Ultra-processed food consumption, cancer risk, and cancer mortality: a prospective cohort study of the UK Biobank

Abstract

BackgroundDietary patterns worldwide are increasingly displaced by many cheap, highly palatable, and ready-to-eat ultra-processed foods (UPFs). Higher UPF consumption has been linked to increased risk for obesity and cardiometabolic diseases, but prospective evidence is limited on cancer outcomes. This study aimed to examine the association between UPF consumption and risk for overall and site-specific cancer incidence and cancer mortality using the UK Biobank cohort. Methods197 426 participants of the UK Biobank from England, Scotland, and Wales with 24-h dietary recall completed between 2009 and 2012 were included. Incident cancer cases were identified through data linkage to national cancer and mortality registries. Food items consumed were categorised according to their degree of food processing using the NOVA classification system. Individual UPF consumption were derived as a percentage of daily food intake. Prospective association was assessed using multivariable Cox proportional hazards models adjusted for baseline sociodemographic and lifestyle characteristics. For female-specific cancers, menopausal status, use of oral contraceptives, hormone replacement therapy, and parity were additionally adjusted. FindingsMean age was 58·0 years (SD 8·0); 54·6% of participants were women. During a median follow-up time of 9·8 years (IQR 9·4–10·6), 15 921 (8·1%) of 197 426 individuals developed cancer and 4009 (2·0%) cancer deaths were encountered. Consumption of UPFs was associated with a higher incidence of overall cancer (hazard ratio per 10% increment in UPF consumption was 1·02 [95% CI 1·01–1·04]; p=0·005) and ovarian cancer in females (1·19 [1·08–1·30]; p<0·001). Positive associations were identified for mortality of overall, breast, and ovarian cancers. InterpretationThis large UK cohort study presents evidence of positive associations between UPF consumption and risks for incidence and mortality of overall and certain site-specific cancers. Limitations include non-representativeness having based on a cohort of middle-aged volunteers, insufficient details for some food items although a conservative approach was undertaken classifying foods to lower processing categories, and potential residual confounding due to the observational nature of the data. Further investigation into the mechanistic pathways is warranted to better identify targets for intervention. FundingCancer Research UK and World Cancer Research Fund.