Anti-Müllerian Hormone Level Predicts Ovulation in Women with Polycystic Ovary Syndrome Treated with Clomiphene and Metformin

Abstract

Objective:

To describe serum anti-Müllerian hormone (AMH) concentrations in a large, well-phenotyped cohort of women with polycystic ovary syndrome (PCOS) and evaluate whether AMH predicts successful ovulation induction (OI) in women treated with clomiphene and metformin.

Design:

Secondary analysis of randomized controlled trial

Subjects:

333 women with anovulatory infertility attributed to PCOS who participated in the double blind randomized trial entitled The Pregnancy in Polycystic Ovary Syndrome I (PPCOS I) study (registration number NCT00068861) who had serum samples from baseline laboratory testing available for further serum analysis were studied.

Main Outcome Measures:

The association between baseline AMH concentration in each of the three treatment groups and ovulation rates, pregnancy rates and live birth rates were assessed.

Results:

322 individuals had a baseline AMH concentration available, of which the mean AMH was 11.7 ± 8.3 ng/mL with a range from 0.1 to 43.0 ng/mL. With each unit (1 ng/mL) increase in baseline AMH, the odds of ovulation decreased by 10 percent (OR 0.90, 95% CI 0.86–0.93); this effect did not differ by treatment group. Women with a high baseline AMH concentration (>8 ng/mL) were significantly less likely to ovulate compared to those with a normal baseline AMH concentration (<4 ng/mL) (OR 0.23, 95% CI 0.05–0.68). This remained statistically significant when controlling for confounders, including age, body mass index, time in study, and Homeostatic Model Assessment for Insulin Resistance (HOMA-IR) score (p=0.008). Ovulation occurred even at very high AMH concentrations; there was no maximum level noted at which no ovulation events occurred. Baseline AMH concentration was not associated with pregnancy or live birth rates when controlling for confounders (p=0.947 and p=0.848, respectively).

Conclusion:

These AMH values of well-phenotyped individuals with PCOS adds to the literature and will aid in identifying AMH criteria for the diagnosis of PCOS. In women with infertility and PCOS, a higher AMH concentration was associated with reduced odds of ovulation with OI with clomiphene, clomiphene+metformin and metformin.

Capsule:

In women with infertility and polycystic ovary syndrome, higher anti-Mullerian hormone concentration was associated with reduced odds of ovulation with ovulation induction therapy.

Introduction

Polycystic ovary syndrome (PCOS) is the most common cause of infertility and anovulation during reproductive age (1, 2). Individuals with PCOS often experience oligo- or anovulation and frequently have an increased pool of ovarian follicles (3). Criteria often used in the diagnosis of PCOS include elevated antral follicle count (AFC) and/or ovarian volume on ultrasound (4). Anti-Müllerian hormone (AMH), a dimeric glycoprotein that is part of the transforming growth factor ß family, is produced by granulosa cells of preantral and antral follicles in the ovary (5). Although AMH is not currently a diagnostic criterion for PCOS, several studies suggest AMH could serve as a proxy for AFC (6–8). Serum concentration of AMH is often noted to be two to four times higher in individuals with PCOS compared to ovulatory controls; however, data on the diversity of AMH concentrations in well-characterized cohorts of patients with PCOS is lacking (2, 7). The most recently published international guideline for PCOS recommends including AMH in the evaluation of anovulation and in the diagnostic criteria for PCOS. However, no clear AMH cutoff value has been established due to limited data on AMH values in persons with PCOS, as well as variation among assays and by age (9).

Excess AMH may be harmful to the process of folliculogenesis and may be involved in the cessation of follicular growth occurring in ovaries affected by PCOS (7, 10). There is evidence that high AMH concentrations can be associated with resistance to clomiphene and letrozole, common ovulation induction (OI) treatments for women with PCOS, perhaps due to impairments in follicular and ovulatory responses (3, 11). However, data on this association as well as the association between AMH concentrations and pregnancy and live birth rates with OI, remains limited.

