Incremental classification yield of androstenedione and DHEAS in women evaluated for polycystic ovary syndrome: a retrospective study

Current guidelines prioritize total and free testosterone for biochemical assessment of hyperandrogenism and suggest considering androstenedione and dehydroepiandrosterone sulfate (DHEAS) when neither is elevated. However, how often these additional androgens change polycystic ovary syndrome (PCOS) classification and Rotterdam phenotype assignment remains uncertain. We quantified the incremental classification yield of adding androstenedione and DHEAS to a pragmatic baseline definition based on documented hirsutism and total testosterone. This single-center retrospective study included 438 women aged 18–35 years evaluated for suspected PCOS or related clinical features and with complete total testosterone, androstenedione, and DHEAS measurements. Hyperandrogenism was assessed using four prespecified, nested definitions: documented hirsutism and/or elevated total testosterone (Model 1); Model 1 plus elevated androstenedione (Model 2); Model 1 plus elevated DHEAS (Model 3); and Model 1 plus elevated androstenedione and/or DHEAS (Model 4). PCOS classification and Rotterdam phenotype assignment were recalculated under each definition. Comparisons used Cochran’s Q and exact McNemar tests with Holm adjustment. Analyses restricted to women without documented hirsutism and with normal total testosterone were exploratory. The proportion classified as hyperandrogenic increased from 86.8% under Model 1 to 93.2% under Model 4, whereas PCOS classification increased from 72.8% to 75.6%. Model 4 changed Rotterdam phenotype assignment in 24 women (5.5%, 95% CI 3.7–8.0). Adding androstenedione changed phenotype assignment more often than adding DHEAS alone (4.3% vs. 1.8%, p = 0.027), and isolated androstenedione elevation was more frequent than isolated DHEAS elevation (14.4% vs. 4.1%). Among 58 women in the exploratory subgroup, 48.3% had elevated androstenedione and/or DHEAS; under Model 4, 20.7% (95% CI 12.3–32.8) were reclassified from non-PCOS to PCOS and 41.4% (95% CI 29.6–54.2) changed phenotype assignment. Adding androstenedione and DHEAS to a pragmatic baseline definition based on documented hirsutism and total testosterone had a greater impact on hyperandrogenism than on overall PCOS classification. Androstenedione contributed more than DHEAS to phenotype reassignments and incremental classification yield. These findings quantify the classification impact of broader androgen assessment but do not establish improved diagnostic accuracy. Prospective studies incorporating guideline-recommended free testosterone assessment and clinical outcomes are needed to determine the clinical relevance of these reclassifications. Not applicable.