Smoking and risk of colorectal cancer sub-classified by tumor-infiltrating T cells

Background Evidence indicates not only carcinogenic effect of cigarette smoking but also its immunosuppressive effect. We hypothesized that the association of smoking with colorectal cancer risk might be stronger for tumors with lower anti-tumor adaptive immune response. Methods During follow-up of 134 981 participants (3 490 851 person-years) in the Nurses' Health Study and Health Professionals Follow-up Study, we documented 729 rectal and colon cancer cases with available data on T-cell densities in tumor microenvironment. Using the duplication-method Cox regression model, we examined a differential association of smoking status with risk of colorectal carcinoma subclassified by densities of CD3 + cells, CD8 + cells, CD45RO (PTPRC) + cells, or FOXP3 + cells. All statistical tests were two-sided. Results The association of smoking status with colorectal cancer risk differed by CD3 + cell density (P heterogeneity =.007). Compared with never smokers, multivariable-adjusted hazard ratios for CD3 + cell-low colorectal cancer were 1.38 (95% confidence interval = 1.09 to 1.75) in former smokers and 1.59 (95% confidence interval = 1.14 to 2.23) in current smokers (P trend =.002, across smoking status categories). In contrast, smoking status was not associated with CD3 + cell-high cancer risk (P trend =.52). This differential association appeared consistent in strata of microsatellite instability, CpG island methylator phenotype, or BRAF mutation status. There was no statistically significant differential association according to densities of CD8 + cells, CD45RO + cells, or FOXP3 + cells (P heterogeneity >.04, with adjusted α of 0.01). Conclusions Colorectal cancer risk increased by smoking was stronger for tumors with lower T-lymphocyte response, suggesting an interplay of smoking and immunity in colorectal carcinogenesis.