| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Departments of Epidemiology (M.R.S., H.Z., D.M., S.H.) and Biostatistics (B.N.), School of Public Health, University of Michigan, and Department of Obstetrics and Gynecology (J.F.R.), University of Michigan Health Sciences System, Ann Arbor, Michigan 48109
Address all correspondence and requests for reprints to: MaryFran Sowers, Ph.D., Department of Epidemiology, School of Public Health, University of Michigan, 109 Observatory, Room 1846, Ann Arbor, Michigan 48109. E-mail: mfsowers{at}umich.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Setting/Participants: Participants were the Michigan Bone Health and Metabolism Study cohort of 629 women, aged 24–44 yr (in 1992/3), with 5757 annual FSH data points over a 14-yr period.
Design/Main Outcome Measures: The study was designed to relate acceleration/deceleration patterns in FSH rate of change to time to final menstrual period (FMP) and chronological age using nonparametric and piecewise regression modeling.
Results: Four major FSH stages, based on rate of FSH change patterns, were identifiable in relation to the FMP. In FSH stage 1, the rate of FSH change increased modestly up to –7 yr prior to the FMP; in FSH stage 2 (–7 to –2 yr prior to FMP), there was a major acceleration in FSH rate of change. FSH stage 3 had an acute increase in FSH rate of change (–2 to +1 yr around the FMP), with average FSH level of 34 mIU/ml. The fourth, or plateau, FSH stage began at 1 yr after FMP when the average FSH level was 54 mIU/ml. During the yr 28–60, there were eight age-specific epochs defined by significant changes of FSH trajectory accelerations or decelerations and rate of change.
Conclusions: Four menopause transition stages bounding the FMP and eight epochs in chronological aging from age 28 to 60 yr were defined by changes of FSH trajectory accelerations/decelerations and rates of change. This timing information, combined with knowledge of FSH levels and menstrual cycle characteristics, can help discern the likely status of women with respect to their reproductive viability and menopause transition stage.
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Intracycle and intercyle variations in FSH (4, 5) and the inhibins (6) have been well characterized and are associated with the selection of a dominant follicle in the ovulatory process (7, 8, 9). However, current reports of FSH levels are limited in their ability to describe the natural history of both diminished ovarian reserve and the stages of the menopause (10). Although Burger et al. (11) was among the first to use longitudinally acquired data to describe FSH levels in relation to chronological age and the final menstrual period (FMP), these investigators reported data from women who were aged 45–55 yr at their first annual examination, precluding their ability to describe women in their late reproductive years with increasingly diminished ovarian reserve and adequately link this information to define stages of the menopause transition.
This report addresses the natural history of change in FSH levels through the late reproductive years and into the menopause stages using FSH data from women who were aged 24–44 yr at their initial annual examination and who have been followed up with annual assessments from 1992–1993 through 2006–2007. We characterized acceleration and deceleration of FSH rates of change, identifying patterns that could be aligned with reproductive senescence and menopause stages. We further described FSH patterns in relation to chronological age and time to FMP and then considered whether smoking behavior, parity, and age at menarche altered these patterns.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The Michigan Bone Health and Metabolism Study is a population-based longitudinal natural history study of reproductive hormones and their relation to the initiation and development of musculoskeletal and metabolic diseases and functional limitations (12, 13). The study was implemented in a cohort of Caucasian women during their young and midadulthood. The 664-woman sample was identified from two sampling frames, the family records of Tecumseh (Michigan) Community Health Study from 1959 to 1985 and a 1992 Tecumseh community listing. In 1992 more 80% of the female Tecumseh offspring, aged 24–44 yr, were recruited from the family records listing (n = 543). Also recruited were 121 women listed in Kohl’s Directory (91%), aged 24–44 yr, who had become Tecumseh community residents after the health study was conducted. The baseline Michigan Bone Health and Metabolism Study eligibility requirements were age (24–44 yr), listing in the sampling frames, and sufficient mobility to attend a research clinic located in the community.
This report includes data from the 14-yr period from 1992–1993 through 2006–2007. There were funding lapses in 1997 and 2003 for 18 and 14 months, respectively, during which neither data nor specimens were collected. For this report, 629 women contributed at least one data point to the 5757 FSH data points available for the longitudinal data analyses. On average, participants contributed more than nine annual FSH data points of a total possible 11 annual points. Blood was not drawn (thereby precluding hormone analyses) if participants were pregnant or lactating at the time of the annual visit, recognizing that lactational amenorrhea is associated with overshoot in FSH levels (14). Over time, 2–7% of women had only interview data available at any given year because participants lived more than 2.5 h from the research clinic, were too ill to contribute to an in-person visit with phlebotomy, or refused phlebotomy (<0.5%). Data were censored at time of death for the 14 participants who have died since the cohort inception (2%).
