Abstract
Objective
To evaluate the ability of concentration of anti-Müllerian hormone (AMH), antral follicle count (AFC), and concentration of follicle stimulating hormone (FSH) to predict the onset of menopause.
Study Design
The Coronary Artery Risk Development in Young Adults Study (CARDIA) Women’s Study was an ancillary study to CARDIA, a population-based study of adults aged 18–30 years followed for 3 decades. For this report, participants were women (n=426) who had attended the CARDIA year 15–16 (2000–2001) examination, had at least one ovary, were not pregnant, and underwent serum AMH and FSH measurement and transvaginal ultrasonography in 2002–2003.
Main Outcome Measures
The probability of menopause in 5 years based upon AMH, FSH, and AFC.
Results
The mean age of the women at the time of AMH, FSH, and AFC assessment was 43 years. The cumulative incidence of menopause at 25 years (or follow-up) was 27% (n=426), and the incidence within 5 years was 13% (n=55). Among women aged 45–49 years, undetectable AMH concentrations were associated with a greater than 60% probability of menopause within 5 years, whereas approximately 1/3 of women with no or just one antral follicle experienced menopause within 5 years. Both low and high concentrations of FSH were associated with greater odds of menopause than intermediate concentrations. Models with multiple markers did not improve the prediction of menopause over that afforded by models with single markers.
Conclusion
The ability to predict onset of menopause was improved with any of the three menopausal markers in addition to age. AMH concentrations were more closely associated with menopause than AFC or FSH.
Keywords: menopause, anti-Müllerian hormone, antral follicle count, follicle stimulating hormone
Introduction
1.1. Significance of the prediction of menopause
Menopause is a major milestone of midlife women, because it represents the loss of fertility and the transition to post-reproductive life [1]. While the average age at menopause is approximately 51 years [2], there is significant variation around this estimate. From a clinical perspective, the likely age at menopause is used to guide decision-making regarding contraception [3, 4] assisted reproduction [5], management of menorrhagia [6], and conditions that may fluctuate with menstrual cycle hormonal variations including migraine and premenstrual dysphoric disorder [7]. Thus, markers that can assist in the prediction of menopause are of interest to women and their providers.
1.2. Rationale for the study
Low serum anti-Müllerian hormone (AMH) [8], high serum follicle stimulating hormone (FSH) levels [9], and low antral follicle count (AFC) [10] are associated with the onset of menopause. However, which of these measures or combination of measures best predicts menopause is not clear. Thus, more data is needed to understand if such markers can improve the prediction of menopause after consideration of age. In addition, more studies are needed which use population-based samples as opposed to selected populations of parous women or women with regular menstrual cycles. Therefore, we examined predictive models for menopause including AMH, FSH, and AFC in a large, multi-center, population-based study of black and white women.
Materials and Methods
2.1. Population and Setting for CARDIA
The Coronary Artery Risk Development in Young Adults (CARDIA) Study and the CARDIA Women’s Study have been previously described in detail.[10, 11] Briefly, CARDIA participants aged 18–30 years were recruited from the populations of Birmingham, Chicago, Minneapolis, and northern California, and baseline examinations were performed in 5115 individuals (2788 women) in 1985–1986. The population was balanced according to age, sex, race and educational attainment at each site. The CARDIA Women’s Study was an ancillary study to CARDIA designed to examine the associations between androgens, polycystic ovaries, and associated clinical features with subclinical atherosclerosis. Eligibility criteria included attendance at the year 15 CARDIA examination (2001–2002), presence of at least one ovary, and lack of current pregnancy. The CARDIA Women’s Study included 1163 participants, and the original protocol did not include AFC. However, with the publication of the Rotterdam criteria for polycystic ovary syndrome in 2003 [12], the protocol was altered to include AFC assessments. Thus, of the 1163 participants, 705 underwent AFC assessments. An institutional review board at each site approved all study procedures, and written informed consent was obtained from study participants prior to assessments.
2.1. Population for the current substudy
For the current report, we included women in the CARDIA Women’s Study. Eligible participants had to have undergone serum measurements of AMH and FSH as well as transvaginal ultrasonography for AFC at the 2002–2003 exam. Participants were excluded if they had bilateral oophorectomy, were pregnant, or did not attend the parent CARDIA exam in 2001–2002. For the current report, women with hysterectomy at any time were excluded, leaving 426 women. As compared with the final sample, excluded women were more likely to have been black (58% vs. 44%, p<0.01) and to have had slightly higher FSH concentrations (7.2 IU/L vs. 6.4 IU/L, p<0.01) but otherwise had similar age, AMH concentrations, and similar values for smoking, body mass index (BMI), and oral contraceptive pill (OCP) use.