Our primary objectives were to 1) expand the published data on AMH values in well-phenotyped PCOS and 2) evaluate whether serum AMH concentration predicted success of OI treatments in women with PCOS. Women were randomly assigned to treatment with clomiphene alone, clomiphene plus metformin or metformin alone. We hypothesized that higher baseline serum AMH concentration would be associated with reduced odds of ovulation in all treatment groups.

Methods

This study utilizes data from the Pregnancy in Polycystic Ovary Syndrome (PPCOS I) trial. The study design, use of infertility screening, power analysis and initial findings of this study are previously described (12–15). This randomized clinical trial was conducted at 12 centers in the United States and was registered on ClinicalTrials.gov as number NCT00068861. All participants gave written informed consent, and the study was approved by the institutional review board at each center. Briefly, 626 women with infertility and PCOS were enrolled in the study. Inclusion criteria included oligomenorrhea and elevated testosterone levels (by predetermined cutoffs that varied by local laboratory site), seeking pregnancy, normal uterine cavity, one or more patent fallopian tubes and current partner with a recent semen analysis with sperm concentration over 20 million per milliliter. Oligomenorrhea was defined as no more than eight spontaneous menses per year. Baseline evaluation included fasting serum levels of testosterone, sex hormone binding globulin (SHBG), proinsulin, insulin and glucose. Standardized interviewer-administered questionnaires elicited medical history, obstetric, gynecologic and infertility history, smoking status, family history and other demographics. Biometric evaluation included height, weight, BMI, waist circumference and hirsutism score by the modified Ferriman-Gallwey method (16). Transvaginal ultrasound was performed to assess polycystic ovary morphology and ovarian volume. Ultrasound information was not included in the PCOS diagnostic criteria as the study was initiated prior to publication of the Rotterdam criteria (14).

Participants were randomly assigned to one of three treatment arms: metformin, clomiphene citrate, or both (14). Progesterone levels were measured weekly or every other week. Treatment was continued until a participant had a positive pregnancy test or completed six cycles of treatment. Serum and plasma from the study participants was stored at −80°C from initial recruitment until batched analysis was performed in 2021. Analysis of participants’ laboratory studies have been previously described (14). In this study, AMH concentration was assessed for available stored baseline serum samples, which were collected in the early follicular phase of the participants’ menstrual cycles. AMH concentration was calculated using an enzyme-linked immunosorbent assay (PCOCheck ELISA, AL-196, Ansh Labs LLC). The assay uses a linear double-sided antibody that has epitope in the N-terminus of AMH. The sensitivity of this assay was 0.029 ng/mL and the interassay coefficient of variation was 3.2% and 2.4% at 1.5 ng/mL and 3.3ng/ml, respectively. Baseline serum samples were available for 333 of the 626 women enrolled in the trial and the data from these participants is presented below. AMH concentration was unable to be obtained for 11 women, of which 8 samples were missing appropriate labeling and 3 had insufficient serum available to complete AMH testing.

The main outcome measure of this analysis was ovulation (cycle progesterone concentration > 5 ng/mL). We also analyzed secondary outcomes of pregnancy (fetal heart motion visualized on transvaginal ultrasound) and live birth (delivery of a viable infant). We also sought to establish an optimal threshold of AMH concentration that predicted ovulation, and to identify any independent predictors of ovulation through logistic regression modeling.

All statistical analyses were conducted by using R Statistical Software version 4.1.2. Descriptive statistics, including mean ± standard deviation and count (percentage) were used to report participant characteristics of the sample. We used Chi-squared tests or Fisher’s exact tests (as appropriate) to assess differences in categorical variables, and the Kruskal-Wallis rank sum test, to test for differences in continuous variables between the three treatment groups. Given that international standards for AMH values do not exist, participants were compared by tertile of baseline AMH level as well as divided into three “clinical” categories based on baseline AMH level: AMH <4 ng/mL, 4–8 ng/mL and >8 ng/mL. These categories were chosen a priori given clinical significance of these values; AMH <4 ng/mL is generally considered normal. While there is not currently an AMH level diagnostic of PCOS, prior meta-analyses have identified cutoff AMH values of 4.7 and 4.9 ng/mL as sensitive and specific in diagnosing PCOS (6, 17, 18). Therefore, an AMH level 4–8 ng/mL was chosen as the intermediate category for our analysis. AMH concentration >8 ng/mL was selected as the highest category; this cut-off was determined based on a previous study which found an AMH threshold of 7.77 ng/mL as predictive of ovulation with clomiphene OI in women with PCOS (19).

