Determine the diagnostic accuracy of ovarian morphology on ultrasonography for polycystic ovary syndrome (PCOS) across the menstrual cycle. Data from 25 women with PCOS (oligo-anovulation + androgen excess) and 40 age- and BMI-matched controls (regular cycles + normal androgens) with consecutive ultrasound scans over ≥2 menstrual cycle phases were included in this retrospective analysis. Phases of interest included: early follicular (1-6 days following menses onset), late follicular (1-7 days prior to ovulation in the presence of a dominant follicle), early luteal (1-7 days following ovulation), and late luteal (1-7 days prior to the onset of menses). Diagnostic accuracy (area under the ROC curve AUC, sensitivity Se, specificity Sp) of mean, maximum, and contralateral (ovary without dominant follicle or corpus luteum) sonographic measures for follicle number per ovary (FNPO), follicle number per single section (FNPS), and ovarian volume (OV) at each phase were determined. DeLong tests determined differences in diagnostic accuracy across phases. FNPO, FNPS, and OV all had significant diagnostic accuracy for PCOS across menstrual cycle phases. OVmean had the highest diagnostic accuracy for PCOS in the early follicular phase (AUC = 0.87, Se = 65%, Sp = 95%), whereas OVcontralateral had the highest accuracy in both the early (AUC = 0.81, Se = 62%, Sp = 92%) and late luteal phases (AUC = 0.93, Se = 100%, Sp = 70%). OVcontralateral outperformed all other measures in the late luteal phase (P < .05). Ultrasonographic evaluations for PCOS may be performed across the menstrual cycle. The presence of a dominant follicle in the late follicular phase did not impact the performance of ovarian markers for PCOS status. By contrast, polycystic ovarian morphology is best defined in the luteal phase by assessments of FNPO and OV in the contralateral ovary.
Carter et al. (Wed,) studied this question.