Real-world performance of the Assessment of Different NEoplasias in the adneXa (ADNEX) model for the pre-operative classification of ovarian tumors

Objective: To predict the risk of malignancy in ovarian tumors, the multi-class risk prediction Assessment of Different NEoplasias in the adneXa (ADNEX) model of the International Ovarian Tumor Analysis group is used internationally. This study aimed to perform an external validation of this model in a real-world clinical setting in the Netherlands. Methods: A multi-center, retrospective diagnostic accuracy study was performed. Women aged >= 18 years who attended the outpatient clinic between May 2020 and December 2021 and had a first ultrasound assessment for an ovarian tumor with application of the International Ovarian Tumor Analysis ADNEX model were included. The reference standard was the pathology result in case of surgery or follow-up with ultrasound according to the current Dutch guideline in case of conservative management. Results: Of the 363 women included, 286 (78.8%) had benign, 24 (6.6%) had borderline, and 53 (14.6%) had malignant pathology. Most (62.5%) of the included patients underwent surgery. The area under the receiver operating characteristic (ROC) curve was 0.92 (95% CI 0.89 to 0.96) if borderline tumors were considered malignant and 0.93 (95% CI 0.89 to 0.96) if borderline tumors were considered benign. Cutoff values between 7% and 14% yielded the highest Youden index. The model was moderately able to differentiate between malignant histologic sub-types. Conclusions: The ADNEX model showed excellent diagnostic performance in daily clinical practice and retained its value in a population in which a proportion of patients were treated with conservative management.