. Metab., Clin. Exp. 38, 662-665. 47. den Tonkelaar, I., Seidell, J. C., van Noord, P. A., Baanders-van Halewijn, E. A., and Ouwehand, I. J. (1990). Fat distribution in relation to age, degree of obesity, smoking habits, parity and estrogen use: A cross-sectional study in 11825 Dutch women participating in the DOM-project. Int. J. Obes. 14, 753-761. 48. Razay, G., and Bolton, C. H. (1992). Coronary heart disease risk factors in relation to the menopause. Q. J. Med. 85, 307-308. 49. Hunter, G. R., Kekes-Szabo, T., Treuth, M. J., Goran, M., and Pichon, C. (1996). Intra-abdominal adipose tissue, physical activity and cardiovascular risk in pre- and postmenopausal women. Int. J. Obes. 20, 860-865. 50. Kotani, K., Tokunaga, K., Fujioka, S., Kobatake, T., Keno, Y., Yoshida, S., Shimomura, I., Tarui, S., and Matsuzawa, Y. (1994). Sexual dimorphism of age-related changes in whole-body fat distribution in the obese. Int. J. Obes. 18, 207-212. 51. Lanska, D. J., Lanska, M. J., Hartz, A. J., and Rimm, A. A. (1985). Factors influencing anatomic location of fat tissue in 52,953 women. Int. J. Obes. 9, 29-38. 52. Troisi, R. J., Wolf, A. M., Mason, J. E., Klingler, K. M., and Colditz, G. A. (1995). Relation of body fat distribution to reproductive factors in pre- and postmenopausal women. Obes. Res. 3, 143-151.

CHAPTER 16 Body Composition, Insulin, and the Menopause 53. Zamboni, M., Armellini, F., Milani, M. E, De Marchi, M., Todesco, T., Robbi, R., Bergamo-Andreis, I. A., and Bosello, O. (1992). Body fat distribution in pre- and postmenopausal women: Metabolic and anthropometric variables and their inter-relationships. Int. J. Obes. Relat. Metab. Disord. 16, 495-504. 54. Bjorkelund, C., Lissner, L., Andersson, S., Lapidus, L., and Bengtsson, C. (1996). Reproductive history in relation to relative weight and fat distribution. Int. J. Obes. 20, 213-219. 55. Rebuff6-Scrive, M., Eldh, J., Hafstr6m, L.-O., and Bj6rntorp, P. (1986). Metabolism of mammary, abdominal and femoral adipocytes in women before and after menopause. Metabol. Clin. Exp. 35, 792-797. 56. Kaye, S. A., Folsom, A. R., Jacobs, D. R., Jr., Hughes, G. H., and Flack, J. M. (1993). Psychosocial correlates of body fat distribution in Black and White young adults. Int. J. Obes. 17, 271-277. 57. Laakso, M. (1993). How good a marker is insulin level for insulin resistance? Am. J. Epidemiol. 137, 959-965. 58. Folsom, A. R., Burke, G. L., Ballew, C., Jacobs, D. R., Haskell, W. L., Donahue, R. P., Liu, K. A., and Hilner, J. E. (1989). Relation of body fatness and its distribution to cardiovascular risk factors in young blacks and whites. Am. J. Epidemiol. 130, 911-924. 59. Lindahl, B., Asplund, K., and Hallmans, G. (1993). High serum insulin, insulin resistance and their associations with cardiovascular risk factors. The northern Sweden MONICA population study. J. Intern. Med. 234, 263-270. 60. Proudler, A. J., Felton, C. V., and Stevenson, J. C. (1992). Ageing and the response of plasma insulin, glucose and C-peptide concentrations to intravenous glucose in postmenopausal women. Clin. Sci. 83, 489494. 61. Wing, R. R., Matthews, K. A., Kuller, L. H., Smith, D., Becker, D., Plantinga, P. L., and Meilahn, E. N. (1992). Environmental and familial contributions to insulin levels and change in insulin levels in middle-aged women. J. Am. Med. Assoc. 168, 1890-1895. 62. Foglia, V. G. (1965). Prediabetes, present concepts. Rev. Invest. 5, 103-111. 63. Shi, K., Mizuno, A., Sano, T., Ishida, K., and Shima, K. (1994). Sexual difference in the incidence of diabetes mellitus in OtsukaLong-Evans-Tokushima-fatty rats: Effects of castration and sex hormone replacement on its incidence. Metab., Clin. Exp. 43, 12141220. 64. Green, I. C., Howell, S. L., E1Seifi, S., and Perrin, D. (1978). Binding of 3H-progesterone by isolated rat islets of Langerhans. Diabetologia 15, 349-355. 65. Winborn, W. B., Sheridan, P. J., and McGill, H. C., Jr. (1987). Localization of progestin receptors in the islets of Langerhans. Pancreas 2, 289 -294. 66. Senoz, S., Direm, B., Gulekli, B., and Gokmen, O. (1996). Estrogen deprivation, rather than age, is responsible for the poor lipid profile and carbohydrate metabolism in women. Maturitas 25, 107-114. 67. Lindheim, S. R., Presser, S. C., Ditkoff, E. C., Vijod, M. A., Stanczyk, F. Z., and Lobo, R.A. (1993). A possible bimodal effect of estrogen on insulin sensitivity in postmenopausal women and the attenuating effect of added progestin. Fertil. Steril. 60, 664-667. 68. Berger, G. M., Maidoo, J., Gounden, N., and Gouws, E. (1995). Marked hyperinsulinaemia in postmenopausal, healthy Indian (Asian) women. Diabetes Med. 12, 788-795. 69. Walton, C., Godsland, I. F., Proudler, A. J., Wynn, V., and Stevenson, J. C. (1993). The effects of the menopause on insulin sensitivity, secretion and elimination on non-obese, healthy women. Eur. J. Clin. Invest. 23, 466-473. 70. Anderson, B., Mattsson, L.A., Hahn, L., Marin, P., Lapidus, L., Holm, G., Bengtsson, B. A., and Bj6rntorp, P. (1997). Estrogen replacement therapy decreases hyperandrogenicity and improves glucose homeostasis and plasm lipids in postmenopausal women with noninsulin-dependent diabetes mellitus. J. Clin. Endocrinol. Metab. 82, 638-643.

259 71. Crook, D., Godsland, I. E, Worthington, M., Felton, C. V., Proudler, A. J., and Stevenson, J. C. (1993). A comparative metabolic study of two low-estrogen-dose oral contraceptives containing desogestrel or gestodene progestins. Am. J. Obstet. Gynecol. 169, 1183-1189. 72. Lemieux, S., Prud'homme, D., Tremblay, A., Bouchard, C., and Despres, J. P. (1996). Anthropometric correlates to changes in visceral adipose tissue over 7 years in women. Int. J. Obes. 20, 618-624. 73. Lemieux, S., Prud'homme, D., Nadeau, A., Tremblay, A., Bouchard, C., and Despres, J. P. (1996). Seven-year changes in body fat and visceral adipose tissue in women. Association with indexes of plasma glucose-insulin homeostasis. Diabetes Care 19, 983-991. 74. Kvist, H., Chowdbury, B., Grangard, U., Tylen, U., and Sjostrom, L. (1988). Total and visceral adipose tissue volumes derived from measurements with computed tomography in adult men and women. Am. J. Clin. Nutr. 48, 1351-1361. 75. Ross, R., Shaw, K. D., Rissanen, J., Martel, Y., de Guise, J., and Avruch, L. (1994). Sex differences in lean and adipose tissue distribution by magnetic resonance imaging: Anthropometric relationships. Am. J. Clin. Nutr. 59, 1277-1285. 76. Meisler, J. G., and St. Jeor, S. (1996). Summary and recommendations from the American Health Foundation's Expert Panel on Healthy Weight. Am. J. Clin. Nutr. 63(3, Suppl.), 474S-477S. 77. Health and Welfare Canada (1988). Promoting healthy weights: A discussion paper. Health Services and Promotion Branch, Health and Welfare, Ottawa. In "Principles of Nutritional Assessment" (R. S. Gibson, ed.), pp. 163-186. Oxford University Press, Oxford. 78. Symons, J. P., Sowers, M. E, and Harlow, S. D. (1997). Relationship of body composition measures and menstrual cycle length. Ann. Hum. Biol. 24, 107-116. 79. Sowers, M. E, Kshirsagar, A., Crutchfield, M., and Updike, S. (1992). Joint influence of fat and lean body composition compartments on femoral bone mineral density in premenopausal women. Am. J. Epidemiol. 136, 257-265. 80. Douchi, T., Oki, T., Nakamura, S., Ijuin, H., Yamamoto, S., and Nagata, Y. (1997). The effect of body composition on bone density in pre- and postmenopausal women. Maturitas 27, 55-60. 81. Brodie, D., Moscrip, V., and Hutcheon, R. (1998). Body composition measurement: A review of hydrodensitometry, anthropometry, and impedance methods. Nutrition 14, 296-310. 82. Frerichs, R. R., Harsha, D. W., and Berenson, G. S. (1979). Equations for estimating percentage of body fat in children 10-14 years old. Pediatr. Res. 13, 170-174. 83. Lohman, T. G., Boileau, R. A., and Massey, B. H. (1975) Prediction of lean body weight in young boys from skinfold thickness and body weight. Hum. Biol. 47, 245-262. 84. Boileau, R. A., and Lohman, T. G. (1977). The measurement of human physique and its effect on physical performance. Orthop. Clin. North Am. 8, 563-581. 85. Siri, W. E. (1956). The gross composition of the body. Adv. Biol. Med. Phys. 4, 239-280. 86. Siri, W. E. (1961). Body composition from fluid spaces and density: Analysis of methods. In "Techniques for Measuring Body Composition" (J. Brozek and A. Henschel, eds.), pp. 223-224. Natl. Acad. Sci., Nat. Res. Counc., Washington, DC. 87. Brozek, J., Grande, F., Anderson, J. T., and Keys, A. (1963) Densitometric analysis of body composition: Revision of some quantitative assumptions. Ann. N. Y. Acad. Sci. 110, 113-140. 88. Heyward, V.H. (1996). Evaluation of body composition. Current issues. Sports Med. 22, 146-156. 89. Cullum, I. D., Ell, P. J., and Ryder, J. R. (1989). X-ray dual photon absorptiometry: A new method for the measurement of bone density. Br. J. Radiol. 62, 587-592. 90. Mazess, R. B., Peppier, W. W., Chestnut, C.H., III, Nelp, W. B., Cohn, S. H., and Zanzi, I. (1981). Total body and lean body mass by dual-

260

91.

92.

93.

94.

95.

96.

SOWERS AND TISCH photon absorptiometry. II: Comparison with total body calcium by neutron activation analysis. Calcif Tissue Int. 33, 361-363. Mazess, R. B., Peppier, W. W., and Gibbons, M. (1984). Total body composition by dual-photon (153Gd) absorptiometry. Am. J. Clin. Nutr. 40, 834-839. Laskey, M. A., Lyttle, K. D., and Barber, R. W. (1992). The influence of tissue depth and composition on the performance of the Lunar dualenergy X-ray absorptiometer whole-body scanning mode. Eur. J. Clin. Nutr. 46, 39-45. Going, S. B., Massett, M. P., Hall, M. C., Bare, L. A., Root, P. A., Williams, D. P., and Lohman, T. G. (1993). Detection of small changes in body composition by dual-energy x-ray absorptiometry. Am. J. Clin. Nutr. 57, 845-850. Boulier, A., Fricker, J., Thomasset, A., and Apfelbaum, M. (1990). Fat-free mass estimation by the two-electrode impedance method. Am. J. Clin. Nutr. 52, 581-585. Lukaski, H. C., and Bolonchuk, W. W. (1988). Estimation of body fluid volumes using tetrapolar bioelectrical impedance measurements. Avia. Space Environ. Med. 59, 1163-1169. Ainsworth, B. E., Stolarczyk, L. M., Heyward, V. H., Berry, C. B.,

97.

98.

99.

100.

101.

Irwin, M. L., and Mussulman, L. M. (1997). Predictive accuracy of bioimpedance in estimating fat-free mass of African-American women. Med. Sci. Sports Exercise 29, 781-787. Stolarczyk, L. M., Heyward, V. H., Goodman, J. A., Grant, D. J., Kessler, K. L., Kocina, E S., and Wilmerding, V. (1995). Predictive accuracy of bioimpedance equations in estimating fat-free mass of Hispanic women. Med. Sci. Sports Exercise 27, 1450-1456. Tichet, J., Vol, S., Balkau, B., LeClesiau, H., and D'Hour, A. (1993), Android fat distribution by age and sex: The waist-hip ratio. Diabetes Metab. 19, 273-276. Heitmann, B. L. (1991). Body fat distribution in the adult Danish population aged 35-65 years: An epidemiological study. Int. J. Obes. 15,535-545. Enzi, G., Gasparo, M., Biondetti, ER., Fiore, D., Semisa, M., and Zurlo, E (1986). Subcutaneous and visceral fat distribution according to sex, age, and overweight, evaluated by computed tomography. Am. J. Clin. Nutr. 44, 739-746. van der Kooy, K., Leenen, R., Seidell, J.C., Deurenberg, E, Droop, A., and Bakker, C. J. (1993). Waist-hip ratio is a poor predictor of changes in visceral fat. Am. J. Clin. Nutr. 57, 327-333.

7HAPTER 1 q

Influence of Estrogen on Collagen R. GALEA AND M. BRINCAT Department of Obstetrics and Gynecology, St. Luke's Hospital Medical School, Gwardamangia MSD 07, Malta

I. II. III. IV.

Introduction Structure of Collagen Age-Related Changes in Collagen Skin

V. Carotid Arteries VI. Urinogenital System VII. Concluding Remarks References


the proteoglycans (long chains of repeating disaccharides attached to specific core proteins).

An intriguing question regarding complex multicelluar organisms concerns what holds their myriad cells together. Certainly one of the primary factors contributing to the solution of this challenge is the group of tough, fibrous, collagenous proteins within connective tissue. Connective tissues of various types (bone, tendon, cartilage, etc.) are sites where the majority of the extracellular matrix resides. The matrix is composed of secreted fibrous proteins embedded in a gellike polysaccharide ground substance. Adhesion proteins that keep cellular components linked to each other are also present. Various types of extracellular matrix occur, such as the basal lamina or basement membrane. Basal laminae form the resting platform of epithelial cells or, alternatively, surround muscle fibers, adipocytes, and peripheral nerves. Differences among the various types of extracellular matrices result from variations on this general repertoire of complex macromolecules. Collagen is the major primary structural protein of the extracellular matrix; elastin and fibrillin are also structural proteins. Other complex macromolecules of extracellular matrix are the specialized proteins (fibronectin, laminin) and MENOPAUSE:


II. S T R U C T U R E OF C O L L A G E N Collagen is the major protein component of most vertebrate connective tissues. In mammals, collagen constitutes about 25-35% of total protein [1]. It is present in virtually every animal tissue and provides an extracellular framework for all metazoan animals. The large family of collagen proteins comprises at least 19 different members in human tissues. These proteins are made up of about 30 distinct polypeptide chains encoded by 30 independent genes. Collagen molecules are characterized by the formation of three lefthanded polypeptide helical chains tightly coiled around each other in a ropelike fashion to form a right-handed supercoil. The collagen helix is more extended than an ce helix because it has three amino acid residues per turn with a pitch of 0.94 mm, giving rise to 0.31 nmol per residue. All collagens contain greater or lesser stretches of this triple helix. (Fig. 1) [2]. Two basic ce chains have been identified in collagen (eel 261


262

FIGURE 1 Collagen fiber and fibril structure, showing putative locations of pores and hole zones. Reprinted with permission from Einhorn [2].

and a2), each consisting of just over 100 amino acids in groups of three, organized in the basic collagen triple helix configuration. The triple helix domains of collagens consist of repeats of the amino acid sequence Gly-X-Y. Hydroxyproline usually occupies position X; position Y is usually occupied by proline. The ring structure of the latter amino acids is responsible for the stabilization of the helical conformations of the polypeptide chains. Hydroxyproline is formed in the endoplasmic reticulum by posttranslational alteration of proline residues by means of the enzyme, prolyl hydoxylase. These proline residues would have already been incorporated into collagen polypeptide chains. The hydroxyl groups of these modified amino acids tend to form hydrogen bonds between adjacent polypeptide chains, providing further stabilization to the triple helix. This very stable complex forms the basic building unit of collagenous structures. Proline and hydroxyproline constitute 20-25% of total amino acids in collagen. Collagen biosynthesis has several unusual features. The first of these is an extensive use of the principle of spontaneous self-assembly seen in the formation of crystals. The three polypeptide chains of the protein fold into a triple helical conformation by a process that begins with the formation of a small nucleus of the triple helix at the carboxyterminal end of the molecule and is followed by propagation of the nucleus in a zipperlike fashion. The self-assembly of collagen monomers into fibers is an entropy-driven process reminiscent of crystallization. Collagens can be divided into two main groups, fibrillar and nonfibrillar. Other small groups also occur, such as the network-forming and anchoring filaments. In the form of fi-

GALEA AND BRINCAT

bers, collagen acts to transmit forces, to dissipate energy, and to prevent mechanical failure in normal tissues. Deformation of collagen fibers involves molecular stretching and slippage, fibrillar slippage, and, ultimately, defibrillation. Type I is the most abundant collagen type, constituting 90% of total body collagen. It predominates in skin and bone. Skin also contains an amount of the very similar type III collagen. Type II collagen is specifc to cartilage. In contrast to glycosaminoglycans, which allow rapid diffusion of water molecules, thereby creating tissue turgor with application of compressive force, collagen fibrils resist tissue stretching. Although the great majority of total body collagen is remarkably stable, a fraction of collagen in all tissues is continuously degraded and replaced throughout life. As these constituents are excreted or metabolized without reincorporation into new collagen, the process of collagen degradation can be tracked by measurement of its breakdown products in urine. For decades, measurement of urinary hydroxyproline was the classical assay for this purpose. However, it was not ideal because only about 10% of the total daily production of hydroxyproline is excreted, the remainder being processed in the liver. Recent interest has centered on nonmetabolized products, such as the pyridinium cross-links, pyridinium (PYD) and deoxypyridinium (D-PYD), which exist primarily in type I collagen of bone. These compounds are formed by the action of the enzyme lysyl oxidase to condense the amino acids lysine and hydroxylysine in adjacent collagen fibers, resulting in the formation of mature nonreducible covalent cross-links PYD and D-PYD. When bone is resorbed, proteolytic degradation of collagen releases free PYD and D-PYD into the circulation for renal clearance and urinary excretion. Measurement of free PYD, D-PYD, and their peptide-bound forms has found recent application in the assessment of patients with bone disorders, for whom the assays may have diagnostic utility and aid in the monitoring of patients on therapy. Another collagen marker that may have utility in management of bone disease is the type I procollagen carboxyterminal extension peptide. This molecule is released as an intact subunit from the intact procollagen molecule during collagen biosynthesis. It circulates in plasma, in which its concentration offers a reflection of bone collagen turnover rate.

III. A G E - R E L A T E D

CHANGES

IN COLLAGEN Age is associated with changes in the quality, type, and amount of collagen. For example, type III collagen is more abundant in the skin of young animals than in that of older animals. This may indicate gene switching comparable to the switch in the bone marrow from fetal hemoglobin to the

CHAPTER 17 Influence of Estrogen on Collagen adult form, hemoglobin A. Growth of connective tissue involves an increased rate of collagen biosynthesis and this is reflected in an increased tissue level of intracellular posttranslation enzyme activities. Both the rates of translation and the levels of these enzymes decrease with age [3]. The menopausal state is certainly characterized by a lack of endogenous estrogen, which is temporally related to reductions in tissue collagen content. Albright [4] speculated that postmenopausal osteoporosis was part of a generalized connective tissue disorder, having observed that the skin of osteoporotic women was thin. Because of reductions in skin collagen and in the density and collagen content of bone, McConkey et al. [5] showed that transparent skin on the back of the hand was most common in women over 60 years of age, and the prevalence of osteoporosis in women with transparent skin was 83% versus 12.5% in women with opaque skin. To investigate the relationship between age, bone mass, and skin thickness, our group evaluated differences in bone density measurements and skin thickness in a group of postmenopausal women. Bone density was assessed using dualenergy X-ray absorptiometry (DXA) of the lumbar spine and proximal femur; skin thickness was measured using high-frequency (22 MHz) ultrasound, a technique that correlates with skin collagen content. Results showed that skin thickness and bone density were much lower in women who had sustained an osteoporotic fracture than in controls (Fig. 2). Women with a fracture had bone mass values that were about 20% below the mean control value. When skin thickness measurements were combined with bone density, the accuracy of predicting the presence of fracture was increased (Fig. 3).

