Gastric cancer staging in the era of neoadjuvant therapy and its prognostic implications.

Abstract

22 Background: Increasingly patients are undergoing neoadjuvant therapy for gastric cancer. The relationship between stage-based prognostic information available prior to treatment (cStage), after surgery (ypStage), and difference between cStage and ypStage (delta) remains unclear. We aim to describe the relationship between cStage and ypStage as relates to survival for gastric cancer patients. Methods: Data from the National Cancer Data Base (NCDB) from 2004-2015 was used for the analysis. Patients with gastric adenocarcinoma who received neoadjuvant therapy then underwent surgery were included. Kaplan Meier curves were used to model survival. Harrell’s C-statistics obtained from Cox Regression models were reported. Results: 9,959 patients met our inclusion criteria. Increases in cStage, ypStage and delta (ypStage-cStage) were associated with worse survival. Median overall survivals for cStages 1-4 were: 53.8, 39.5, 29.2, 20.9 months (logrank test, p<0.0001). Median survivals for ypStage 0-4 were: 95.4, 89.7, 36.9, 23.4, 16.0 months (logrank test, p<0.0001). Survival was further stratified by delta. A representative table comparing cStage 2 and ypStage 2 is shown below. A cox regression model with cStage as predictor of survival yielded a Harrell’s C-statistic of 0.555; when delta was added to the model, the C-statistic increased to 0.638. Separately, a Cox-regression model with ypStage as predictor yielded a C-statistic of 0.632; when delta was added to this model, the C-statistic increased negligibly to 0.638. Conclusions: Prognostic accuracy using cStage prior to treatment improved when tumor responsiveness was considered while this was not the case for ypStage. Pre-surgical prognostic information should be provided with a caveat that treatment response will influence survival. Post-surgery, the clinical stage is less relevant and ypStage can be used alone in providing prognostic information. [Table: see text]