Analyzing 7,285 circulating protein aptamers in 4,372 EPIC cohort participants, researchers identified specific proteins that may mechanistically connect plant-based diet quality to chronic disease risk. Using Mendelian randomization and colocalization to filter for genetic causality, the team found 8 proteins linked to healthful plant-based diet scores (hPDI) and 12 linked to unhealthful plant-based patterns (uPDI) with strong disease associations — including EGFR with breast cancer, MMP10 with endometrial cancer, NCAN with type 2 diabetes, and PCSK9 with cardiovascular disease.
This research represents a meaningful methodological advance over purely observational diet-disease studies. By combining high-throughput SomaScan proteomics with Mendelian randomization, investigators attempt to move beyond correlation toward biological plausibility, an approach increasingly standard in precision nutrition research. The PCSK9-CVD link is particularly compelling — PCSK9 inhibition is already a validated therapeutic target, lending credibility to the pathway. However, several caveats apply: the 4,372-person sample is modest for proteomics at this scale, and Mendelian randomization assumes instrument validity that can fail with pleiotropic variants. The distinction between healthful and unhealthful plant-based diets — a nuance often lost in public messaging — is a genuine strength. As a preprint not yet peer-reviewed on medRxiv, these findings remain provisional and the specific protein candidates should be interpreted cautiously until independent replication and full expert scrutiny confirm the methodology. Still, this represents incremental but genuinely useful progress in mechanistic nutrition science.
