Background: Global dietary patterns have shifted dramatically with marked increases in the consumption of ultra-processed foods (UPFs), formulations typically high in free sugars, sodium, saturated fats and additives. While higher UPF intake has been associated with increased risk of colorectal cancer, breast cancer, ovarian cancer, and lung cancer, key evidence gaps impede the development of public health action. Understanding the biological mechanisms through which UPFs affect cancer risk is critical to strengthening causal inference, identifying biomarkers for more accurate UPF exposure assessment, and revealing targets most relevant for prevention and policy. The gut microbiome is shaped by dietary intake and may mediate metabolic and inflammatory changes relevant to cancer, while circulating metabolites reflect both diet, microbial function and host response, providing a unique opportunity to uncover biomarkers of UPF exposure and functional mechanisms linking UPF to cancer risk.
Objectives: To improve our understanding of UPF exposures and host- microbiome/metabolic interactions, we conducted an integrative analysis of plasma metabolite and gut metagenomic data from 800 participants. Specifically, we aimed to identify functional host and microbiome signatures of UPF consumption.
Methods: We conducted a cross-sectional study of 800 participants without cancer who completed questionnaires, including a past-year food frequency questionnaire (CDHQ-III), provided a blood and stool sample. Food processing was categorized using NOVA, and expressed as a percentage of total energy intake. Plasma metabolites were measured using untargeted mass spectrometry and stool samples were assayed with whole metagenome shotgun sequencing. Metabolite data were analyzed with multivariate regression and orthogonal partial least squares discriminatory analysis while differential abundance analysis was used to assess microbial composition. Integration of metabolite and metagenomic data was conducted in DIABLO. 
Results: 1,080 metabolites were detected, of which 61, including pipecolate and phosphatidylethanolamines, were inversely associated with UPF consumption while 7 metabolites, including bile acid metabolites, were positively associated (q<0.05). Distinct metabolite profiles were observed between high (quartile 1) and low (quartile 4) UPF consumption including metabolites in pathways implicated in cancer development such as arginine biosynthesis, nucleotide metabolism and the TCA cycle-related energy pathways. Functional analyses revealed inverse associations between UPF consumption and gut commensals involved in short-chain fatty acid metabolism and fiber fermentation alongside positive associations between UPF consumption and taxa involved in pro-inflammatory pathways. Integrative analyses demonstrated strong correlations between metabolite and microbial profiles. Pathways discriminating high vs low UPF consumption largely reflected single omics analysis, with top metabolite features including pipecolate, bile acid metabolites and 13-carboxy-γ-tocopherol, while top microbial features were Faecalibacterium prausnitzii, Bifidobacterium adolescentis and Agathobacter rectalis.
Conclusions/Implications: Dietary patterns characterized by higher UPF intake are associated with distinct metabolite and microbial profiles, some of which play roles in pathways related to cancer. These findings may help to improve estimates of UPF exposure by providing objective measures of UPF consumption. Prospective and interventional studies are needed to validate results, assess whether pathways can be modified through dietary change, and determine the relevance of biomarkers and pathways for policy-relevant prevention strategies.