263

a

.925 .92

~ 0.915

.915 .91 .905 "

~

.9 .895

.~

.89 .885 -

p-70% stenosis of one or more coronary arteries), Rosano et al. [64] reported that estrogen administration (1 mg Estrace) 40 min before a treadmill test increased total exercise time and time to ST depression, and reduced symptoms on exertion in the women on estrogen compared to those given placebo. In the Leisure World Study [40], a prospective study, hormone users with a history of angina or myocardial infarction at baseline had approximately a 35% decreased risk of mortality compared to those not taking hormones (mean duration of hormone use at baseline was 8 years). Simi-

548 larly, in the Lipid Research Clinics cohort [36], among women with prevalent cardiovascular disease, those taking hormones (n = 74) had an 80% lower cardiovascular death rate than nonusers (n = 162) (death rates = 13.8/10,000 and 66.3/10,000, respectively; relative risk = 0.21; approximate 95% confidence interval, 0.03-1.6). Kim et al. [65] followed 293 women after coronary angioplasty, and compared the 100 subjects who used hormones both before angioplasty and during follow-up to 193 subjects who had never used hormones. Survival at 4 and 7 years was 98 and 95% in the estrogen group versus 90 and 78% in nonusers. In addition, survival or freedom from infarction at 7 years was 89% for women taking hormones and 66% for nonusers. In a similar study of 1091 women after coronary artery bypass surgery, Sullivan et al. [66] reported that hormone users had improved survival at 5 and 10 years (98.8 vs. 80.7% and 69.3 vs. 46.3%, respectively). After adjusting for confounding factors such as age and number of diseased vessels, hormone use was still inversely related with survival (relative risk = 0.34; p = 0.001). Finally, Sullivan et al. [48] conducted a study of 2268 women presenting for angiography: 446 with no detectable coronary artery disease, 644 with mild to moderate disease, and 1178 with severe disease. Among those with no disease, the 5-year survival of hormone users was the same as nonusers. However, among those with mild to moderate disease, estrogen users had better 5-year survival (98 vs. 91%). The difference was even more marked for those with severe disease (97 vs. 81%). Thus, the most substantial benefit was for women with the worst disease at baseline. Results from the Heart and Estrogen/progestin Replacement Study (HERS) [67], the first large-scale randomized trial of hormone therapy and cardiovascular disease, have been reported. A total of 2763 women with coronary disease were randomized to 0.625 mg of oral conjugated estrogen combined with 2.5 mg of continuous medroxyprogesterone acetate (n = 1380) or placebo (n = 1383). Overall, among these women with coronary disease, there was no overall protection against second cardiovascular events for women assigned to treatment compared to those given placebo (RR = 0.99; 95% CI, 0.80-1.22). However, there was a strong relation between duration of treatment and risk of major coronary events; in particular, there was a decreasing risk of heart disease with increasing duration of hormone use (p-trend = 0.009). In the first year of the trial, the risk of major coronary disease increased 52% among treated women, with most of this risk concentrated in the first 8 months of treatment; in the second year, there was no relation between treatment and disease (RR = 1.00), and in the third year the relative risk was 0.87. By the fourth to fifth years of the trial, rates of coronary events were 33% lower in women assigned to hormone therapy; the average duration of treatment was 4.1 years and 25% of women assigned to treatment had discontinued hormone use at the end

GRODSTEIN AND STAMPFER

of 3 years, thus the decreased risk observed during years 4 5 is likely a substantial underestimate of the benefits of hormone therapy. Although this decreased risk of coronary disease with long-term hormone therapy is consistent with the large body of observational evidence on primary prevention, the increased risk of second coronary events in the shortterm was unexpected; indeed, there are no other available data equivalent to those from HERS showing the short-term effects of hormone use on clinical disease. Perhaps a susceptible group of women experience adverse effects (i.e., thrombotic complications) of hormone therapy in the short-term. An increase in venous thrombosis is consistently supported by both observational studies and the HERS trial. Finally, for secondary prevention of stroke, in a trial of aspirin and stroke in patients who had experienced transient ischemic attacks [58], hormone users had a relative risk of 0.16 (p = 0.01) for stroke or death from any cause and a relative risk of 0.23 (p = 0.06) for stroke, compared to nonusers.

IV. A S S O C I A T I O N

OF HORMONES

WITH LOWER RISK OF CHD: CAUSE AND EFFECT

OR SELECTION?

The findings from the observational studies that hormone users are at generally lower risk from cardiovascular disease do not necessarily imply cause and effect. Women and their physicians decide on hormone therapy. Often the health status of women will have an important influence on this decision and on the results of studies that examine them. Thus, some have argued that hormone use is merely a marker rather than a cause of good health. Most of the observational studies reviewed here have provided some information bearing on this critical point. One way to judge the evidence for this position is to examine results of studies in which all the women were judged eligible by their physicians to receive hormone therapy. Only two small studies of primary prevention of CHD meet that criterion [30,31 ]; the summary relative risk from those two studies was 0.22 (95% CI, 0.06-0.88). These findings do not support the hypothesis that selection of healthy women for hormone use can explain the lower rate of CHD among users. With a similar intent, the Nurses' Health Study [33] tried to evaluate whether increased medical care of women using postmenopausal hormones might be responsible for the benefit observed. In an analysis limited to women who reported regular physician visits (50% of the cohort), results were similar to those found in the larger population of all subjects: the relative risk for major coronary heart disease was 0.52 (95% CI, 0.37-0.74) for current hormone use. Another approach is to examine the risk profile of hormone users and nonusers to determine if there is a consistent

549

CHAPTER 37 Cardiovascular Disease and Hormone Therapy

pattern of higher risk among the nonusers, and to assess whether the differences, if any, are sufficient to explain the large decrease in risk among hormone users. Barrett-Connor [70] observed that, in a cohort of postmenopausal women, those taking hormones reported more intensive health care behavior, including frequent screening tests such as blood cholesterol measurement and mammograms. An examination of determinants of hormone therapy in 9704 women participating in a large, multicenter study of osteoporotic fractures [71 ] found that hormone users tended to be better educated, less obese, and drank alcohol and participated in sports more often than nonusers. Similarly, in a prospective study of randomly selected premenopausal women, Matthews et al. [72] observed a better cardiovascular risk factor profile prior to hormone use among the women who subsequently took hormones at menopause than among women who did not. However, many of the large studies reviewed here are based in homogeneous groups, chosen because of their common profession or community. In the Nurses' Health Study, all women are registered nurses with access to health care and knowledge, and the distribution of established coronary risk factors was similar among current and never users of estrogens [33]. The same findings were observed in the Lipid Research Clinics Program Follow-up Study [36] (Table I). In both of these investigations, multivariate control for risk fac-

TABLE I

Risk Factor Profiles of Estrogen Users and Nonusers

Study of osteoporotic fractures a % Risk factor

Current

Past

Never

15.8

18.2

27.0

43.2 75.5

43.5 73.5

76.1

BMI --> 27.3 (kg/m2) d

Waist/hip ratio > 0.84

-- once per week BMI --> 29 (kg/m 2)

aCauley et al. [71]. bStampfer e t al. [32]. CBush e t al. [36]. dBMI, Body mass index. eMI, Myocardial infarct.

Lipid Research Clinics Program c %

Nurses' Health Study b % Risk factor

Current/past Never

, L m > O0

E

0 "0 e-

c"

N

Placebo

0.2-1

0.5-2.5

1-5

AND JUDD

1-10

Norethindrone acetate,mgEthinyl estradiol, l.tg

1

2.5

5

10

Ethinyl estradiol, #g

FIGURE 3 Meanendometrial severity score _ SE after 2 years of treatment. The asterisks indicate statistically significant (P -< 0.045) differencesamong unopposed ethinyl estradiol groups and ethinyl estradiol dose-matched

CHAPTER41 HRT and Risk of Endometrial Cancer of endometrial hyperplasia. The mean endometrial severity score shown in Fig. 3 rose with increasing doses of EE 2, to a mean of nearly 4, which corresponded to markedly proliferative endometrium with the 10-~g/day dosage. It is now clear that in direct comparisons to placebo, two of these preparations (CEE, 0.625 mg; EE 2, 10/xg), which are the lowest doses that prevent bone loss without calcium supplementation, also induce statistically significant increases of endometrial hyperplasia in women given these medications for 1 to 3 years. The third trial was conducted by the Menopause Study Group and will be discussed later in this chapter. The trial also assessed 0.625 mg CEEs. Thus, in the two reports that studied CEEs, the total percentages of women who developed hyperplasia at 1 year of therapy were similar [Menopause Study Group, 57 of 283 total biopsies (20.1%) [57], vs. PEPI, 25 of 115 biopsies (21.7%)] [58]; however, the percentages who developed the more serious diagnoses of complex or atypical hyperplasia were different [Menopause Study Group, 2 adenomatous and 0 atypical of 283 biopsies (0.7%) [57], vs. PEPI, 12 complex (adenomatous) and 3 atypical of 115 biopsies (13.4%)] [58]. There are various potential explanations for this difference. In PEPI, the women were sampled without regard to the day of the hormonal cycle, whereas the Menopause Study group sampled women between days 22 and 28 of the hormonal cycles. Second, in PEPI, seven subjects had worse hyperplasia on either a D and C or hysterectomy specimen and these diagnoses were substituted for the biopsy results. In the Menopause Study Group trial, only biopsy samples were reported. Third, PEPI was sponsored by NIH and the resuits were calculated in the absence of corporate sponsors. The Menopause Study Group was sponsored by WyethAyerst Laboratories. Fourth, the PEPI study included several women whom the arbiter pathologist had diagnosed with simple hyperplasia with atypia in the CEE-only group but who had not been identified by the central or clinic pathologists. The Menopause Study Group reported no one with atypical hyperplasia. Last, this difference between the two studies may not be statistically significant and may have occurred only by chance. Several conclusions can be drawn from the two placebocontrolled trials. First, estrogen only therapy at dosages of 0.625 mg/day of CEEs and 10/zg/day of EE 2 unequivocally and significantly increase the occurrence of endometrial hyperplasia over that seen with the use of placebos. This conclusion was reinforced by the inclusion only of women with normal (PEPI) or atrophic (CHART) endometrium at baseline. This conclusion is also supported by the many observational studies and small clinical trials that have shown an exaggerated occurrence of hyperplasia with estrogen-only therapy. Second, if women on estrogen-only therapy develop simple (cystic) hyperplasia and are continued on the medication, a minority (20%) will develop a more serious form of hyperplasia in the next 1-2 years. Third, the occurrence

599 of hyperplasia was steady across the 3 years of the PEPI trial. If the same rate of occurrence that was seen during the first 3 years continued, it would be anticipated that more than 50% of women would have complex or atypical hyperplasia after 5 years of therapy. Fourth, there was a dose-response relationship with estrogen (EE2) and the occurrence of endometrial hyperplasia in the CHART study with the highest dose (10 ~g/day) necessitating early termination of this group. Fifth, in PEPI, the numbers of unscheduled biopsies and D and Cs necessary to follow the women on CEEs alone were significantly greater than those required to follow the women on placebo. Sixth, the lowest doses of CEEs (0.625 mg/day) and EE 2 (10/xg/day) that prevent bone loss from the spine and hip (PEPI) [64,65] and the spine (CHART) [59] would be anticipated to result in sufficient occurrence of hyperplasia to necessitate discontinuation after a few years of estrogen-only therapy in a majority of women if routine endometrial biopsies were performed on a yearly basis.

II. E S T R O G E N AND PROGESTIN T H E R A P Y Several studies published in the 1980s suggested that the combined administration of estrogen with a progestin would reduce the occurrence of endometrial cancer. Of particular note was a study from the Wilford Hall Air Force Hospital, which reported that the incidence of endometrial carcinoma was 390.06 per 100,000 women-years for those who used unopposed estrogens compared with 245.5 per 100,000 women-years for nonhormone users [66]. In women who took an estrogen and a progestin, the incidence was 49 cases per 100,000 women-years. These data were interpreted to indicate that the addition of a progestin not only avoids the enhanced risk of those who take estrogen but reduces the risk to levels substantially lower than those seen in nonhormone users. Although this study has been quoted frequently, it suffered from several methodologic problems. Of particular concern was the likelihood that the groups were not equivalent for the risk of developing endometrial cancer when they were assigned therapy. The risk of cancer in the untreated women was high (245.5 per 100,000 women-years) compared with the national figures (approximately 100 per 100,000 women-years). This indicated that women with some risk factors of this cancer were not given estrogen replacement. Results of the seven studies that provided data on risk for endometrial cancer among users of estrogen plus progestin are presented in Table V. In two of these studies [29,40], three or fewer endometrial cancers occurred; one included only hypoestrogenic women [29] and another did not adjust

600 the results for the confounding effects of age [30]. Although the overall summary RR for endometrial cancer among women who took estrogen plus progestin was 0.8 (95% CI, 0.6-1.2), the direction of the effect was different in cohort versus case-control studies: 0.4 (95% CI, 0.2-0.6) and 1.8 (95% CI, 1.1-3.1), respectively. Two studies provided information on the effect of the number of days per month that the progestin was used with estrogen. In one, women who took progestin for fewer than 10 days per month had an RR for endometrial cancer of 2.0, compared to an RR of 0.9 for those who took progestin for at least 10 days per month [25]. In the other, the RR for endometrial cancer was 1.8 and did not vary by the number of days per month that the progestin was used [28]. In terms of the so-called protective effect of progestin on the endometrium, the type, potency, dosage, monthly duration, and frequency of administration all appear to be important variables; however, limited data are available about these issues. For type, the first is progesterone. This can be administered as a micronized formulation for oral use or as a vaginal suppository. Intramuscular injections are available but are rarely used for replacement therapy. The second is a C-21 progestin. These are derivatives of progesterone that have been modified chemically so that they can pass through the gastrointestinal tract and be active when given orally. These include MPA and megestrol acetate. The third is a 19nortestosterone derivative. These are derivatives of testosterone in which carbon at position 19 has been removed and an ethinyl group has been placed at position 17. These substances have both progestin and androgen effects, with the progestin actions predominating. In the United States, these include norethindrone, norethindrone acetate, and levonorgestrel. The potency of a number of progestins to protect the endometrium has been studied [67]. The end point was the ability to elicit an endometrial response comparable with that of a secretory phase endometrium. The relative potencies were shown to be levonorgestrel (8.000), norethindrone (1.000), MPA (0.090), and progesterone (0.002). Some questions have been raised about these results because of bioavailability problems with the MPA in this study. Two groups of investigators have compared the biochemical and histologic changes elicited by various progestins in regard to dose [68,69]. Both groups treated women with 0.625 or 1.25 mg of CEEs. The women were then given different doses of progestin, and their endometrium was sampled by biopsy at a specific interval after the progestin was begun. The British group performed the biopsies on the subjects on day 6 of progestin [68]. These investigators showed that the 1-mg dose of norethindrone was equally effective as the 5-mg dose in reducing DNA replication and estradiol receptor formation, but that it was not as efficient in forming a secretory endometrium. The 0.15-mg dose of levonorgestrel was equally effective as the 0.50-mg dose at

AGARWAL AND JUDD

altering the biochemical and histologic changes of progestin. For MPA, there was a dose-dependent suppression of DNA synthesis between 2.5 and 10.0 mg [70]. The levels of estradiol receptor were reduced by the 10.0-mg dose into the range seen in the secretory phase of the cycle but were not statistically different from values observed in younger women during the proliferative phase of their cycles. For histologic changes, the 10.0-mg dose elicited suboptimal responses. American investigators performed biopsies of subjects on day 11 of MPA [69]. They reported that the 5- and 10-mg doses suppressed estradiol receptor equally but that only the 10-mg dose resulted in a homogeneous secretory pattern within the endometrium. The duration of progestin use each month is another important variable. At a cellular level the action of progestin is rapid. Norethindrone (5 mg) has been shown to reduce thymidine labeling and nuclear estradiol receptors while raising estradiol dehydrogenase and isocitric dehydrogenase activity of the endometrium after 3 days of administration [45]. Maximal effects were seen at 6 days. Clinically, one observational study showed that the occurrence of hyperplasia was 12% with estrogen alone, 2% with estrogen plus 7 to 10 days of a progestin, and 0% with estrogen plus 13 days of a progestin each month [54]. This was after a mean treatment time of 9.7 months. Another observational study reported 32 and 18% hyperplasia with high-dose and low-dose estrogen therapy, 18% with lowdose estrogen therapy, and 3 - 4 % with sequential estrogen and progestin replacement, with the progestin being given for 6 to 10 days per month [55]. The mean treatment time was 15.1 months. Several small drug trials evaluating the endometrial response to single or multiple regimens of estrogen and progestin or comparing the impact of estrogen alone with that of an estrogen and a progestin have been published [71-74]. For the most part, these studies have been prospective, randomized, and double blind. Each has shown that the addition of a progestin to estrogen replacement has reduced the occurrence of hyperplasia. There are now three large clinical trials that have compared the occurrence of hyperplasia in women given estrogen or estrogen and a progestin [57-59]. The methodology of the PEPI and CHART trials have already been reviewed. The third trial, which was by the Menopause Study Group, was a prospective double-blind, parallel study conducted with healthy postmenopausal women at 99 sites in the United States and Europe [57]. Eligibility criteria were women between 45 and 65 years of age with an intact uterus, who were at least 12 months since their last menstrual period, had a serum FSH level higher than the lower limit found in postmenopausal women for a given laboratory (most were between 25 and 35 mIU/ml), and had not used estrogen or progestin medications for at least 2 weeks. At all sites approval of Institutional Review Boards was obtained and the

CHAPTER41 HRT and Risk of Endometrial Cancer participants gave informed consent before enrollment in the study. The participants were randomly assigned to one of five treatment groups for 13 cycles (1 year). All women were given CEEs at a dosage of 0.625 mg/day and MPA or a matching MPA placebo daily. Two of the groups received MPA (2.5 or 5.0 mg) daily. Two other groups received 5.0 or 10.0 mg for the last 14 days of each 28-day hormonal cycle. The last group was given CEEs plus placebo for MPA daily. The study was not controlled with a double-placebo group. All medications were supplied by Wyeth-Ayerst Laboratories (Philadelphia, Pennsylvania; CEEs, Premarin; MPA, Cycrin). Endometrial biopsies were performed at baseline and between days 22 and 28 of cycles 6 and 13. Other biopsies were taken at any time if medically indicated. If hyperplasia developed, the participant was withdrawn from the study. Approximately 75% of biopsies were obtained with the Pipelle suction curette, the Novak curette, or the Vabra aspirator. The remaining 25% were obtained with 14 other curettes. With a hypothesized incidence of endometrial hyperplasia of 7.5% in the CEE-alone group and 2% in the CEE/ MPA groups, a sample size of 215 patients per treatment group was estimated to provide 80% power to detect at least one statistically significant difference at the 0.0125 level (Bonferroni adjustment for four multiple comparisons). To allow for a 20% discontinuation rate, an enrollment of about 270 patients per group was planned. To ensure uniformity of interpretation, all endometrial biopsy specimens were evaluated by a single pathologist at The Johns Hopkins Hospital (Baltimore, Maryland). The terminology used to report endometrial hyperplasia in this study (cystic or adenomatous hyperplasia without atypia; cystic or adenomatous hyperplasia with atypia) corresponds to the alternative classifications of simple hyperplasia, complex hyperplasia; simple atypical or complex atypical hyperplasia, respectively. The criteria and terminology used have been described in the literature and were similar to the criteria and terminology used in the PEPI trial [75,76]. If a subject developed hyperplasia, she was terminated from the study. A total of 1724 postmenopausal women were enrolled. The population that completed this 1-year study with endometrial biopsy data valid for analysis was composed of 1385 subjects. Four women were 44 years old, but their data were included in this report. The prestudy characteristics of the subjects were not different statistically (p > 0.05) among treatment groups for any demographic characteristic. Among women taking the continuous combined regimens, approximately 20% of the biopsy specimens had either no tissue or no endometrial tissue identified. Those who took the sequential regimens had either no tissue or no endometrial tissue in only 10% of biopsy specimens, and those in the conjugated estrogens-alone