2.2. Outcome
The primary outcome was menopause within the 5 years after ovarian reserve assessment. Participants completed a self-administered questionnaire during the CARDIA 2005–2006 (Year 20) and 2010–2011 (Year 25) examinations which enquired after cessation of menses and whether cessation occurred naturally, surgically, or through other means such as chemotherapy [10]. In a validation study, of the 18 women who reported that their menses had ceased at the 2002–2003 visit, 78% had FSH levels ≥ 40 IU/L, and among 202 women who reported both their age and calendar date of their final menstrual period, 94% reported a date which matched within one year of their self-reported age at menopause. At each visit, standardized questionnaires and examinations included assessment of lifestyle behaviors, including smoking and oral contraceptive use, information on regularity of menses, and anthropometrics. BMI was calculated in kg/m2.
2.3. FSH, AMH, AFC
Blood samples were drawn on the day of the transvaginal ultrasound. Visits were timed to the follicular phase (days 1–10 of the menstrual cycle) among regularly cycling women; women using OCPs or women with irregular menstrual cycles were scheduled for a visit when convenient. Within one hour of blood draw, samples were processed into aliquots and frozen at −70° C. FSH was measured in serum using a sequential two-step immunoenzymatic “sandwich” assay (Beckman Coulter, Brea, CA) by the University of Alabama Birmingham Diagnostic Laboratory. The intra-assay coefficients of variation ranged from 3.8 – 5.4% and interassay coefficients of variation ranged from 3.9 – 4.2%. In 2014, AMH was measured in stored samples from the 2002–3 visit using Ansh Laboratories (Webster, TX) ultra-sensitive AMH ELISA platform at the clinical reference laboratory, ReproSource, (Woburn, MA). The clinical lower limit of detection was 0.02 ng/dl; intra-assay coefficients of variation ranged between 3–7% and interassay coefficients of variation ranged between 5–10%.
Transvaginal ultrasound was performed by American Registry for Diagnostic Medical Sonography certified sonographers. Thirteen sonographers (a range of 2–5 sonographers per CARDIA field center) performed transvaginal ultrasounds. A 5–7.5 MHz vaginal probe was used to search for each ovary. If no ovary was found after 2 minutes, iliac vessels were photographed. When an ovary was found, antral follicles (follicles measuring 2–10 mm) were counted. AFC was defined as the total number of follicles on both ovaries. The CARDIA Women’s Study included a quality assurance protocol for ultrasonography. For transvaginal ultrasound, data from a repeat transvaginal ultrasound by a sonographer was available for a subset of 28 participants. For AFC ≤ 4 vs. > 4, interrater reliability was high (κ=0.85) between the repeat ultrasound and the original ultrasound.
2.4. Statistical Analysis
2.4.1. Classification of age, AMH, AFC, and FSH
Age, AMH, AFC, and FSH were treated as categorical variables when calculating descriptive statistics and then as continuous variables when examining the relationship with menopause onset at 5 years. Currently, there is no consensus with these variables as to which categories should be used to predict age at menopause. Therefore, for the descriptive analyses, we used cutpoints commonly used in clinical practice and which maximized the sensitivity or specificity of that particular measure. For AMH, cutpoints of <0.09 ng/dl (the lower limit of detection of the AMH assay) and ≥ 2.0 ng/dl were used. For AFC, cutpoints of < 1 follicle vs. > 10 follicles were examined. Although an FSH concentration of 40 IU/L is commonly considered consistent with postmenopause [13], values this high were rare in this premenopausal population (n=6), consistent with controversy as to which FSH value should be used in the premenopausal stage to predict age at menopause. Thus, extremes of FSH were defined as the lowest quartile (< 5.3 IU/L) vs. the uppermost quartile ( > 13.0 IU/L). Spearman’s correlation was used to estimate the associations among continuously measured AMH, AFC, and FSH.
2.4.2. Multivariable models
Multivariable logistic regression was use to predict the cumulative probability of menopause within 5 years of measurement. Age, AMH, AFC, and FSH were modeled as continuous covariates using restricted cubic splines to allow for non-linear associations. We considered 8 prediction models and compared model fit using the C-Index (area under the receiver operator characteristic curve), Akaike Information Criterion (AIC) [14], and likelihood ratio tests indicating whether goodness-of-fit differed between models. Since age is known to be a key predictor of menopause, an “age only” model was considered our baseline model for comparison. Bootstrap cross validation (150 replications) was used to provide corrected indices of model fit. Initially, we used Cox regression models to examine the association between each marker and incident natural menopause. However, the proportional hazards assumption was violated, therefore, we used logistic regression models which focused on the odds of menopause within 5 years. Odds were modeled flexibly using regression splines and translated into probability = odds/(1+odds) for figure plotting.
2.4.3. Sensitivity analyses
To make the analyses applicable to a general population, our main analyses included all women, including women using OCPs and women with irregular menses. However, we conducted multiple sensitivity analyses focusing on the following subpopulations: 1) women not using OCPs, 2) women with regular menses who had marker assessment during days 1–10 of their menstrual cycle, and 3) women without a single dominant antral follicle (>20 mm). All statistical analyses were performed using R statistical software (version 3.3.0).
