Accuracy of IOTA Simple Rules, IOTA ADNEX Model, RMI, and Subjective Assessment for Preoperative Adnexal Mass Evaluation: The Experience of a Tertiary Care Referral Hospital
Objectives: The aim of this study was to evaluate the accuracy of IOTA Simple Rules (SR), IOTA ADNEX model, Risk of Malignancy Index (RMI), and subjective assessment (SA) which is used for adnexal mass assessment in our institution. Design: This is a prospective observational study. Participants/Materials, Setting, Methods: We included patients with at least one adnexal mass who needed elective surgical evaluation based on clinical and laboratory findings. Patients admitted to Clinic for Gynecology and Obstetrics, University Clinical Center of Serbia, were recruited for the study between January 2019 and June 2021. Level II ultrasonographers performed a gray scale and Doppler exam for each patient. Preoperative classification of adnexal masses (benign or malignant) was performed by SA, the International Ovarian Analysis Group (IOTA) SR, IOTA ADNEX model, and Risk of Malignancy Index (RMI). Postoperatively obtained histological findings were used as a reference. Results: During the study period, we enrolled 179 premenopausal and 217 postmenopausal patients, representing 396 patients in our sample. Prevalence of malignant disease in pre- and postmenopausal groups was 16.2% (29/179) and 41% (89/217), respectively. Malignant disease was diagnosed in 29.8% (118/396) of patients. SA achieved the highest discrimination accuracy between benign and malignant tumors (area under the curve [AUC] of 0.928, 95% CI [0.898–0.952]). For SA, the overall diagnostic accuracy, sensitivity, specificity, positive likelihood ratio (LR+), and negative likelihood ratio (LR−) were 91.4%, 88.1%, 92.8%, 12.25, and 0.13. The AUC for Simple Rules with subjective assessment in inconclusive cases (SR + SA) was 0.912 (95% CI [0.880–0.938]). Regarding SR + SA, diagnostic accuracy, sensitivity, specificity, LR+, and LR− were 92.4%, 88.1%, 94.2%, 15.31, and 0.13. The ADNEX model had the AUC of 0.914 (95% CI [0.882–0.940]). Binary classification using the ADNEX model at a cut-off value of 10% for malignancy had the sensitivity, specificity, LR+ and LR− of 92.4%, 73.0%, 3.42, and 0.10. This resulted in the lowest overall accuracy of 78.8%. The AUC for RMI was 0.854 (95% CI [0.815–0.887]), with overall accuracy, sensitivity, specificity, LR+ and LR− of 82.3%, 73.7%, 86.0%, 5.26, and 0.31. There was no difference in the AUCs of the SA and IOTA models for the whole group, premenopausal, and postmenopausal groups. RMI performed worse compared to SA and the IOTA models. The ADNEX model achieved the highest accuracy at the cut-off value of 35%. Limitations: The data generalizability is limited by a single institution-dependent sampling. Conclusions: The IOTA SR and ADNEX model were reliable and comparable with the SA and performed better than the RMI. The IOTA SR model offers the potential for immediate and reliable diagnosis, even in the hands of less experienced ultrasonographers. Both IOTA models studied can be a valuable adjunct to a clinician’s decision-making process.