601 group had either no tissue or no endometrial tissue in approximately 15% of biopsy specimens. Endometrial hyperplasia developed in 23 (2%) of 1469 patients included in the evaluation of the 6-month data (Tables IV and V). The incidence with each of the CEE/ MPA regimens was significantly lower than with CEE alone. There were no statistically significant differences between the lower dose and higher dose MPA regimens or between the continuous combined and the sequential regimens. The evaluation of the 12-month data (including 6-month data) showed that endometrial hyperplasia had developed in 62 (4%) of the 1385 patients included in this analysis. It should be mentioned that all subjects were counted just once. The incidence of endometrial hyperplasia was significantly lower with each of the CEE/MPA regimens than with CEE alone. There were a total of five cases of endometrial hyperplasia in the two lower dose MPA groups and no cases of hyperplasia in the two higher dose groups. This difference approached statistical significance (P = 0.06). During the course of this study, two women developed endometrial cancer during the thirteenth cycle. One of them received CEE and sequential MPA (10 mg), and the other was given CEE alone. Both women underwent hysterectomy. The second large clinical trial that compared estrogen alone with estrogen/progestin (E/P) was the PEPI trial [58]. The methodology and placebo and estrogen-only results were reviewed above. In PEPI, among women taking one of the three E/P regimens, 16 (13.6%) taking cyclic MPA, 9 (7.5%) taking continuous MPA, and 14 (11.7%) taking cyclic MP had unscheduled biopsy rates that were similar to those of women receiving placebo (P = 0.338), but significantly lower than women receiving estrogen only. Women receiving the E/P regimens underwent zero to two D and Cs per regimen, similar to the rate of women receiving placebo (P = 0.43), but lower than those receiving CEEs alone (Table II). Five women receiving the E/P regimens had hysterectomies, one for atypical hyperplasia, one for persistent vaginal bleeding, two for uterine leiomyomas, and one for an ovarian cystadenoma. Table VI summarizes the endometrial histology results of the women taking one of the E/P regimens. Ten cases of simple (cystic) hyperplasia, two of complex (adenomatous) hyperplasia, and one of atypical hyperplasia were distributed among the three E/P groups. There was no difference in the occurrence of abnormal biopsy specimens between the women who received placebo and those who received one of the three E/P regimens (P = 0.16). However, the incidence of hyperplasia was significantly less in the women taking E/P than in women receiving CEEs alone. The last large trial was the CHART study [59]. Again, the methodology and the placebo and EE2-only results were reviewed earlier. A significantly greater percentage of subjects developed endometrial hyperplasia while receiving 10/zg of EE e compared with the 1 mg NA/10/zg ofEE e combination

602

AGARWAL AND JUDD

TABLE V I

Temporal Induction of Endometrial Hyperplasia with Estrogen/Progestin in Large Clinical Trials a Follow-up (months)

Study

Regimen

Baseline number

Menopause Study Group [57]

CEE, 0.625 mg + MPA, 10 mg • 14 days MPA, 5 mg • 14 days MPA, 5 mg four times daily MPA, 2.5 mg four times daily

345 345 345 345

CEE, 0.625 mg + MPA, 10 mg • 12 days MPA, 2.5 mg four times daily MPA, 200 mg • 12 days

118 120 120

NA mg/EE 2/xg 0.2/1 0.5/2.5 1/5 1/10

139 136 146 145

PEPI [581

CHART [59]

6

12

0/292 1/293 0/291 1/295

0/272 3/277 0/272 2/279

18

1/117 0/115 2/115 0/84 0/75 0/92 0/92

0/80 0/69 9/65 0/71

0/67 0.50 0/70 0.63

24

36

1/112 0/111 2/110

4/108 1/109 2/110

1/69 0/57 0/65 0/65

aThe incidence of endometrial hyperplasia is expressed as the number of cases per number of evaluable subjects at each evaluation time point.

(Tables IV and V). As the dose of unopposed EE 2 increased, there were increased percentages of subjects with hyperplasia. This contrasted to the N A - E E 2 combination groups, which had no significant dose trend at any time point with or without the placebo group included. The protective effect of the addition of NA to EE 2 on the endometrium was demonstrated by the life table analysis, which yielded a probability of developing hyperplasia of 0.24 in the 10-/xg EE 2 treatment group by 19 months in contrast to 0 probability for the 1-mg NA/10-p~g EE 2 group at the end of the study. The probability of hyperplasia over time was statistically significantly greater for the 10-/xg EE 2 treatment group compared with the 1-mg NA/10-/xg EE 2 treatment group (P < 0.001).

III. ENDOMETRIAL

SURVEILLANCE

The traditional and definitive procedure for diagnosis of endometrial hyperplasia or cancer has been the fractional dilatation and curettage. Under general anesthesia, this has been shown to produce adequate material for histological evaluation in up to 94% of cases [77]. Major problems with this diagnostic procedure include cost, possible medical complications, and inconvenience, and it is for these reasons that alternative screening procedures have been sought. Less expensive and invasive office endometrial sampling procedures such as those employing the Pipelle device, Novak curette, and Vabra aspirator are replacing dilatation and curettage. The real test of these devices is their sensitivity at detecting endometrial cancer. In a study of 40 patients al-

ready known to have endometrial carcinoma as diagnosed by other procedures, 39 (97.5%) were confirmed as having endometrial carcinoma by Pipelle biopsy specimens [78]. The Vabra aspirator, Novak curette, and dilation and curettage have also been shown to have similar diagnostic capabilities [79]. In general, the Pipelle device is best tolerated by patients. Although the sensitivitities for these office procedures are similar to that reported for dilatation and curettage, isolated reports of women with endometrial cancer missed by these instruments continue to trouble gynecologists. There are two situations when ultrasound-based evaluation of abnormal postmenopausal bleeding is particularly valuable: (1) to screen for discreet abnormalities of the endometrium such as polyps, which may not be sampled adequately by office biopsy and may require hysteroscopic excision, and (2) in the presence of cervical stenosis, rendering office endometrial biopsy impossible, in which case a reassuring sonographic evaluation may render a surgical diagnostic procedure unnecessary. A number of studies have been published using transvaginal sonography to differentiate women at risk for the presence of endometrial hyperplasia or cancer from those with a normal or atrophic endometrium [80-83]. Most have focused on the measurement of endometrial thickness and have demonstrated that an endometrial thickness of greater than 5 mm is a valid cutoff. Women with an endometrial thickness of greater than 5 mm should undergo endometrial sampling to exclude cancer or precancerous lesions [80,81]. Reports have indicated that the use of color doppler with transvaginal ultrasonography improves specificity without compromising sensitivity. One study attempted to character-

CHAPTER41 HRT and Risk of Endometrial Cancer

603

FIGURE 4 Ultrasoundimages during saline infusion sonography. (A) During early saline infusion and (B) after maximal uterine distention, demonstrating the presence of a subserosal fibroid. Both pictures are from the same woman and illustrate how saline infusion sonography can help improve measurementof endometrial thickness and detection of focal intrauterine lesions.

ize uterine tumors by their color flow patterns [84]. They found the mean intratumoral resistance index value for endometrial carcinomas to be 0.34 and for leiomyomata 0.58, and suggested that a value of less than 0.4 should be regarded as one of malignancy. In the absence of other confirmatory reports, these data remain of interest, but are considered research tools at the present time. Not surprisingly, the use of saline infusion sonography (SIS) in the evaluation of this common clinical situation has been the focus of a number of reports. Figure 4 shows how SIS can assist with the ultrasonic measurement of endometrial thickness and with the detection of focal intrauterine lesions. A direct comparison of SIS to hysteroscopy in 47 asymptomatic postmenopausal women with the sonographic finding of a thickened endometrium ( > 15 mm) was conducted by Wolman et al. [85]. A prospective, doubleblind design was used in which all subjects underwent SIS and then hysteroscopy a week later by an examiner unaware of the SIS findings. An additional three patients failed SIS due to cervical stenosis. In summary, the authors detected a sensitivity of 86% and a specificity of 100% for SIS in this population. Similar results have been obtained by other investigators, and further refinements such asthe concomitant use of color doppler with pulsatility and resistance indices are currently under evaluation.

IV. P H Y S I O L O G Y

OF

ENDOMETRIAL SHEDDING An exciting and rapidly developing area of research involves the role of matrix metalloproteinases (MMPs) in endometrial shedding. This family of enzymes, which is found in endometrium and other tissues, is involved in the degradation of most components of the extracellular matrix as seen with menstruation. The balance between MMP produc-

tion and that of specific tissue inhibitors of MMP action regulates their impact. It has been shown that endometrial MMPs are expressed in menstrual cycle-specific patterns, consistent with regulation by steroid hormones. Although long-term hormone replacement is advocated for the protection of bone and the heart, a majority of women commencing with HRT stop within the first 2 years. In those with a uterus, vaginal bleeding is commonly cited as a reason for discontinuing this potentially beneficial therapy. It is possible, therefore, that therapies may be developed that directly modulate MMP activity so as to decrease vaginal bleeding seen with E/P, thereby increasing compliance.

V. SUMMARY Based on the available literature, the lowest doses of estrogen that prevent bone loss from the spine and hip have been partially established. These include CEE (0.625 mg) [63,64], piperazine estrone sulfate (1.25 mg) [86], and transdermal estradiol (0.05 mg) [87]. The lowest doses of two other estrogens that prevent bone loss from the spine have been partially established. These include EE 2 (10/zg) [59] and micronized estradiol (0.5 mg) [88]. From the review just completed, it is clear that estrogenonly therapy at doses that will prevent postmenopausal bone loss will stimulate endometrial hyperplasia. Because longterm therapy with estrogens is necessary to prevent osteoporotic fractures, it is expected that many older women will use estrogen for several to many years. It is also clear that the addition of a progestin, whether it be MPA, MP, or NA, given at the doses studied, with either continuous or sequential regimens, will reduce the occurrence of hyperplasia seen with estrogen-only therapy to that occurring with the administration of placebo. It is possible, even likely, that the same or different progestins could be

604

AGARWAL AND JUDD

administered using the same or different doses, the same or different schedules, or the same or different regimens, which could also reduce the occurrence of estrogen-stimulated endometrial hyperplasia. One possible example may be the use of a levonorgestrel-releasing intrauterine device. If this is the case, then future studies of the nature of those reviewed here will be needed to establish this claim.

18.

19.

20.

References 21. 1. Landis, S. H., Murray, T., Bolden, S., and Wingo, P. A. (1998). Cancer statistics. Ca-- Cancer J. Clin. 48, 6-30. 2. Smith, D., Prentice, R., Thompson, D., and Hermann, W. (1975). Association of exogenous estrogen and endometrial carcinoma. N. Engl. J. Med. 293, 1164-1167. 3. Ziel, H., and Finkle, W. (1975). Increased risk of endometrial carcinoma among users of conjugated estrogens. N. Engl. J. Med. 293, 1167-1170. 4. Mack, T. M., Pike, M. C., Henderson, B. E., Pfeffer, R. I., Gerkins, V. R., Arthur, M., and Brown, S. E. (1976) Estrogens and endometrial cancer in a retirement community. N. Engl. J. Med. 294, 1262-1267. 5. Gray, L., Christopherson, W., and Hoover, R. (1977). Estrogens and endometrial carcinoma. J. Am. Coll. Obstet. Gynecol. 49, 385-389. 6. McDonald, T., Annegers, J., O'Fallon, W., Dockerty, M., Malkasian, G., and Kurland, L. (1977). Exogenous estrogen and endometrial carcinoma: Case-control and incidence study. Am. J. Obstet. Gynecol. 127, 572-579. 7. Horwitz, R., and Feinstein, A. (1978). Alternative analytic methods for case-control studies of estrogens and endometrial cancer. N. Engl. J. Med. 299, 1089-1094. 8. Hoogerland, D., Buckler, D., Crowley, J., and Carr, W. (1978). Estrogen use-risk of endometrial carcinoma. Gynecol. Oncol. 6, 451-458. 9. Wigle, D. T., Grace, M., and Smith, E. S. (1978). Estrogen use and cancer of the uterine corpus in Alberta. Can. Med. Assoc. J. 118, 12761278. 10. Antunes, C. M., Stolley, P. D., Rosensheim, N. B., Davies, J. L., Tonascia, J. A., Brown, C., Burnett, L., Rutledge, A., Pokempner, M., and Garcia, R. (1979). Endometrial cancer and estrogen use. N. Engl. J. Med. 300, 9-13. 11. Jick, H., Watkins, R. N., Hunter, J. R., Dinan, B. J., Madsen, S., Rothman, K. J., and Walker, A. M. (1979). Replacement estrogens and endometrial cancer. N. Engl. J. Med. 300, 218-222. 12. Weiss, N., Szekely, D., English, D., and Schweid, A. (1979). Endometrial cancer in relation to patterns of menopausal estrogen use. JAMA, J. Am. Med. Assoc. 242, 261-264. 13. Hulka, B. S., Fowler, W. C., Jr, Kaufman, D. G., Grimson, R. C., Greenberg, B. G., Hogue, C. J., Berger, G. S., and Pulliam, C. C. (1980). Estrogen and endometrial cancer: Cases and two control groups from North Carolina. Am. J. Obstet. Gynecol. 137, 92-101. 14. Jelovsek, E R., Hammond, C. B., Woodard, B. H., Draffin, R., Lee, K. L., Creasman, W. T., and Parker, R. T. (1980). Risk of exogenous estrogen therapy and endometrial cancer. Am. J. Obstet. Gynecol. 137, 85-91. 15. Salmi, T. (1980). Endometrial carcinoma risk factors with special reference to the use of oestrogens. Acta Endocrinol. (Copenhagen) 233, 37-43. 16. Spengler, R. E, Clarke, E. A., Woolever, C. A., Newman, A. M., and Osborn, R. W. (1981). Exogenous estrogens and endometrial cancer: A case-control study and assessment of potential biases. Am. J. Epidemiol. 114, 497-506. 17. Stavraky, K. M., Collins, J. A., Donner, A., and Wells, G. A. (1981). A comparison of estrogen use by women with endometrial cancer, gyne-

22. 23.

24.

25.

26. 27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

cologic disorders, and other illnesses. Am. J. Obstet. Gynecol. 141, 547-555. Kelsey, J. L., LiVolsi, V. A., Holford, T. R., Fischer, D. B., Mostow, E. D., Schwartz, P. E., O'Connor, T., and White, C. (1982). A casecontrol study of cancer of the endometrium. Am. J. Epidemiol. 116, 333-343. Henderson, B., Casagrande, J., Pike, M., Mack, T., Rosario, I., and Duke, A. (1983). The epidemiology of endometrial cancer in young women. Br. J. Cancer 47, 749-756. La Vecchia, C., Franceschi, S., Decarli, A., Gallus, G., and Tognoni, G. (1984). Risk factors for endometrial cancer at different ages JNCI, J. Natl. Cancer Inst. 73, 667-671. Shapiro, S., Kelly, J. P., Rosenberg, L., Kaufman, D. W., Helmrich, S. P., Rosenshein, N. B., Lewis, J. L., Jr., Knapp, R. C., Stolley, P. D., and Schottenfeld, D. (1985). Risk of localized and widespread endometrial cancer in relation to recent and discontinued use of conjugated estrogens. N. Engl. J. Med. 313, 969-972. Buring, J. E., Bain, C. J., and Ehrmann, R. L. (1986). Conjugated estrogen use and risk ofendometrial cancer.Am. J. Epidemiol. 124, 434-441. Ewertz, M., Schou, G., and Boice, J. D., Jr. (1988). The joint effect of risk factors on endometrial cancer. Eur. J. Cancer Clin. Oncol. 24, 189-194. Rubin, G., Peterson, H., Lee, N., Maes, E., Wingo, P., and Becker, S. (1990) Estrogen replacement therapy and the risk of endometrial cancer: Remaining controversies. Am. J. Obstet. Gynecol. 162,148-154. Voigt, L. E, Weiss, N. S., Chu, J., Daling, J. R., McKnight, B., and van Belle, G. (1991). Progestagen supplementation of exogenous oestrogens and risk of endometrial cancer. Lancet 338, 274-277. Jick, S., Walker, A., and Jick, H. (1993). Estrogens, progesterone, and endometrial cancer. Epidemiology 4, 20-24. Levi, F., La Vecchia, C., Gulie, C., Franceschi, S., and Negri, E. (1993). Oestrogen replacement treatment and the risk of endometrial cancer. An assessment of the role of covariates. Eur. J. Cancer 29, 1445-1449. Brinton, L., and Hoover, R. (1993). Estrogen replacement therapy and endometrial cancer risk: Unresolved issues. Obstet. Gynecol. 8 1 , 2 6 5 271. Hammond, C., Jelovsek, E, Lee, K., Creasman, W., and Parker, R. (1979). Effects of long-term estrogen replacement therapy. II. Neoplasia. Am. J. Obstet. Gynecol. 133, 537-547. Gambrell, R. Jr, Massey, F., Castaneda, T., Ugenas, A., Ricci, C., and Wright, J. (1980). Use of the progestogen challenge test to reduce the risk of endometrial cancer. Obstet. Gynecol. 55, 732-738. Petitti, D. B., Perlman, J. A., and Sidney, S. (1987). Noncontraceptive estrogens and mortality: Long-term follow-up of women in the Walnut Creek Study. Obstet. Gynecol. 70, 289-293. Ettinger, B., Golditch, I. M., and Friedman, G. (1988). Gynecologic consequences of long-term unopposed estrogen replacement therapy. Maturitas 10, 271-282. Paganini-Hill, A., Ross, R., and Henderson, B. (1989). Endometrial cancer and patterns of use of oestrogen replacement therapy: A cohort study. Br. J. Cancer 59, 445-447. Persson, I., Adami, H. O., Bergkvist, L., Lindgren, A., Pettersson, B., Hoover, R., and Schairer, C. (1989). Risk of endometrial cancer after treatment with oestrogens alone or in conjunction with progestogens: Results of a prospective study. Br. Med. J. 298, 147-151. Hoover, R., Framueni, J. E, and Everson, R. (1976). Cancer of the uterine corpus after hormonal treatment for breast cancer. Lancet 1, 885887. Vakil, D., Morgan, R., and Haliday, M. (1983). Exogenous estrogens and development of breast and endometrial cancer. Cancer Detect. Prev. 6, Hunt, K., Vessey, M., McPherson, K., and Coleman, M. (1987). Longterm surveillance of mortality and cancer incidence in women receiving hormone replacement therapy. Br. J. Obstet. Gynaecol. 94, 620-635. Grady, D., Gebretsadik, T., Kerlikowske, K., Ernster, V., and Petitti, D.

605

CHAPTER 41 HRT and Risk of Endometrial Cancer

39. 40.

41.

42.

43. 44.

45.

46.

47.

48.

49.

50.

51.

52. 53.

54.

55.

56.

57.

