EP34.13: Ultrasound diagnostic accuracy for ovarian borderline tumours in a tertiary centre using IOTA simple rules and ADNEX models

Abstract

Objectives: To evaluate transvaginal ultrasound (TVUS) diagnostic accuracy and the performance of the International Ovarian Tumour Analysis (IOTA) and ADNEX models in the preoperative evaluation of ovarian masses described on ultrasound as borderline ovarian tumours (BOT) in a gynecological oncology centre in Portugal. Methods: Retrospective study with identification of all women who underwent TVUS examination with suspected diagnosis of BOT between 2010 and 2018 and were submitted to surgical excision. All images were reviewed and classified according to IOTA and ADNEX models. Results: 27 cases with preoperative TVUS diagnostic suspicion of BOT were included in the final analysis. Median age at diagnosis was 41 years (interquartile range [IQR] 34-49) and median BMI was 25 kg/m 2 (IQR 24-32). Most cases were asymptomatic (n = 17, 71%). Average size of maximal diameter of the lesion was 105mm . CA125 values at diagnosis ranged between 3,9-82,3 UI/mL (in 6 cases above the cutoff value). CA 19.9 values at diagnosis ranged between 1,7-1100 UI/mL (in 3 cases above the cutoff value). Regarding IOTA simple rules classification, 17 were classified as inconclusive, 9 as malignant and 1 as benign. Histological classification confirmed 8 cases of BOT (3 serous and 5 mucinous) and the remaining 19 were benign. Six of BOT cases had been classified by IOTA simple rules as malignant and 2 as unclassifiable, rendering a sensitivity of 75% (6/8) and a positive predictive value for malignancy of 66.7% (6/9). Among the 4 cases in which a tumour subtype was described in TVUS (mucinous vs. serous) there was an agreement with histology in 3 cases. ADNEX model sensitivity was 87,5% (7/8) and specificity was 63,2% (12/19). Conclusions: TVUS impression led to an excessive number of cases of BOT suspicion; IOTA simple rules and ADNEX models performed better than subjective assessment at the differential diagnosis, although ADNEX outperformed in sensitivity.