We used logistic regression to model the association between baseline AMH and the three outcomes: ovulation, pregnancy and live birth. AMH was modeled as continuous, as tertiles, and as the “clinical” categories described above. Unadjusted models evaluated baseline AMH and treatment group alone. As the original PPCOS I study found a difference in outcomes by treatment group with higher live birth rates in the clomiphene and clomiphene + metformin groups compared to the metformin alone group, we adjusted for treatment group in our analyses. Age, BMI, race, ethnicity, waist to hip ratio, hirsutism, LH, total testosterone, insulin sensitivity, androgen index, combined ovarian volume, polycystic ovary morphology, duration of treatment, infertility diagnosis, history of pregnancy and smoking status were individually evaluated as possible confounders in subsequent logistic regression models. Age was included in all adjusted models given the known association between increasing age and reduced AMH concentration (20, 21). Additionally, we tested for effect modification of baseline AMH concentration by treatment group by including interactions terms in the models; statistical significance of the interaction was tested using the analysis of deviance. Specifically, we assessed the interactions between AMH concentration and race, and AMH concentrations and BMI. A receiver operating curve was utilized to identify an optimal threshold of AMH concentration predicting ovulation by optimizing sensitivity and specificity. Odds ratios and 95% confidence intervals are reported from logistic regression models. A two-tailed p-value of ≤ 0.05 defined statistical significance.

Results

Baseline serum was available for 333 of the 626 women enrolled in the PPCOS trial. This included 124 participants randomized to OI with clomiphene alone, 124 participants to clomiphene and metformin, and 85 participants to metformin alone. Demographic characteristics are shown by AMH category in Table 1 and by treatment arm in Supplemental Table 1. The three treatment arms were similar with respect to all characteristics with the exception of smoking status; there was a higher proportion of current smokers in the clomiphene alone and clomiphene and metformin groups compared to the metformin alone group (Supplemental Table 1). 322 participants had serum available for AMH testing; among these women the mean baseline AMH concentration was 11.7 ng/mL with a standard deviation of 8.3 (range 0.1–43.0 ng/mL). Stratified by clinical baseline AMH, 39 individuals had an AMH concentration <4 ng/mL, 90 had an AMH concentration 4–8 ng/mL and 193 had an AMH concentration >8 ng/mL.

Table 1.

