Association between primary tumor site, perioperative CEA ratio, smoking status, and overall survival in patients with colorectal cancer.

Abstract

e15156 Background: There are differences in the incidence, clinical presentation, molecular pathogenesis and outcome of colorectal cancer (CRC) based on tumor location. Emerging research suggests that perioperative carcinoembryonic antigen (CEA) ratio (post-op/pre-op CEA) is a prognostic factor for CRC patients. We aimed to determine the association between tumor location, CEA ratio, smoking status and overall survival (OS) among patients with CRC. Methods: We analysed 323 patients who underwent resection for CRC at KUMC. After excluding those without pre- or post-operative CEA data, 162 patients were classified as either high ( > = 0.5) or low ( < 0.5) ratio. Primary outcomes were: 1) OS stratified by tumor location; 2) OS stratified by CEA ratio; and 3) whether the association between CEA ratio and OS differed by tumor location, after adjusting for stage and smoking status. Kaplan-Meier method was used to estimate survival rates, and Cox proportional hazards models for multivariate analysis. Results: The median age was 63 years (inter-quartile range 53-72), 61% male, 43% smokers, 73% left-sided tumors, median pre-operative CEA was 3.0 (IQR 1.5-7), and 64% had CEA ratio > = 0.5. The OS rates were 85.7% and 91.9% in patients with left-sided vs right-sided tumors respectively (log-rank p-value = 0.9). The OS rates were 83.5% and 91.5% in patients with high vs low CEA ratios respectively (log-rank p-value = 0.3). The effect of CEA ratio on OS was significantly different when stratified by tumor location (p-value for interaction < 0.001). However, in the stratified analysis, the n was too small to permit further inferential analysis. In multivariate analysis, both tumor location (HR 0.6; p = 0.5) and CEA ratio (HR 1.5; p = 0.5) were not significantly associated with OS after adjusting for smoking status and tumor stage. Smoking was significantly associated with higher rates of death (HR 3.9; p = 0.04) when adjusted for tumor location, CEA ratio, and tumor stage. Conclusions: There was no difference in OS between left versus right-sided tumors. The association between CEA ratio and OS was significantly modified by tumor location. However, to attribute this modification to left vs right warrants validation in a larger cohort as our sample size was limited. Smoking increases mortality irrespective of right vs left sided CRC.