Meghan Smith, M.D. (a), Jacqueline R. Ho, M.D. (a), Lynda K McGinnis, Ph.D. (a), Lihong Ma, M.D. (a), Miryoung Lee, Ph.D. (b), Stefan A. Czerwinski, Ph.D. (b), Tanya Glenn, M.D. (c), David R. Cool, Ph.D. (c), Pascal Gagneux, Ph.D. (d), Frank Z Stanczyk, Ph.D. (a), and Steven R. Lindheim, M.D., M.M.Mc (c, e)
(a) Division of Reproductive Endocrinology and Infertility, Department of Obstetrics and Gynecology, University of Southern California, Los Angeles, California
(b) Department of Epidemiology, Human Genetics, and Environmental Sciences, The University of Texas Health Science Center at Houston, School of Public Health, Brownsville, Texas
(c) Division of Reproductive Endocrinology and Infertility, Department of Obstetrics and Gynecology, Wright State University Boonshoft School of Medicine, Dayton, Ohio
(d) Department of Cellular and Molecular Medicine, University of California San Diego
(e) Shanghai Jiaotong University School of Medicine, Shanghai, People’s Republic of China
Support provided by the Goldhirsh-Yellin Foundation Research Award, the Dayton Area Graduate Medical Education Community Research Grant, and the National Institutes of Health (grant R01HD12252).
Reproductive endocrinologists have long searched for the holy grail in predicting oocyte quality and success (i.e., live birth rate) in assisted reproduction. In this pursuit, several hormones that relate to ovarian function have been identified, including antimüllerian hormone (AMH) and inhibin B. AMH is produced by the granulosa cells of the developing pre-antral to small antral follicles in the ovary. It has now become well established as a biochemical marker for ovarian reserve in adult, infertile women (1). Within the realm of assisted reproduction, AMH has been shown to aid in predicting response to ovarian stimulation with exogenous gonadotropins (1). However, a population-based study has shown that AMH cannot be reliably used to predict the chance of spontaneous conception as measured by positive pregnancy testing (2). Has the utility of AMH been over-promised? Or have we just begun to scratch the surface of this hormone’s role in the physiology and pathophysiology of disorders of the hypothalamic-pituitary-ovarian axis, including precocious puberty, polycystic ovarian syndrome (PCOS), and premature ovarian insufficiency?
AMH levels have been found to be significantly elevated in women with PCOS, presumably because a larger number of follicles are arrested in the pre-antral phase in these women (3, 4). PCOS is one of the most common endocrine disorders in reproductive aged women, with an estimated prevalence of 6% to 10% when using the United States National Institutes of Health criteria, and as high as 15% when broader Rotterdam criteria are applied (5, 6). Work has been done to see if AMH can be adjunctive in the diagnosis of PCOS. A meta-analysis of 10 studies revealed that an AMH cut-off level of 4.7 ng/mL had a sensitivity and specificity of 82.8% and 79.4%, respectively, in diagnosing PCOS (3). Furthermore, given that PCOS is associated with long-term health consequences, including insulin resistance and metabolic syndrome (MetS), several investigators have examined whether the level of AMH can reliably predict the presence of metabolic dysfunction in those with PCOS (7, 8).
In a cross-sectional study of women aged 18 to 46 years with PCOS, Feldman et al. (7) reported that while AMH levels were significantly correlated with body mass index (BMI), blood pressure, and total testosterone, among others, they found that a lower AMH in women with PCOS was associated with an increased risk of MetS (7). In contrast, Nardo et al. (9) found that higher AMH values were associated with elevated androgens and insulin resistance in adult women with PCOS. Pinola et al. (10) demonstrated in adolescent girls at age 16 years a significant correlation between AMH, testosterone, and incidence of oligoovulation or anovulation but did not reveal any correlation between AMH and metabolic markers.
Thus, to continue to elucidate the role of AMH in predicting anovulation and metabolic dysfunction, we sought to use longitudinal data to examine changes in AMH from childhood to adulthood and its association with anovulation, sex steroids, marker of metabolic dysfunctions, and MetS.
Our primary objective was to determine if higher AMH values were associated with the presence of oligoovulation or anovulation, MetS, or other markers of metabolic dysfunction, including hypertension, elevated fasting glucose, and dyslipidemia. Secondary objectives included examining if elevated AMH values were associated with ovarian hyperandrogenism, based on total and free testosterone levels, and whether AMH, total or free testosterone (T), or dehydroepiandrosterone sulfate (DHEAS) could predict the presence of MetS.