(1995). Hormone replacement therapy and endometrial cancer risk: A meta-analysis. Obstet. Gynecol. 85, 304-313. Greenland S. (1987). Quantitative methods in the review of epidemiologic literature. Epidemiol. Rev. 9, 1-30. Nachtigall, L. E., Nachtigall, R. H., Nachtigall, R. D., and Beckman, E. (1979). Estrogen replacement therapy II: A prospective study in the relationship to carcinoma and cardiovascular and metabolic problems. Obstet. Gynecol. 54, 74-79. Horwitz, R. I., and Feinstein, A. R. (1986). Estrogens and endometrial cancer. Response to arguments and current status of an epidemiologic controversy. Am. J. Med. 81, 503-507. Gordon, J., Reagan, J. W., Finkle, W. D., and Ziel, H. K. (1977). Estrogen and endometrial carcinoma. An independent pathology review supporting original risk estimate. N. Engl. J. Med. 297, 570-571. Noyes, R. W., Hertig, A. T., and Rock, J. (1950). Dating the endometrial biopsy. Ferti. Steril. 1, 3-25. King, R. J. B., Townsend, P. T., Whitehead, M. I., Young, O., and Taylor, R. W. (1981). Biochemical analyses of separated epithelium and stroma from endometria of premenopausal and postmenopausal women receiving estrogen and progestins. J. Steroid Biochem. 14, 978-987. Whitehead, M. I., Townsend, P. T., Pryse-Davies, J., Path, E R. C., Ryder, T. A., and King, R. J. B. (1981). Effects of estrogens and progestins on the biochemistry and morphology of the postmenopausal endometrium. N. Engl. J. Med. 305, 1599-1605. Murphy, L. J., Gong, Y., and Murphy, L. C., (1991). Growth factors in normal and malignant uterine tissue. Ann. N.Y. Acad. Sci. 622, 383391. Gardner, R. M., Verner, G., Kirkland, J. L., and Stancel, G. M. (1989). Regulation of uterine epidermal growth factor (EGF) receptors by estrogen in the mature rat and during the estrous cycle. Steroid Biochem. 32, 339-343. Reynolds, R. K., Talavera, F., Roberts, J. A., Hopkins, M. P., and Menon, K. M. J. (1990). Regulation of epidermal growth factor and insulin-like growth factor 1 receptors by estradiol and progesterone in normal and neoplastic endometrial cell cultures. Gynecol. Oncol. 38, 396-406. Nelson, K. G., Takahashi, T., Bossert, N. L., Walmer, D. K., and McLachlan, J. A. (1991). Epidermal growth factor replaces estrogen in the stimulation of female genital-tract growth and differentiation. Proc. Natl. Acad. Sci. U.S.A. 88, 21-25. Gusberg, S. B., and Kaplan, A. (1963). Precursors of corpus cancer IV. Adenomatous hyperplasia as stage 0 carcinoma of endometrium. Am. J. Obstet. Gynecol. 87, 662-678. Gusberg, S. B., Chen, S. Y., and Cohen, C. J. (1974). Endometrial cancer: Factors influencing the choice of treatment. Gynecol. Oncol. 2, 308-313. Sherman, A. I., and Brown, S. (1979). The precursors of endometrial carcinoma. Am. J. Obstet. Gynecol. 135, 947-956. Kurman, R. J., Kaminski, P. T., and Norris, H. J. (1985). The behavior of endometrial hyperplasia: A long-term study of "untreated" hyperplasia in 170 patients. Cancer (Philadelphia) 56, 403-412. Sturdee, D. W., Wade-Evans, T., Paterson, M. E. L., Thom, M., and Studd, J. W. W. (1978). Relations between bleeding pattern, endometrial histology, and oestrogen treatment in menopausal women. Br. Med. J. 1, 1575-1577. Whitehead, M. I., McQueen, J., King, R. J. B., and Campbell, S. (1979). Endometrial histology and biochemistry in climacteric women during oestrogen and oestrogen/progestogen therapy. J. R. Soc. Med. 72, 322327. Paterson, M. E. L., Wade-Evans, T., Sturdee, D. W., Thom, M., and Studd, J. W. W. (1980). Endometrial disease after treatment with oestrogens and progestogens in the climacteric. Br. Med. J. 1, 822-824. Woodruff, J. D., and Pickar, J. H. for the Menopause Study Group (1994). Incidence of endometrial hyperplasia in postmenopausal women taking conjugated estrogens (Premarin) with medoxyprogester-

58.

59.

60.

61. 62.

63.

64.

65. 66. 67. 68.

69.

70.

71.

72.

73.

74.

75.

76. 77. 78.

one acetate or conjugated estrogens alone. Am. J. Obstet. Gynecol. 170, 1213-1223. Writing Group for the PEPI Trial (1996). Effects of hormone replacement therapy on endometrial histology in postmenopausal women. JAMA, J. Am. Med. Assoc. 275, 370-375. Speroff, L., Rowan, J., Symons, J., Genant, H., and Wilborn, W., for the CHART Study Group (1996). The comparative effect on bone density, endometrium, and lipids of continuous hormones as replacement therapy. (CHART Study). JAMA, J. Am. Med. Assoc. 276 (17), 13971403. Writing Group for the PEPI Trial (1995). The postmenopausal estrogen/progestin interventions (PEPI) trial: Rationale, design and conduct (I). J. Controlled Clin. Trials 16 (Suppl.), 3S-19S. Hendrickson, M., and Kempson, R. (1980). "Major Problems in Pathology," Vol. 12, pp. 285-318. Saunders, Philadelphia. Writing Group for the PEPI Trial (1995). The postmenopausal estrogen/progestin interventions (PEPI) trial: Baseline characteristics of participants (IV). J. Controlled Clin. Trials 16 (Suppl.), 54S-72S. Writing Group for the PEPI Trial (1995). Effects of estrogen or estrogen/progestin regimens on heart disease risk factors in postmenopausal women. JAMA, J. Am. Med. Assoc. 273, 199-208. Lindsay, R., Hart, D. M., and Clark, D. M. (1984). The minimum effective dose of estrogen for prevention of postmenopausal bone loss. Obstet. Gynecol. 63, 759-763. Writing Group for the PEPI Trial (1996). Effects of hormone therapy on bone mineral density. JAMA, J. Am. Med. Assoc. 276, 1389-1396. Gambrell, R. D. (1989). Prevention of endometrial cancer with progestogens. Maturitas 8, 1-21. King, R. J. B., and Whitehead, M. I. (1986). Assessment of potency of orally administered progestins in women. Fertil. Steril. 46, 1062-1066. Fraser, D., Whitehead, M. I., Endacott, J., Morton, J., Ryder, T. A., and Pryse-Davies, J. (1989). Are fixed-dose oestrogen/progestogen combinations ideal for all HRT users? Br. J. Obstet. Gynaecol. 96, 776 -782. Gibbons, W. E., Moyer, D. L., Lobo, R. A., Roy, S., and Mishell, D. R. (1986). Biochemical and histologic effects of sequential estrogen/progestin therapy on the endometrium of postmenopausal women. Am. J. Obstet. Gynecol. 154, 456-461. Lane, G., Siddle, N. C., Ryder, T. A., Pryse-Davies, J., King, R. J. B., and Whitehead, M. I. (1986). Is Provera the ideal progestogen for addition to postmenopasual estrogen therapy? Fertil. Steril. 45, 345-352. Gelfand, M. M., and Ferenczy, A. (1989) A prospective 1-year study of estrogen and progestin in postmenopausal women: Effects of endometrium. Obstet. Gynecol. 4, 398-402. Clisham, P. R., Cedars, M. I., Greendale, G., Fu, Y. S., Gambone, J., and Judd, H. L. (1992). Long-term transdermal estradiol therapy: Effects of endometrial histology and bleeding patterns. Obstet. Gynecol. 79, 196-201. Williams, D. B., Voigt, B. J., Fu, Y. S., Schoenfeld, M. J., and Judd, H. L. (1994). Assessment of less than monthly progestin therapy in postmenopausal women given estrogen replacement. Obstet. Gynecol. 84 (5), 787-793. Ettinger, B., Selby, J., Citron, J., Vangessel, A., Ettinger, M., and Hendrickson, M. (1994). Cyclic hormone replacement therapy using quarterly progestin. Obstet. Gynecol. 83, 693-700. Novak, E. R., and Woodruff, J. D. (1979). "Novak's Gynecologic and Obstetric Pathology with Clinical and Endocrine Relations," 8th ed., pp. 171-237. Saunders, Philadelphia. Kurman, R. J., ed. (1987). "Blaustein's Pathology of the Female Genital Tract," 3rd ed., pp. 257-372. Springer-Verlag, New York. Grimes, D. A. (1982). Diagnostic dilation and curettage: A reappraisal. Am. J. Obstet. Gynecol. 142, 1-6. Stovall, T. G., Photopulos, G. J., Poston, W. M., Ling, F. W., and Sandles, L. G. (1991). Pipelle endometrial sampling in patients with known endometrial carcinoma. Obstet. Gynecol. 77, 954-956.

606 79. Stovall, T. G., Solomon, S. K., and Ling, E W. (1989). Endometrial sampling prior to hysterectomy. Obstet. Gynecol. 73, 405-409. 80. Fleischer, A. C., Kalemeris, G. C., Machin, J. E., Entman, S. S., James, A. E., Jr. (1986). Sonographic depiction of normal and abnormal endometrium with histopathologic correlation. J. Ultrasound Med. 5, 445 -452. 81. Granberg, S., Wikland, M., Karlsson, B., Norstr6m, A., and Friberg, L. G. (1991). Endometrial thickness as measured by endovaginal ultrasonography for identifying endometrial abnormality. Am. J. Obstet. Gynecol. 164, 47-52. 82. Lin, M. C., Gosink, B. B., Wolf, S. I., Feldesman, M. R., Stuenkel, C. A., Braly, E S., and Pretorius, D. H. (1991). Endometrial thickness after menopause: Effect of hormone replacement. Radiology 180, 427432. 83. Nasri, M. N., and Coast, G. J. (1989). Correlation of ultrasound findings and endometrial histopathology in postmenopausal women. Br. J. Obstet. Gynecol. 96, 1333-1338.

AGARWAL AND JUDD 84. Kurjak, A., and Zalud, I. (1991). The characterization of uterine tumors by transvaginal color doppler. Ultrasound Obstet. Gynecol. 1, 50-52. 85. Wolman, I., Jaffa, A. J., Hartoov, J., Bar-Am, A., and David, M. E (1996). Sensitivity and specificity of sonohysterography for the evaluation of the uterine cavity in perimenopausal patients. J. Ultrasound Med. 15, 285-288. 86. Harris, S. T., Genant, H. K., Baylink, D. J., Gallagher, J. C., Karp, S. K., McConnell, M. A., Green, E. M. and Stoll, R. W. (1991). The effect of estrone (Ogen) on spinal bone density of postmenopausal women. Arch. Intern. Med. 151, 1980-1984. 87. Field, C. S., Ory, S. J., Wahner, H. W., Herrmannn, R. R., Judd, H. L., and Riggs, B. L. (1993). Preventive effects of transdermal 17/3estradiol on osteoporotic changes after surgical menopause: A 2-year placebo-controlled trial. Am. J. Obstet. Gynecol. 168, 114-121. 88. Ettinger, B., Genant, H., Steiger, E, and Madvig, E (1992). Lowdosage micronized 17/3-estradiol prevents bone loss in postmenopausal women. Am. J. Obstet. Gynecol. 166, 479-488.

2 H A P T E R 4~

Risk of Pulmonary Embolism/Venous Thrombosis CAROLYN WESTHOFF

Columbia University, College of Physicians and Surgeons, New York, New York 10032

IV. Incidence and Prognosis V. Clinical Recommendations References

I. Clinical Entities II. Diagnosis and T r e a t m e n t III. P a t h o p h y s i o l o g y

I. C L I N I C A L

in the lower extremity only, and the remaining one-third presents as a pulmonary embolus with or without a symptomatic clot in the leg. Clots in the leg may be silent, recognized only after a pulmonary embolism has occurred, if then, or they may come to clinical attention due to swelling and pain in the extremity. In clinically recognized cases of deep vein thrombosis of the leg the main symptoms are swelling (88%), pain (56%), and tenderness (55%). Symptoms such as redness, palpable venous cord, or a positive Homan's sign are present in only a minority of confirmed cases [2]. Pulmonary emboli present with dyspnea (80%), pleuritic chest pain (60%), and cough (41%) [3]. On ausculation, 60% of angiogram-confirmed cases have audible crackles. There is no important difference between men and women in the frequency of these presenting complaints. Because of difficulties in diagnosing these conditions, it is likely that many milder cases may be unrecognized. The sequelae of venous thrombosis are immediate death due to pulmonary embolism, recurrent symptomatic or asymptomatic thrombosis with repeated risk of embolism, and the development of postphlebitic syndrome. The

ENTITIES

Venous thrombosis mainly presents in the deep veins of the leg or in the lung, but can also occur elsewhere, including the brain, retina, liver, mesentery, and upper extremities. Clots originating in the upper extremity are relatively rare, and are usually sequelae of medical procedures such as catheter placement [ 1]. Thrombophlebitis is a clinical syndrome of pain and swelling in the leg that originates as a valve pocket thrombus, most often in the soleal vein or posterior tibial or popliteal veins, during some period of stasis. Such a thrombus may propagate upward to the femoral and iliac veins where pieces then break off into the venous circulation to become trapped in the lung as a pulmonary embolus. It is thought that distal leg clots do not embolize until after upward propagation to the thigh; however, clots may originate in the thigh or pelvis rather than in the leg. Further propagation in the lung or showers of emboli can be fatal. In most cases of pulmonary embolism it is possible to identify an apparent source clot in the lower extremity. About two-thirds of clinically recognized cases of venous thromboembolism (VTE) present with a venous thrombosis MENOPAUSE: BIOLOGY AND PATHOBIOLOGY

607


608

CAROLYN WESTHOFF

postphlebitic syndrome encompasses chronic swelling, skin changes including ulceration, and chronic pain interfering with ambulation [4]. These chronic symptoms can also be associated with superficial phlebitis. The main goal of treatment of deep-vein thrombophlebitis (DVT) is to prevent pulmonary embolism (PE), prevent recurrence of thrombosis, and prevent the postphlebitic syndrome.

II. D I A G N O S I S

AND TREATMENT

The method of diagnosis of clots depends primarily on the location of the suspected clot. Proximal thrombi (that is, above the knee) are diagnosed by real-time B-mode compression ultrasound with high sensitivity and high positive predictive value. This test is safe, noninvasive, and widely available. If the patient has had clots previously, compressability may be abnormal, and thus sonography will be less helpful for diagnosis of recurrent clots; it is also less useful for distal clots. Contrast venography is still the standard for diagnosis of thrombi, and is more accurate than sonogram when distal thrombi are suspected. Because venography is invasive and more difficult to perform, its use is limited to cases in which sonography is not helpful. Magnetic resonance imaging is proving to be useful for diagnosis, but has not replaced other modalities, except perhaps in pregnant women [5]. A suspected pulmonary embolus is first evaluated by a chest radiograph plus perfusion scan or by a ventilationperfusion (V/Q) scan in which anatomical mismatches between air flow and blood flow in the lung may indicate the location of an embolus. The advantage of doing the perfusion scan as the first diagnostic test is that it is noninvasive and does not require highly specialized skills to carry out. A multicenter study, Prospective Investigation of Pulmonary Embolism Diagnosis (PIOPED), was carried out to assess the performance characteristics of V/Q scans used to assess 931 adults suspected of PE [3]. A high-probability perfusion scan in women has a sensitivity of 41% and a positive predictive value for a PE of 86%. An intermediate- or high-probability V/Q scan has a sensitivity of 84% and a positive predictive value of 44%. Thus, scans that are intermediate probability or nondiagnostic usually require proceeding to a pulmonary angiogram for a definitive diagnosis in order to avoid unnecessary treatment for those patients who will prove, by angiogram, to be negative for PE. Conventional treatment of DVT and PE has been intravenous unfractionated heparin during hospitalization followed by 3 to 6 months of oral anticoagulant therapy (OAT). Despite the ongoing risk of DVT recurrence, OAT is typically not continued past 6 months due to the risk of hemorrhage. Initial thrombolytic therapy in the acute phase is not widely used due to the risks of hemorrhage. Use of thrombectomy or inferior vena cava filters is considered in selected patients with large thrombi or when adequate anticoagula-

tion has not prevented recurrences. Also, treatment has been initiated with subcutaneous low-molecular-weight heparin (LMWH) used at home without laboratory dose adjustment. This appears to be at least similar to unfractionated heparin in the prevention of thrombus extension, death, hemorrhage, and recurrence of VTE events [6]. An advantage of LMWH is that it can be used at home without laboratory monitoring; a disadvantage is its greater expense. The duration of subsequent therapy with OAT is extended, possibly for life, in patients who have a continuing risk for VTE, such as those with cancer and those with an inherited predisposition.

III. PATHOPHYSIOLOGY The underlying causes of venous thrombosis have not been better described than by Virchow, who proposed in the nineteenth century that hypercoagulability, injury to the vessel wall, and circulatory stasis cause intravascular thrombosis. The disease was not described in antiquity, and Dexter proposes that the emergence of venous thrombosis as a major medical problem is due to the modern increase in chairsitting [7]. Multiple factors are required for a venous clot to occur, because each factor alonemthat is, stasis or injury or hypercoagulabilitymis only rarely associated with a clinically evident thrombus. Most of the recognized risk factors for VTE have a an obvious connection to stasis, injury, or hypercoagulability. The coagulation system is a dynamic balance between procoagulant and anticoagulant factors. The coagulation system factors related to arterial thrombosis appear to be distinct from those related to venous thrombosis. The risk of venous clots has been linked to deficiencies of several anticoagulant factors. In particular, inherited deficiencies of protein C, protein S, and antithrombin III confer an increased risk of clinically apparent venous clots. Acquired or inherited resistance to activated protein C (APC resistance) also leads to increased risk of thromboses. APC resistance can be measured genotypically or phenotypically. Genetic assays aim to identify in factor V an autosomal recessive point mutation called the Leiden mutation [8], which causes APC resistance. Phenotypic assays measure the resistance to cleavage of the factor V complex in the presence of activated protein C; this phenotype favors coagulation. The APC-resistent phenotype is present in individuals with the Leiden factor V mutation [9], and also, in a brief report from the World Health Organization Monitored Investigation of Cardiac Disease (MONICA) survey, the APC-resistent phenotype has been associated with oral contraceptive users, hormone replacement therapy users, and those with a body mass index greater than 30 kg/m 2 [10]. Screening of asymptomatic individuals for these inherited predispositions is not indicated on any routine basis. A genetically determined abnormality in prothrombin (a procoagulant) has been identified in thrombophilic families

CHAPTER 42 Pulmonary Embolism/Venous Thrombosis [ 11 ]. This and other specific inherited or acquired abnormalities of the coagulation system are suspected to increase the risk of venous thrombosis. Additional genetic studies will continue to pinpoint specific mutations in clotting factors that increase susceptibility to venous thrombosis, but these are likely to be rare and are currently not well established.

IV. INCIDENCE

AND PROGNOSIS

Several cohort studies have quantified the incidence of venous thrombosis and pulmonary embolism by identification of all new cases arising in a defined population. The first of these was the Tecumseh Community Health Study, whereby 9000 individuals of all ages in a city in Michigan were questioned and examined repeatedly in several cycles from 1957 to 1969 [12]. Most of the 169 first thrombotic events and the 63 recurrent events were identified by report of the participants without further validation. The incidence of these events increased with age, and men and women over age 40 experienced similar rates of DVT and PE. Because there were only 45 cases reported by women over age 40, the incidence rates are imprecise, but nonetheless they are similar to those obtained subsequently in larger studies with more uniform and complete surveillance of the population. These investigators estimated there would be about 250,000 clinically recognized cases annually in the United States, of which nearly half would occur in women over age 40 years. The Worcester DVT study calculated incidence rates for VTE based on discharge diagnoses from all 16 hospitals covering the Worcester Standard Metropolitan Statistical Area (1985 population, 379,953) from July, 1985 through December, 1986. All medical records with an eligible discharge diagnosis were reviewed to identify cases living in the catchment area and to confirm the diagnoses [2]. In total there were 615 cases included in the a n a l y s i s - - 4 0 5 initial episodes and 210 recurrent episodes. When extrapolated to the United States population, this study predicts about 170,000 initial and 99,000 recurrent cases annually, with nearly half of these in women over 40 years old. This study also identified an increase in risk with age, and similar incidence rates for men and women. Age-specific incidence rates for first events in women over 40 are given in Table I; the number of subjects used to calculate these rates was estimated from information in the paper. Overall, 12% of patients with a first episode died in the hospital, 5% after DVT, and 23% after PE. Case fatality was similar for men and women and increased with age. Immediate case fatality was only 2% below age 40 years, but was 10% from ages 40 to 59 years and 11% from ages 60 to 79 years. All patients in the Worcester cohort who were discharged alive were followed for 2 to 3.5 years, and 30% of the patients died during the follow-up interval. Long-term survival depended on age, but not on sex or whether the initial event was a PE or a DVT.

