Background: A diet comprising low fiber intake, high consumption of red meat, and alcohol drinking, are all established risk factors for colorectal cancer (CRC). Identifying metabolic fingerprints associated with data-driven dietary patterns may provide insight into potential mechanisms linking diet and risk of CRC. Objective: The aim of this study was to first identify robust data-driven dietary factors derived from food frequency data and to assess their relationships with CRC risk. Second, to investigate plasma metabolic profiles in relation to CRC-associated diet factors. Methods: We used self-reported food frequency data and fasting plasma samples from 860 incident CRC cases (median 12 years from sampling to diagnosis) individually matched to controls by cohort, sex, age at diagnosis, year of blood draw and data collection (+/-1yr), fasting status and number of sample freeze thaw cycles from the population-based Northern Sweden Health and Disease Study (NSHDS). Diet factors were identified from combined exploratory/confirmatory factor analysis on random half-splits of the dietary data. Factors that appeared reproducible between repeated half-splits were retained, and their associations with CRC risk were assessed by multivariable conditional logistic regression, adjusted for age, BMI, smoking, physical activity, total energy intake, education and marital status. Factors associated with CRC risk were related to LC-MS metabolic profiles using random forest regression. Preliminary Results: We identified 12 dietary factors that were robustly associated with CRC risk: red meat, alcohol, snacks, fish, healthy breakfast, fast food, spread, full fat, low-fat, smoked fish/meat, vegetables, and an additional factor consisting of rice and fruit/syrup soup. Both alcohol (OR 1.16, 95% CI 1.00-1.34, p=0.047, fdr=0.283) and the healthy breakfast factor (OR 0.89, 95%CI 0.80-0.99, p=0.030, fdr=0.283) were associated with CRC in the adjusted model. Of these two factors, alcohol was also weakly associated with the metabolic profile (Q2=0.19). Conclusion: The data-driven factor analysis successfully identified dietary exposures, of which several corresponded to established CRC risk factors. Alcohol and healthy breakfast habits associated with CRC risk after adjustment for potential confounders. We also identified a metabolic profile in plasma potentially linked to alcohol consumption which merits further investigation.

Citation Format: Stina Bodén, Rikard Landberg, Sophia Harlid, Ingegerd Johansson, Marc J. Gunter, Richard Palmqvist, Rui Zheng, Bethany Van Guelpen, Carl Brunius. Data-driven dietary patterns and their association with metabolic fingerprints and risk of colorectal cancer - a nested case-control study from Northern Sweden [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr LB-156.