MON-041 Diagnostic Performance of Ovarian Morphology for Anovulatory Conditions in Lean and Overweight Women

Introduction: Sonographic evidence of polycystic ovarian morphology (PCOM) is a cardinal feature of polycystic ovary syndrome (PCOS), a condition which reflects a spectrum of phenotypes. The criteria to define PCOM are based on an upper threshold for follicle number and ovarian size in lean, regularly cycling women. Whether ovarian features can be used to distinguish between distinct anovulatory conditions is unclear, as is any impact of body mass index (BMI) on the diagnostic performance of ovarian morphology for anovulatory conditions. Objectives: To determine whether ovarian morphology can discriminate between women with regular cycles, normoandrogenic anovulation (NA-Anov), and hyperandrogenic (HA-Anov) anovulation. Any impact of BMI-specific thresholds to improve the diagnostic performance of ovarian morphology for anovulatory conditions was determined. Methods: Women with HA-Anov (biochemical and/or clinical hyperandrogenism and irregular cycles; N=53), NA-Anov (irregular cycles in the absence of hyperandrogenism; N=42), and normoandrogenic women with regular cycles (Controls, N=41) underwent a reproductive health history, physical exam, transvaginal ultrasound scan of their ovaries and fasting blood tests for reproductive hormones. Follicle number per ovary (FNPO, 2-9mm) and ovarian volume (OV) were determined. The diagnostic performance of sonographic markers for anovulatory conditions was tested using receiver operating characteristic curves. Results: FNPO and OV discriminated between HA-Anov and Controls when all women were considered (area under the curve [AUC]=0.82, sensitivity [Se] 72%, specificity [Sp] 90% and AUC=0.84, Se 77%, Se 81%, respectively). The diagnostic accuracy (AUC = 0.87) and sensitivity (83%) of FNPO improved in lean women (BMI < 25kg/m(2)), whereas specificity (93%) improved for women with overweight/obesity (BMI > 25kg/m(2)). By contrast, the diagnostic performance of OV declined when BMI was considered. FNPO discriminated between HA-Anov and NA-Anov in lean women (AUC=0.77) whereas OV discriminated between anovulatory conditions in women with overweight or obesity (AUC=0.76). FNPO, but not OV, differentiated between NA-Anov and controls – albeit thresholds were lower for women in lean (>22 follicles) versus overweight categories (>38 follicles). Conclusion: Ovarian morphology has diagnostic potential for anovulatory conditions – but its performance is impacted by BMI status. OV differentiated between HA and NA status, whereas follicle counts discriminated anovulatory conditions from controls suggesting differential roles for FNPO and OV in reproductive dysfunction. Consideration of BMI improved diagnostic performance in some cases, however the overlap in morphological features between NA- and HA-Anov is too extensive to propose condition-specific thresholds at this time.