609 TABLE I Incidence of Venous Thrombosis and Pulmonary Embolism in Women > 4 0 Years Old a Venous thrombosis

Pulmonary embolism

Age (years)

Worcester

Malm6

Worcester

40 -49 50-59 60-69 70-79 Cases (N)

1.0 4.2 9.9 21.1 129b

9.7 10.3 21.7 42.9 189

0 1.1 6.2 6.9 540

NHS 1.3 1.8 3.2 344

a Cases per 10,000 women per year. Rates from Worcesterinclude only first events; rates from Malm6 and the NHS may include patients with recurrent thromboses or emboli. bThe numbers of cases from Worcester were calculated from data in the paper.

A similar study evaluated DVT in Malm6, Sweden, a city of 230,000. Subjects included all patients referred for phlebography (venogram) for suspected DVT in 1987 [13]. There was only one source of diagnostic testing for this population, and phlebography was the test of choice at the time of the study. There were 366 tests positive for DVT of 1009 referrals, including 189 women aged 40 years or older. Some of these cases may have been diagnosed and treated solely as outpatients, in contrast to most other studies that have considered only inpatient cases. Patients presenting with a primary pulmonary embolus were not included in this study, but PE was clinically suspected in 5% of the cases with DVT. As in the previous cohorts, the incidence rates increased with age and were similar in men and women. The incidence rates in Malm6 may appear higher than those in Worcester because of the possible inclusion of outpatients, the inclusion of a few patients with PE in the DVT group, and because new (76%) and recurrent (24%) cases were not separated. Fatality rates were not presented. Participants in the Nurses' Health Study (NHS) were evaluated for the occurrence of PE between 1976 and 1992 [ 14]. Medical records were examined for all of the 280 PEs that were reported by questionnaire or by death certificate. All but 36 of these cases occurred in women 40 years or older. Age-specific incidence rates are shown in Table I; a rate for women older than 70 years is not calculated because very few members of the cohort had yet reached that age. Mortality rates were not presented. Based on the Tecumseh, Worcester, Malm6 and NHS cohorts, about one-quarter to one-third of all cases of PE and VTE among women over 40 years old are recurrences. The risk of a recurrence, even after 6 months of anticoagulant treatment, is much greater than the risk of a first event. In an 8-year study of 355 consecutive, new VTE cases in Padua, Italy, where the cases were diagnosed from 1986 to 1987,

610

CAROLYN WESTHOFF

78 patients experienced a recurrent venous thrombotic event during follow-up, including 9 fatal pulmonary emboli [4]. The recurrences accrued gradually with a cumulative recurrence rate of 30% at 8 years. Those cases whose primary VTE was secondary to surgery or trauma had a decreased risk of recurrence. Whether sex or age influenced recurrence risk was not presented. Mortality at 8 years was 30% and was mainly related to cancer. After 8 years of follow-up, 29% of subjects developed post-thrombotic syndrome. In a similar study from Cleveland [15], 124 cases with venogramconfirmed VTE diagnosed in 1984 and 1985 were followed until recurrence or death; follow-up ended in 1992. During follow-up, 42% of the cases died; the main predictors of death were cancer, stroke, or age greater than 75 years at the initial diagnosis. Chronic symptoms of pain, swelling, or discoloration in the affected leg were reported by 42%, and a recurrence was diagnosed in 15% of patients. These followup studies indicate that chronic symptoms and recurrence are common after a first VTE even in patients who receive a full 6-month course of anticoagulation. Little else is known about the distribution of VTE in the general population. A study of all 23,000 VTE hospital discharges in California from 1991 through 1994 found that about 75 % of cases are idiopathic and 25% are secondary to cancer, surgery, trauma, or another medical event [ 16]. This study also revealed that the risk among African-Americans was higher than among whites for idiopathic events [relative risk (RR), 1.3; 95% confidence interval (CI), 1.1-1.5]. The risk for Hispanics for idiopathic events was lower (RR, 0.6; 95% CI, 0.5-0.7) as was the risk for Asians (RR, 0.26; 95% CI, 0.2-0.3). The risk for secondary VTE was similar for whites, African-Americans, and Hispanics, but was lower for Asians.

prevalence is about 3%, and is even higher among patients with recurrent VTE. The relative risk of VTE from family studies is about 8 and from a population-based study is 6.5, which shows good agreement [19-21]. DVT of the leg is the most common manifestation, and one-half of predisposed individuals will not experience a first clinical event until after age 40 years. The prevalence of protein S deficiency has not been described in the general population. Among patients with a first VTE episode, the prevalence is about 1-2%, and seems to be higher among patients with recurrent thrombosis [21]. There may be phenotypic variations in protein S deficiency, and the relative risks for VTE for individuals with this class of clotting abnormality have not been precisely quantified. Antithrombin III deficiencies are rare, occurring in only 0.02% of unselected blood donors [ 18], in whom the relative risk for VTE may be as high as 50 [21,22], and the first VTE event will often occur before age 25 years, which makes this the most serious of the inherited predispositions. The most recently described and most common of the inherited predispositions is APC resistance due to a point mutation in factor V [23]. The population prevalence is between 3 and 6% and is greater among whites than in other racial groups [24]. The relative risk of VTE in individuals with this mutation is about 8 compared to the unaffected population, and the risk may be substantially higher for homozygotes [25]. A first event in an affected individual may occur at any age. Altogether, about 30% of individuals with a first VTE event occurring in the absence of a predisposing clinical situation will have one of these underlying genetically determined abnormalities of the coagulation system. The APC resistance due to the Leiden mutation is the most common of these underlying abnormalities [8].

B. R i s k s o f V T E w i t h C l i n i c a l l y A. R i s k s o f V T E w i t h C o n g e n i t a l

Identified Risk Factors

or A c q u i r e d C a u s e s o f H y p e r c o a g u l a b i l i t y Familial clustering of VTE has allowed the investigation and recognition of several inherited disorders of coagulation that are associated with a dramatically increased risk of VTE. In population-based studies, however, the risk of thrombosis in the presence of specific genetically determined abnormalities is sometimes lower than the risk as estimated in studies of affected families; this suggests that for certain abnormalities, members of strongly affected families may have additional, unrecognized abnormalities that generate higher risks than seen in otherwise unselected individuals with the same defect [17]. Bearing in mind these discrepancies, the frequency of each of the major genetic predispositions and estimates of the associated risk will be presented. The prevalence of protein C deficiency has been estimated in large studies of blood donors to be about 0.2-0.4% [18]. Among patients experiencing a first VTE episode, the

The strongest risk factors for VTE have been readily identified clinically without the use of laboratory or epidemiological studies to establish risk. Immobilization with or without injury is the hallmark of these risk factors. VTE associated with any of these antecedent factors is generally referred to as secondary VTE because the clot is presumed to have arisen as a result of a specific precipitating event. In most series of consecutive cases, from 40 to 80% of all subjects are considered to have a VTE that is secondary to a clinically recognized precipitating event. In studies to assess more subtle causes of VTE, those VTE patients with any of these major risk factors are excluded. Due to this approach, it is essentially unknown how cofactors might increase risk in the large proportion of VTE patients who have a precipitating factor. Studies of clinical risk factors have not presented separate analyses regarding menopausal women, but there is no a p r i o r i reason to suspect important differences in risk factors by age or sex. Overall, there is an increase in

CHAPTER42 Pulmonary Embolism/Venous Thrombosis TABLE II

Risk

Risk Factors for VTE in Hospitalized Women Aged > 4 0 Years Old a Reason for hospitalization

40%

Major surgery, illness, or trauma with past history of VTE Fracture or orthopedic surgery of pelvis, hip, or leg Pelvic or abdominal surgery for cancer Lower limb paralysis (e.g., stroke) or amputation

a

Adapted from THRIFT Consensus Group [26].

VTE risk after age 40 years, but much of this increase is probably due to the increased prevalence of the underlying medical risk factors with increasing age. Many VTE events occur after hospitalization for another problem, and are not themselves the primary reason for the hospitalization. Risk of VTE occurring in patients who are already hospitalized has been well characterized. A risk classification for women over age 40 modified from the Thromboembolic Risk Factors (THRIFT) Consensus Group [26] is presented in Table II. Women who fall into the low-risk group do not warrant prophylactic anticoagulation, but for those in the moderate- or high-risk groups routine prophylaxis is advisable. A meta-analysis of the incidence of DVT following general surgery indicates that effective prophylactic measures include low-dose heparin, graduated elastic compression stockings, and intermittent pneumatic compression [27]. In the PIOPED study several risk factors were present among women who were referred for evaluation for possible PE [3]; these included surgery within 3 months before the onset of symptoms (38%), immobilization (35%), malignancy (28%), a history of previous phlebitis (18%), stroke (7%), and trauma (7%). After evaluation, these factors were associated with a 50% or greater increased risk of a positive angiogram among the women in the study population. An increased risk of either deep vein thrombosis or pulmonary embolus has been identified in such patients in numerous studies [28].

C. Other Risk Factors for Primary or Idiopathic VTE Few studies have evaluated causes of VTE after excluding cases with strong precipitating factors. The main exposures of interest have been weight, smoking, chronic medical conditions, and use of exogenous hormones. Use of hor-

611 mones will be discussed separately below. Some studies have identified modestly increased risks for VTE in women with chronic medical conditions such as hypertension, diabetes mellitus, and gall bladder disease [29,30]. Because evaluation of hormone use has often been a primary goal of the analyses, women with these conditions have often been excluded in order to control for confounding. Overall, evaluation of risk associated with the common chronic medical conditions has not been illuminating. Obesity, when defined as a body mass index, has generally been found to be a risk factor for VTE. Obesity had no effect on VTE risk only in the Walnut Creek Contraceptive Drug Study, in which obesity was defined as weight 15% above the cohort mean [31 ]; this study included 38 cases, of which 23 were 40 years or older. In the Nurses' Health Study there were 125 women who reported an idiopathic pulmonary embolism. Among these women, obesity (defined as a BMI of 29 kg/m 2) was associated with a relative risk of 2.9 (95% confidence interval, 1.5-5.4). Most of these cases were older than 40 years, and the increased risk associated with obesity held for all ages [ 14]. In a United Kingdom cohort that included 292 female VTE cases aged 5 0 - 7 9 years, a BMI of 26 kg/m 2 or greater had a relative risk for VTE of 2.0 (95% confidence interval, 1.4-2.9) compared to women with a BMI of 25 kg/m 2 or less [30]. Large studies of oral contraceptives and thrombosis included younger study populations, but all have identified obesity as a risk factor for VTE [32-34]. The evidence supports an increased risk of VTE in obese women in the menopausal age group as well as in younger women. There are no data concerning VTE and obesity in men. The data for smoking are less consistent. Studies in younger women find little or no increased risk of VTE in current smokers [32-34]. The large United Kingdom cohort of women aged 5 0 - 7 9 years identified a relative risk of 1.2 (95% confidence interval, 0.9-1.7) for VTE among current smokers compared to never smokers [30]. Among the 125 idiopathic PE cases from the NHS, there was a RR of 1.9 (95% confidence interval, 0.9-3.7) for women who smoked 2 5 - 3 4 cigarettes daily and a RR of 3.3 (95% confidence interval, 1.7-6.5) for women who smoked 35 or more cigarettes daily compared to nonsmokers [ 14]. In sum, cigarette smoking is not an important explanatory variable for VTE in women, regardless of age.

D. Hormonal Risk Factors for Venous Thromboembolism Case reports followed very soon by epidemiologic studies showed that use of the old, high-dose oral contraceptives (OCs) were associated with an increased risk of venous thrombosis, including pulmonary embolus [35]. As the dose of estrogen in the OC decreased there was a concomitant decrease in VTE risk [36]. This led to the conclusion that the

612

CAROLYN WESTHOFF

TABLE III Current HRT Use and VTE Risk The Early "Negative" Studies Study

Years

Number of cases

Adjustedrelative risk (95% CI)

BCDSP [39] Nachtigall [40] Petitti [ 2 9 ] Devor [ 4 1 ]

1972 1969-1978 1969-1976 1980-1987

18 30 17 121

2.3 (0.6-8.0) a 0.85 (ns; CI not provided) 0.7 (0.2-2.5) b 0.6 (0.2-1.8)

a Relative risk and CI not provided in original publication, but subsequently estimatedby Douketis [54]. NS, not significant. b90% CI.

risk of VTE in OC users was linked to the estrogen dose, a concept that was little challenged over the past two decades [37]. Because the dose of estrogen in hormone replacement therapy (HRT) is so much lower than the dose in even the lowest dose OCs, it seemed unlikely that HRT would be unlikely to cause any problem with venous clots [38]. The original studies (Table III) that attempted to assess risk of VTE in HRT users supported this line of thought. The first report came from the Boston Collaborative Drug Surveillance Program [39] and reported on 18 cases of VTE. Of the cases, 14% were HRT users, and of the controls, 8% were users; if the study were larger these percentages would have indicated an increased risk. The authors correctly concluded "significant associations were not present"; however, the small number of cases and rare use of HRT meant that the study did not have statistical power to detect an association. The next report came from the Walnut Creek Contraceptive Drug Study [29,31], and included 17 cases, of whom 12% used HRT; 15% of the controls used HRT. This does not indicate any hint of increased risk, but again, due to the small number of cases and limited use of HRT the study lacked statistical power to detect an association. Both of these studies excluded the majority of cases to focus on idiopathic VTE only. In contrast, Nachtigall [40] excluded none of the cases; in this randomized controlled trial of chronically ill, permanently hospitalized women, 84 women were randomized to an oral HRT regimen for 10 years, and 84 controls received placebo. These 168 women experienced 30 VTE e v e n t s - - a number that is at least 10 times greater than the largest expected number calculated from incidence rates in Table I. The high rate of VTE in this population may have been due to limited mobility of the subjects during their chronic hospitalization. There are also cases of superficial thrombophlebitis included, and no requirement for confirmatory diagnostic tests. There was no indication of increased VTE risk among the women randomized to HRT; however, it is difficult to interpret this finding in this unusual population. Devor [41 ] also included all cases of VTE in a hospitalbased case-control analysis. Based on 121 cases, there was no evidence of excess risk in the HRT users. The overall

analysis showed no increase in risk associated with HRT, but when subsets of cases were excluded (such as women with a previous thrombosis) the relative risk increased. As is true for other consecutive, unselected series of VTE cases, most of the cases in this study were secondary to a known predisposing factor. These four early evaluations of the H R T - V T E association did not indicate an increased risk, but the first two studies were too small for a precise or sophisticated analysis and the randomized trial included an unusual population. Finally, the larger case-control study included so many women with secondary VTE and other serious diseases that it is difficult to assess whether HRT would have been given to these women by the physicians who were caring for their medical illnesses. With the clarity of hindsight, it appears that none of these early studies had a design or number of cases appropriate to address the question.

E. S t u d i e s o f t h e H R T - V T E

Association

Starting in 1996, a series of new large studies began to be published (Table IV), suggesting an increased risk of both deep vein thrombosis and pulmonary embolus among current users of hormone replacement therapy. These new studies include data from two large prospective cohort studies, the Nurses' Health Study [42] and the Oxford-Family Planning Association cohort [43], from an historical cohort study of the Group Health Incorporated (GHI) medical plan database [44], from an historical cohort study using record linkage with the United Kingdom General Practice Research Database [30], from a record linkage study using medical databases in Italy [45], from a hospital-based case-control study in the United Kingdom [46], and, finally, from a randomized controlled trial of women with heart disease, the Heart and Estrogen/progestin Replacement Study (HERS) trial, in the United States [47]. All of these studies in Table IV consider just those cases with a first episode of VTE. All of the studies excluded cases and controls with risk factors for VTE (e.g.,

TABLE IV Risk of Primary VTE among Current HRT U s e r s m S u b s e q u e n t Studies Adjusted relativerisk (95% CI) Number of cases Currentuse First-yearuse

Study

Years

Grodstein [42] Daly (letter) [43] Daly [ 4 3 ] Jick [44] Gutthann [30] Varas [ 4 5 ] Hulley [ 4 7 ]

1976-1992 1982-1993 1993-1994 1980-1994 1991-1994 1991-1995 1993-1997

a Pulmonary embolus only.

123a 18 103 42 292 171 46

2.1 (1.2-3.8) 2.2 (0.6-7.9) 3.5 (1.8-7.0) 3.6 (1.6-7.8) 2.1 (1.4-3.2) 2.3 (1.0-5.3) 2.9 (1.5-5.6)

Not calculated Not calculated 6.7 (2.1-21.3) 6.7 (1.5-30.8) 4.6 (2.5-8.4) 2.9 (1.2-6.9) 3.3 (1.1-10.1)

CHAPTER 42 Pulmonary Embolism/Venous Thrombosis cancer, fracture, and stroke), as well as excluding cases with superficial phlebitis. All of the studies described specific diagnostic criteria that were required for inclusion as a case. Although the methodology differed and the stringency of the diagnostic criteria varied somewhat between studies, the relative risks shown in Table IV are nonetheless extremely similar, and indicate a risk among current HRT users that is two to three times the risk among nonusers. Unlike previous smaller studies, these studies were able to assess risks in some subgroups of HRT users. The most consistent finding was that the increased risk of VTE was greatest in or limited to the first year of HRT use. Until recently, the usual approach to analysis was to look for increased risk as the time of exposure increased, e.g., the risk of lung cancer increases with increasing years of cigarette smoking. The notion of looking for an early, transient risk is more recent and an understanding of these consistent results is not yet established. Other studies indicate that the increased risk of VTE in OC users may also be early and transient [34,48]. Other subgroup analyses have attempted to assess conventional markers of risk such as dose. In these analyses, there was an indication of an increased risk of VTE with higher estrogen doses in one study [32], but no d o s e - r e s p o n s e effect in the others that were able to assess dose [42,46,49]. The studies did generally agree on little or no excess risk in past users [42,45,46,49]. The populations that were studied often included only a few HRT users or a limited range of HRT regimens; therefore, the investigators had a limited ability to compare risks between regimens. In addition to the studies listed above, the Postmenopausal Estrogen/Progestin Intervention (PEPI) trial also evaluated phlebitis in women randomly assigned to five different HRT regimens; although there was a suggestion of an increased risk of VTE among the subjects receiving active treatment, this outcome was too rare to allow comparisons between the groups [50]. Taken together, these studies have provided little evidence to distinguish the risks of opposed versus unopposed estrogen, of oral versus other routes of administration, or between different formulations. Overall, there appears to be a small, perhaps transient, increase in risk that is not limited to or avoided by any particular HRT regimen. The majority of VTE cases in women in the menopausal age group are secondary to other identifiable risk factors. The incidence of cases that might be attributed to HRT use can be calculated from the cohort studies. In the GHI cohort there were about two extra VTE cases per 10,000 HRT users per year [44]. An estimate based on United Kingdom incidence data similarly suggested about two additional cases per 10,000 HRT users per year [46]. In the HERS study the excess risk was about 7 cases per 1000 in year 1 and about 2 cases per 1000 in years 4 and 5 [47]; however, women were selected for that randomized trial based on diagnosed cardiovascular disease, and they appear to have a much higher baseline risk for VTE than do women in

613 the general population. The HERS results do suggest that women with an increased baseline risk of VTE may experience many more cases of VTE if using HRT than is seen among low-risk women. All of these studies have looked at traditional estrogenbased hormone replacment regimens. Some data are accumulating about the risk of VTE in women using the other, newer selective estrogen receptor modulators. Tamoxifen has the longest and widest use of these and has been associated with increased risk of VTE at a magnitude similar to the studies of estrogen [51]. Tibolone, a nonestrogen treatment for hot flashes, was evaluated in some of the European studies with relative risks for VTE slightly lower than those seen for estrogen; however, the number of users of tibolone in these studies was small and therefore the relative risk estimate thus far must be considered imprecise [46]. Raloxifene use in placebo-controlled clinical trials has been associated with a relative risk of VTE of about 3.0 [52], and this excess is noted in the product labeling [53].

