Polycystic Ovary Syndrome in Adolescents: Which MR Imaging–based Diagnostic Criteria?

Purpose To evaluate the validity and reproducibility of magnetic resonance (MR) imaging-based ovarian morphologic measurements for diagnosis of polycystic ovary syndrome (PCOS) in adolescents. Materials and Methods This case-control study included 110 adolescent girls (age range, 13-17 years) who underwent pelvic MR imaging in 2006-2015. The case group included girls with high (n = 40, hyperandrogenism and oligomenorrhea or amenorrhea), intermediate (n = 8, hyperandrogenism), or low (n = 7, oligomenorrhea or amenorrhea) suspicion of PCOS. Control subjects were 55 age-matched (± 2 years) girls with no clinical hyperandrogenism, oligomenorrhea, or amenorrhea. The validity (sensitivity, specificity, and area under the receiver operating characteristic curve [AUC]) of the number of follicles per ovary (FPO) measuring 9 mm or smaller (FPO-9) and FPO measuring 5 mm or smaller (FPO-5), ovarian volume (OV), sphericity index, peripheral distribution of follicles, and absence of a dominant follicle were determined, with girls who were highly suspected of having PCOS compared with control subjects as the reference. Two radiologists independently measured these criteria in 50 girls who were suspected of having PCOS to assess reproducibility (κ and intraclass correlation coefficients [ICCs]). Results All criteria except sphericity index and absence of a dominant follicle were significantly associated with the level of suspicion of PCOS (P ≤ .05). The AUCs for FPO-9 (0.78; 95% confidence interval [CI]: 0.68, 0.87), FPO-5 (0.73; 95% CI: 0.62, 0.83), and OV (0.77; 95% CI: 0.68, 0.87) were significantly greater than 0.5; that was not true for sphericity index (AUC, 0.58; 95% CI: 0.47, 0.70). Sensitivity and specificity for peripheral distribution of follicles were 33% (95% CI: 19%, 49%) and 95% (95% CI: 85%, 99%), respectively; for absence of a dominant follicle, they were 90% (95% CI: 76%, 97%) and 27% (95% CI: 16%, 41%), respectively. Reproducibility was almost perfect for OV (ICC, 0.89), substantial for absence of a dominant follicle (κ, 0.74), moderate for FPO-9 (ICC, 0.54) and FPO-5 (ICC, 0.61), and fair for peripheral distribution of follicles (κ, 0.37). Conclusion The most accurate MR imaging-based diagnostic criteria for PCOS were OV, FPO-9, and peripheral distribution of follicles; however, reproducibility of these measures was moderate, except that for OV (ICC, 0.89). © RSNA, 2017 Online supplemental material is available for this article.