Smoking and molecular subtypes of colorectal cancer

From the Peters Group, Division of Public Health Sciences

In 2019, there was an incidence of 145,600 colorectal (CRC) cases and 51,020 deaths in the US. CRC is one of the most common fatal cancers in the US. CRC can be characterized by its genetic variability and molecular heterogeneity, and the molecular classifications of CRC lead to various prognostic and treatment conclusions. To better understand if smoking an established risk factor for CRC is associated with specific subtypes of CRC Dr. Riki Peters’ Group, in the Division of Public Health Sciences assessed the associations between smoking and CRC subtypes defined by microsatellite instability (MSI) status, CpG island methylator phenotype (CIMP) status, and mutations in well-known oncogenes (KRAS and BRAF). As the team harmonized individual level data across multiple studies, they were not only able to look at individual tumor characteristics but also the sample sizes to investigate combinations of all 4 markers. This paper was published in JNCI Cancer Spectrum. 

The study used data from 11 observational studies including 9,789 CRC patients and 11,231 controls from the Genetics and Epidemiology of Colorectal Cancer Consortium (GECCO) and the Colon Cancer Family Registry.  Cases were confirmed via medical record, pathology report, or death certificate.  Subtype classifications were created by combining markers: MSI, CIMP, BRAF, and KRAS status. Demographic and environmental factors were collected during in-person interviews or self-administered questionnaires.  Multinomial logistic regression models were used to measure the odds ratios and 95% confidence intervals for the association of smoking with the risk of CRC subtypes.  All statistical tests were adjusted for multiple testing.

As expected, they found overall that smoking was associated with higher risk of CRC. Looking at each of the four tumor characteristics separately they observed that the association between smoking and CRC risk differed statistically significantly by MSI status, CIMP status and BRAF mutation, with the biggest differences observed for CIMP-positive vs CIMP-negative and BRAF-mutated vs and BRAF-non-mutated. For instance, compared with never-smokers, smokers in the fourth quartile of pack-years had a 90% higher risk of CIMP-positive CRC but only 35% higher risk for CIMP-negative CRC. When they combined all four markers, they found that the association was stronger in tumors that were CIMP positive, MSI high, or KRAS wild type when combined.

Figures 1 and 2 provides a detailed analysis of smoking differences by molecular subtypes based on the combined analysis of all 4 markers.   

Graphical representation of Smoking CRC associations differ by molecular subtype
Smoking CRC associations differ by molecular subtype Image from Dr. Peters
Graphical Representation of Smoking-CRC associations differ between groups (case-case analysis)
Smoking-CRC associations differ between groups (case-case analysis) Image from Dr. Peters

 This is the largest study to date. Dr. Peters concluded: “By bringing together colorectal tumor characteristics from a large number of well characterized studies, we were not only able to establish associations with specific individual tumor characteristics, but also able to analyze the combination of these markers. This allowed us to identify that smoking is particularly strong associated with tumor developed through the serrated pathway. This has important implication for screening as those cancer and particularly its precursor lesion tend to be more difficult to be detected by colonoscopy screening”.

Tabitha Harrison, a GECCO team member and leader of this project added. “We are running similar analyses for other colorectal cancer risk factors (such as BMI, diet, hormone use, alcohol intake, etc.) to see if we can learn more about whether the effects of these risk factors vary by tumor subtypes. This work is thanks to a large consortium of studies that have collaborated together for over a decade, allowing for analyses using sizeable datasets.”

This research was supported by the National Institutes of Health, National Cancer Institute, US Department of Health and Human Services.

Fred Hutch/UW Cancer Consortium members Polly A. Newcomb, Amanda Phipps, Wei Sun, Li Hsu, Tabitha Harrison and Riki Peters contributed to this work. Xiaoliang (Wendy) Wang is a former post-doc at the Fred Hutch and currently working as researcher at FlatIron, New York.  Tabitha Harrison is now PhD student at the University of Washington

Wang X, Amitay E, Harrison TA, Banbury BL, Berndt SI, Brenner H, Buchanan DD, Campbell PT, Cao Y, Chan AT, Chang-Claude J. Association between Smoking and Molecular Subtypes of Colorectal Cancer. JNCI Cancer Spectrum. 2021 Jun 14. DOI:10.1093/jncics/pkab056