Materials and Methods
Subjects
This was a retrospective cohort study which included all patients who presented to our fertility clinic for infertility evaluation and treatment and were found to have a random serum AMH level >5 ng/mL between April 2009 and May 2013. We reviewed the medical charts of all patients seen during this time period (n = 1681) and excluded from this study all women with AMH level <5 ng/mL (n = 1547). A total of 134 women were found to have AMH level >5 ng/mL and were included in the study. Women were divided into 3 groups according to serum AMH level: 5 to 10 ng/mL (n = 84), >10 to 14 ng/mL (n = 30), and >14 ng/mL (n = 20). As part of their infertility workup, all women underwent evaluation for PCOS because of their elevated serum AMH level. The women’s medical histories were reviewed including the following parameters: age, body mass index (kg/m 2 ), serum serum follicle-stimulating hormone (FSH), luteinizing hormone (LH), AMH, total testosterone, dehydroepiandrosterone sulfate (DHEAS), 17-OH-progesterone, thyroid-stimulating hormone, and prolactin. The diagnosis of PCOS was made when at least 2 of the following 3 criteria existed, as proposed by the Rotterdam Consensus Meeting: oligomenorrhea or amenorrhea, clinical hyperandrogenism and/or hyperandrogenemia, and polycystic ovaries. No PCOS patient had evidence of hyperprolactinemia, thyroid disease, Cushing’s syndrome, congenital adrenal hyperplasia, or androgen-secreting tumors. This study was approved by our institutional review board.
Ovarian stimulation
Women were treated using a stimulation protocol that either included downregulation using a gonadotropin-releasing hormone (GnRH) agonist in a long protocol or a GnRH antagonist to prevent premature ovulation. Ovarian stimulation was performed using a combination of recombinant FSH and human menopausal gonadotropin. Ultrasound for follicular tracking and blood sampling for estradiol levels were performed every 1-3 days. When at least 6 follicles with a diameter of 16 mm were detected, either 5000 or 10,000 IU human chorionic gonadotropin (hCG) was administered, depending on the estimated risk for hyperstimulation. Oocytes were retrieved 35 hours following hCG under transvaginal sonographic guided needle puncture. In patients who were deemed to be at high risk for ovarian hyperstimulation syndrome (OHSS), all embryos were cryopreserved. The implantation rate was calculated as the number of gestational sacs divided by the number of embryos transferred. Clinical pregnancy was defined as the presence of a gestational sac on ultrasound performed at 6 weeks after embryo transfer. The clinical pregnancy rate was calculated as the number of clinical pregnancies divided by the number of embryo transfer procedures.
AMH assay
Random serum AMH levels for each woman, unrelated to the day of the menstrual cycle, were measured by enzyme linked immunosorbent assay at Reprosource Inc. (Woburn, MA) as previously described. Intra- and interassay coefficients of variation with serum controls were approximately 5-9% and 7-12%, respectively.
Statistical analysis
For comparison of clinical characteristics and PCOS phenotypes, women were divided into 3 groups according to serum AMH level: 5 to 10 ng/mL (n = 84), >10 to 14 ng/mL (n = 30), and >14 ng/mL (n = 20). For comparing ART cycle characteristics and outcomes, women who underwent ART of the original population were divided into 2 groups according to serum AMH level: 5 to 10 ng/mL (n = 37), and >10 ng/mL (n = 21). In this case, all women with AMH >10 ng/mL were grouped together to ensure sufficient numbers for statistical analysis. Data are expressed as mean ± standard deviation. Mean differences between multiple groups were compared using 1-way analysis of variance; otherwise, medians were tested using Kruskal-Wallis test. The Bonferroni correction was applied for multiple comparisons in post hoc tests. Mean differences between 2 groups were compared using Student t test or Mann-Whitney test, as appropriate. Proportions were compared between groups using χ 2 test. Because of a skewed distribution, AMH values were log transformed, and log AMH was compared between groups. The degree of correlation between AMH level and various continuous variables was determined using Pearson’s correlation tests. Receiver operating characteristic (ROC) curves were generated to investigate the diagnostic test performance of random AMH level for amenorrhea. The sensitivity and specificity were calculated for the optimal AMH cutoff level determined by ROC curve analysis. All significance tests were 2-tailed and P < .05 was noted to be statistically significant. SigmaPlot (Systat Software Inc., Chicago, IL) and MedCalc software (Mariakerke, Belgium) were used for statistical analysis.