Results
3.1. Participant characteristics
Participant characteristics are shown in Table 1. At the year 15–16 examination, women were approximately 43 years of age and 44% were black. Approximately 20% smoked, and the majority was overweight or obese. While most women had regular menstrual cycles, 15% of women were not able to predict their period for half of their cycles. Twenty-seven percent (n=117) had achieved natural menopause by year 25 follow-up, and approximately 55 women achieved natural menopause within 5 years of their AMH, AFC, or FSH measure.
Table 1.
Characteristics of women at assessment of anti-Müllerian hormone (AMH), follicle stimulating hormone (FSH), and antral follicle count (AFC) (n=426), at the year 15–16 examination. Median and interquartile range or n (%) shown.
| Age at year 15 exam (years) | 43 (39–45) |
| White (n,%) | 240 (56%) |
| Education (years) | 15 (13–16) |
| Body mass index (kg/m2) | 28 (24–34) |
| Current oral contraceptive pill use (n, %) | 67 (16%) |
| Current cigarette use (n,%) | 83 (19%) |
| Regular menses (n,%) | 364 (85%) |
| AMH (ng/dl) | 0.77 (0.22, 2.00) |
| AMH categories (n, %) | |
| Undetectable | 57 (13%) |
| 0.09–1.99 ng/dl | 262 (62%) |
| ≥ 2.0 ng/dl | 107 (25%) |
| AFC (number of follicles) | 5 (2–9) |
| AFC categories (n,%) | |
| 0–1 follicles | 91 (21%) |
| 2–9 follicles | 235 (55%) |
| ≥ 10 follicles | 100 (23%) |
| FSH (IU/L) | 6.4 (4.6, 9.1) |
| FSH categories (n,%) | |
| 0–5.39 IU/L | 103 (24%) |
| 5.4–12.9 IU/L | 259 (61%) |
| ≥13.0 IU/L | 64 (15%) |
3.2. Distribution and correlation of AMH, AFC, and FSH
Continuous values of AMH and AFC were positively correlated (r=0.51, p<0.01), whereas FSH was negatively and similarly correlated with both AMH and AFC (r=−0.22 and r=−0.20 respectively, both p<0.01). Table 2 shows the proportion of women with low and high AFC and FSH values by category of AMH. The middle quartiles of menopausal measures were combined for simplicity of presentation. Women with undetectable AMH concentrations were more likely to have low follicle counts and elevated FSH concentrations, and women with AMH ≥ 2.0 ng/dl were more likely to have high follicle counts and lower FSH concentrations (p<0.01). Similarly, women with low follicle counts were more likely to have high FSH concentrations and women with high follicle counts were more likely to have low FSH concentrations (p<0.01).
Table 2.
Number and proportion of women by anti-Müllerian hormone (AMH), antral follicle count (AFC), and follicle stimulating hormone (FSH) categories.
| AMH (ng/dl) | |||
|---|---|---|---|
| Undetectable | 0.09–1.9 | > 2.0 | |
| AFC (follicle number) | |||
| 0–1 | 30 (53%) | 54 (21%) | 7 (7%) |
| 2–10 | 27 (47%) | 173 (66%) | 35 (33%) |
| ≥10 | 0 | 35 (13%) | 65 (61%) |
| AMH (ng/dl) | |||
| FSH (IU/L) | Undetectable | 0.09–1.9 | > 2.0 |
| 0–5.39 | 5 (9%) | 63 (24%) | 35 (33%) |
| 5.4–12.9 | 22 (38%) | 165 (63%) | 72 (67%) |
| ≥ 13.0 | 30 (52%) | 34 (13%) | 0 |
| AFC (follicle number) | |||
| FSH (IU/L) | 0–1 | 2–9 | >10 |
| 0–5.39 | 22 (24%) | 58 (25%) | 23 (23%) |
| 5.4–12.9 | 40 (44%) | 144 (61%) | 75 (75%) |
| ≥ 13.0 | 29 (32%) | 33 (14%) | 2 (2%) |
3.3. AMH, AFC, FSH and cumulative incidence of menopause within 5 years
Figure 1 shows the percentages of women who were menopausal within 5 years of their AMH, AFC, and FSH measure. Few women 34–39 years of age experienced their final menstrual period in the next 5 years, and thus AMH, AFC, and FSH could not distinguish which women would experience menopause in this age category. Regardless of age, no women with AMH concentrations ≥ 2.0 ng/dl experienced menopause within 5 years (Figure 1a). Among women aged 45–49 years, 60% of women with undetectable AMH concentrations experienced menopause within 5 years. Women with high AFC were less likely to experience menopause, although approximately 1/5 of women with high follicle counts i.e. ≥ 10 still underwent menopause. Among women with only 0–1 follicles, approximately 1/3 underwent menopause within 5 years (Figure 1b). The relationship between FSH and probability of menopause was U-shaped, where both high and low FSH were associated with a higher incidence of menopause: although a greater proportion of women with FSH ≥ 13.0 IU/L experienced menopause within 5 years than women with lower concentrations, approximately 45% of women in the lowest quartile of AMH underwent menopause within 5 years, limiting the utility of FSH for distinguishing between women at higher vs. lower risk. This pattern of results for AMH, AFC, and FSH was similar among women not using OCPs, women menstruating regularly and with marker measurement during their follicular phase, and women without a single dominant follicle (results not shown).