This study was approved by the University of Michigan Institutional Review Board and informed consent was obtained from all participants.
Measures of menopausal transition status
Menopause status was based on the regularity of menstrual bleeding in the year before the study visit. A woman was classified as premenopausal if she had no increase in menstrual irregularity in the previous year. Perimenopause was defined as having menstrual irregularity and having nine or fewer menstrual cycles in a 12-month time period. Postmenopause was characterized as having at least 12 consecutive months of amenorrhea associated with no other medical cause. FMP was defined retrospectively as 12 months of amenorrhea with no alternative physiologically normal explanation such pregnancy or lactation.
Surgical menopause, including hysterectomy and oophorectomy, was verified by medical record abstraction. Hormone therapy and oral contraceptive use were assessed at each visit. Information included preparation components and duration of use. For purposes of these analyses, data from women with hysterectomy/oophorectomy were censored at the time of surgery and data at time of hormone therapy use were censored for that time point.
Sex steroid hormones
Blood and urine specimens were collected in a fasting state on d 2–7 of the follicular phase of the menstrual cycle. If a woman was postmenopausal or sufficiently advanced into the menopause transition that phlebotomy could not be linked to a menstrual bleed, specimens were collected on the anniversary of her study enrollment ± 15 d. Specimens were aliquoted and stored at –80 C without thaw until assay. Serum FSH concentrations were measured with a two-site chemiluminescence (sandwich) immunoassay, which uses constant amounts of two antibodies that have a specificity for the intact FSH molecule a polyclonal sheep anti-FSH antibody labeled with acridinium ester and a monoclonal mouse anti-FSH antibody covalently coupled to paramagnetic particles. Separation, aspiration and deionized water wash steps separate bound from free. A direct relationship exists between the amount of FSH in the sample and the relative light units detected by the luminometer (photomultiplier tube). The test results are determined from a calibration curve using standards obtained from Bayer Diagnostics, which are referenced to the World Health Organization second international reference preparation 78/549 standard. The coefficients of variation (percent) at locations along the standard curve (in parentheses) were as follows: 7.8% (3.3 mIU/ml), 3.2% (9.9 mIU/ml), 5.1% (18.2 mIU/ml), 4.4% (22 mIU/ml), and 3.3% (60.8 mIU/ml). The lower limit of detection was 1.05 mIU/ml.
Other measures
Height (centimeters) and weight (kilograms) were measured at each annual study visit with a stadiometer and balance-beam scale, respectively. Body mass index was calculated by dividing the weight (kilograms) by height (meters) squared. Body composition was measured using bioelectrical impedance. Women were interviewed annually about selected aspects of their personal behavior and reproductive history. Smoking history and practice was ascertained annually. Participants were classified as never, past, or current smokers. Parity was described based on the number of live births over 28 wk of age. Age at menarche was self-reported.
Data analysis
Variable distributions were examined for normality, the presence of nonplausible outliers, and/or changing variability over time. Univariate analyses were used to decide whether transformations of outcome measures were necessary for satisfying model assumptions such as normality and constant variance. In analyses, FSH values were log transformed (natural) but back transformed to facilitate ease of communication.
A multiple-step process, more fully described in Appendix I (supplemental data, published as supplemental data on The Endocrine Society’s Journals Online Web site at http://jcem.endojournals.org), was used to organize FSH rates of change into stage or epochs related to the FMP and chronological age. In the first step, because relationships between FSH values and time to FMP or chronological age could not be appropriately modeled by incorporating quadratic or cubic terms, a nonparametric stochastic mixed model was fit (15). The 95% confidence intervals were fit around the mean FSH values in relation to time to FMP and chronological age. In the second step, to estimate the rate of change and acceleration, which are the first- and second-order derivatives of log(FSH), respectively, we solved differential equations associated with the cubic spline function. The curvature of mean profile of logFSH over time, representing the degree of bend in the line, was approximated by integrating both rate of change and acceleration/deceleration. The 95% confidence bands of these characteristics were obtained using bootstrapping of 100 samples. In the third step, and informed by steps one and two, data were organized into epochs by setting nodes for piecewise linear mixed models using inflection points defined by differentiating the cubic spline smoothing functions (16). This estimated the mean rate of change at each segments in relation to time to FMP and chronological age. The population mean and SEs for logFSH at turning time points were presented in backtransformed scale for clinical understanding using 
-method Taylor series approach.
Analyses were implemented in SAS version 9.1 and SAS macro language (SAS Institute, Cary, NC), or Matlab7.0 (The MathWorks, Inc., Natick, MA).
