A comparative analysis of prediction models for complete gross resection in secondary cytoreductive surgery for ovarian cancer

Abstract

ObjectiveWe sought to examine compliance and outcomes using Memorial Sloan Kettering “(MSK) criteria” to predict complete gross resection (CGR) and compare them with the validated Tian and AGO models. MethodsPatients who underwent SCS for recurrent platinum-sensitive ovarian cancer from 5/2001–6/2014 were identified. The AGO and Tian models were applied to the study population; appropriate statistical tests were used to determine ability to predict CGR. Results214 SCS cases were identified. Since the implementation of MSK criteria, the CGR rate has been 86%. The AGO model had a 49% accuracy rate in predicting CGR, and predicted gross residual disease (RD) in 51%; however, CGR was achieved in 86%. The Tian model had an 88% accuracy rate. Of the 4% scored as Tian high risk for gross RD, 33% achieved a CGR. Comparing models, McNemar's p-value was 0.366 between the Tian and MSK models and <0.001 between AGO and MSK criteria. Median PFS was 21.3 (95%CI, 18.2–24.5), 22.5 (95%CI, 19.4–25.3), and 14.1months (95%CI, 9.7–22.1) for the entire cohort, for those achieving CGR, and for those left with RD, respectively (p=0.013). OS was 82.2 (95%CI, 60.2–123.3), 95.6 (95%CI, 63.6–NE), and 57.5months (95%CI, 27.5–113.9), respectively (p=0.014). ConclusionCGR during SCS is associated with extended PFS and OS. We report a high rate of CGR using MSK criteria. There was good concordance between the Tian and MSK models; however, the latter has fewer variables and is more user-friendly. Tian criteria may be applied to intermediate MSK cases for further stratification.