Age-stratified thresholds of anti-Müllerian hormone improve prediction of polycystic ovary syndrome over a population-based threshold

Objective: Due to its consistent elevation in polycystic ovary syndrome (PCOS) and correlation with polycystic ovarian morphology (PCOM), anti-Mullerian hormone (AMH) has been proposed as a marker of the syndrome. However, prior studies reporting thresholds of AMH for a PCOS diagnosis have been limited by small sample size, inappropriate controls, and heterogeneous AMH assays. We sought to evaluate the suitability of a standardized AMH assay as a biomarker of PCOS. Design: Cross-sectional study at academic medical centres across the United States. Patients: Women with PCOS were diagnosed by Rotterdam criteria and included 282 subjects from the multisite PPCOS II trial and 109 patients from a tertiary academic centre's multidisciplinary PCOS clinic. Controls included 245 participants in the ovarian ageing (OVA) study, a community-based cohort of ovulatory women not seeking treatment for fertility. Measurements: Determination of AMH by a central laboratory. Receiver-operating characteristic (ROC) analyses were used to investigate the accuracy of AMH thresholds for prediction of PCOS diagnosis with stratification by age. Results: The optimal threshold of AMH to distinguish PCOS from controls was 55.36 pmol/L (sensitivity: 0.82, specificity: 0.78, J: 0.60). When examining the population by age groups, the optimal AMH threshold decreased with increasing age. Conclusions: AMH is an effective biomarker of PCOS. Age-stratified thresholds more accurately predicted PCOS than an overall population-based threshold.