Diagnostic Performance of International Ovarian Tumor Analysis Logistic Regression Model LR2 for Adnexal Masses Classification at a Tertiary Gynecology Center in Singapore

Abstract

Background: The International Ovarian Tumor Analysis (IOTA) LR2 model has been shown to provide a reasonably accurate preoperative classification of ovarian tumors. The purpose of this study was to evaluate the diagnostic performance of the IOTA LR2 model in distinguishing benign and malignant adnexal masses in the Singapore population. Methods: This was a retrospective study in a tertiary referral center. Women who attended the Gynecology Unit at Singapore General Hospital with evidence of adnexal tumor on ultrasound examination were evaluated using the IOTA LR2 protocol. The LR2 model was then used to calculate the probability of malignancy. Likelihood ratio of malignancy greater than 10% classifies the mass as malignant. The preoperative diagnosis of women who underwent surgery within 120 days of ultrasound examination was correlated with the final histopathological result. Results: Of the 353 women included in the final study population, 223 had benign disease, 29 had borderline malignant, and 101 had invasive cancer. The IOTA LR2 model had a sensitivity of 79.2% (95% confidence interval (CI), 71.2-85.8%) and a specificity of 79.4% (95% CI, 73.5-84.5%). The area under the receiver-operating characteristics curve was 0.84 (95% CI, 0.80 - 0.89). Conclusions: The IOTA LR2 model maintained its overall diagnostic accuracy when used in our local population. Although it is useful as a first-step test for triaging women with ovarian masses for surgery, a second-stage test would be required to minimize the number of women with benign disease being offered surgery for suspected ovarian malignancy. J Clin Gynecol Obstet. 2021;10(3):67-72 doi: https://doi.org/10.14740/jcgo758