A diagnostic scoring system for differentiating between benign and malignant cystic ovarian tumors, utilizing imaging features and biomarkers.

Objective: Our study aims to establish and validate a diagnostic scoring system for distinguishing malignant cystic ovarian tumors (MCOTs) from benign cystic ovarian tumors (BCOTs). Methods: The study population was sourced from two independent hospitals. The subjects included 159 patients with 196 masses (137 in the training cohort and 59 in the validation cohort) who had undergone MRI or CT examinations with pathologically confirmed BCOT or MCOT. Four clinical characteristics, four biomarkers, and 16 imaging features were collected. Univariate analyses and multivariate logistic regression analyses were conducted to identify independent predictors for differentiating MCOTs from BCOTs. The independent predictors were weighted based on regression coefficients to construct a scoring system. The overall score distribution was categorized into three groups to illustrate the diagnostic probability of MCOTs. Results: The scoring system consisted of four independent predictive factors, including CA125, and three imaging features: texture, septum thickness, and enhancement degree. The area under the curve (AUC) for the scoring system was 0.956 (95% CI 0.926-0.986; p < 0.001), comparable to that of the primary predictive model at 0.971 (95% CI 0.949-0.993; p < 0.001). Utilizing 6.5 points as the cut-off value, a sensitivity of 86.6% and a specificity of 91.4% were achieved. The number of patients with MCOT in the three groups significantly increased with higher scores. Conclusion: The established scoring system is reliable and convenient for distinguishing between MCOTs and BCOTs by utilizing elevated CA125 levels, cystic-solid components, septum thickness≥4 mm, and a moderate or prominent degree of enhancement.