Accuracy of Magnetic Resonance Imaging for Identifying Ovarian Cancer in a Community-Based Setting

Introduction: Many ovarian or adnexal masses have an indeterminate appearance on ultrasound that can raise concerns about cancer. Although magnetic resonance imaging (MRI) has been reported to reliably distinguish between benign and malignant masses, studies evaluating the accuracy of MRI in community-based practice settings are lacking. Methods: Women who underwent MRI to further evaluate an ultrasound-detected adnexal mass in 2016-2017 within a large community-based health system were identified. MRI reports were classified as favoring malignancy, benign disease, or indeterminate, blinded to pathological outcome. With a minimum of 2 years of follow-up, all ovarian cancers and borderline tumors were identified, and the accuracy of MRI assessment was determined. Results: Among 338 women who had MRI to evaluate an adnexal mass, 144 (42.6%) subsequently underwent surgery. MRI favored malignancy in 7 (4.9%) cases, benign disease in 89 (62.2%) cases, and was indeterminate in 48 (33.6%) cases. Of the seven cases in which MRI favored malignancy, two cancers and five benign tumors were found. An additional 10 cases of cancer or borderline tumor were found among women who had MRI reports that were read as indeterminate (n=6) or that favored benign disease (n=4). The sensitivity, specificity, positive predictive value, and negative predictive value of an MRI favoring malignancy were 16.7%, 96.2%, 28.5%, and 92.7%, respectively. Discussion: In a large community-based setting, an MRI favoring malignancy was more likely to be associated with benign disease than cancer and identified only 16.7% of true malignant cases. The findings suggest that the ability of MRI to differentiate between benign and malignant adnexal masses in community-based practice settings is currently limited.