Diagnostic value of calculated testosterone indices in the assessment of polycystic ovary syndrome.

BACKGROUND: One of the main criteria to establish a diagnosis of polycystic ovary syndrome (PCOS) is hyperandrogenemia. Recent observations suggest that total testosterone may not be a sensitive marker for the detection of androgen excess. The aim of the present study was to compare the value of different androgen determinations for diagnosis of PCOS. METHODS: Untreated PCOS patients (n=133; mean age 28 years) and healthy control women (n=54; mean age 28 years) were included in the study. Measurements of total testosterone and sex hormone-binding globulin (SHBG), luteinizing hormone (LH), follicle-stimulating hormone (FSH), androstendione, dehydroepiandrosterone sulfate (DHEAS) and albumin were performed. In addition, the free androgen index (FAI), free and bioavailable testosterone were calculated. Clinical signs of hyperandrogenism were evaluated by physical examination. The area under the receiver operating characteristic curve (AUC-ROC) was used to compare the sensitivity and specificity of different androgen determinations to detect PCOS, defined as clinical hyperandrogenism and irregular cycles compatible with the National Institutes of Health criteria of chronic anovulation and clinical or biochemical hyperandrogenism. RESULTS: All biochemical parameters of hyperandrogenism were significantly higher in PCOS patients than in controls (all p<0.0001). The highest AUC-ROC was found for bioavailable testosterone (0.852) followed by FAI (0.847) and free testosterone (0.837). Lower AUC-ROC was found for SHBG, total testosterone and androstendione (0.765, 0.799 and 0.706, respectively). When FAI=4.97 was taken as a cutoff value, sensitivity was 71.4% and specificity was 85.2%. A cutoff of 0.78 nmol/L for bioavailable testosterone had even higher sensitivity of 75.9%, but slightly lower specificity of 83.3%. FAI and bioavailable testosterone correlated significantly (all p<0.05) with total testosterone, androstendione, LH/FSH ratio and DHEAS. In addition, free testosterone, bioavailable testosterone and FAI correlated significantly with hirsutism scores, and ovarian volume and follicle count. CONCLUSIONS: ROC analysis provided evidence that calculated testosterone indices (bioavailable testosterone, FAI, free testosterone) are useful parameters for the discrimination of PCOS patients and healthy controls.