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Other characteristics of the population are shown in Table 1
. Not surprisingly, over the time period, the number of women who were premenopausal (and not using exogenous hormones) declined from 70.6% at baseline to 23.8% at the 2006–2007 visit; surgical menopause frequency changed from 4.4% of the cohort at baseline to 20.2% at the 2006–2007 visit. Almost 15% of the cohort remained nulliparous across time and more than half of the cohort reported never smoking cigarettes.
|
Figure 1 shows the rising mean population logFSH levels in relation to the FMP with 95% confidence intervals (CI). Four different stages were identifiable in relation to time to the FMP using measures of acceleration and deceleration and piece-wise modeling of FSH rates of change (see Fig. 2). Tables showing the regression β-coefficients and their 95% confidence intervals are shown in the Appendix.
|
|
.
|
Patterns of FSH with chronological age
Figure 3 shows the mean population values of FSH in relation to chronological age with 95% confidence intervals. Using nonparametric stochastic modeling and piece-wise linear mixed modeling, we identified eight epochs of accelerations and decelerations in the rates of FSH change over the chronological age period from 28 to 60 yr. Each modeling approach identified similar patterns (see Table 3) as demonstrated by a relative difference in FSH levels of less than 3% between the two methods at matching time points. The eight distinct chronological age epochs in FSH rates of change identified between ages 28 and 60 were 28–33, 33–40, 40–42, 42–45, 45–50, 50–52, 52–55, and 55–60 yr.
|
|
. A marked shift in the population mean FSH level is notable at the node for the age 45–50 yr segment in which the mean FSH value at the beginning of the segment was 11.1 mIU/ml (SE = 0.44). At the node for the age 50–52 yr segment, the mean FSH was 22.7 mIU/ml (SE = 1.04) at the beginning of the segment, whereas at the node for the age 52–55 yr segment, the mean FSH was 35.9 mIU/ml (SE = 1.96). At the node for the age 55–60 segment, the mean FSH was 59.8 mIU/ml (SE = 4.19) and, at the age 60+ yr segment, the mean FSH was 67.6 (SE = 11.23), respectively.
Figure 4 shows the rates of change at different ages as well as the rates of acceleration or deceleration in those rates of change according to chronological age. The mean rates of change at each stage are shown in Table 3 and Appendix. The age segments that were associated with statistically significant accelerations greater than zero (P < 0.05) were observed at ages 40, 42, 45, and 50 yr. After age 40 yr, logFSH increased rapidly and then achieved maximum rate of change between ages 50 and 52 yr. At age 52 yr, the FSH rate of change was more than 4 times faster than the rate of change observed in the stable period when women were in their mid-30s. After age 55 yr, the rate of change in logFSH was slightly positive but no longer statistically significant (P = 0.55).
|
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
FSH has long been considered a candidate biomarker for describing reproductive and menopause transition stages (18) for two reasons. First, FSH stimulates folliculogenesis, a key in the dynamics of ovarian aging. Parallel with the age-related decline in the oocyte quality and quantity (19), there are progressively higher follicular phase FSH levels observed (20) in older ovulatory women, compared with younger ovulatory women, believed to be due to the diminished restraint associated with declining inhibin-B levels, the dominant inhibin of the small antral follicles (21, 22, 23, 24). Second, the rise in FSH levels precedes significant declines in ovarian steroid secretion, including that of estradiol (25).
Although the centrality of FSH to the ovulatory process has been long acknowledged and FSH cut points have been used clinically to characterize subfertility as a criterion to initiate treatment, a clinical interpretation of actual FSH values in relation to menopause transition staging has been a source of concern. The well-known pulsatile, intracycle, and intercycle variation in FSH levels make it difficult to interpret values without referencing biological landmarks including menstrual bleeding, ovulation, or the LH surge. This measurement issue is coupled with the understanding that FSH levels are really an indirect indicator reflecting the diminishing restraint of inhibin-B levels of the antral follicles (26, 27) and not a direct measure of follicle quantity or quality.
This study focuses on assigning stages based on rates of FSH change rather than absolute FSH levels and then in turn identifies mean ages and amplitudes of FSH associated with intervals with accelerating or decelerating rates of change. If one would transfer the staging vocabulary currently applied to menstrual cycle-defined staging (which is also based on variability), FSH stage 1 would represent the premenopause and be characterized by minimal change in variability of annual FSH changes. In FSH stage 2, there was evidence of a major acceleration of FSH rate of change, potentially reflecting an increasing decline in oocyte quality and quantity as more directly assessed by antimullerian hormone (AMH) and inhibin-B (28, 29). In many of the current naming conventions, this might be considered the early perimenopause and according to our observations, this stage would last, on average, 5 yr. In FSH stage 3, there was an acute acceleration in the FSH rate of change, which may coincide with the frequent failure to establish a dominant follicle and subsequent corpus luteum and deterioration in the reciprocal relationship of FSH with inhibin-B and inhibin-A, as reviewed by Welt et al. (24). This stage, which might be labeled the late perimenopause, lasted, on average, at least 3 yr including approximately 2 yr before the FMP and 1 yr after the FMP. Finally, in FSH stage 4, commencing 1 yr after the FMP, there was a plateau in FSH variability, indicating that folliculogenesis was no longer a viable physiological process. It is important to note that these FSH stages were developed in relation to a menstrually defined criterion, the FMP; a one-to-one correspondence between endocrine-based classifications of the menopause transition with a menstrual bleeding-based classification should not be expected because menstrual bleeding reflects an extensive uteroovarian biology.
