Background: Thus far, the association of tumor size with prognosis in colon cancer has not been considered and has remained unclear. This study, therefore, aimed to investigate the association between tumor size as a continuous variable and prognosis in colon cancer using Cox models with restricted cubic splines.
Methods: Using the Surveillance, Epidemiology, and End Results database, we selected 128,369 patients with colon cancer who underwent surgery. Overall survival and colon cancer-specific survival were separately analyzed, and tumor size was separately evaluated as a continuous variable and a categorical variable. To investigate the relationship after adjusting for covariates, we used the proportional hazards models. The restricted cubic splines model was used to determine the presence of nonlinear or linear association and flexibly visualize the association.
Results: The adjusted covariate model showed that the hazard ratio of colon cancer rapidly increased with a tumor size of 4 cm and slowly increased with a tumor size larger than 4 cm. When tumor size was analyzed as a categorical variable, the multivariable-adjusted model demonstrated a nearly linear relationship between tumor size and hazard ratio regardless of overall survival or cancer-specific survival, and the hazard ratio of group 5 (4.1-5 cm) was nearly a turning point. Subgroup analysis with respect to lymph node metastasis showed that the relationship between tumor size and prognosis in colon cancer was evident in lymph node metastasis.
Conclusion: There was a strong negative relationship between tumor size and prognosis in colon cancer. However, when tumor size was less than 4 cm, the relationship between tumor size and prognosis was steep compared with that when tumor size was larger than 4 cm, especially in lymph node metastasis.
Copyright © 2020. Published by Elsevier Inc.