V. C L I N I C A L

RECOMMENDATIONS

Venous thromboembolism is a common problem among women in the menopausal age group, and is most likely to occur among women with predisposing medical problems. In large part VTE increases with age because the predisposing problems become more common with advancing age. Among healthy, low-risk women, even after age 40, the risk of VTE is probably about 1 - 2 new cases per 10,000 women per year. In this population, the risk of VTE for women using HRT may increase two- to threefold from this low level, yielding somewhere between 1 and 6 additional cases per 10,000 HRT users. This is comparable both in relative and absolute terms to the increased risk seen among younger women who use oral contraceptives. As with younger women and OCs, this risk of VTE in HRT users needs to be communicated to potential HRT users so that they can weigh this along with all other risks and benefits in making a decision to use HRT. Among menopausal women with chronic medical problems or women who will be having surgery or hospitalization for some other reasons, the baseline risk of a VTE is substantially higher. The results from the HERS trial [47] indicate that the women in that trial with known heart disease had a baseline risk of VTE of at least 20/10,000 women per year and that HRT use in these women increased that risk threefold. The proportional increase was the same as that seen in low-risk women, but because their baseline risk of VTE was higher, this increase in risk may translate into 2 0 40 additional cases per 10,000 users per year. Because most of the studies agree that the excess risk is concentrated in the first 1-2 years of HRT use, this information is most important for women who are considering whether to initiate HRT.

614

CAROLYN WESTHOFF

Although there are no specific relevant data regarding the risk of VTE among hospitalized HRT users, it appears prudent to consider discontinuing oral HRT in women who are going to undergo major surgery or who develop a medical problem that is associated with immobilization or other disease that is associated with a high risk of VTE. Among women who are currently being treated with anticoagulants, there are no data to indicate whether HRT use would modify their risk of a recurrent event.

References 1. Ault, M., and Artal, R. (1998). Upper extremity DVT: What is the risk? Arch. Intern. Med. 158, 1950-1951. 2. Anderson, E A., Jr., Wheeler, H. B., Goldberg, R. J., Hosmer, D. W., Patwardhan, N. A., Jovanovic, B., Forcier, A., and Dalen, J. E. (1991). A population-based perspective of the hospital incidence and case-fatality rates of deep vein thrombosis and pulmonary embolism. The Worcester DVT Study. Arch. Intern. Med. 151,933-938. 3. Quinn, D., Thompson, B., Terrin, M., Thrall, J., Athanasoulis, C., McKusick, K., Stein, E, and Hales, C. (1992). A prospective investigation of pulmonary embolism in women and men. JAMA, J. Am. Med. Assoc. 268, 1689-1696. 4. Prandoni, R, Lensing, A. W., Cogo, A., Cuppini, S., Villalta, S., Carta, M., Cattelan, A. M., Polistena, P., Bernardi, E., and Prins, M. H. (1996). The long-term clinical course of acute deep venous thrombosis. Ann. Intern. Med. 125, 1-7. 5. Baker, W. (1998). Diagnosis of deep venous thrombosis and pulmonary embolism. Curr. Concepts Thromb. 82, 475-495. 6. Haas, S. (1998). Treatment of deep venous thrombosis and pulmonary embolism. Current recommendations. Curr. Concepts Thromb. 82, 495-510. 7. Dexter, L. (1973). President's Address: The chair and venous thrombosis. Trans. Am. Clin. Climatol. Assoc. 84, 1-15. 8. Dahlb/ack, B., Hillarp, A., Rosen, S., and Z611er, B. (1996). Resistance to activated protein C, the FV:Q 5~ allele, and venous thrombosis. Ann. Hematol. 72, 166-176. 9. Bertina, R., Koeleman, B., Koster, T., Rosendall, E, Dirven, R., de Ronde, H., van der Velden, E, and Reitsma, E (1994). Mutation in blood coagulation factor V associated with resistance to activated protein C. Nature (London) 369, 64-67. 10. Lowe, G. D. O., Rumley, A., Woodward, M., and Reid, E. (1996). End of the line for "third-generation" pill controversy? Lancet 349, 11131114. 11. Poort, S., Rosendaal, F., Reitsma, P., and Bertina, R. (1996). A common genetic variation in the 3'-untranslated region of the prothrombin gene is associated with elevated plasma prothrombin levels and an increase in venous thrombosis. Blood 88, 3698-3703. 12. Coon, W., Willis, E, and Keller, J. (1973). Venous thromboembolism and other venous disease in the Tecumseh community health study. Circulation 47, 839-846. 13. Nordstr6m, M., Lindblad, B., Bergqvist, D., and Kjellstrom, T. (1992). A prospective study of the incidence of deep-vein thrombosis within a defined urban population. J. Intern. Med. 232, 155-160. 14. Goldhaber, S., Grodstein, F., Stampfer, M., Manson, J., Colditz, G., Speizer, F., Willett, W., and Hennekens, C. (1997). A prospective study of risk factors for pulmonary embolism in women. JAMA, J. Am. Med. Assoc. 277, 642-645. 15. Beyth, R., Cohen, A., and Landefield, C. (1995). Long-term outcomes of deep-vein thrombosis. Arch. Intern. Med. 155, 1031-1037.

16. White, R., Zhou, H., and Romano, E (1998). Incidence of idiopathic deep venous thrombosis and secondary thromboembolism among ethnic groups in California. Ann. Intern. Med. 128, 737-740. 17. Rosendaal, E (1997). Risk factors for venous thrombosis: Prevalence, risk, and interaction. Semin. Hematol. 34, 171-187. 18. Tait, R. C., Walker, I. D., Reitsma, E H., Islam, S. I., McCall, F., Poort, S. R., Cankie, J. A., and Bertina, R. M. (1995). Prevalence of protein C deficiency in the healthy population. Thromb. Haemostasis 73, 87-93. 19. Allaart, C., and Bri~t, E. (1994). Familial venous thrombophilia. In "Hemostasis and thrombosis" (A. Bloom, C. Forbes, D. Thomas, et al., eds.), pp. 1349-1360. Churchill-Livingstone, New York. 20. Bovill, E. G., Bauer, K. A., Dickermann J. D., Callas, E, and West, B. (1989). The clinical spectrum of heterozygous protein C deficiency in a large New England kindred. Blood 73, 712-717. 21. Koster, T., Rosendaal, F. R., Briet, E., van der Meer, E J., Colly, L. E, Trienekens, E H., Poort, S. R., Reitsma, E H., and Vandenbroucke, J. E (1995). Protein C deficiency in a controlled series of unselected outpatients: An infrequent but clear risk factor for venous thrombosis (Leiden thrombophilia study). Blood 85, 2756-2761. 22. Hiejboer, H., Brandjes, D. P., Buller, H. R., Sturk, A., and Ten Cate, J. W. (1990). Deficiencies of coagulation-inhibiting and fibrinolytic proteins in outpatients with deep-vein thrombosis. N. Engl. J. Med. 323, 1512-1516. 23. Dahlb~ick, B., Carlsson, M., and Svensson, E (1993). Familial thrombophilia due to a previously unrecognised mechanism characterized by poor anticoagulant response to activated protein C: Prediction of a cocafactor to activated protein C. Proc. Natl. Acad. Sci. U.S.A. 90, 1004-1008. 24. Ridker, E M., Hennekens, C. H., Lindpainter, K., Stampfer, M. J., Eisenberg, P. R., and Miletich, J. E (1995). Mutation in the gene coding for coagulation factor V and the risk of myocardial infarction, stroke, and venous thrombosis in apparently healthy men. N. Engl. J. Med. 332, 912-917. 25. Vandenbroucke, J., and Helmerhorst, E (1996). Risk of venous thrombosis with hormone-replacement therapy. Lancet 348, 972. 26. THRIFT Consensus Group (1992). Risk of and prophylaxis for venous thromboembolism in hospital patients. Br. Med. J. 305, 567-574. 27. Clagett, G., and Reisch, J. (1988). Prevention of venous thromboembolism in general surgical patients. Results of meta-analysis. Ann. Surg. 208, 227-240. 28. Salzman, E., and Hirsch, J. (1994). The epidemiology, pathogenesis, and natural history of venous thrombosis. In "Hemostasis and Thrombosis: Basic Principles and Clinical Practice" (R. Coleman, J. Hirsch, V. Marder, and E. Salzman, eds.), pp. 1275-1296. Lippincott, Philadelphia. 29. Petitti, D., Wingerd, J., Pellegrin, E, and Ramcharan, S. (1979). Risk of vascular disease in women: Smoking, oral contraceptives, noncontraceptive estrogens, and other factors. JAMA, J. Am. Med. Assoc. 242, 1150-1154.

30. Gutthann, S. P., Garcia Rodriguez, L. A., Castellsague, J., and Oliart, A. D. (1997). Hormone replacement therapy and risk of venous thromboembolism: Population based case-control study. Br. Med. J. 314, 796-800. 31. Petitti, D., Wingerd, J., Pellegrin, E, and Ramcharan, S. (1978). Oral contraceptives, smoking, and other factors in relation to risk of venous thromboembolic disease. Am. J. Epidemiol. 108, 480-485. 32. Jick, H., Jick, S., Gurewich, V., Myers, M. W., and Vasilakis, C. (1995). Risk of idiopathic cardiovascular death and nonfatal venous thromboembolism in women using oral contraceptives with differing progestagen components. Lancet 346, 1589-1592. 33. World Health Organization Collaborative Study of Cardiovascular Disease and Steroid Hormone Contraception (1995). Effect of different progestagens in low oestrogen oral contraceptives on venous thromboembolic disease. Lancet 346, 1582-1588.

615

CHAPTER 42 P u l m o n a r y E m b o l i s m / V e n o u s Thrombosis 34. Spitzer, W. O., Lewis, M. A., Heinemann, L. A., Thorogood, M., and MacRae, K. D. (1996). Third generation oral contraceptives and risk of venous thromboembolic disorders: And international case-control study. Br. Med. J. 312, 83-88. 35. Inman, W. H., Vessey, M. E, Westerholm, B., and Engelund, A. (1970). Thromboembolism disease and the steroidal content of oral contraceptives: A report to the Committee on Safety of Drugs. Br. Med. J. 2, 203209. 36. Gerstman, B. B., Piper, J. M., Tomita, D. K., Fergus, W. J., Stadel, B. V., and Lundin, F. E. (1991). Oral contraceptive estrogen dose and the risk of deep venous thromboembolic disease. Am. J. Epidemiol. 133, 32-37. 37. Realini, J., and Goldzieher, J. (1985). Oral contraceptives and cardiovascular disease: A critique of the epidemiological studies. Am. J. Obstet. Gynecol. 152, 729-798. 38. Lobo, R. (1992). Estrogen and the risk of coagulopathy. Am. J. Med. 92, 283-285. 39. Boston Collaborative Drug Surveillance Program (1974). Surgically confirmed gallbladder disease, venous thromboembolism, and breast tumors in relation to postmenopausal estrogen therapy. N. Engl. J. Med. 290, 15-19. 40. Nachtigall, L., Nachtigall, R., Nachtigall, R., and Beckman, E. (1979). Estrogen replacement therapy II: A prospective study in the relationship to carcinoma and cardiovascular and metabolic problems. Obstet. Gynecol. 54, 74-79. 41. Devor, M., Barrett-Connor, E., Renvall, M., Feigal, D., Jr., and Ramsdell, J. (1992). Estrogen replacement therapy and the risk of venous thrombosis. Am. J. Med. 92, 275-282. 42. Grodstein, F., Stampfer, M., Manson, J., Colditz, G., Willett, W., Rosner, B., Speizer, E, and Hennekens, C. (1996). Postmenopausal estrogen and progestin use and the risk of cardiovascular disease. N. Engl. J. Med. 335, 453-461. 43. Daly, E., Vessey, M. P., Painter, R., and Hawkins, M. (1996). Casecontrol study of venous thromboembolism risk in users of hormone replacement therapy. Lancet 348, 1027. 44. Jick, H., Derby, L., Myers, M., Vasilakis, C., and Newton, K. (1996). Risk of hospital admission for idiopathic venous thromboem-

45.

46.

47.

48. 49.

50.

51.

52.

53. 54.

bolism among users of postmenopausal oestrogens. Lancet 348, 981-983. Varas-Lorenzo, C., Garcia-Rodriguez, L., Cattaruzzi, C., Troncon, M., Agostinis, L., and Perez-Gutthann, S. (1998). Hormone replacement therapy and the risk of hospitalization for venous thromboembolism: A population-based study in Southern Europe. Am. J. Epidemiol. 147, 387-390. Daly, E., Vessey, M. P., Hawkins, M. M., Carson, J. L., Gough, P., and Marsh, S. (1996). Risk of venous thromboembolism in users of hormone replacement therapy. Lancet 348, 977- 980. Hulley, S., Grady, D., Bush, T., Furberg, C., Herrington, S., Riggs, B., and Vittinghoff, E. (1998). Randomized trial of estrogen plus progestin for secondary prevention of coronary heart disease in postmenopausal women. JAMA, J. Am. Med. Assoc. 280, 605-613. Lidegaard, O. (1998). Thrombotic diseases in young women and the influence of oral contraceptives.Am. J. Obstet. Gynecol. 179, $62-$67. Grodstein, F., Stampfer, M., Goldhaber, S., Manson, J., Colditz, G., Speizer, F., Willett, W., and Hennekens, C. (1996). Prospective study of exogenous hormones and risk of pulmonary embolism in women. Lancet 348, 983-987. Writing group for the PEPI Trial (1995). Effects of estrogen or estrogen/progestin regimens on heart disease risk factors in postmenopausal women. JAMA, J. Am. Med. Assoc. 273, 199-208. Fisher, B., Costantino, J., Redmond, C., Poisson, R., Bowman, D., Couture, J., Dimitrov, N. V., Wolmark, N., Wickerham, D. L., Fisher, E. R. et al. (1989). A randomized clinical trial evaluating tamoxifen in the treatment of patients with node-negative breast cancer who have estrogen-receptor-positive tumors. N. Engl. J. Med. 320, 479-484. Delmas, P., Mitlak, B., and Christiaansen, C. (1998). Effects of raloxifene in post menopausal women (letter). N. Engl. J. Med. 338, 13131314. Eli Lilly and Company (1998). "Evista| Product Information." Eli Lilly and Company. Douketis, J. D., Ginsberg, J. S., Holbrook, A., Crowther, M., Duku, E. K., and Burrows, R. E (1997). A reevaluation of the risk for venous thromboembolism with the use of oral contraceptives and hormone replacement therapy. Arch. Intern. Med. 157(14), 1522-1530.

~HAPTER

4":

Combined Estrogen/ Androgen Replacement Therapy: Benefits and Risks BARBARA B.

SHERWlN

Department of Psychology and Department of Obstetrics and Gynecology,

McGill University, Montreal, H3A 1B 1 Canada

IV. Risks of E/A Replacement Therapy V. Future Directions References

I. Introduction

II. Historical Background III. Benefits of E/A Replacement Therapy

II. H I S T O R I C A L B A C K G R O U N D


Although combined estrogen/androgen replacement therapy for postmenopausal women is still not a common treatment, the concept of combined therapy actually developed over 60 years ago. Soon after testosterone propionate had been synthesized in the mid-1930s, it was used to treat a variety of gynecological disturbances such as menorrhagia, mastitis, dysmenorrhea, and menopausal symptoms in oophorectomized women [1]. In one of the earliest studies, menopausal women received estrogen and 25 to 50 mg of testosterone propionate daily [2]. The therapy resulted in the serendipitous finding that sexual appetite and response were significantly greater than that experienced with estrogen alone. Following this observation, many studies originally undertaken to assess the efficacy of various therapeutic agents for the management of menopausal symptoms almost universally reported increased libido as an effect of exogenous androgen [3-5]. Several investigators reported, however, that the effects of androgen administration on libido

Controlled studies have prodided a voluminous amount of information on the benefits and risks of estrogen replacement therapy in menopausal women, yet there is still a paucity of data bearing on the efficacy of combined estrogen/androgen (E/A) replacement therapy. Although it is not clear why this is true, it is likely that outdated notions of female sexuality, the historical attribution of menopausal symptoms to neurotic illness, and the clinical sequelae of androgen-excess endocrinopathies in women may all have played a role in the apparent reluctance of medical scientists and practitioners to investigate and prescribe combined E/A preparations as replacement therapy to selected postmenopausal women. Historically, there has also been a reluctance, on behalf of the medical community, to acknowledge that there was any physiological role for androgens in women. This chapter reviews the benefits and risks of combined E/A therapy in the postmenopause and then underlines methodological issues that bear on research in this area.

M E N O P A U S E : B I O L O G Y AND PATHOBIOLOGY

617


618 were dependant on previous sexual functioning; women who described "very little or no libido" in the past reported no change in sexual desire following testosterone pellet implants, whereas those who "had once had libido but lost it" reported a marked increase after treatment [6,7]. Greenblatt et al. assessed the relative efficacy of hormones administered singly or in combination to menopausal women [8]. Satisfactory relief of all symptoms was reported by 96.9% of patients who received estrogen alone, and by 89.6% who received an estrogen/androgen combination (diethylstilbestrol 0.25 mg/day and methyltestosterone, 5 mg/day). Only 23.5 % of those who received androgen alone reported symptom relief whereas 83.3% of the subjects in the placebo group reported no improvement of symptoms. Most noteworthy, 66.6% of the patients stated a preference for the estrogen/androgen combination because of the increased well-being and libido they experienced while on this regimen. These findings were confirmed in a study by Caldwell and Watson, who found that postmenopausal women who received a combined estrogen/androgen preparation showed improved physical capacity, increase in weight, and improvement in memory and in ability to learn new material [9]. Despite the consistency in the findings of these early studies, they lacked control groups, used unsystematic methods of data collection, and the majority were merely anecdotal in nature. A second (though somewhat weak) source of support for the libido-enhancing effects of androgen comes from the clinical reports of its administration in the treatment of estrogen-dependant breast cancers [10,11]. This treatment was based on the belief that large doses of androgen would oppose the effects of estrogen and thus halt spread of the disease. These women spontaneously reported increased libido as a result of the hormone therapy. However, the massive doses of testosterone propionate used in those cases (1200 mg/week), the confounds inherent in this gravely ill subject sample, and the uncontrolled nature of the reports detract from the meaningfulness of these findings. A third series of studies undertaken in the late 1950s and early 1960s similarly concerned women who underwent several surgical procedures in an effort to halt the continuing spread of metastatic breast cancer. Following mastectomy and oophorectomy, no changes in sexual desire were obvious [12]. However, following adrenalectomy, carried out because of progression of the malignant disease, 14 out of 17 patients reported the sudden absence of all sexual desire. In a later study of seven patients who had had oophorectomy 1 to 7 years before the adrenals were removed, all reported almost total loss of sexual feelings and responsivity after adrenalectomy [ 13]. The authors concluded that the radical decrease in libido in these women following their adrenalectomies was due largely to their near total loss of any source of endogenous androgen production. Despite the fact that these patients were gravely ill and the data were uncontrolled and

BARBARA B. SHERWIN

anecdotal, for over 20 years these studies were frequently cited as evidence of androgen's libido-enhancing properties in women.

III. BENEFITS OF E/A REPLACEMENT

THERAPY A. E n e r g y L e v e l

It has been known for a long time that androgens have anabolic properties. In hypogonadal men treated with testosterone, lean body mass increased by 11% and muscle mass in the upper arm and the thigh increased by 21% and 12%, respectively [14]. There is little question that androgenic steroid administration to prepubertal boys, to hypogonadal men, and to women results in increased nitrogen retention and increased muscle mass. The observation that androgens have energizing properties may be a result of the general anabolic effects of this steroid hormone on the central nervous system. There is a dearth of information on whether exogenous androgen administered along with estrogen to postmenopausal women influences their energy level and sense of wellbeing. The sole study that measured these parameters systematically reported an increase in ratings of energy level and wellbeing in premenopausal women who had had their uterus and ovaries removed and randomly received a combined E/A preparation or androgen alone postoperatively [15]. This did not occur in women who were given estrogen alone or placebo.