Results
To estimate the prevalence of ultrahigh serum AMH levels (>10 ng/mL) in the general population of reproductive age women, we examined our data from a large fertility focused US-based laboratory (Reprosource Inc.). Of 27,637 total women who were tested for serum AMH level as part of their infertility workup between 2012-2013, 23,862 women (86.3%) had AMH <5 ng/mL; 3003 women (10.9%) had AMH between 5 to 10 ng/mL; and 772 women (2.8%) had AMH >10 ng/mL. Random serum AMH level >5 ng/mL was chosen to identify women at high-risk of polycystic ovarian morphology (PCOM), consistent with previously suggested threshold. In total, 134 women in our study population were found to have random serum AMH level >5 ng/mL. The subjects were divided into 3 groups according to serum AMH level: 5 to 10 ng/mL (n = 84); >10 to 14 ng/mL (n = 30); and >14 ng/mL (n = 20). The patients’ clinical and biochemical characteristics are shown in Table 1 . Although no significant age differences were noted between groups, BMI was significantly greater in women with AMH >10-14 ng/mL compared with the other 2 groups. LH was significantly higher in both AMH >10-14 and >14 ng/mL groups as compared with women with AMH 5-10 ng/mL. Moreover, as AMH increased, the LH/FSH ratio significantly increased from 0.98 in the 5-10 ng/mL group, to 1.6 in AMH >10-14 group, and 2.2 in >14 ng/mL group. Similarly, as AMH increased, testosterone level demonstrated a gradual increase across AMH groups and was highest in those with AMH >14 ng/mL. DHEAS was also highest in women with AMH >14 ng/mL. Overall, women in the AMH >14 ng/mL group had significantly greater rate of hyperandrogenemia (80%) as compared with 5-10 ng/mL (38%) and >10-14 ng/mL (47%) groups. In addition, women with hyperandrogenism (HA) had significantly greater serum AMH level compared with women without HA ( Figure 1 , A).
| Variable | Serum AMH, ng/mL | P value | ||||
|---|---|---|---|---|---|---|
| 5-10 (n = 84) | >10-14 (n = 30) | >14 (n = 20) | 5-10 vs >10-14 | 5-10 vs >14 | >10-14 vs >14 | |
| AMH, ng/mL | 6.8 (1.5) | 11.65 (1.1) | 22.95 (10.1) | |||
| BMI, kg/m 2 | 24.3 (5.0) | 27.2 (5.7) | 24.6 (4.5) | .01 | NS | .04 |
| Age, y | 30.2 (5.2) | 30.1 (3.9) | 29.5 (4.6) | NS | NS | NS |
| FSH, IU/L | 5.4 (2.2) | 5.4 (1.5) | 5.2 (1.6) | NS | NS | NS |
| LH, IU/L | 5.3 (3.1) | 8.6 (5.5) | 11.9 (7.9) | .02 | .002 | NS |
| LH/FSH ratio | 0.98 (0.6) | 1.6 (1.0) | 2.2 (1.1) | .01 | .001 | .04 |
| Testosterone, ng/dL | 42.8 (20.9) | 56.2 (28.4) | 75.9 (22.8) | .04 | < .001 | .02 |
| DHEAS, μg/dL | 201.3 (93.6) | 188.1 (85) | 249.2 (104) | NS | NS | .05 |
| Hyperandrogenemia, % | 38 | 47 | 80 | NS | < .001 | .03 |
| Polycystic ovaries, % | 54.2 | 97 | 100 | < .001 | < .001 | NS |
| Menstrual regularity | ||||||
| Regular periods, % | 49.4 | 17 | 15 | .002 | .005 | NS |
| Oligomenorrhea, % | 49.4 | 77 | 55 | .009 | NS | NS |
| Amenorrhea, % | 1.2 | 6.7 | 30 | NS | < .0001 | .03 |
| PCOS diagnosis, % | 51.8 | 97 | 100 | < .001 | < .001 | NS |