Figure 1.
Percentage of women who achieve menopause within 5 years after assessment of AMH (Panel A), AFC (Panel B), and FSH (Panel C).
3.4. Odds of menopause by AMH, AFC, and FSH in multivariable models
Figure 2 shows the estimates of the probability of menopause within the next 5 years from logistic regression models by age at AMH measurement and AMH concentration. No matter how low the AMH concentration, the probability of menopause was low for women in their thirties (bottom line on Figure 2). Similarly, no matter how high the AMH concentration, the probability of menopause was high for women who were 50 years of age (top line on Figure 2). The probability of menopause for women over 45 years of age was high if their AMH levels were undetectable. The relationship between age, AMH concentrations, and probability of menopause was logarithmic, such that even slightly higher AMH in the range between undetectable concentrations and 0.5 ng/dl had a significant impact on menopause probability. Estimates of probability for women not using OCPs (Appendix Figure 1) and among women with regular menses and follicular phase marker assessment (Appendix Figure 2) also showed a logarithmic relationship between AMH concentrations and probability of menopause within 5 years. OCP use (p=0.91) did not modify probability curves; regular menses had borderline significance for modifying the curves (p=0.07), but predicted probabilities were close to 0 and estimates of confidence were wide.
Figure 2.
Probability of menopause within 5 years by age and anti-Müllerian hormone (AMH) concentration based on logistic regression models. Lower limit of AMH detection is 0.09 ng/dl.
Table 3 shows the odds of menopause within 5 years for each menopausal marker. Given that the odds ratios changed depending upon women’s ages and with the range of menopausal markers, the odds ratios shown focus on the group of women aged 45–49 years, with AMH levels between 0.09 and 0.5 ng/dl, with FSH <14 IU/L, and with AFC between 0–10 follicles. After adjustment for age, AMH was the only menopausal marker significantly associated with odds of menopause. Table 3 also shows estimates of model fit for logistic regression models examining the odds of menopause within 5 years. Models containing both age and AMH had the highest C-index and lowest AIC (consistent with better model quality), followed by models including age and other markers. When compared to models including age alone, model fit was improved with addition of any menopausal marker or combination of markers. When compared to models containing age and AMH, addition of FSH and/or AFC did not lead to improved model fit. Further adjustment for covariates including covariates in Table 1 did not change this pattern (results not shown). Again, these patterns were similar in sensitivity analyses among women not using OCPs, women with regular menses and follicular phase measurement, and women without a single dominant follicle (Appendix Table 1).
Table 3.
Comparisons of model fit for models containing age and menopausal markers (anti-Müllerian hormone (AMH), antral follicle count (AFC), and follicle stimulating hormone (FSH). Odds ratios (95% confidence intervals) for age, AMH, AFC, and FSH shown; bold type indicates statistical significance at p<0.05.*
| Odds ratios (95% confidence intervals) |
C-statistic | Akaike Information Criterion (AIC) |
Likelihood ratio test vs. models with age alone |
Likelihood ratio test vs. models with age and AMH |
|
|---|---|---|---|---|---|
| Age alone | 0.83 | 254.9 | - | - | |
| Age (years) | 7.4 (4.1, 13.6) | ||||
| Age and AMH | 0.91 | 199.4 | <0.001 | ||
| Age (years) | 4.1 (2.1, 7.9) | ||||
| AMH (ng/dl) | 0.13 (0.05, 0.34) | ||||
| Age and AFC | 0.84 | 250.1 | 0.009 | ||
| Age (years) | 6.5 (3.5, 12.0) | ||||
| AFC (number) | 0.33 (0.13, 8.2) | ||||
| Age and FSH | 0.85 | 239.4 | <0.001 | 0.077 | |
| Age (years) | 6.1 (3.3, 11.3) | ||||
| FSH (IU/L) | 0.79 (0.18, 3.6) | ||||
| Age, AMH, FSH | 0.92 | 198.3 | <0.001 | ||
| Age (years) | 3.9 (2.0, 7.8) | ||||
| AMH (ng/dl) | 0.18 (0.07, 0.43) | ||||
| FSH (IU/L) | 0.47 (0.10, 2.3) | ||||
| Age, FSH, AFC | 0.86 | 238.5 | <0.001 | 0.275 | |
| Age (years) | 5.7 (3.0, 10.6) | ||||
| AFC (number) | 0.47 (0.19, 1.2) | ||||
| FSH (IU/L) | 0.77 (0.17, 3.4) | ||||
| Age, AMH, AFC | 0.91 | 200.2 | <0.001 | ||
| Age (years) | 4.1 (2.1, 8.0) | ||||
| AMH (ng/dl) | 0.12 (0.04, 0.31) | ||||
| AFC (number) | 2.0 (0.57, 6.9) | ||||
| Age, AMH, FSH, AFC | 0.91 | 198.3 | <0.001 | 0.070 | |
| Age (years) | 3.9 (2.0, 7.8) | ||||
| AMH (ng/dl) | 0.15 (0.06, 0.39) | ||||
| FSH (IU/L) | 0.44 (0.09, 2.2) | ||||
| AFC (number) | 2.4 (0.69, 8.6) |
For estimation of odds ratios, model assumptions are shown for women aged 45–49 years, with AFC 0–10 follicles, FSH 0–13.9 IU/L, and AMH 0.09–0.50 ng/dl.
