Anti-Müllerian hormone as a diagnostic tool for PCOS under different diagnostic criteria in an unselected population

Is anti-Müllerian hormone (AMH) a valid tool to diagnose polycystic ovary syndrome (PCOS) according to different subsets of criteria among an unselected group of women?

In this cross-sectional study, AMH concentrations were measured in an unselected group of women. The ability of AMH to diagnose PCOS according to National Institutes of Health (NIH), Rotterdam-2003 and Androgen Excess and PCOS Society (AE-PCOS) criteria was tested by using frozen serum aliquots (n = 392) that had been collected from a previous prevalence study of PCOS.

The respective age and body mass index adjusted area under the curve (aAUC, 95% confidence interval) values were 0.80 (0.71-0.89), 0.74 (0.67-0.81) and 0.71 (0.64-0.79). When the definition of polycystic ovary morphology (PCOM) was set to an antral follicle count (AFC) of 20 instead of 12, the prevalence of syndrome dropped from 19.9% to 10.2% and from 15.3% to 8.9% according to Rotterdam-2003 and AE-PCOS criteria, respectively. In patients with Phenotype A, who had hyperandrogenism, ovulatory dysfunction and PCOM, AMH had an aAUC of 0.85 (0.77-0.92) to diagnose the syndrome. In Phenotypes B (hyperandrogenism + ovulatory dysfunction), C (hyperandrogenism + PCOM) or D (ovulatory dysfunction + PCOM), AMH had poor to fair ability to diagnose the syndrome.

AMH has poor to fair validity to diagnose PCOS among an unselected group of women, except for patients bearing all features of the syndrome (Phenotype A). This finding is valid using the NIH, Rotterdam-2003 and AE-PCOS criteria and even after revising the definition of PCOM as AFC ≥20.