B. S e x u a l F u n c t i o n i n g Effects of testosterone on various components of mating behavior have been studied intensively in female nonhuman primates. On the whole, these studies show that the administration of estrogen alone to ovariectomized rhesus monkeys was associated with a decrease in proceptive behavior (i.e., attempts to solicit mounts from the male, a behavioral homology of sexual desire in women). Implantation of minute amounts of testosterone, via a cannula, into the anterior hypothalamus of estrogen-treated ovariectomized and adrenalectomized unreceptive female rhesus monkeys resulted in restoration of their proceptivity without affecting other aspects of sexual behavior, such as attractivity [ 16]. These studies on testosterone and sexual behavior in female nonhuman primates serve to underline two points. One is that there is a specificity of action of testosterone on compoments of sexual behavior such that it enhances proceptivity (the animal's motivation to engage in sexual behavior) but has no effect on its attractivity or its receptivity to males.

CHAPTER43 Estrogen/Androgen Replacement Therapy Second, the fact that a very small dose of testosterone implanted in the hypothalamus was effective in restoring sexual desire in rhesus monkeys [ 16] suggests that testosterone exerts an effect on sexual desire in female rhesus directly on the brain and not via an influence on peripheral tissues. Several correlational studies have tested the association between circulating levels of this sex steroid and aspects of sexual behavior in postmenopausal women. Leiblum and colleagues [ 17] reported that neither estradiol nor testosterone discriminated between sexually active and inactive untreated postmenopausal women, but sexually active women had less vaginal atrophy compared to the inactive women. In a longitudinal study of perimenopausal women, plasma testosterone levels were positively associated with coital frequency [18]. Moreover, a positive correlation occurred between testosterone levels and sexual desire and sexual arousal in premenopausal women over the age of 40 years [19]. Other epidemiological studies that have investigated changes in sexual functioning in peri- and postmenopausal women failed to measure circulating levels of hormones [20,21]. One recent population-based study in middle-aged women failed to find an association between testosterone levels and any aspect of sexual functioning [22]. Another and perhaps more powerful paradigm for investigating the role of testosterone in women involves administering hormone replacement therapy to women who have just undergone total abdominal hysterectomy (TAH) and bilateral salpingo-oophorectomy (BSO). When both ovaries are removed from premenopausal women, circulating testosterone levels decrease significantly within the first 2 4 - 4 8 hr postoperatively [23]. The fact that these women are deprived of ovarian androgen production following this surgical procedure has provided a rationale for administering both estrogen and androgen as replacement therapy. In Britain and Australia, subcutaneous implantation of pellets containing estradiol and testosterone has been used as a treatment for menopausal symptoms for several decades. This route of sex-steroid administration results in a slow constant release of the sex hormones over a period of at least 6 months. Women complaining of decreased libido despite treatment with estrogen received subcutaneous implants of 40 mg estradiol and 100 mg testosterone [24]. Patients reported a significant increase in libido by the third postimplantation month. These findings gained support from a double-blind study of women complaining of loss of libido despite treatment with oral estrogens [25]. They randomly received a subcutaneous implant containing either estrogen alone or estrogen plus testosterone. After 6 weeks, the loss of libido in the estrogen-alone implant group remained, whereas the combined estrogen/testosterone group showed significant symptomatic relief. In a recent prospective, 2-year, single-blind randomized trial, 34 postmenopausal women received either estradiol 50-mg implants or estradiol 50-mg plus testosterone 50-mg

619 implants administered every 3 months [26]. Women who received the combined implant had a significantly greater improvement compared with those receiving estrogen alone, in sexual activity, satisfaction, pleasure, and orgasm. We undertook a series of studies to investigate effects of E/A replacement therapy on aspects of sexual functioning. In one study, premenopausal women who needed to undergo total abdominal hysterectomy and bilateral salpingooophorectomy for benign disease were tested preoperatively [27]. Postoperatively, women randomly received either a combined E/A preparation, estrogen alone, androgen alone, or placebo. These drugs were all depot preparations and were administered intramuscularly once a month. Those who received one of the androgen-containing preparations (E/A or androgen alone) had higher scores on measures of sexual desire, sexual arousal, and number of sexual fantasies compared to women who received estrogen alone or placebo. This occurred despite the fact that estrogen treated women had supraphysiological levels of estrogen [15], had no hot flashes [ 15], and their mood was equally as positive as those who received the combined E/A drug [28]. These findings suggested that testosterone enhanced the cognitive, motivational aspects of sexuality in women such as sexual desire and interest and that it did so directly and not secondary to its enhancement of other aspects of physical and/or psychological functioning. These results were confirmed in a longitudinal study of women who had undergone TAH and BSO 4 years earlier and had been maintained for at least 2 years on either the combined E/A depot preparation or on the same dose of estrogen alone administered intramuscularly [29]. A third group had undergone TAH and BSO and remained untreated. Levels of sexual desire and interest were greatest in the women treated with E/A preparation and covaried with their plasma testosterone level throughout the treatment month as the intramuscular drug was being metabolized. The androgenic enhancement of sexual motivation in women treated with the combined intramuscular drug has been shown to persist with long-term chronic administration of monthly injections that cause an initial surge in estradiol and testosterone levels and metabolize slowly over a period of several weeks [30]. Twenty women who were dissatisfied with their estrogen or estrogen/progestin therapy received a placebo for 2 weeks and then were randomized to either 1.25 mg esterified estrogens combined with 2.5 mg methyltestosterone or to 1.25 mg esterified estrogen alone for 8 weeks [31]. Women who received the combined E/A drug reported improved sexual sensation and sexual desire compared both to their previous estrogen therapy and to the postplacebo baseline assessments, whereas no changes in sexual functioning occured in the women treated with estrogen alone. Taken together, findings from the subcutaneous implant pellet studies and from the prospective studies on surgically menopausal women given an intramuscular E/A preparation

620

BARBARA B. SHERWIN

provide compelling evidence that E/A replacement therapy acts to increase overall energy level over and above estrogen alone and to enhance sexual desire and sexual arousal in postmenopausal women. Frequency of sexual activity was less affected by combined E/A therapy perhaps because its determinants are more complex and involve couple issues. Rather, these findings strongly suggest that, in women, just as in men [32] testosterone has its major impact on the cognitive motivational, or libidinal, aspects of sexual behavior such as desire and fantasies. Moreover, studies on nonhuman primates suggest the likelihood that testosterone exerts this effect on sexual desire via mechanisms that impact directly on the brain rather than by an effect on peripheral tissues [ 16]. Finally, the results of these studies also allow the observation that although estrogen is important for the integrity of vaginal tissues, it does not maintain sexual desire and interest in postmenopausal women.

C. B o n e M e t a b o l i s m Androgen-specific receptors have been demonstrated in osteoblastic cells in women [33] and androgens directly stimulate bone cell proliferation [34]. In a cross-sectional study of hormone levels and bone mineral density (BMD) in premenopausal women, a significant positive correlation was evident between the percentage of free testosterone and BMD, but not between BMD and free estradiol after controlling for body wieght [35]. There is some evidence that testosterone has anabolic effects on protein metabolism. When women receiving 0.625 mg esterified estrogen/day were compared to those receiving the same dose of estrogen plus 1.25 mg/day methyltestosterone (E/A), urinary excretion markers of bone resorption decreased equally in both groups [35]. Whereas the women treated with estrogen alone had a reduction in serum markers of bone formation, those who received the combined E/A drug had an increase in their serum markers of bone formation. There are very few prospective randomized trials of effects of E/A preparations on B MD. In one study, women either received a subcutaneous implant of 50 mg of estradiol or an implant containing the same amount of estradiol plus 100 mg of testosterone [36]. After 3 years, there was no change in BMD in the estrogen-alone group but a 2.5 % increase in B MD occurred in the combined E/A implant group. Unfortunately, in this study, only the metacarpal bone density, which is not at risk for osteoporotic fracture, was measured. In a prospective, randomized study, women received either 1.25 mg esterified estrogen/day or the same amount of estrogen plus 2.5 mg methyltestosterone/day for 2 years [37]. Both treatment regimens prevented bone loss at the spine and hip. However, only the combined E/A drug was associated with a significant increase in spinal bone mineral

density compared with pretreatment baseline. In summary, these findings indicate that whereas exogenous estrogen prevents bone loss in postmenopausal women, the addition of testosterone to an estrogen replacement regimen actually increases bone density.

IV. RISKS OF E/A REPLACEMENT

THERAPY

A. Effects o n L i p o p r o t e i n L i p i d s A meta-analysis of observational studies demonstrated a 50% reduction of heart disease risk in postmenopausal women taking estrogen [38]. It has been estimated that approximately 50% of the cardiovascular risk reduction provided by exogenous estrogen is mediated by lipoprotein changes [39]. Although the coadministration of a progestin, either continuously or cyclicly, attenuates the beneficial effect of estrogen on high-density lipoprotein (HDL) cholesterol, the lipid profile in hormone-treated women is preferable to that in placebo-treated women [40]. The addition of testosterone to an estrogen replacement therapy regimen may partially offset the beneficial effects of estrogen on risk factors for cardiovascular disease. Oral estrogen/androgen replacement reduced both total cholesterol and low-density lipoproteins (LDLs) more than oral estrogen alone [37,41]. However, the addition of testosterone to the regimen also decreased HDL levels relative to estrogen alone, thereby resulting in a detrimental increase in the ratio of HDL/total cholesterol [37,41 ]. However, combined preparations also lower triglyceride levels compared to those with estrogen alone, which is thought to be beneficial. In a prospective, controlled study, 291 surgically menopausal women received either 0.625 or 1.25 mg conjugated equine estrogen (CEE) or 0.625 or 1.25 mg esterified estrogen (EE) combined with 1.25 or 2.5 mg methyltestosterone, respectively [42]. At 6 and 12 months, decreases in triglycerides and HDL cholesterol were observed for both doses of estrogen/androgen, whereas increases were observed in women who were given CEE. Total cholesterol and LDL decreased in all treatment groups by 6 months. Furthermore, no adverse effects on serum creatinine or liver function tests were observed in any treatment group. Studies of nonoral routes of administration of estrogen/ androgen drugs tell a somewhat different story. Subcutaneous implantation of estradiol (40 mg) and testosterone (100 mg) did not cause any changes in cholesterol, triglycerides, or HDL cholesterol from pretreatment levels [43]. Farish and colleagues [44] likewise found that subcutaneous pellets of estradiol (50 mg) and testosterone (100 mg) had no effect on HDL fractions, but testosterone appeared to enhance slightly the LDL cholesterol-lowering effect of estradiol.

CHAPTER43 Estrogen/Androgen Replacement Therapy Following 2 years of treatment with an intramuscular combined estrogen/androgen depot preparation, Sherwin and associates [45] reported that combined replacement therapy did not adversely affect the lipoprotein cholesterol profile in these women compared with patients treated with parenteral estrogen alone and surgically menopausal women who were untreated. Other evidence suggests that the route of administration modulates the response to hormone therapy. For example, percutaneous and vaginal administration of estradiol do not cause the increases in triglycerides and very lowdensity lipoprotein (VLDL) observed during oral therapy [46]. Thus, it seems likely that parental routes of administration of combined estrogen/androgen drugs in the postmenopause may not cause the detrimental effect on HDL seen with oral preparations, as parenteral routes of administration bypass the so-called hepatic first-pass effect. Based on the evidence available to date, it would seem that, compared to exogenous estrogen administered alone, oral estrogen/androgen preparations cause a decrease in HDL cholesterol levels and in triglyceride levels. On the other hand, parenteral routes of estrogen/androgen administration seem not to change lipid and lipoprotein levels compared to effects of parenterally administered estrogen only. However, it must also be acknowledged that estrogens administered via nonoral routes have a lesser beneficial effect on lipid metabolism compared to oral preparations. It also now seems clear that non-lipoprotein-mediated effects of estrogen provide cardioprotection. Examples of such mechanisms are reduced atherosclerotic plaque formation unrelated to HDL cholesterol levels [47], and estrogen-induced vasodilatation [48]. In this regard, it is both interesting and potentially important that treatment of ovariectomized cynomolgus monkeys with oral estrogens improved endothelium-mediated vasodilatation of their atherosclerotic coronary arteries, and the addition of oral methyltestosterone did not alter this response [49]. Moreover, administration of an oral combined estrogen/androgen preparation caused an increase in vaginal blood flow measured with laser doppler velocimetry equal to that which occurred with estrogen alone [50]. This suggests that the addition of androgen to an estrogen replacement regimen does not compromise peripheral blood flow. Moreover, prospective clinical trials that measured blood pressure found no changes in either systolic or diastolic blood pressure in women treated with a combined estrogen/androgen drug compared to pretreatment baseline and compared to women given estrogen only [37,42,44,48-50].

B.

Symptoms of Virilization

There is scanty empirical data from controlled studies on the incidence of hirsutism in postmenopausal women treated with combined estrogen/androgen preparations. A single ex-

621 ception is a recent study that assessed hair growth on the upper lip, chin, and sideburn area using a modified Ferriman-Gallwey scale [42]. The incidence of facial hirsutism was 3% among CEE recipients and 6% for women who had received estrogen and androgen. Interestingly, in this study, which had four treatment groups of different doses of CEE (0.625 and 1.25 mg) and of a combined drug (0.625 and 1.25 mg EE plus methyltestosterone (MT) 1.25 and 2.5 mg, respectively), all the cases of hirsutism (mild or moderate severity) occurred in women taking the high dose of either CEE or of the combined E/A drug. In that study, acne was also reported by 3 % of the women taking the combined E/A drug. Indeed, safety surveillance data on Estratest and Estratest HS (Solvay Pharmaceuticals, Marietta, Georgia)indicate that, of all adverse events reported between 1989 and 1996, the most commonly reported were alopecia (11.1%), acne (5.8%), and hirsutism (4.5%) [51]. Surprisingly, the percentage of adverse events characterized as virilization was similar for both the higher dose (1.25 mg EE + 2.5 mg MT) and the lower dose (0.625 mg EE + 1.25 mg MT) combined drugs. In contrast, our own clinical experience with an intramuscular combined E/A drug suggests that approximately 20% of women who receive 150 mg testosterone enanthate intramuscularly every 4 weeks along with estrogen will develop mild hirsutism manifested by an increased growth of hair on the chin and/or upper lip. When the dose is reduced to 75 mg testosterone enanthate per month, less than 5% of women have increased hair growth. Moreover, hair growth decreases or usually stops entirely when the patient is switched to treatment with estrogen alone. There is good reason to believe that, in women, hirsutism is a dosedependent side effect of exogenous testosterone. Its development would depend also on the amount of estrogen given in combination, because both sex steroids influence the production of sex hormone-binding globulin (SHBG), which, in turn, determines the concentration of free, or biologically available, testosterone. It is also noteworthy that voice alteration constituted only 0.5% of all adverse events reported with Estratest. Neither has deepening of the voice occurred with the doses of the injectible E/A preparation used in our own studies.

C. Effects on E n d o m e t r i a l H i s t o l o g y Of course, the postmenopausal uterus must be protected from excessive estrogenic stimulation during estrogen replacement therapy. The relevant question, therefore, is whether the addition of androgen to an estrogen replacement regimen should be coadministered with more or less progestin than would be used to prevent endometrial hyperplasia with estrogen alone. In a study that compared the effect of oral estrogen alone with an oral combined estrogen/androgen preparation on endometrial histology, similar

622

BARBARA B. SHERWIN

changes in estrogen-stimulated proliferative growth occurred in both groups after 6 months of treatment [35]. The majority of women developed a moderately proliferative endometrial histology pattern and no woman displayed hyperplastic changes. Others have found endometrial hyperplasia both in women treated with estrogen alone and in those given a combined preparation [52]. It would seem, therefore, that the addition of testosterone neither facilitates nor antagonizes the stimulatory effect of estrogens on the endometrium. This suggests that a progestational agent should be added to the hormone regimen when a combined estrogen/ androgen preparation is used.

V. F U T U R E

DIRECTIONS

Although controlled studies on the clinical efficacy of combined estrogen/androgen preparations in postmenopausal women are relatively few in number, the consistency of their findings makes them compelling. All of them demonstrate that the addition of testosterone to an estrogen replacement regimen increases the cognitive aspects of sexuality such as desire and interest, and also heightens energy level and well-being over and above treatment with estrogen alone. These findings support the notion that in women, just as in men, testosterone is important for the maintenance of libido, or sexual desire. Because human sexuality is complex and multidetermined, this statement assumes that physiological levels of testosterone are one determinant of sexual desire in premenopausal women and deficits in behavior may occur sometime after the menopause when the production of testosterone decreases. This is predicated on the findings that exogenous testosterone increased circulating levels of the hormone and restored sexual desire in several studies in both naturally and surgically menopausal women. At the present time, there are no data available to address whether the administration of testosterone to premenopausal women complaining of decreased sexual desire and who have normal ovarian and adrenal function would be efficacious. Although such studies have never been done, it is likely that the sexual dysfunction in healthy premenopausal women would not be hormonally related and testosterone treatment geared toward inducing supraphysiological levels would be unwise. Rather, the available literature supports the opinion that the treatment of postmenopausal women with combined estrogen/ androgen drugs should strive to reinstate physiological levels of both sex hormones when clinically indicated. If it is true that testosterone maintains sexual desire in women, as research findings indicate, then it would be expected that postmenopausal women might experience an even greater loss of libido when treated with estrogen alone. This is because estrogens increase SHBG whereas androgens reduce it, with the consequence being a reduction in bioavailable androgens. When normally postmenopausal women

were given 2 mg/day of oral micronized estradiol, testosterone levels decreased by 42%, dehydroepiandrosterone sulfate (DHEAS) levels fell by 23%, and dehydroepiandrosterone (DHEA) levels decreased by 11% [53]. Free testosterone was reduced even more profoundly because SHBG levels increased 160% following treatment with micronized estradiol. Therefore, the decrease in testosterone production by the ovarian stoma during perimenopause causes a deficiency in testosterone that is compounded in women given exogenous estrogen, which reduces the bioavailability of the already diminished pool of androgens. Unfortunately, most of the studies on combined estrogen/androgen preparations failed to measure plasma levels of hormones or of SHBG, and so it is not known what the optimal dose of each, given in combination, should be in order to maintain SHBG levels within the physiological range. Route of administration of combined estrogen/androgen preparations is also an important consideration in view of the lack of effect of parenteral preparations on the lipid profile compared to oral combined preparations. Ideally, of course, one would want to maintain the beneficial effects of estrogen on HDL while attenuating any HDL-lowering effects of the testosterone. Virilizing side effects of combined estrogen/androgen therapy in the postmenopause occur in some proportion of women. Unfortunately, the occurrence of hirsutism, acne, or alopecia with combined preparations has rarely been studied systematically. Therefore, it is not known what absolute doses of estrogen and testosterone or, perhaps more importantly, what dose ratio of these two hormones would have the least probability of inducing undesirable androgenic effects. Intuitively, a dose of testosterone that induces physiological (and not supraphysiological) levels of circulating testosterone in postmenopausal women would likely minimize the risk of virilizing side effects. Maintaining SHBG levels within the normal female range by balancing the dose of each hormone might also be important in this regard. Finally, it has been suggested that ethnic differences might play a role in determining response to combined estrogen/ androgen therapies. Conventional wisdom holds that women of Mediterranean origin who are dark-haired and darkskinned may be more likely to experience hirsutism compared to light-haired, light-skinned women of Northern European origin following treatment with estrogen/androgen drugs. Once again, this possibility has not been studied systematically and remains within the realm of clinical report. Clinical experience with combined estrogen/androgen drugs has provided some guidelines for identifying the subset of postmenopausal women who might benefit from this therapy. As has already been mentioned, sexual function is multidetermined and it is quite likely that sexual desire and interest may be diminished during the perimenopause secondary to other symptoms that frequently occur at that time such as hot flashes, sleep disturbance, and atrophic vaginitis.