Discussion
4.1. Summary of findings: AMH
In a population-based cohort, measurement of AMH best improved the prediction of menopause probability within 5 years compared to age alone and compared to other menopausal markers. Due to the low probability of menopause within 5 years for younger women, ovarian reserve markers had higher predictive value for older women. AMH concentrations > 2.0 ng/dl were associated with minimal probability of menopause within the next 5 years regardless of a woman’s age, and conversely, AMH concentrations below detection level were associated with a greater than 60% probability of menopause among women aged 45 years and older. Of note, even small changes in AMH above undetectable levels correlated with large decreases in probability of menopause, particularly among women aged greater than 45 years of age.
4.2. Summary of findings: AFC and FSH
Although higher AFC was associated with a lower proportion of women experiencing menopause than a lower AFC, only about one-third of women aged 45–49 years with 0–1 follicles underwent menopause, suggesting that low AFC counts had weaker predictive value than low AMH for the onset of menopause. While models including AFC improved menopause prediction compared to models with age alone, AFC values themselves were not significantly associated with odds of menopause. While FSH concentrations improved the prediction of menopause after consideration of age, the relationship between FSH and the onset of menopause was U-shaped and thus difficult to use for estimation of menopause probability. Approximately 45% and 65% of women aged 45–49 years in the lowest and highest quartile of FSH, respectively, experienced menopause within 5 years.
4.3. Previous studies reporting upon AMH, AFC, and FSH for the prediction of menopause
Previous reports, including those by our group as well as others, have noted that measures of ovarian reserve can predict age at menopause [8, 10, 15–20]. However, implementation in the clinical setting has been limited by several factors. First, the detection limits of AMH varied widely with the first generations of assays [21], making cutpoints difficult to determine. Second, the logistical difficulties inherent in assessing AFC in a uniform manner across centers [22] and the invasiveness of transvaginal ultrasound have limited widespread use of AFC. Third, the fluctuation of FSH between extreme elevations and depressions in the perimenopause [23], even when assessed in the follicular phase of the menstrual cycle and in the absence of estrogen use, have limited the use of FSH for these purposes. Finally, the strength of the association between markers and age menopause weakens over time; specifically, the longer the time to menopause, the weaker the association between menopausal marker and age at menopause [15]. Such changes in the strength of association with time impair the value of these markers for menopause prediction.
4.4. Methodologic strengths of the current report
However, our current report adds to the previous literature in several respects. First, the present report of 426 women is the largest containing more than a single marker of ovarian reserve. Thus, we were able to compare the associations between several markers of ovarian reserve and time to menopause. Second, by examining the proportion of women who reached menopause within a specific time frame, we were able to present results in a manner which acknowledges that risk of menopause changes over time. Along these lines, we present the odds of menopause within 5 years for measures consistent with high ovarian reserve as well as for low ovarian reserve. Information is displayed graphically for AMH to aid ease of interpretation. Third, the cohort included women who were drawn from a population-based sample, whereas previous cohorts focused upon women with regular menstrual cycles, women undergoing infertility evaluations, and women not using OCPs, interfering with the interpretation of results from women who did not fit these criteria. Similarly, approximately half of the population in our report was African-American, an important characteristic to consider given that the CARDIA study [8] and at least one other report have noted racial/ethnic differences in AMH concentrations between blacks and whites [24]. Finally, our study addresses several of the limitations of ovarian reserve measurements mentioned in the previous paragraph, including imprecise and relatively less sensitive AMH assays; AFC estimates generated from single-center reports; and the inaccuracy of risk models which assume uniform risks over time.
4.5. Clinical relevance
Although historically used to guide assisted reproductive therapy, markers of ovarian reserve have potential application to the larger population of women in perimenopause. Such symptoms include menorrhagia, which in the Study of Women’s Health Across the Nation affected almost one-third of late reproductive-stage women [25] and which lowers quality of life [26]. The choice of therapy for such symptoms, including intrauterine device insertion, endometrial ablation, hysterectomy, and systemic hormone and anti-platelet therapy, are informed by expected length of time with symptoms, i.e. time to final menstrual period as well as individual factors. Such symptoms also include those associated with the hormone fluctuation characterizing the menopause, such as menstrual migraines, which affect over one in 10 perimenopausal women [27] and may be treated with estrogen therapy. However, estrogen therapy obscures the menstrual cycle, and currently therapy must be punctuated by trials off of medication, which in turn are associated with exacerbation of symptoms. Finally, ovarian reserve markers may be of use in the prediction of conditions associated with the postmenopause, for which current risk markers are limited. For example, bone density testing for the identification of women with osteopenia is not recommended until 65 years of age [28], although such guidelines have the potential for missing the subset of women with markedly accelerated bone loss preceding the final menstrual period [29]. Thus, estimates of the length of time to menopause are of value aside from estimates of fecundity.