Materials and Methods:
We analyzed the serum of 85 participants in the Fels Longitudinal Study (11). In 1929 this study began in Yellow Springs, Ohio as a way to follow child growth and development during the depression in order to help protect the health of children. Following this time, additional data were collected in order to determine the risk factors for various conditions including cardiovascular disease and obesity, as well as growth charts and maturation. From its beginning, participants have not been selected based on health status or any other obesity-, cardiovascular disease -, or diabetes-related trait. Participants are not examined when menstruating, pregnant, or having other transient conditions that could affect data quality.
Participants included in the analysis were between the ages of 10 and 19.2 years at their baseline visit and were followed over time to a maximum age of 39 years. The younger patients were included to examine AHM changes through puberty. Subjects completed 1 to 10 serial visits (median 3, total 298 visits) from 1990 to 2015. At each study visit, demographics, anthropometrics, fasting glucose and insulin, DHEAS, and lipids were determined, and blood samples were collected for later analysis. Serum were frozen and stored at -80°C until assayed. The de-identified serum samples were shipped on dry ice to the Reproductive Endocrine Research Laboratory at the University of Southern California for the biomarker assays. The Ultrasensitive AMH ELISA from Ansh Labs was utilized to determine AMH levels, with an assay sensitivity of 60 pg/mL. Inter-assay coefficients of variation were 9.7% at 1.6 ng/mL and 12.0% at 4.5 ng/mL, respectively. Testosterone was measured by a competitive Elisa Immunoassay following an extraction step using ethyl acetate, hexane. The assay sensitivity was 0.03 ng/dL and the inter-assay Coefficient of Variations (CVs) were 6.9% at 36 ng/dL and 7.1% at 67 ng/dL. For sex hormone binding globulin (SHBG), the sensitivity was 1 nmol/L and the inter-assay CVs were 2.7% at 4 nmol/L and 4.9% at 84 nmol/L. Free T was calculated using a validated algorithm.
Participants were grouped after puberty according to presence of oligoovulation or anovulation, biochemical hyperandrogenism, the presence of any one marker of metabolic dysfunction, and MetS. Markers of metabolic dysfunction included abdominal obesity (waist > 88 cm for adults or ≥ 90th percentile for children), hypertension (systolic >130 mmHg, diastolic > 85 mm Hg or taking hypertensive medication), low high-density lipoprotein (< 40 mg/dL for age < 16 y or <50 mg/dL for age ≥16 y), hypertriglyceridemia (>150 mg/dL or taking any lipid lowering medication), and elevated fasting glucose (>100 mg/dL or taking any glucose lowering medication). MetS in adults was defined by the American Heart Association and National Heart Lung and Blood Institute 2005 criteria, which requires three out of the above five cardiometabolic abnormalities. In children, MetS was defined by the International Diabetes Federation consensus criteria, which differs in its criteria based on the child’s age. We also assessed for obesity (BMI > 30 kg/m2 for adults or ≥ 95th percentile for children) in participants. Self-reported oral contraceptive use was included as either present or absent. Body composition was determined by age-specific BMI percentile and waist to height ratio. While we had data on biochemical hyperandrogenism and the presence of regular menstrual cycles, we did not have information on clinical hyperandrogenism or ultrasonographic appearance of the ovaries and, therefore, could not analyze the presence of PCOS in participants in this data set.
Hormones were logarithmically transformed for statistical analysis. A generalized mixed effect model adjusting for serial correlation was used to examine how individual hormones related to the outcome variables (i.e., MetS). To satisfy linear model assumptions, AMH values were natural log transformed (AMHlog) for analyses. This model with a random intercept and unstructured covariance matrix was used for analysis between AMHlog levels and metabolic syndrome (yes or no) as outcome. We also examined the serial association between AMHlog and hormone levels while adjusting for covariates such as age and body composition. Best model selection was based on the approach using the information criteria such as Akaike information criterion and Bayseian information criterion. All data were analyzed using SAS (SAS Institute Inc, Cary North Carolina USA) version 9.4. P values of <.05 were set as statistically significant.
Overall, mean values of the analytes were AMH 5.2 ng/mL (standard deviation, 3.7), total T 29.8 ng/dL (17.5), free T 5.2 pg/mL (4.1), DHEAS 1.4 ug/mL (0.9), and SHBG 72.3 nmol/L (56.2). In the cohort, 21.1% were obese and 34.1% had a large waist circumference at some point during their follow-up period. Markers of metabolic dysfunction were noted in the cohort as follows: hypertension 8.2%; low levels of high-density lipoprotein 30.9%; hypertriglyceridemia 31. 6%; and elevated fasting glucose 3.5%. In total, 10.6% had MetS. Descriptive characteristics of participants at their first visit is represented in Table 1.
