This study analyzed molecular links between plant-based diet quality and chronic diseases using circulating protein measurements and genetic instruments in the European Prospective Investigation into Cancer and Nutrition (EPIC). The analysis included 4,372 participants and measured 7,285 aptamers using the SomaScan platform to profile circulating proteins. The authors aimed to connect diet-associated proteomic signatures to risk of overall and site-specific cancers, type 2 diabetes (T2D), and cardiovascular disease (CVD) by integrating proteomic discovery with genetic prioritization.
Two previously defined plant-based diet indices were evaluated: a healthful plant-based diet index (hPDI) and an unhealthful plant-based diet index (uPDI). These indices quantify the quality of plant-based dietary patterns and were used to identify proteins whose circulating concentrations correlated with healthier versus less healthy plant-forward diets.
The study used complementary statistical and machine-learning approaches to screen the 7,285 SomaScan-measured aptamers for associations with the hPDI and uPDI dietary scores. These discovery-stage methods aimed to generate a set of diet-associated proteins for further genetic interrogation. The source reports that this two-step strategy combined conventional association testing with machine-learning selection to prioritize proteins linked to diet quality in the EPIC sample.
To evaluate whether diet-associated proteins may play causal roles in chronic disease risk, the authors applied cis-protein quantitative trait loci (cis-pQTL) Mendelian randomization (MR). MR leverages genetic variants that influence circulating protein levels to test for associations with disease outcomes that are less susceptible to confounding or reverse causation. The study complemented MR with colocalization analyses to assess whether the same genetic signals underlie both protein concentration and disease association at a locus. This combined cis-pQTL MR plus colocalization approach was used to prioritize proteins with genetic evidence linking them to overall and site-specific cancers, T2D, and CVD.
Among the diet-associated proteins that underwent MR analysis, the authors report a subset with robust colocalization between protein loci and disease signals. In total, 8 hPDI-related and 12 uPDI-related protein–disease associations showed strong colocalization. The manuscript highlights several illustrative pairs with biological and clinical relevance, including EGFR associated with breast cancer, MMP10 associated with endometrial cancer, NCAN associated with T2D, and PCSK9 associated with CVD. These prioritized proteins are presented as candidate mediators that may help explain how plant-based diet quality relates to risk of specific chronic diseases.
The integrated proteomic and genetic analyses nominate circulating proteins that could plausibly link plant-based diet quality to disease pathways for cancer, T2D, and CVD. By identifying proteins with both diet associations and genetic evidence for disease relevance (via cis-pQTL MR and colocalization), the study provides mechanistic hypotheses that may inform future functional, clinical, or interventional research. The authors emphasize that these findings offer biological insights relevant to the potential health benefits of healthful plant-based diets, which has implications for public health and planetary health discussions. As noted in the source, this work is reported as a preprint and has not undergone peer review, so findings should not yet guide clinical practice.
The project received ethical approval from the IARC Ethics Committee and the relevant committees of participating EPIC centres. The authors confirmed that participant consent and required institutional approvals were obtained. All authors declared no competing interests. The source states that data produced in the study are available on reasonable request to the authors. Funder information recorded in the manuscript includes support by the National Research Foundation of Korea (NRF) among other funders acknowledged by bioRxiv/medRxiv for platform support.
Note on scope and limitations
The summary above is drawn from the preprint abstract and accompanying declarations. Detailed methods, full lists of identified proteins, effect estimates, statistical thresholds, and other analytic specifics were reported in the manuscript but are not reproduced verbatim here. Because the article is a preprint, results have not been peer reviewed; readers should consult the full manuscript and subsequent peer-reviewed publications for complete methodological detail and validation.