Patient characteristics by AMH (anti-Müllerian hormone) category

Characteristic*	AMH <4ng/mL (n=39)||	AMH 4–8ng/mL (n=90)	AMH >8ng/mL (n=193)	p-value†
Treatment Group				0.335
Clomiphene	18 (46%)	37 (41%)	63 (33%)
Clomiphene + Metformin	12 (31%)	29 (32%)	81 (42%)
Metformin	9 (23%)	24 (27%)	49 (25%)
Age, years	30.2 ± 4.4	28.2 ± 3.8	27.6 ± 3.6	0.002
BMI, kg/m2	36.9 ± 7.5	36.8 ± 9.0	33.0 ± 8.2	0.001
Waist circumference, cm	106.5 ± 16.9	106.2 ± 18.8	98.9 ± 21.4	0.001
Waist to hip ratio	0.87 ± 0.08	0.87 ± 0.10	0.85 ± 0.08	0.076
Ethnicity, n (%)				0.149
Hispanic/Latina	16 (41%)	28 (31%)	50 (26%)
Non-Hispanic/Latina	23 (59%)	62 (69%)	143 (74%)
Race
American Indian/Alaska Native	10 (26%)	7 (7.8%)	16 (8.3%)	0.009
Asian	2 (5.1%)	3 (3.3%)	8 (4.1%)	0.841
Black	6 (15%)	10 (11%)	19 (9.8%)	0.588
White	21 (54%)	70 (78%)	151 (78%)	0.005
Ultrasonographic Findings
Combined ovarian volume, cm3	17.6 ± 6.7	19.6 ± 9.3	25.7 ± 12.1	<0.001
Morphologic features of polycystic ovary in at least 1 ovary, n (%)	26 (72%)	77 (90%)	178 (94%)	0.001
Ferriman-Gallwey hirsutism score	15.3 ± 9.5	15.1 ± 6.9	13.7 ± 7.6	0.279
Baseline fasting serum testing
Insulin, mU/mL	28.4 ± 27.1	25.6 ± 29.7	19.7 ± 24.3	<0.001
Proinsulin, pmol/L	24.7 ± 18.3	27.4 ± 24.7	19.9 ± 19.0	<0.001
Glucose, mg/dl	91.3 ± 20.3	90.5 ± 15.7	88.0 ± 16.1	0.163
Luteinizing hormone, mIU/mL	7.7 ± 5.6	11.9 ± 7.6	14.1 ± 9.0	<0.001
SHBG, nmol/L	28.1 ± 13.2	26.0 ± 17.1	30.7 ± 16.7	0.010
Total testosterone, ng/dl	50.2 ± 25.7	55.9 ± 24.3	68.1 ± 31.2	<0.001
Free androgen index‡	7.3 ± 4.4	9.5 ± 6.4	10.1 ± 7.3	0.075
HOMA-IR§	6.8 ± 7.4	6.0 ± 7.8	4.6 ± 7.4	<0.001
Days in Study	177 ± 72	190 ± 63	184 ± 65	0.638
Previous Diagnosis of Infertility	33 (85%)	70 (78%)	178 (92%)	0.003
Previous History of Pregnancy	18 (46%)	29 (32%)	64 (33%)	0.259
Smoking Status				0.942
Never Smoker	26 (67%)	54 (60%)	118 (61%)
Ever Smoker	7 (18%)	22 (24%)	29 (15%)
Current Smoker	6 (15%)	14 (16%)	46 (24%)

Abbreviations: AMH=anti-Müllerian hormone, BMI = body mass index, SHBG = sex hormone binding globulin, HOMA-IR = Homeostatic Model Assessment for Insulin Resistance

^*Data are presented as mean ± standard deviation or number (percentage).

^†p-values are calculated by Kruskal-Wallis rank sum test, Pearson’s chi-squared test or Fisher’s exact test as appropriate.

^‡Free androgen index was calculated with the following formula: (total testosterone [nanomoles per liter]/SHBG [nanomoles per liter]) x 100

^§HOMA-IR was calculated with the following formula: (insulin x glucose)/405

^||Data for baseline AMH concentrations were missing in 11 participants: 6 in the clomiphene arm, 2 in the clomiphene + metformin arm, and 3 in the metformin arm.

Women who ovulated had significantly lower BMI, waist circumference, insulin, proinsulin, and Homeostatic Model Assessment for Insulin Resistance (HOMA-IR) than those who did not ovulate; women who ovulated were also slightly older on average (Supplemental Table 2).

In the unadjusted analysis, baseline AMH concentration was significantly associated with odds of ovulation with OI during the study period; for each one unit increase in baseline AMH, the odds of ovulation decreased by 8 percent (OR 0.92, 95% CI 0.89–0.95; p<0.001; Supplemental Table 3). Comparison of AMH concentration among those who ovulated versus those who did not ovulate is depicted in Figure 1A with division by treatment group in Figure 1B. There was no maximum AMH concentration noted above which no participants ovulated. A sub-analysis of those who ovulated demonstrated that those who ovulated after two cycles of treatment had a baseline AMH concentration 2.2 ng/mL greater than those who ovulated in the first cycle of treatment (95% CI 0.17–4.3, p=0.034). However, when cycles of treatment to ovulation was treated as categorical, with the referent group experiencing ovulation in the first cycle of treatment, there was no significant difference noted.