When adjusting for age, BMI, and contraceptive use, AMH was not associated with the presence of oligoovulation or anovulation or levels of total T, or SHBG (Table 2). Moreover, AMH, total and free T, and DHEAS were not independently related to the presence of MetS. However, AMH was significantly associated with free Tlog (β = 0.133, P =.040), and marginally associated with DHEASlog (β = 0.178, P =.054).
Based on our findings, AMH does not appear to be a useful clinical marker for those at risk of ovulatory disorders or MetS in the pubertal transition and early adulthood. In the study by Pinola et al. (10) mentioned earlier, the 16-year-old females who reported oligomenorrhea or amenorrhea (n=200) had significantly higher AMH levels compared with those reporting normal cycles (n=200). Compared to our study, the number of those studied was not only much larger, but also the age was restricted to 16 years. The study by Pinola et al. (10) also showed that AMH is not a good marker for metabolic factors, which is consistent with our findings.
Our data showed a significant correlation between AMH and free T levels. Serum AMH levels at age 16 correlated significantly with total T levels in the study by Pinola et al. (10) but free T levels were not measured in that study. Abnormal serum levels of total T are found in about 30% of PCOS patients, but abnormal free T is found in about 60% of these patients (12). Free T appears to be the single most predictive marker of hyperandrogenemia in PCOS patients. Our study also showed that AMH was marginally associated with DHEAS. This androgen is elevated in approximately 30% of women with PCOS (12). Although we could not analyze the presence of PCOS in the participants in the present study, our findings with AMH, free T, and DHEAS warrant further longitudinal studies in which these markers are evaluated in adolescent girls with PCOS.
At present, caution should be used when counseling non-infertile patients on any broader implications that an AMH level may have on their natural fertility, risk for PCOS, or jeopardy for other metabolic disorders. Thus, while the search for the holy grail continues, where we seek other novel markers of ovarian function, the story of AMH presently reveals that care should be taken when trying to extrapolate findings to a broader context.
1. Broer SL, Dólleman M, van Disseldorp J, Broeze KA, Opmeer BC, Bossuyt PM, et al. Prediction of an excessive response in in vitro fertilization from patient characteristics and ovarian reserve tests and comparison in subgroups: an individual patient data meta-analysis. Fertil Steril 2013;100(2):420-9.e7.
2. Steiner AZ, Pritchard D, Stanczyk FZ, Kesner JS, Meadows JW, Herring AH, Baird DD. Association between biomarkers of ovarian reserve and infertility among older women of reproductive age. JAMA 2017;318(14):1367-76.
3. Iliodromiti S, Kelsey TW, Anderson RA, Nelson SM. Can anti-Mullerian hormone predict the diagnosis of polycystic ovary syndrome? A systematic review and meta-analysis of extracted data. J Clin Endocrinol Metab 2013;98(8):3332-40.
5. Zawadski J, Dunaif A. Diagnostic criteria for polycystic ovary syndrome: Towards a rational apporach. In: Dunaif A, Givens J, Haseltine F, Merriam G, eds Polycystic ovary syndrome. Boston: Blackwell Scientific Publications; 1992:377-84
6. 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.
8. Hillman JK, Johnson LN, Limaye M, Feldman RA, Sammel M, Dokras A. Black women with polycystic ovary syndrome (PCOS) have increased risk for metabolic syndrome and cardiovascular disease compared with white women with PCOS. Fertil Steril 2014;101(2):530-5.
9. Nardo LG, Yates AP, Roberts SA, Pemberton P, Laing I. The relationships between AMH, androgens, insulin resistance and basal ovarian follicular status in non-obese subfertile women with and without polycystic ovary syndrome. Hum Reprod 2009;24(11):2917-23.
10. Pinola P, Morin-Papunen LC, Bloigu A, Puukka K, Ruokonen A, Järvelin MR, et al. Anti-Müllerian hormone: correlation with testosterone and oligo- or amenorrhoea in female adolescence in a population-based cohort study. Hum Reprod 2014;29(10):2317-25.
Growth, maturation, and body composition: The Fels Longitudinal Study 1929–1991. 1st ed. Cambridge: Cambridge University Press; 1992..
12. Huang A, Brennan K, Azziz R. Prevalence of hyperandrogenemia in the polycystic ovary syndrome diagnosed by the NIH 1990 criteria. Fertil Steril 2010;93:1838-941.