EP23.35: External validation of the O‐RADS lexicon and the IOTA two‐step strategy to stratify tumours into O‐RADS risk groups in a North American academic centre

To investigate the diagnostic performance of the O-RADS lexicon and the IOTA-2 step strategy in a North American Academic Network Hospital System. Retrospective external diagnostic validation study based on data from one academic medical centrw and three network community-based hospitals. A total of 205 adult patients presenting with an adnexal mass between January 1, 2014, and March 15, 2023, treated either conservatively, or with surgery, were included in this diagnostic study. Studies were excluded that did not meet predefined quality criteria. Median age of the 139 patients was 45 (IQR, 35-57 years). Of the observed ovarian lesions, 104 (74.8%) were benign, 15 (10.8%) were malignant, and 20 (14.5%) were uncertain classification. Applying the O-RADS lexicon resulted in a 1.42% observed prevalence of malignancy in O-RADS 2, 1.4% in O-RADS 3, 30% in O-RADS 4, and 75% in O-RADS 5. Two patients in O-RADS 3 had lesions which were observed due to severe surgical comorbidities with slow growth but no evidence of increasing classification up to 4 years following initial evaluation. The mean risk of malignancy generated by the ADNEX model without CA125 was significantly higher than in malignancy versus benign lesions for the academic medical centre (77.3% vs 4.6%), and for the three community-based hospitals (841% versus 15.9%). The findings of this external diagnostic validation study suggest that both the O-RADS lexicon and the IOTA-2 step strategy can be utilised to stratify patients into risk groups. We confirm the observed malignancy rate is not below 1% as previously demonstrated for O-RADS 2. While the majority of the studies from the community-based hospitals met the predefined data quality criteria and the results of risk stratification were similar to the academic medical centre evaluation, more of these women required further imaging to further define the lesion.