CHAPTER 43 Estrogen/Androgen Replacement Therapy

B e c a u s e e s t r o g e n controls hot flashes and restores the integrity o f the vaginal tissues, e s t r o g e n r e p l a c e m e n t t h e r a p y s h o u l d be first-line t r e a t m e n t for w o m e n c o m p l a i n i n g o f sexual d y s f u n c t i o n s at the time o f m e n o p a u s e . S h o u l d the c o m p l a i n t persist after other s y m p t o m s h a v e abated f o l l o w ing t r e a t m e n t with e s t r o g e n alone, then t r e a t m e n t with a c o m b i n e d p r e p a r a t i o n s h o u l d be attempted. A w o m a n ' s prem e n o p a u s a l history is also relevant in treatment. W o m e n w h o report that a d e c r e a s e in sexual desire c o i n c i d e d in t i m e with the h o r m o n a l c h a n g e s that c h a r a c t e r i z e the m e n o p a u s e are m o r e likely to r e s p o n d s u c c e s s f u l l y to a c o m b i n e d drug. In contrast, w o m e n w h o r e p o r t a l o w level o f sexual desire or interest that is lifelong are less likely to r e s p o n d c o m p l e t e l y to a h o r m o n a l i n t e r v e n t i o n alone and m a y require marital or c o u p l e t h e r a p y in addition to h o r m o n e treatm e n t to r e s o l v e the p r o b l e m . Finally, s o m e p o s t m e n o p a u s a l w o m e n w h o are b e i n g treated with e s t r o g e n c o m p l a i n o f severe fatigue that i m p e d e s t h e m f r o m c a r r y i n g out their norm a l activities. T h e s e w o m e n m a y also benefit f r o m c o m b i n e d e s t r o g e n / a n d r o g e n t h e r a p y a l t h o u g h they n e e d to be i n f o r m e d that the t r e a t m e n t m a y also stimulate sexual feelings that m a y not be w e l c o m e d by some. T h e r e is g r o w i n g a c c e p t a n c e o f the fact that c o m b i n e d e s t r o g e n / a n d r o g e n drugs are a useful addition to the a r m a m e n t a r i u m o f t r e a t m e n t options for p o s t m e n o p a u s a l w o m e n . C o m b i n e d p r e p a r a t i o n s h a v e a s u p e r i o r efficacy c o m p a r e d to e s t r o g e n alone on i n c r e a s i n g b o n e density, on the alleviation o f fatigue, and in r e s t o r i n g sexual desire and interest in postm e n o p a u s a l w o m e n . H o w e v e r , m u c h w o r k r e m a i n s to be d o n e to d e t e r m i n e the safest and m o s t effective doses and r o u t e o f a d m i n i s t r a t i o n o f such p r e p a r a t i o n s as well as their p o s s i b l e l o n g - t e r m effects.

References 1. Greenblatt, R. B. (1942). Androgenic therapy in women. J. Clin. Endocrinol. 2, 665-666. 2. Shorr, E., Papanicolaou, G. N., and Stimmel, B. F. (1938). Androgens in postmenopausal women. Proc. Soc. Exp. Biol. Med. 38, 759-768. 3. Carter, A. C., Cohen, E. J., and Shorr, E. (1947). The use of androgens in women. Vitam. Horm. (N. Y.) 5, 317-391. 4. Groome, J. R. (1939). Androgens in women. Lancet 2, 722-724. 5. Silberman, D., Radman, H. M., and Abarnel, A. R. (1940). Hormone therapy for menopause. Am. J. Obstet. Gynecol. 39, 332-338. 6. Greenblatt, R. B, Mortara, F., and Torpin, R. (1942). Sexual libido in the female. Am. J. Obstet. Gynecol. 44, 658-663. 7. Kupperman, H. S, and Studdiford, W. E. (1953). Endocrine therapy in gynecologic disorders. Postgrad. Med. 14, 410- 425. 8. Greenblatt, R. B, Barfield, W. E, Garner, J. F, Calk, G. L, and Harrod, J. P. (1950). Evaluation of an estrogen, androgen, estrogen-androgen combination and a placebo in the treatment of the menopause. J. Clin. Endocrinol. 10, 1547-1558. 9. Caldwell, B. M, and Watson, R. (1952). An evaluation of psychologic effects of sex hormone administration in aged women. J. Gerontol. 7, 228-244. 10. Foss, G. L. (1951). The influence of androgens on sexuality in women. Lancet 1, 667-669.

623 11. Kennedy, B. J. (1973). Effects of massive doses of sex hormones on libido. Med. Aspects Hum. Sex 7, 67-75. 12. Waxenberg, S. E, Drellich, M. G, and Sutherland, A. M. (1959). Changes in female sexuality after adrenalectomy. J. Clin. Endocrinol. 19, 193-202. 13. Drellich, M. G, and Waxenberg, S. E. (1966). Erotic and affectional components of female sexuality. In "Science of Psychoanalysis" (J. Masserman, ed.), pp. 45-55. Grune & Stratton, New York. 14. Bhasin, S., Storer, T., Strakova, J., Phillips, J., Phillips, C., Berman, N., Bunnell, T., and Casaburi, R. (1994). Testosterone increases lean body mass, muscle size and strength in hypogonadal men. Clin. Res. 42, 74A. 15. Sherwin, B. B, and Gelfand, M. M. (1985). Differential symptom response to parenteral estrogen and/or androgen administration in the surgical menopause. Am. J. Obstet. Gynecol. 151, 153-160. 16. Everitt, B. J., and Herbert, J. (1975). The effects of implanting testosterone propionate in the central nervous system on the sexual behavior of the female rhesus monkey. Brain Res. 86, 109-120. 17. Leiblum, S., Bachmann, G., Kemmann, E., and Colburn, D. (1983). Vaginal atrophy in the postmenopausal woman: The importance of sexual activity and hormones. JAMA, J. Am. Med. Assoc. 249, 21952198. 18. McCoy, N., and Davidson, J. (1985). A longitudinal study of the effects of menopause on sexuality. Maturitas 7, 203-210. 19. F16ter, A., Nathorst-B66s, J., Carlstr6m, K., and von Schoultz, B. (1997). Androgen status and sexual life in perimenopausal women. Menopause 4, 95- 100. 20. Dennerstein, L., Smith, A. M. S., and Morse, C. A. (1994). Sexuality and the menopause. J. Psychosom. Obstet. Gynaecol. 15, 59-66. 21. Hallstrom, T. (1977). Sexuality in the climacteric. Clin. Obstet. Gynecol. 4, 227-239. 22. Dennerstein, L., Dudley, E. C., Hopper, J. L., and Burger, H. (1997). Sexuality, hormones and the menopausal transition. Maturitas 26, 83-93. 23. Longcope, C. (1981). Metabolic clearance and blood production rates of estrogen in postmenopausal women. Am. J. Obstet. Gynecol. 111, 779-785. 24. Burger, H. G, Hailes, J., Manelaus, M., Nelson, J., Hudson, B., and Balazs, N. (1984). The management of persistent menopausal symptoms with oestradiol testosterone implants: Clinical, lipid and hormonal results. Maturitas 6, 351-358. 25. Burger, H. G., Hailles, J., Nelson, J., and Menelaus, M. (1987). Effects of combined implants of estradiol and testosterone on libido in postmenopausal women. Lancet 294, 936-937. 26. Davis, S. R., McClaud, P., Strauss, B. J. G., and Burger, H. (1995). Testosterone enhances estradiol's effects on postmenopausal bone density and sexuality. Maturitas 21, 227-236. 27. Sherwin, B. B., Gelfand, M. M., and Brender, W. (1985). Androgen enhances sexual motivation in females: A prospective cross-over study of sex steroid administration in the surgical menopause. Psychosom. Med. 7, 339-351. 28. Sherwin, B. B. (1988). Affective changes with estrogen and androgen replacement therapy in surgically menopausal women. J. Affective Disord. 14, 177-187. 29. Sherwin, B. B., and Gelfand, M. M. (1987). The role of androgen in the maintenance of sexual functioning in oophorectomized women. Psychosom. Med. 49, 397-409. 30. Sherwin, B. B. (1985). Changes in sexual behavior as a function of plasma sex steroid levels in postmenopausal women. Maturitas 7, 225-233. 31. Sarrel, P., Dobay, B., and Wiita, B. (1998). Estrogen and estrogenandrogen replacement in postmenopausal women dissatisfied with estrogen-only therapy. J. Reprod. Med. 43, 129-134. 32. Bancroft, J., and Wu, F. C. W. (1983). Changes in erectile responsiveness during androgen replacement therapy. Arch. Sex. Behav. 12, 59-66.

624 33. Colvard, D. S., and Eriksen, E. F., Keeting, E E. Wilson, E. M., Lubahn, D. B., French, F. S., Riggs, B. L., and Spelsberg, T. C. (1989). Identification of androgen receptors in normal human osteoblast-like cells. Proc. Natl. Acad. Sci. U.S.A. 86, 854-857. 34. Kasperk, C. H., Wergedal, J. E., Farley, J. R., Linkhart, T. A., Turner, R. T., and Baylink, D. G. (1989). androgens directly stimulate proliferation of bone cells in vitro. Endocrinology (Baltimore) 124, 15761578. 35. Raisz, L. G., Wiita, B., and Arctic, A. (1996). Comparison of estrogen alone and estrogen plus androgen on biochemical markers of bone formation and resorption in postmenopausal women. J. Clin. Endocrinol. Metab. 81, 37-43. 36. Barlow, D. H., Abdalla, H. I., Roberts AD. (1986). Long-term hormone therapy: Hormonal and clinical effects. Obstet. Gynecol. 67, 321325. 37. Watts, N. B., Notelovitz, M., Timmons, M. C., Addison, W. A., Wiita, B., and Downey, L. J. (1995). Comparison of oral estrogens and estrogens plus androgen on bone mineral density, menopausal symptoms, and lipid-lipoprotein profiles in surgical menopausal women. Obstet. Gynecol. 85, 529-537. 38. Grady, D., Rubin, S. M., Petitti, D. B., Fox, C. S., Black, D., Ettinger, B., Ernster, V. L., and Cummings, S. R. (1992). Hormone therapy to prevent disease and prolong life in postmenopausal women. Ann. Intern. Med. 117, 1016-1037. 39. Bush, T. L., Barrett-Connor, E., Cowan, L. D., Criqui, H. H., Wallace, R. B., Suchindran, C. M., Tyroler, H. A., and Rifkind, B. M. (1987). Cardiovascular mortality and noncontraceptive use of estrogen in women: Results from the Lipid Research Clinics Program Follow-up Study. Circulation 75, 1102-1109. 40. Writing Group for the PEPI Trial (1995). Effects of estrogen or estrogen/progestin regimens on heart disease risk factors in postmenopausal women. JAMA, J. Am. Med. Assoc. 273, 199-208. 41. Hickok, L. R., Toomey, C., eroff, L. (1993). A comparison of esterified estrogens with and without methyltestosterone: Effects on endometrial histology and serum lipoproteins in postmenopausal women. Obstet. Gynecol. 82, 919-924. 42. Barrett-Connor, E., Timmons, C., Young, R., and Wiita, B. (1996). Interim safety analysis of a two-year study comparing oral estrogen-

BARBARA B. SHERWIN

43.

44.

45.

46.

47.

48. 49.

50.

51.

52.

53.

androgen and conjugated estrogens in surgically menopausal women. J. Women's Health 5, 593-602. Teran, A. Z., and Gambrell, R. D., Jr. (1988). Androgens in clinical practices. In "Androgens in the Menopause" (L. Speroff, ed.), pp. 1422. McGraw-Hill, New York. Farish, E., Fletcher, C. D., Hart, D. M. Azzawi, E. A., Abdalla, H. I., and Gray, C. E. (1984) The effect of hormone implants on serum lipoproteins and steroid hormones in bilaterally oophorectomized women. Acta Endocrinol. (Copenhagen) 106,116-123. Sherwin, B. B., Gelfand, M. M., Schucher, R., and Grabor, J. (1987). Postmenopausal estrogen and androgen replacement and lipoprotein lipid concentrations. Am. J. Obstet. Gynecol. 156, 414-419. Mischell, D. R., Jr., Moore, R. E., Roy, S., Brenner, C. F., and Cage, M. A. (1978). Clinical performance and endocrine profiles with contraceptive vaginal rings containing a combination of estradiol and dnorgestrel. Am. J. Obstet. Gynecol. 130,155-161. Adams, M. R., Clarkson, T. B., Koritnik, D. R., and Nash, H. A. (1987). Contraceptive steroids and coronary artery atherosclerosis of cynomolgus macaques. Fertil. Steril. 47, 1010-1018. Sarrel, P. (1990). Ovarian hormones and the circulation. Maturitas 590, 287-298. Honor6, E. H., Williams, J. K., Adams, M. R., Ackerman, D. M., and Wagner, J. D. (1996). Methyltestosterone does not diminish the beneficial effects of estrogen replacement therapy on coronary artery reactivity in cynomolgus monkeys. Menopause 3, 20-36. Sarrel, P. (1998). Ovarian hormones and vaginal blood flow: Using laser Doppler velocimetry to measure effects in a clinical trial of postmenopausal women. Int. J. Impotence Res. 10, $91-$93. Phillips, E., and Bauman, C., (1997). Safety surveillance of esterified estrogens - methyltestosterone (Estratest-Estratest HS ) replacement therapy in the United States. Clin. Ther. 19, 1070-1084. Gelfand, M. M., Ferenczy, A., and Bergeron, C. (1989). Endometrial response to estrogen-androgen stimulation. In "Menopause Evaluation Treatment and Health Concerns" (C. B. Hammond, F. B. Haseltine, and I. Seniff, eds.), pp. 29-40. Alan R. Liss, New York. Casson, P. R., Elkind-Hirsh, K. E., Buster, J. E., Hornsby, E J., Carson, S. A., and Snables, M. C, (1997). Effect of postmenopausal estrogen replacement on circulating androgens. Obstet. Gynecol. 90, 995-998.

~HAPTER 4 ~

DHEA: Biology and Use Therapeutic Intervention JOHN E.

BUSTER AND P E T E R R. CASSON 9

IV. DHEA Replacement V. Conclusions References

I. Introduction

II. Androgens: Origin and Control III. DHEA, DHEAS, and Androgen Metabolites: Increase during Childhood and Decline with Age


elderly individuals could retard maladies of age, including cardiovascular disease, neoplasia, diabetes, immunosenescence, osteoporosis, muscular wasting, decreased libidinal interest, and depression [6-8]. Despite burgeoning interest in this field, it is not clear when or how to administer DHEA to menopausal women. Oral DHEA is nontheless widely consumed in the United States as a food supplement. This chapter reviews work describing the origin and regulation of DHEA and DHEAS, the fate of their downstream metabolites, and the impact of declining androgen production with age. It further examines conditions that accelerate age-related decline of DHEA and DHEA sulfate production and examines the case for replacing DHEA in elderly women.

Dehydroepiandrosterone (DHEA) and dehydroepiandrosterone sulfate (DHEAS) are 19-carbon steroids originating principally from the zona reticularis of the human adrenal cortex. They circulate in abundance and have production rates far higher than any other circulating steroid. Ubiquitous tissue sulfatases rapidly interconvert DHEAS to DHEA, which has a higher clearance rate and shorter halflife than DHEAS. DHEA has virtually no androgenic activity of its own but is converted intracellularly to bioactive androgens and estrogens. Measurement of circulating DHEAS, a stable clinical marker of adrenal androgen secretion, provides a rough index of the circulating pool of available prohormone [1-3]. Circulating DHEAS concentrations and production decline with advancing age. In the elderly, DHEAS concentrations and production are about 10% of reproductive age peaks. This decline, sometimes called "adrenopause," occurs concurrently with involution of the zona reticularis of the adrenal cortex [1-5]. Adrenopause represents a senescent endocrine deficiency similar to menopause. If it is truly a deficiency state, restitution of adrenal androgen levels in

MENOPAUSE: BIOLOGY AND PATHOBIOLOGY

Department of Obstetrics and Gynecology, Baylor College of Medicine, Houston, Texas 77030

II. ANDROGENS: ORIGIN AND C O N T R O L The literature traditionally identifies five androgens (Table I) as clinically important: DHEAS, DHEA, androstenedione ( A 4 A ) , testosterone (T), and dihydrotestosterone (DHT). In women, these androgens have widely differing

625


626

BUSTER AND CASSON

Anter,oPitu,taryr 1Lt

TABLE I Androgen Concentrations and Relative Biological Potency in Reproductive-Age Women Using Testosterone as a Standard of 1.0 a

Androgen

Relative potency bioas say

Dihydrotestosterone Testostrone Androstenedione DHEA DHEAS

5 1 0.1 0.01 0.001

DHEA-S

>90~

/ 7.r A

+ACTH+c~176~ AnteriOrPituitary ~ 9 DHEA-S

, Adrenal Gland

serum concentrations, production rates, potencies, and origins (Table I). Although the basic 19-carbon (C~9) androstane nuclear structure is held in common (Fig. 1), effects on target tissues differ. As examples, DHEAS is associated with immunomodulation, enhancement of insulin effect, and osteoporosis protection, whereas libidinal drive, sex hair development, seborrhea, and acne are more associated with T [6-10]. Interconversions between DHEAS, T, and other androgens blur these distinctions [9-12]. The sources, regulation, and interconversion of these five androgens, as influenced by aging and menopausal status, are depicted in Figs. 2 and 3. They are described in detail in the following discussions.

A. Dehydroepiandrosterone and Dehydroepiandrosterone Sulfate DHEA and DHEAS, collectively designated DHEA(S), are prohormones without known receptors or specific target tissues. DHEAS and DHEA are secreted daily in milligram amounts by the zona reticularis (ZR) of the human adrenal cortex [11]. The ZR, the sole secretory source of DHEAS, is unique to humans and higher primates [4,13]. At 8 to

18 16 14

2 4

15

6

FIGURE 1 Androstane structural nucleus (19 carbons) common to the five clinically important androgens.

~ LH~

)-Estradi

a Adapted from Casson and Carson [6].

19 ~

Ovary

ANDROSTENEDIONE ") 5o% so*/. TESTOSTERONE ) 100% (-~-HYDROTESTOSTERONE)

Gland

Representative concentration ng/dl ng/dl ng/dl ng/dl ng/dl

DHEA

co

Adrenal

~, +ANDROSTENEDIONE )

Menopause: Biology and Pathobiology

Nitric Oxide: Biology and Pathobiology

The Liver: Biology and Pathobiology

Nitric Oxide Biology and Pathobiology

The Liver: Biology and Pathobiology 5th Edition

Menopause and Culture

Fish Parasites: Pathobiology and Protection

Menopause and Culture

Understanding Menopause

Fish Parasites: Pathobiology and Protection

Fish Parasites: Pathobiology and Protection

Fish Parasites: Pathobiology and Protection

Portal Hypertension: Pathobiology, Evaluation, and Treatment

Menopause (Herbal Health)

Menopause (At Your Fingertips)

Menopause For Dummies

Hormone Replacement Therapy and the Menopause

What Nurses Know ... Menopause

Menopause For Dummies

Menopause: Natural & Medical Solutions

Menopause: Endocrinology and Management (Contemporary Endocrinology)

Complete Idiot's Guide to Menopause

Hot and bothered: women, medicine, and menopause in modern America

The Complete Idiot's Guide to Menopause

Systems Biology and Synthetic Biology

Hot and Bothered: Women, Medicine, and Menopause in Modern America

The Only Menopause Guide You'll Need

The Complete Idiot's Guide to Menopause

Menopause: A Mental Health Practitioner's Guide

Menopause: Biology and Pathobiology

Nitric Oxide: Biology and Pathobiology

The Liver: Biology and Pathobiology

Nitric Oxide Biology and Pathobiology

The Liver: Biology and Pathobiology 5th Edition

Menopause and Culture

Fish Parasites: Pathobiology and Protection

Menopause and Culture

Understanding Menopause

Fish Parasites: Pathobiology and Protection

Fish Parasites: Pathobiology and Protection

Fish Parasites: Pathobiology and Protection

Portal Hypertension: Pathobiology, Evaluation, and Treatment

Menopause (Herbal Health)

Menopause (At Your Fingertips)

Menopause For Dummies

Hormone Replacement Therapy and the Menopause

What Nurses Know ... Menopause

Menopause For Dummies

Menopause: Natural & Medical Solutions

Menopause: Endocrinology and Management (Contemporary Endocrinology)

Complete Idiot's Guide to Menopause

Hot and bothered: women, medicine, and menopause in modern America

The Complete Idiot's Guide to Menopause

Systems Biology and Synthetic Biology

Hot and Bothered: Women, Medicine, and Menopause in Modern America

The Only Menopause Guide You'll Need

The Complete Idiot's Guide to Menopause

Menopause: A Mental Health Practitioner's Guide

Recommend Documents