Evaluation of the IOTA ADNEX Model, Two-Step Strategy and RMI in Routine Gynaecologic Care in Denmark and Implications for Implementation: A Prospective Multicenter Cohort Study.

Abstract

To evaluate whether the IOTA ADNEX model and the Two-Step Strategy improve triage and referral of adnexal masses in routine gynaecologic care compared with the RMI and to identify an appropriate malignancy-risk threshold. Prospective multicenter cohort study. Thirteen non-tertiary hospitals and clinics and one tertiary referral hospital in Denmark. A complete-case cohort of 966 patients with adnexal masses. Malignancy risk was estimated using prospectively collected clinical data, ultrasound findings, and CA125 levels. Reference standard was histopathology or ≥ 12 months of clinical follow-up. Performance was evaluated across predefined thresholds (1%-30% for ADNEX/Two-Step Strategy (modified benign descriptors + ADNEX); ≥ 200 for RMI), stratified by centre type. Negative and positive predictive values (NPV, PPV), sensitivity, referral rates to assess correct and incorrect referrals. In non-tertiary centres, NPVs were ≥ 96% for IOTA models versus 95% for RMI; corresponding values in the tertiary centre were 82%-100% versus 78%. PPVs increased with higher thresholds and approached RMI at 20% threshold. In non-tertiary centres, where referral decisions are made, a 15% threshold provided the most favourable balance between sensitivity (~63%) and referral rate (~14%). At thresholds ≥ 25%, referral rates were similar to RMI (~8%), with only marginal gains in sensitivity (~50% vs. 39%). Most additionally detected tumours were stage I ovarian cancers or borderline tumours. For masses classified as benign by modified benign descriptors, ADNEX showed high NPVs but low PPVs and negligible net benefit, providing limited additional diagnostic value over the Two-Step Strategy. IOTA-based models improve early detection but increase referral rates. A 15% risk threshold appears to offer a clinically reasonable balance between early detection of malignancy and referral burden, based on observed trade-offs between detection and referral rates. ClinicalTrials.gov identifier: NCT04188652.