Figure 1A.

Box plot depicting no ovulation (red) vs ovulation (blue) by anti-Müllerian hormone concentration. 1B. Box plot depicting no ovulation (red) and ovulation (blue) by anti-Müllerian hormone concentration in each treatment group. 1C. Predicted probability of ovulation with ovulation induction treatment by baseline AMH concentration for first quartile of BMI (28.6 kg/m2), median BMI (34.5 kg/m2), and third quartile of BMI (40.6 kg/m2).

There was no significant association between baseline AMH concentration and clinical pregnancy or live birth (unadjusted OR 1.00, 95% CI 0.97–1.03, p=0.939 and unadjusted OR 1.01, 95% CI 0.98–1.04, p=0.674, respectively; Supplemental Table 3). Treatment group was not a significant effect modifier of the association between baseline AMH concentration and ovulation, clinical pregnancy, or live birth (p=0.075, p=0.2, p=0.10, respectively). Race was not noted to be a significant effect modifier of the association between baseline AMH concentration and ovulation (p=0.54). BMI was, however, a significant effect modifier of the association between baseline AMH concentration and ovulation, with women with higher BMI having a steeper decline in probability of ovulation with increasing baseline AMH concentration compared to those with lower BMI (p=0.009; Figure 1C).

The following variables were shown to be potential confounders in univariate logistic regression models assessing association with ovulation: BMI, days in study and HOMA-IR (Supplemental Table 4). There was no association between hyperandrogenism and odds of ovulation, with hyperandrogenism measured by Ferriman-Galwey hirsutism score, total testosterone, and free androgen index. Age was included in all adjusted models given the established inverse association with AMH concentration. For clinical pregnancy, possible confounders were BMI, LH concentration, free androgen index, days in study, race (categorized as black, white, or other) and current smoking status. For live birth, potential confounders were BMI, LH concentration, baseline HOMA-IR, Ferriman-Gallwey hirsutism score, free androgen index, days in study, and race. When adjusting for the aforementioned confounders, a significant inverse association between baseline AMH concentration and ovulation remained such that for each one unit in baseline AMH, the odds of ovulation decreased by 10 percent (OR 0.90, 95% CI 0.86–0.93, p=0.004; Supplemental Table 3).

When participants were divided into clinically relevant categories by baseline AMH concentration, those in the highest AMH group (>8 ng/mL) were significantly less likely to ovulate with OI treatment compared to those in the normal AMH group (<4 ng/mL) (OR 0.23, 95% CI 0.05–0.68; p=0.019; Table 2). This difference remained statistically significant when adjusting for treatment arm (OR 0.20 for ovulation in >8 ng/mL group compared to <4 ng/mL group, 95% CI 0.05–0.61; p=0.012) and when adjusting for confounders for ovulation (OR 0.17 for ovulation in >8 ng/mL group compared to <4 ng/mL group, 95% CI 0.04,0.56; p=0.008; Table 2).

Table 2.

AMH concentration categories and association with ovulation

Baseline AMH concentration category	Number of participants	Unadjusted Odds Ratio for Ovulation (95% CI; p-value)	Odds Ratio for Ovulation Adjusted for Treatment Arm (95% CI, p-value)*	Odds Ratio for Ovulation Adjusted for Potential Confounders (95% CI, p-value)†
<4 ng/mL	39	Reference	Reference	Reference
4–8 ng/mL	90	0.75 (0.16,2.69; p=0.70)	0.76 (0.16,2.76; p=0.70)	0.90 (0.18,3.47; p=0.90)
>8 ng/mL	193	0.23 (0.05,0.68; p=0.019)	0.20 (0.05,0.61; p=0.012)	0.17 (0.04,0.56; p=0.008)

^*The model included adjustment for treatment arm.

^†The model included adjustment for potential confounders identified with logistic regression for ovulation: age, BMI, baseline HOMA-IR and days in study.

