Background. Alcohol consumption is associated with increased risk of several cancers, including hepatocellular carcinoma, upper aero-digestive tract, including oesophageal, breast, and colorectal cancers. While ethanol and its metabolites are known to have genotoxic effects, the biological mechanisms linking alcohol intake to cancer development remain poorly understood. Large-scale proteomic profiling in prospective epidemiological studies offers an opportunity to investigate molecular pathways that may underlie alcohol-related carcinogenesis.
Objectives. To derive a proteomic signature of alcohol intake using plasma proteomics data and to examine the association of this signature, as well as its individual proteins, with risk of alcohol-related cancers.
Methods. Within a case-cohort study within the European Prospective Investigation into Cancer and nutrition (EPIC), self-reported alcohol intake were compared to plasma proteomics measured using the 7K SomaScan platform in 15,170 participants. In a representative subcohort (4,115 participants), proteins linearly associated with alcohol intake were first identified. A 10-fold cross-validated LASSO regression model was used to construct a proteomic signature of alcohol consumption. The association between the derived signature and alcohol intake was evaluated using nested cross-validation. Associations between the proteomic signature, individual proteins, and cancer risk were estimated using Cox proportional hazards models.
Results. The proteomic signature comprised 342 proteins and was moderately correlated with alcohol intake (Pearson r = 0.66; 95% CI: 0.64-0.68). After correction for multiple testing (false discovery rate lower than 0.05), the proteomic signature was statistically significantly associated with risk of squamous-cell oesophageal cancer (N=36, hazard ratio, HR = 4.92 ; 95% confidence interval, CI: 2.53-9.57) and upper aerodigestive tract cancers (N=146, HR = 1.62; 95% CI: 1.04-2.52). The proteomic signature was not statistically significantly associated with colon (N=658, HR = 1.18; 95% CI: 0.92-1.51), rectal (N=319, HR = 1.14; 95% CI: 0.80-1.61), breast (N=970, HR = 0.96; 95% CI: 0.74-1.25), or hepatocellular carcinoma (N=50, HR = 0.75; 95% CI: 0.33-1.71). Within the signature, 1 protein was significantly associated with breast cancer risk, 4 with oesophageal cancer risk, 4 with colon cancer risk, 38 with upper aerodigestive tract cancer risk, and 109 with hepatocellular carcinoma risk after correction for multiple testing. Several proteins included in the signature, such as APOA1, ME1, and UBC, are involved in metabolic and signalling pathways, including those related to the peroxisome proliferator-activated receptor (PPAR) signalling, which may indicate alteration of lipid homeostasis and metabolic stress associated with alcohol intake.
Conclusions/Implications. We identified a proteomic signature of alcohol intake that was associated with risk of multiple alcohol-related cancers in a large prospective cohort. The proteins contributing to this signature highlighted metabolic and signalling pathways that may be relevant to alcohol-associated carcinogenesis. These findings provide insights into potential biological mechanisms linking alcohol consumption to cancer risk.