| Infertility cause if present in addition to PCOS | ||||||
| Male factor, % | 40 | 36.7 | 33.3 | NS | NS | NS |
| Tubal factor, % | 10 | 10 | 5 | NS | NS | NS |
| Endometriosis, % | 1.2 | 3.3 | 0 | NS | NS | NS |
The rate of PCOM was significantly greater in women with AMH >10-14 ng/mL (97%) and >14 ng/mL (100%) compared with AMH 5-10 ng/mL (54.2%). Similarly, PCOS diagnosis was significantly more prevalent in the AMH >10-14 ng/mL (97%) and >14 ng/mL (100%) groups compared with 5-10 ng/mL group (51.8%) ( Table 1 ). Women with PCOM had significantly greater serum AMH level compared with those without PCOM ( Figure 1 , B).
A significantly greater proportion of women with AMH 5-10 ng/mL had regular periods (49.4%) compared with women in >10-14 ng/mL (17%) and >14 ng/mL (15%) groups, whereas the proportion of women with amenorrhea was significantly greater in women with AMH >14 ng/mL (30%) compared with 5-10 ng/mL (1.2%) and >10-14 ng/mL (6.7%) groups ( Table 1 ). Women with amenorrhea demonstrated significantly greater serum AMH level than oligoovulatory or normoovulatory women ( Figure 1 , C). ROC curve analysis revealed that AMH had strong diagnostic ability for amenorrhea in our study population ( Figure 2 ). The AUC for AMH in predicting amenorrhea was 0.87 (95% confidence interval, 0.80–0.92; P < .0001). The largest Youden index and optimal specificity (91.7%) and sensitivity (79.4%) for the prediction of amenorrhea was obtained when the threshold AMH concentration was 11.4 ng/mL.
The correlation analyses showed that serum AMH levels correlated positively with LH (r = 0.225, P = .049) total testosterone (r = 0.323, P = .002) and DHEAS (r = 0.195, P = .038). In contrast, AMH levels did not correlate with age (r = −0.03, P = .7), BMI (r = −0.001, P = .99) or FSH (r = −0.02, P = .89).
Patients who underwent ovarian stimulation for ART were divided into 2 groups according to serum AMH level: 5-10 ng/mL (n = 37) and >10 ng/mL (ultrahigh AMH) (n = 21). ART patient and cycle characteristics were compared between groups and are summarized in Table 2 . The mean AMH levels for the 5-10 ng/mL and ultrahigh AMH groups were 6.4 ± 1.4 ng/mL and 17.9 ± 9.5 ng/mL, respectively. Age and BMI were not significantly different between the 2 groups. Gonadotropin dose and estrogen on day of hCG administration were comparable between groups. The number of oocytes retrieved was increased in the ultrahigh AMH group compared with the 5-10 ng/mL group, but it did not reach statistical significance. However, the number of good quality embryos was significantly greater (approximately 2-fold) in the ultrahigh AMH group compared with the 5-10 ng/mL AMH group. In addition, women with ultrahigh AMH levels developed OHSS at significantly greater rates (>3-fold) compared with women with AMH 5-10 ng/mL (28.6% vs 8.1%, P = .04) ( Table 2 ). Embryo transfers were canceled because of OHSS risk in 3 women with AMH 5-10 ng/mL, and 5 women with ultrahigh AMH levels.
