Efficacy of predictive models for polycystic ovary syndrome using serum levels of two antimüllerian hormone isoforms (proAMH and AMHN,C)

Abstract

To compare total antimüllerian hormone (AMH), proAMH, AMHN,C, and the ratio of the two forms in predictive models for polycystic ovary syndrome (PCOS) diagnosis. Total AMH consists of proAMH (inactive precursor) and AMHN,C (receptor-competent), but neither isoform has been tested individually for their ability to predict PCOS diagnosis. Cross-sectional study using biobanked samples collected between July 2008 and January2010. Not applicable. Overweight, premenopausal women aged 18-45years with PCOS (n = 45, with 21 fulfilling National Institutes of Health diagnostic criteria and 24 fulfilling European Society for Human Reproduction and Embryology/American Society for Reproductive Medicine (ESHRE) criteria, but not National Institutes of Health criteria) and without PCOS (n = 23 controls). None. Serum concentrations of proAMH and total AMH (proAMH and AMHN,C combined) were determined by immunoassay. The AMHN,C concentrations were calculated by subtraction ([AMHN,C] = [total AMH] - [proAMH]). Relative levels of proAMH were expressed as the AMH prohormone index (API = [ProAMH]/[Total AMH] × 100). In women with PCOS, total AMH, proAMH, and AMHN,C levels were higher, and the API was lower (P=.010), than in controls indicating increased conversion of proAMH to AMHN,C. Receiver-operating characteristic analysis for proAMH (area under the curve [AUC] = 0.82), AMHN,C (AUC = 0.86), and API (AUC = 0.70) did not improve the prediction for PCOS when compared with total AMH (AUC = 0.86). The proAMH and AMHN,C do not appear to improve the ability to predict a diagnosis of PCOS beyond total AMH assays. However, the ratio of inactive proAMH precursor to receptor-competent AMHN,C (API) differs in women with PCOS relative to unaffected controls indicating that AMH signaling mechanisms may be altered in women with PCOS.