Abstract Objectives To design a diagnostic nomogram model integrating the Ultrasound Ovarian-Adnexal Reporting and Data System (US O-RADS) and serum indexes for predicting malignancy in patients with ovarian masses. Methods This was a retrospective study including 201 benign ovarian masses patients median age : 36(29-53) and 136 ovarian cancer (OC) patients median age : 53(46-60). Before surgical resection, all patients underwent ultrasound examination, and US O-RADS were assessed. Meanwhile, cancer antigen 125 (CA125) and neutrophil to lymphocyte ratio (NLR) were tested. All patients were pathologically diagnosed. The differences in these indexes between two groups were analysed. Meanwhile, ROC curves were firstly graphed to determine their optimal cut-off values. Logistic regression analyses were used to identify CA125, NLR and US O-RADS for OC. Then, the nomogram model was established. Finally, we evaluated the model and performed internal validation. Results CA125, NLR and US O-RADS in the malignant group were higher compared to the benign group. These parameters were incorporated to develop a nomogram model with an area under the ROC curve (AUC) of 0.950 (95% confidence interval CI, 0.927-0.973), 94.1% sensitivity and 84.6% specificity. The calibration curve displayed a satisfactory fitting degree. Meanwhile, Decision curve analysis (DCA) provided a net benefit for a range of threshold probabilities. Conclusions This nomogram model yielded a favourable diagnostic accuracy for predicting malignancy in patients with ovarian masses. Advances in knowledge This was the first nomogram model integrating US O-RADS and serum indexes, which showed satisfactory discrimination, calibration abilities and clinical net benefit for predicting malignancy in patients with ovarian masses.
Gu et al. (Wed,) studied this question.