4.6. Limitations
Limitations of this report include lack of more precise estimates of the proportion of women reaching their final menstrual period, i.e. within 1 year or 6 months. Such estimates would require more precise methods of determining the age at the final menstrual period rather than the self-report employed in CARDIA, as well as a larger population-based sample that would allow for estimation before and after covariates such as BMI and genetic variation that also determines ovarian reserve [16]. Such a cohort is unlikely to be amassed given the length of follow-up time required as well as the invasive nature of transvaginal ultrasounds and difficulty of timing FSH measures to the menstrual cycle. We may have underestimated the predictive capability of FSH, as this marker fluctuates throughout the menstrual cycle and is typically measured on the 3rd day of menstruation. However, for those women with irregular menstrual cycles or who use sex steroids, measurement on day 3 is difficult. In addition, the non-linear relationship between FSH and time to menopause limits its use for menopause prediction. The length of follow-up did not allow comparison of markers for women of younger reproductive age or reproductive ages greater than 49 years, and our results do not apply to these age extremes. Finally, estimates of time to menopause used only a single measure of AMH, and it is possible that AMH levels decline at different rates between women [30].
4.7. Conclusions
We conclude that among late reproductive-age women, AMH concentrations can predict the odds of menopause within the next 5 years, and specifically, AMH greater than 2.0 ng/dl is associated with an extremely low probability of menopause whereas undetectable concentrations predict a higher risk of menopause among women in their forties. These estimates may be of value to women and their physicians in the management of reproductive disorders. Other markers of ovarian reserve may also be useful even after consideration of a woman’s age. However, due to non-linearity in FSH concentrations with probability of menopause, need for timing to the follicular phase of the menstrual cycle, and the necessity of transvaginal ultrasound for the assessment of AFC, FSH and AFC may be more difficult to apply to menopause prediction in the clinical setting. Future investigations should also explore the value of AMH in risk prediction for disorders associated with menopause, such as osteopenia and atherosclerosis, as well as the management of conditions associated with the management of menorrhagia in the perimenopause.
Highlights.
Using data from a population-based study of adults, we evaluated the ability of several measures of ovarian age to predict the onset of menopause.
Among women aged 45–49 years, those with undetectable anti-Müllerian hormone concentrations had a 60% probability of menopause within 5 years.
Among women aged 45–49 years, approximately one-third of women with no or just one antral follicle underwent menopause within 5 years.
The concentration of follicle stimulating hormone had a non-linear relationship with the likelihood of menopause within 5 years.
Acknowledgments
Funding
This work was supported by the National Institutes of Health (HHSN268201300025C, HHSN268201300026C, HHSN268201300027C, HHSN2682013000- 28C, HHSN268201300029C, HHSN268200900041C, AG005, R01-HL-065611, K23-HL-87114, R03-HL-135453, and DP3-DK-098129).
Abbreviations
- AFC
antral follicle count
- AMH
anti-Müllerian hormone
- BMI
body mass index
- CARDIA
Coronary Artery Risk Development in Young Adults Study
- ELISA
enzyme linked immunosorbent assay
- FSH
follicle stimulating hormone
- OCP
oral contraceptive pill
Appendix
Appendix Figure 1.
Probability of menopause within 5 years for women with regular menses.
Appendix Figure 2.
Probability of menopause within 5 years for women who are not using oral contraceptives.
Ancillary Table 3.
Comparisons of model fit for models containing age and menopausal markers (anti-Müllerian hormone or AMH, antral follicle count or AFC, follicle stimulating hormone or FSH).
| C-statistic for models with all women (n=426) |
C-statistic for models in women not using estrogen. (n=359) |
C-statistic for models with women with regular menses only. (n=327) |
C-statistic for models excluding any woman with a dominant follicle. (n=306) |
|
|---|---|---|---|---|
| Age alone | 0.83 | 0.82 | 0.81 | 0.86 |
| Age and AMH | 0.91 | 0.91 | 0.91 | 0.94 |
| Age and AFC | 0.84 | 0.84 | 0.82 | 0.86 |
| Age and FSH | 0.86 | 0.86 | 0.84 | 0.89 |
| Age, AMH, FSH | 0.91 | 0.92 | 0.91 | 0.94 |
| Age, FSH, AFC | 0.86 | 0.87 | 0.84 | 0.89 |
| Age, AMH, AFC | 0.91 | 0.91 | 0.91 | 0.95 |
| Age, AMH, FSH, AFC | 0.91 | 0.92 | 0.91 | 0.95 |
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Author Contributions
Cathy Kim: wrote manuscript, interpreted analyses
James Slaughter: conceived and performed analyses, revised manuscript
Eric Wang: interpreted analyses, revised manuscript
Duke Appiah: interpreted analyses, revised manuscript
Pam Schreiner: collected data, interpreted analyses, revised manuscript
Ben Leader: interpreted analyses, revised manuscript
Ronit Calderon-Margalit: interpreted analyses, revised manuscript
Barbara Sternfeld: collected data, revised manuscript
David Siscovick: collected data, revised manuscript
Melissa Wellons: conceived and designed analyses, revised manuscript
Conflict of interest
The authors declare that they have no conflict of interest.
