Association between environmental modifiable risk score and molecular subtypes in colorectal cancer patients

Abstract

Background: Colorectal cancer (CRC) is associated with several modifiable environmental risks, including dietary and non-dietary factors like calcium intake and smoking habits. It is unknown if the association between overall environmental risk and CRC is differentially associated with certain molecular subtypes.Methods: A total of 6389 cases and 6835 controls of European ancestry from 11 observational studies were included. We combined 13 modifiable risk factors of CRC to create an Environmental Risk Score (ERS) that characterizes overall environmental predisposition to CRC. Four somatic colorectal tumor markers were assessed individually and in combination, including CpG Island Methylator phenotype (CIMP), oncogenic mutations in KRAS and BRAF, and the presence of microsatellite instability (MSI). A multinomial logistic regression was used to assess the association between ERS and risk of CRC molecular subtypes, adjusting for age, sex, study, and energy intake. We also stratified analyses by sex as a sensitivity analysis. Results: CRC risk was positively associated with the ERS [odds ratio (OR)=2.67 (95% CI: 2.40, 2.97)]. Associations between ERS and CRC risk were modestly stronger for KRAS-wildtype CRC compared to KRAS-mutated CRC (Pdifference = 0.063), particularly in females (Pdifference = 0.029). Associations between ERS and CRC risk were consistent in regard to BRAF mutation, CIMP, and MSI status. All molecular subtypes defined by combinations of tumor markers showed uniformity in their association with the ERS and CRC risk, although Jass type 3 tumors (MSI-low/stable, CIMP-low/negative, BRAF-wildtype, KRAS-mutated) in females and Jass type 5 tumors (MSI-high, CIMP-low/negative, BRAF-wildtype, KRAS-wildtype) in males differed significantly in their association with the ERS from the predominant Jass type 4 (Pdifference = 0.044 and 0.015, respectively). Discussion: The ERS was strongly associated with CRC risk overall and across most subgroups of CRC defined by tumor characteristics. Our results suggest that KRAS-wildtype tumors are more strongly associated with environmental factors than are KRAS-mutated tumors, particularly in women.