Can serum anti-Müllerian hormone (AMH) replace polycystic ovary morphology (PCOM) determined by ultrasound as a diagnostic component of polycystic ovary syndrome (PCOS)? Despite good correlations between serum AMH and PCOM, the use of a high serum AMH as a proxy for PCOM resulted in the reclassification of PCOS in 5% of study participants, with the main effect being more women identified, although some women previously classified as having PCOS were no longer classified as such. AMH has been proposed as an alternative to PCOM as a diagnostic component of PCOS. Previous studies are limited by poorly defining PCOS, use of infertile women as comparators, measurement of hormones by immunoassay that lack precision in the female range, low-resolution ovarian ultrasound and inconsistent handling and storage of serum samples. This is an Australian cross-sectional study of 163 non-healthcare-seeking women. Serum AMH was measured by both the Ansh picoAMH assay and the Beckman Coulter Access 2 (BA2) assay, in parallel with androgens measured by liquid chromatography-tandem mass spectrometry, in blood samples of women, not pregnant, breast feeding or using systemic steroids, who also underwent high-resolution ovarian ultrasound. PCOS was determined by the Rotterdam criteria with PCOM defined by the Androgen Excess-PCOS Taskforce recommendation of ≥25 follicles in at least one ovary. Cut-off serum concentrations that best identified women as having PCOM were identified by receiver operator characteristic (ROC) curves. A total of 163 women, mean (SD) age 32.5 (5.5) years, who provided a blood sample and had both ovaries visualized on ultrasound were included in the analysis. Women with isolated PCOM had higher median (range) Ansh AMH and BA2 AMH concentrations than those with no PCOS characteristics [56.9 pmol/l (34.6, 104.2) versus 18.7 (3.2, 50.9), P = 0.002 and 38.5 pmol/l (22.2, 100.2) versus 16.7 (3.5, 38.9), P = 0.002, respectively]. An AMH ≥ 44.0 pmol/l, suggested by the ROC curve, identified 80.6% of women with PCOM, falsely identified 15.2% of women without PCOM as having PCOS and had a positive predictive value of 55.6%. The negative predictive value was 94.9%. An AMH BA2 assay cut-off of ≥33.2 pmol/l provided a sensitivity of 80.6%, a specificity of 79.5% and a positive predictive value for PCOM of 48.1%. The negative predictive value was 94.6% for PCOM. When serum AMH was used in the place of PCOM as a diagnostic criterion for PCOS, the Ansh assay resulted in an additional seven women classified as having PCOS and no longer classified one woman as having PCOS. For the BA2 assay, eight additional and two fewer women were classified as having PCOS. Overall, both assays resulted in six more women being classified as having PCOS. Women with functional hypogonadotrophic hypogonadism were not excluded and may have been misclassified as having an oligo-amenorrhoea-PCOM phenotype. As study participants were predominantly Caucasian/White, our findings cannot be generalized to women of other ethnicities. Although serum AMH reflects the number of developing ovarian follicles, the absolute values vary between assays and specific reference ranges for individual assays are required. Irrespective of the assay used, replacing PCOM with serum AMH to diagnose PCOS in a community-based sample altered the number of women classified as having or not having PCOS. Consequently, although overall the risk of women being identified as having PCOS would be increased, some women would no longer be classified as having this condition. The study was supported by the Norman Beischer Research Foundation and the Grollo-Ruzzene Foundation. S.R.D. is an NHMRC Senior Principal Research Fellow (Grant No. 1135843). S.R.D. reports unrelated support that includes grants from the NHMRC Australia, personal fees for educational activities from Besins Healthcare, Abbott Chile, BioFemme and Pfizer Australia, personal Advisory Board/consultancy fees from Theramex, Abbott Laboratories, Astellas, Mayne Pharmaceuticals, Roche Diagnostics, Lawley Pharmaceuticals and Que Oncology and has received institutional grant funding from Que Oncology and Ovoca research. The other authors declare no conflicts of interest. N/A.
Read full abstract