When participants were divided into tertiles of baseline AMH concentrations, those in the highest tertile (AMH>12.9 ng/mL) were significantly less likely to ovulate than those in the lowest tertile (AMH<7.0 ng/mL) (OR 0.21, 95% CI 0.10–0.44; p<0.001, Table 3). This difference persisted when adjusting for treatment arm (OR 0.20 for highest vs lowest tertile, 95% CI 0.09–0.41; p<0.001), and when adjusting for confounders for ovulation (OR 0.12 for highest vs lowest tertile, 95% CI 0.05–0.28, p<0.001).

Table 3.

AMH concentration by tertile and association with ovulation

Baseline AMH concentration tertile	Unadjusted Odds Ratio for Ovulation (95% CI; p-value)	Odds Ratio for Ovulation Adjusted for Treatment Arm (95% CI, p-value)*	Odds Ratio for Ovulation Adjusted for Potential Confounders (95% CI, p-value)†
<7.0 ng/mL	Reference	Reference	Reference
7.0–12.9 ng/mL	0.70 (0.30,1.58; 0.40)	0.53 (0.22,1.24; 0.15)	0.47 (0.19,1.16; p=0.106)
>12.9 ng/mL	0.21 (0.10,0.44; <0.001)	0.20 (0.09,0.41; <0.001)	0.12 (0.05,0.28; p<0.001)

^*The model included adjustment for treatment arm.

^†The model included adjustment for potential confounders for ovulation: age, BMI, baseline HOMA-IR and days in study.

A receiver operative curve was utilized and identified an optimal threshold AMH concentration of 11.18 ng/mL for prediction of ovulation with ovulation induction with an area under the curve of 0.715 (Supplemental Figure 1). This threshold of AMH produced a specificity of 73.0 percent and a sensitivity of 69.5%.

Discussion

In women with anovulatory infertility and PCOS, we found that higher baseline AMH concentration was associated with reduced odds of ovulation with OI treatment, regardless of treatment method (clomiphene, clomiphene+metformin, or metformin). Each one unit increase in AMH concentration was associated with a 10 percent reduction in odds of ovulation in our adjusted analysis. Women with AMH concentration >8 ng/mL were significantly less likely to ovulate than those with normal AMH concentration of <4 ng/mL. This difference persisted when adjusting for both OI agent and confounders associated with ovulation, suggesting that AMH concentration is a critical predictor of whether a patient with PCOS-related infertility will achieve ovulation with OI. Other independent predictors of odds of ovulation with OI in our study included BMI and baseline HOMA-IR score, with both having an inverse association with odds of ovulation. We did not find an association between hyperandrogenism as measured by Ferriman-Galwey hirsutism score, total testosterone, or free androgen index and odds of ovulation. We noted that the relationship between baseline AMH concentration and ovulation was dependent on BMI, with predicted probability of ovulation declining more steeply by AMH concentration in women in with higher BMI. No association between baseline AMH concentration and pregnancy or live birth rates was noted in our cohort.

Our findings are generally consistent with the prior studies of OI treatment among women with PCOS. Mumford et al. conducted a secondary analysis of data from the PPCOS II randomized clinical trial of 748 women and found a lower mean AMH concentration among women who ovulated with OI with clomiphene or letrozole (3). In their analysis, each one unit increase in AMH concentration was associated with a 6 percent reduction in odds of ovulation, comparable to our finding of a 10 percent reduction with each unit increase in AMH. Interestingly, the mean baseline AMH concentration in our study was higher than that of women in the PPCOS II trial with our mean AMH concentration being 11.7 ng/mL, whereas the fourth quartile of AMH in the PPCOS II cohort was >10.23 ng/mL. Further, Xi et al. prospectively studied 81 anovulatory women with PCOS treated with clomiphene and found that AMH concentration independently predicted ovulation with a threshold AMH concentration of 7.77 ng/mL identified using a receiver operating curve, while we identified a threshold AMH concentration of 11.18 ng/mL as predictive of ovulation (19). This may suggest our study includes women with a more severe PCOS phenotype, or that modern AMH assays are trending towards higher values.