Ethical approval
The work described has been carried out in accordance with The Code of Ethics of the World Medical Association (Declaration of Helsinki) for experiments involving humans and Uniform Requirements for manuscripts submitted to Biomedical journals. Informed consent was obtained for experimentation with human subjects.
References
- 1.Ayers B, Forshaw M, Hunter M. The impact of attitudes towards the menopause on women’s symptom experience: a systematic review. Maturitas. 2010;65(1):28–36. doi: 10.1016/j.maturitas.2009.10.016. [DOI] [PubMed] [Google Scholar]
- 2.Gold E. The timing of the age at which natural menopause occurs. Obstet Gynecol Clin North Am. 2012;38(3):425–40. doi: 10.1016/j.ogc.2011.05.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Mendoza N, Soto E, Sanchez-Borrego R. Do women aged over 40 need different counseling on combined hormonal contraception. Maturitas. 2016;87:79–83. doi: 10.1016/j.maturitas.2016.02.008. [DOI] [PubMed] [Google Scholar]
- 4.Allen R, Cwiak C. Contraception for midlife women. Menopause. 2016;23(1):111–3. doi: 10.1097/GME.0000000000000584. [DOI] [PubMed] [Google Scholar]
- 5.MacArthur T, Bachmann G, Ayers C. Menopausal women requesting egg/embryo donation: examining health screening guidelines for assisted reproductive technology. Menopause epub ahead of print, May 16, 2016. 2016 doi: 10.1097/GME.0000000000000622. [DOI] [PubMed] [Google Scholar]
- 6.Bradley L, Gueye N. The medical management of abnormal uterine bleeding in reproductive-aged women. Am J Obstet Gynecol. 2016;214(1):31–44. doi: 10.1016/j.ajog.2015.07.044. [DOI] [PubMed] [Google Scholar]
- 7.Ibrahimi K, Couturier E, MaassenVanDenBrink A. Migraine and perimenopause. Maturitas. 2014;78(4):277–80. doi: 10.1016/j.maturitas.2014.05.018. [DOI] [PubMed] [Google Scholar]
- 8.Nair S, Slaughter J, Terry J, Appiah D, Ebong I, Wang E, Siscovick D, Sternfeld B, Schreiner P, Lewis C, Kabagambe E, Wellons M. Anti-mullerian hormone (AMH) is associated with natural menopause in a population-based sample: the CARDIA Women’s Study. Maturitas. 2015;81:493–8. doi: 10.1016/j.maturitas.2015.06.026. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Greendale G, Ishii S, Huang M, Karlamangla A. Predicting the timeline to the final menstrual period: the Study of Women’s Health Across the Nation. J Clin Endocrinol Metab. 2013;98(4):1483–91. doi: 10.1210/jc.2012-3732. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Wellons M, Bates G, Schreiner P, Siscovick D, Sternfeld B, Lewis C. Antral follicle count predicts natural menopause in a population-based sample: the CARDIA Women’s Study. Menopause. 2013;20(8):825–30. doi: 10.1097/GME.0b013e31827f06c2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Friedman G, Cutter G, Donahue R, Hughes G, Hulley S, Jacobs D, Jr, Liu K, Savage P. CARDIA: study design, recruitment, and some characteristics of the examined subjects. J Clin Epidemiol. 1988;41:1105–16. doi: 10.1016/0895-4356(88)90080-7. [DOI] [PubMed] [Google Scholar]
- 12.Rotterdam ESHRE/ASRM-Sponsored PCOS Consensus Workshop Group. Revised 2003 consensus on diagnostic criteria and long-term health risks related to polycystic ovary syndrome. Fertil Steril. 2004;81(1):19–25. doi: 10.1016/j.fertnstert.2003.10.004. [DOI] [PubMed] [Google Scholar]
- 13.Randolph J, Crawford S, Dennerstein L, Cain K, Harlow S, Little R, Mitchell E, Nan B, Taffe J, Yosef M. The value of follicle-stimulating hormone concentration and clinical findings as markers of the late menopausal transition. J Clin Endocrinol Metab. 2006;91(8):3034–40. doi: 10.1210/jc.2006-0243. [DOI] [PubMed] [Google Scholar]
- 14.Akaike H. A new look at the Bayes procedure. Biometrika. 1978;65:53–9. [Google Scholar]
- 15.Depmann M, Eijkemans M, Broer S, Scheffer G, van Rooij I, Laven J, Broekmans F. Does anti-Mullerian hormone predict menopause in the general population? Results of a prospective ongoing cohort study. Hum Reprod epub ahead of print, May 2016. 2016 doi: 10.1093/humrep/dew112. [DOI] [PubMed] [Google Scholar]
- 16.Depmann M, Broer S, van der Schouw Y, Tehrani F, Eijkemans M, Mol B, Broekmans F. Can we predict age at natural menopause using ovarian reserve tests or mother’s age at menopause? A systematic literature review. Menopause. 2015;23(2):224–232. doi: 10.1097/GME.0000000000000509. [DOI] [PubMed] [Google Scholar]