Based on published (6, 20, 29) describing the declines of AMH and inhibin-B, it is possible to attribute the increasing rate of change in FSH stage 2 to the loss of restraint by declining inhibin-B levels. However, a definitive explanation of the acute change in FSH in the time period 2 yr before the FMP has not been satisfactorily articulated, although there are several possible explanations. First, this acute increase in FSH rate of change may be related to a decline in the production of inhibin-A from the corpus luteum. A cross-sectional (30) and longitudinal study (11, 29) suggest that inhibin-A levels eventually decline as women approach the menopausal transition; however, there has been no characterization of an inhibin-A threshold or rate of decline permissive for the ultimate loss of restraint on FSH regulation leading to this observed acute increase in FSH rate of change. A recent study relating inhibin-A levels in women classified into various menopause transition states reported, not unexpectedly, that inhibin-A levels measured in the follicular phase were at or near the limits of assay detection and noninformative (31). Another regulator of FSH is activin A, but recent evidence does not indicate that activin A regulates this acute FSH change (20, 31). Alternatively, the loss of regulation could be a function of hypothalamic-pituitary aging rather than the singular loss of control via ovarian-related peptides (32).
Chronological aging represents a population average of the many individual trajectories leading to FMP at variable ages. However, there is additional information to be learned from considering FSH levels in relation to chronological age. For example, these data showed the remarkable stability achieved in FSH levels to age 33 yr after which time there is a slight increase in the level and variability until age 40 yr after which time, there is increasing compromise in oocyte quality associated with increasing probability of subfertility and spontaneous abortion. This stability is useful in providing a baseline to appreciate the magnitude of the rate of FSH change at ages 48–52 yr.
The strengths of this report include a large cohort of women representative of a general population group rather than a selected clinical subsample. There was excellent participation over a 14-yr observation period that encompassed the middle and late reproductive periods as well as the menopause transition. The same FSH assay was used across the time frame with no change in the antibody over time. Furthermore, samples were collected annually and specimens were collected in the early follicular phase of the menstrual cycle. The early follicular phase allows for the comparison of FSH values in a standardized time frame. This data collection protocol, however, precludes evaluation of FSH variation during the later follicular phase, which may be more reflective of actual ovulatory events, or the luteal phase in which other ovarian peptides, including inhibin-A, may be differentially expressed reflecting endocrine control. Furthermore, these longitudinally acquired data do not include information about hypothalamic-pituitary sensitivity. The population is limited to Caucasian women, so findings may not be generalizable to women of other race/ethnic groups. Finally, it would be highly desirable to have these models reproduced in other longitudinally acquired data.
This report indicates that the natural history of FSH through the late reproductive years and into the menopause transition can be segmented into eight age-specific epochs and four ovarian aging-specific stages by evaluating acceleration and deceleration patterns and grounding these patterns in the underlying biology of oocyte number and folliculogenesis events. These data would indicate that there is a shift from premenopause FSH rate of change stability to early perimenopause stages at age 40 and 42 yr and that at age 45 yr, there is a major acceleration in the rate of FSH change thought to be the signal that folliculogenesis is increasingly compromised. The report identifies four FSH stages and the FSH levels and ages that clinicians can use, along with menstrual cycle characteristics, to help interpret the likely status of women with respect to reproductive viability and menopause stage.
|
Footnotes |
|---|
Disclosure Statement: The authors have nothing to disclose.
First Published Online July 22, 2008
Abbreviations: AMH, Antimullerian hormone; FMP, final menstrual period; IQR, interquartile range.
Received March 3, 2008.
Accepted July 14, 2008.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  N. Soto, G. Iniguez, P. Lopez, G. Larenas, V. Mujica, R. A. Rey, and E. Codner Anti-Mullerian hormone and inhibin B levels as markers of premature ovarian aging and transition to menopause in type 1 diabetes mellitus Hum. Reprod., November 1, 2009; 24(11): 2838 - 2844. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M.F.R. Sowers, H. Zheng, D. McConnell, B. Nan, C.A. Karvonen-Gutierrez, and J.F. Randolph Jr Testosterone, sex hormone-binding globulin and free androgen index among adult women: chronological and ovarian aging Hum. Reprod., September 1, 2009; 24(9): 2276 - 2285. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
