Abstract
BACKGROUND: Prior studies are inconclusive regarding the effect of obesity on mortality in persons with colorectal cancer (CRC). We sought to determine the association of pre-diagnosis body mass index (BMI) trajectories on mortality after CRC diagnosis. METHODS: Utilizing the Multiethnic Cohort, we included adults aged 18-75 between 1 January 1993 and 1 January 2019 with a diagnosis of CRC and at least three available BMIs. The primary exposure, BMI, was subjected to group-based trajectory modeling (GBTM). We evaluated all-cause and CRC-specific mortality, using Cox proportional hazard (PH) models. RESULTS: Of 924 persons, the median age was 60 years, and 54% were female. There was no statistically significant association between pre-cancer BMI trajectory and either all-cause or cancer-specific mortality. In competing risk analysis, the risk of CRC-specific mortality was higher for African Americans (HR = 1.56, 95% CI [1.00-2.43], p = 0.048) and smokers (HR = 1.59, 95% CI [1.10-2.32], p = 0.015). Risk of all-cause mortality was higher for Hawaiian persons (HR = 2.85, 95% CI [1.31-6.21], p = 0.009) and persons with diabetes (HR = 1.83, 95% CI [1.08-3.10], p = 0.026). CONCLUSIONS: Pre-diagnosis BMI trajectories were not associated with mortality after CRC diagnosis, whereas race/ethnicity, diabetes, and smoking were associated with an increased risk of death. Our findings suggest the obesity paradox alone does not account for mortality after CRC diagnosis.