Ultrasound-based ADNEX model for differentiating between benign, borderline, and malignant epithelial ovarian tumors

Abstract

Abstract Background: The purpose of this study was to evaluate the ability of the International Ovarian Tumor Analysis-Assessment of Different NEoplasias in the adneXa (IOTA-ADNEX) model to distinguish between benign, borderline, and malignant epithelial ovarian tumors(BeEOTs, BEOTs, and MEOTs, respectively). Methods: The study included 813 patients with BeEOTs, BEOTs, and MEOTs who underwent ultrasound examinations and pelvic operations. Comparisons were made between the clinical information and ultrasonographic features of the three patient groups. Three clinical variables and six ultrasound variables were used to estimate malignancy risk. The sensitivity, specificity, positive predictive value, negative predictive value, and AUC (the area under the receiver operating characteristics [ROC] curve) of the ADNEX model were calculated. Results: Of the 813 patients, 257 (31.6%) had BeEOTs, 114 (14.0%) had BEOTs, and 442 (54.4%) had MEOTs. The most common type, serous and mucinous epithelial tumors, accounted for 81.3% of the total cases. In the MEOTs group, serous, mucinous, endometrioid, and clear cell tumors accounted for 85.3%, 2.5%, 4.5%, and 7.7% of the total cases, respectively. For a cut-off value of 10% to identify the overall risk for ovarian cancer (OC), the sensitivity and specificity were 99.1% and 73.2%, respectively. According to the ROC curves, the AUC was 0.987 (95% CI: 0.981–0.993) for BeEOTs compared with MEOTs, 0.820 (95% CI: 0.768–0.872) for BeEOTs compared with BEOTs, 0.912 (95% CI: 0.876–0.948) for BeEOTs compared with stage I OC, and 0.995 (95% CI: 0.992–0.998) for BeEOTs compared with stages II–IV OC. The AUC was 0.614 (95% CI: 0.519–0.709) for BEOTs compared with stage I OC, 0.903 (95% CI: 0.869–0.937) for BEOTs compared with stages II–IV OC, and 0.851 (95% CI: 0.800–0.902) for stage I OC compared with stages II–IV OC. Conclusions: The IOTA-ADNEX model demonstrated good diagnostic performance for the three categories of EOTs and can be helpful for clinical treatment management.