| Characteristic | AMH 5-10 ng/mL | AMH >10 ng/mL | P value |
|---|---|---|---|
| No. of cycles | 37 | 21 | |
| Age | 30.9 ± 5.4 | 30.2 ± 3.8 | NS |
| BMI, kg/m 2 | 24.6 ± 3.6 | 27.3 ± 6.1 | NS |
| AMH, ng/mL | 6.4 ± 1.4 | 17.9 ± 9.5 | |
| AMH range, ng/mL | 5.0–9.9 | 10–48 | |
| No. of oocytes retrieved | 13.5 ± 4.8 | 16 ± 8.2 | NS |
| No. of good quality embryos | 2.1 ± 1.2 | 4.3 ± 3.5 | .03 |
| Gonadotropin dose, IU | 2184 ± 1059 | 2249 ± 992 | NS |
| E2 on day of hCG, pg/mL | 2875 ± 770 | 3048 ± 991 | NS |
| OHSS, % | 8.1 (3/37) | 28.6 (6/21) | .04 |
As shown in Table 3 , although the mean number of embryos transferred and fertilization rate were comparable between women with AMH 5-10 ng/mL and ultrahigh AMH levels, the clinical pregnancy rate was significantly increased in women with ultrahigh AMH levels compared with women with AMH 5-10 ng/mL (75.0% vs 44.1%, P = .04). Moreover, implantation rate and multiple pregnancy rate were also increased in the ultrahigh AMH group compared with the 5-10 ng/mL group (51.6% vs 33.8% and 31.2% vs 20.6%, respectively), although both differences were not statistically significant.
| Variable | AMH 5-10 ng/mL | AMH >10 ng/mL | P value |
|---|---|---|---|
| No. of embryo transfer procedures | 34 | 16 | |
| Mean no. of transferred embryos | 1.9 ± 0.7 | 1.9 ± 0.6 | NS |
| Fertilization rate, % | 63.3 (316/499) | 67.6 (227/336) | NS |
| Multiple pregnancy rate, % | 20.6 (7/34) | 31.2 (5/16) | NS |
| Implantation rate, % | 33.8 (22/65) | 51.6 (16/31) | NS |
| Clinical pregnancy rate, % | 44.1 (15/34) | 75.0 (12/16) | .04 |
Results
To estimate the prevalence of ultrahigh serum AMH levels (>10 ng/mL) in the general population of reproductive age women, we examined our data from a large fertility focused US-based laboratory (Reprosource Inc.). Of 27,637 total women who were tested for serum AMH level as part of their infertility workup between 2012-2013, 23,862 women (86.3%) had AMH <5 ng/mL; 3003 women (10.9%) had AMH between 5 to 10 ng/mL; and 772 women (2.8%) had AMH >10 ng/mL. Random serum AMH level >5 ng/mL was chosen to identify women at high-risk of polycystic ovarian morphology (PCOM), consistent with previously suggested threshold. In total, 134 women in our study population were found to have random serum AMH level >5 ng/mL. The subjects were divided into 3 groups according to serum AMH level: 5 to 10 ng/mL (n = 84); >10 to 14 ng/mL (n = 30); and >14 ng/mL (n = 20). The patients’ clinical and biochemical characteristics are shown in Table 1 . Although no significant age differences were noted between groups, BMI was significantly greater in women with AMH >10-14 ng/mL compared with the other 2 groups. LH was significantly higher in both AMH >10-14 and >14 ng/mL groups as compared with women with AMH 5-10 ng/mL. Moreover, as AMH increased, the LH/FSH ratio significantly increased from 0.98 in the 5-10 ng/mL group, to 1.6 in AMH >10-14 group, and 2.2 in >14 ng/mL group. Similarly, as AMH increased, testosterone level demonstrated a gradual increase across AMH groups and was highest in those with AMH >14 ng/mL. DHEAS was also highest in women with AMH >14 ng/mL. Overall, women in the AMH >14 ng/mL group had significantly greater rate of hyperandrogenemia (80%) as compared with 5-10 ng/mL (38%) and >10-14 ng/mL (47%) groups. In addition, women with hyperandrogenism (HA) had significantly greater serum AMH level compared with women without HA ( Figure 1 , A).
