Nomogram based on CT and clinical features to predict R0 resection in patients with stage IIB–IV epithelial ovarian cancer: a multi-center study
To develop a nomogram based on CT and clinical features to predict R0 resection in patients with stage IIB-IV epithelial ovarian cancer (EOC). 209 patients with stage IIB-IV EOC from three independent medical institutions were stratified into training cohort (from institutions 1 and 2, n = 144) and independent validation cohort (from institution 3, n = 65). Univariate and multivariate logistic analyses of CT and clinical features obtained within two weeks before debulking surgery were used to determine the independent predictors of R0 resection in the training cohort. Nomogram was developed based on the predictors. Receiver operating characteristic (ROC) curves and calibration curves were performed to evaluate the predictive performance of the nomogram. R0 resection was achieved in 66.00 and 61.50% patients in the training and validation cohorts, respectively. In the training cohort, overall peritoneal cancer index based on CT (CT-PCI) (OR 1.245, P < 0.001), serum human epididymis protein4 (HE4) level (OR 1.003, P = 0.012), and neutrophil-to-lymphocyte ratio (NLR) (OR 1.272, P = 0.031) were independent predictors of R0 resection in patients with stage IIB-IV EOC. Nomogram based on them achieved areas under the ROC curves of 0.908 (95% CI 0.848-0.950) and 0.779 (95% CI 0.659-0.873) in the training and independent validation cohort, respectively. The calibration curves showed good agreement between the nomogram predictions and the actual observations in both cohorts. The nomogram based on overall CT-PCI, serum HE4 level, and NLR could be reliable in predicting R0 resection in EOC patients with stage IIB-IV.