Preoperative prediction of suboptimal resection in advanced ovarian cancer based on clinical and CT parameters

Abstract

Background : Cytoreduction is important as a survival predictor in advanced ovarian cancer.

Purpose : To determine the prediction of suboptimal resection (SOR) in advanced ovarian cancer based on clinical and computed tomography (CT) parameters.

Material and Methods : Between 2007 and 2015, 327 consecutive patients with FIGO stage III–IV ovarian cancer and preoperative CT were included. During 2007–2012, patients were assigned to a derivation dataset (n = 220) and the others were assigned to a validation dataset (n = 107). Clinical parameters were reviewed and two radiologists assessed the presence or absence of tabulated parameters on CT images. Logistic regression analyses based on area under the receiver-operating characteristic curve (AUROC) were performed to identify variables predicting SOR, and generated simple score using Cox proportional hazards model.

Results : There was no statistical difference in patients’ characteristics in both datasets, except for residual disease (P = 0.001). Optimal resection improved from 45.0% (99/220) in the derivation dataset to 64.4% (69/107) in the validation dataset. Logistic regression identified that Eastern Cooperative Oncology Group-performance status (ECOG-PS 2), involvements of peritoneum, diaphragm, bowel mesentery and suprarenal lymph nodes, and pleural effusion were independent variables of SOR. Overall AUROC for score predicting SOR was 0.761 with sensitivity, specificity, and positive and negative predictive values of 70.6%, 73.2%, 68.7%, and 91.9%, respectively. In the derivation dataset, AUROC was 0.792, with sensitivity of 71.4% and specificity of 74.3%, and AUROC of 0.758 with sensitivity of 69.2% and specificity of 72.8% in the validation dataset.

Conclusion : CT may be a useful preoperative predictor of SOR in advanced ovarian cancer.