In contrast, Sachdeva et al. found in their prospective observational study of 164 women with PCOS-related infertility treated with clomiphene that Ferriman-Gallwey score was the best predictor of non-responsiveness of clomiphene treatment, and androstenedione was another independent predictor (22). In our analysis, Ferriman-Gallwey score and biochemical hyperandrogenism (by total testosterone and free androgen index) were not associated with the odds of ovulation. This discrepancy could be due to the addition of metformin in our study; the majority of participants (two of three treatment arms) were treated with either combination therapy with clomiphene and metformin or metformin alone. Moreover, our study consisted of a larger study population with randomization to OI treatment arms.

Our study has several strengths, including a large study population of over 300 individuals who were well characterized with respect to PCOS diagnosis, including demographics, biomarkers, and ultrasonographic data. Additionally, this was a randomized trial and baseline testing was similar between the three OI treatment arms. Participants were monitored prospectively throughout their participation with frequent measurement of serum progesterone levels, providing accurate assessment of ovulation. A limitation of our study is that baseline serum samples for analysis were available for only 333 of the 626 participants in the PPCOS I trial; however, this sub-group had similar outcomes with OI treatment as the original trial and therefore likely is a representative sample of the original study population. Additionally, the serum samples had been stored for several years prior to AMH concentration being analyzed for this study; however, we expect given their storage at −80 degrees that concentrations should remain stable over time (23, 24). Finally, since letrozole was not routinely used for OI treatment at the time of the PPCOS I study, we were unable to evaluate for associations between AMH and this treatment method.

Our data corroborate and strengthen the findings of the PPCOS II study, by again demonstrating the association between elevated AMH levels and reduced odds of ovulation in a well-characterized cohort of women with PCOS. Baseline AMH concentrations seem to be more important than other biometric and phenotypic factors when predicting who will ovulate with OI treatment. Additionally, women with higher BMI were noted to have a steeper decline in probability of ovulation with increasing baseline AMH concentration, indicating that BMI significantly modified the effect of baseline AMH on odds of ovulation. AMH values using the Ansh assay in this PCOS cohort yielded high AMH values compared to the normal values by age currently published (25). Newer data have suggested that AMH concentration can serve as a proxy for AFC, and in 2023, international guidelines have called for inclusion of AMH in the evaluation and diagnosis of PCOS (6–9). However, no clear AMH cutoff value has been established due to limited data on AMH values in persons with PCOS, and thus data from this cohort is an essential contribution toward achieving clear PCOS criteria. New studies are essential to add to the body of literature describing AMH concentrations because AMH assays continue to change over time. Further, it is unlikely that a study as large and well described will be reproduced in the near future and so our data add to the literature by demonstrating the wide range of AMH concentrations seen in women with infertility and PCOS, which will be helpful in guiding both decision-making for diagnosis and treatment of women with these conditions. In addition, studies describing AMH concentrations will be important to the multidisciplinary team caring for persons with PCOS to start interpreting a lab that they are unlikely to be familiar with; in particular primary care physicians, medical endocrinologists, and general gynecologists may need guidance on the clinical management of AMH values. Interpretation of AMH results will be an increasingly important skill for clinicians, and ultra-high AMH levels should prompt a referral to specialists in reproductive endocrinology.

Conclusions

Our study demonstrates that increased baseline AMH concentration was associated with reduced odds of ovulation with clomiphene and metformin OI treatment for PCOS-related infertility even when considering age, BMI, hyperandrogenism, insulin resistance, and ultrasonographic findings of PCOS. We evaluated AMH continuously and in clinically meaningful categories and demonstrated that the odds of ovulation were lower in women with higher AMH levels. This significant finding will guide counseling for women with PCOS-related infertility regarding their odds of treatment success with OI. Finally, our findings should encourage early referral to reproductive endocrinology in women with very high AMH values, and in women for whom initial attempts at low-dose OI are unsuccessful.

Supplementary Material

1

Supplemental Figure 1. Receiver operating curve for prediction of ovulation with ovulation induction treatment by baseline AMH concentration.

Data Sharing Statement:

Data that underlie the results reported in this article, after deidentification, will be made available to the editors upon request.