- 17.Sowers M, Eyvazzadeh A, McConnell D, Yosef M, Jannausch M, Zhang D, Harlow S, Randolph J., Jr Anti-mullerian hormone and inhibin B in the definition of ovarian aging and the menopause transition. J Clin Endocrinol Metab. 2008;93(9):3478–83. doi: 10.1210/jc.2008-0567. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Tehrani F, Solaymani-Dodaran M, Tohidi M, Goharai M, Azizi F. Modeling age at menopause using serum concentration of anti-Mullerian hormone. J Clin Endocrinol Metab. 2013;98:729–35. doi: 10.1210/jc.2012-3176. [DOI] [PubMed] [Google Scholar]
- 19.Freeman E, Sammel M, Lin H, Gracia C. Anti-mullerian hormone as a predictor of time to menopause in late reprodutive age women. J Clin Endocrinol Metab. 2012;97(5):1673–80. doi: 10.1210/jc.2011-3032. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Dolleman M, Depmann M, Eijkemans M, Heimensem J, Broer S, van der STroom E, Laven J, van Rooij I, Scheffer G, Peeters P, van der Schouw Y, Lambalk C, Broekmans F. Antimullerian hormone is a more accurate predictor of individual time to menopause than mother’s age at menopause. Hum Reprod. 2014;29(3):584–591. doi: 10.1093/humrep/det446. [DOI] [PubMed] [Google Scholar]
- 21.Sui H, Sammel M, Homer M, Bui K, Haunschild C, Stanczyk F. Comparability of antimullerian hormone levels among commerically available immunoassays. Fertil Steril. 2014;101(6):1766–72. doi: 10.1016/j.fertnstert.2014.02.046. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Iliodromiti S, Anderson R, Nelson S. Technical and performance characteristics of anti-Mullerian hormone and antral follicle count as biomarkers of ovarian response. Hum Reprod Update. 2015;21(6):698–710. doi: 10.1093/humupd/dmu062. [DOI] [PubMed] [Google Scholar]
- 23.Sowers M, Zheng H, McConnell D, Nan B, Harlow S, Randolph J., Jr Follicle stimulating hormone and its rate of change in defining menopause transition stages. J Clin Endocrinol Metab. 2008;93(10):3958–64. doi: 10.1210/jc.2008-0482. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Bleil M, Gregorich S, Adler N, Sternfeld B, Rosen M, Cedars M. Race/ethnic disparities in reproductive age: an examination of ovarian reserve estimates across four race/ethnic groups of healthy, regularly cycling women. Fertil Steril. 2014;101(1):199–207. doi: 10.1016/j.fertnstert.2013.09.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Paramsothy P, Harlow S, Greendale G, Gold E, Crawford S, Elliott M, Lisabeth L, Randolph J., Jr Bleeding patterns during the menopausal transition in the multi-ethnic Study of Women’s Health Across the Nation (SWAN): a prospective cohort study. BJOG. 2014;121(12):1564–73. doi: 10.1111/1471-0528.12768. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Lukes A, Baker J, Eder S, Adomako T. Daily menstrual blood loss and quality of life in women with heavy menstrual bleeding. Womens Health (Lond) 2012;8(5):503–11. doi: 10.2217/whe.12.36. [DOI] [PubMed] [Google Scholar]
- 27.Marvin V, Pavlovic J, Fanning K, Buse D, Reed M, Lipton R. Perimenopause and menopause are associated with high frequency headache in women with migraine: results of the American Migraine Prevalence and Prevention Study. Headache. 2016;56(2):292–305. doi: 10.1111/head.12763. [DOI] [PubMed] [Google Scholar]
- 28.U.S. Preventive Services Task Force. Screening for osteoporosis: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med. 2011;154(5):356–64. doi: 10.7326/0003-4819-154-5-201103010-00307. [DOI] [PubMed] [Google Scholar]
- 29.Sowers M, Zheng H, Greendale G, Neer R, Cauley J, Ellis J, Johnson S, Finkelstein J. Changes in bone resorption across the menopause transition: effects of reproductive hormones, body size, ethnicity. J Clin Endocrinol Metab. 2013;98(7):2854–63. doi: 10.1210/jc.2012-4113. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.de Kat A, van der Schouw Y, Eijkemans M, Herber-Gast G, Visser J, Verschuren W, Broekmans F. Back to the basics of ovarian aging: a population-based study on longitudinal anti-Mullerian hormone decline. BMC Med. 2016;14(1):151. doi: 10.1186/s12916-016-0699-y. [DOI] [PMC free article] [PubMed] [Google Scholar]