Plasma proteins associated with plant-based diets: Results from the Atherosclerosis Risk in Communities (ARIC) study and Framingham Heart Study (FHS)

Abstract

Background & AimsPlant-based diets are associated with a lower risk of chronic diseases. Large-scale proteomics can identify objective biomarkers of plant-based diets, and can improve our understanding of the pathways that link plant-based diets to health outcomes. This study investigated the plasma proteome of four different plant-based diets [overall plant-based diet (PDI), provegetarian diet, healthful plant-based diet (hPDI), and unhealthful plant-based diet (uPDI)] in the Atherosclerosis Risk in Communities (ARIC) Study and replicated the findings in the Framingham Heart Study (FHS) Offspring cohort. MethodsARIC Study participants at visit 3 (1993-1995) with completed food frequency questionnaire (FFQ) data and proteomics data were divided into internal discovery (n=7,690) and replication (n=2,543) data sets. Multivariable linear regression was used to examine associations between plant-based diet indices (PDIs) and 4,955 individual proteins in the discovery sample. Then, proteins that were internally replicated in the ARIC Study were tested for external replication in FHS (n=1,358). Pathway overrepresentation analysis was conducted for diet-related proteins. C-statistics were used to predict if the proteins improved prediction of plant-based diet indices beyond participant characteristics. ResultsIn ARIC discovery, a total of 837 diet-protein associations (PDI=233; provegetarian=182; hPDI=406; uPDI=16) were observed at false discovery rate (FDR) <0.05. Of these, 453 diet-protein associations (PDI=132; provegetarian=104; hPDI=208; uPDI=13) were internally replicated. In FHS, 167/453 diet-protein associations were available for external replication, of which 8 proteins (PDI=1; provegetarian=0; hPDI=8; uPDI=0) replicated. Complement and coagulation cascades, cell adhesion molecules, and retinol metabolism were over-represented. C-C motif chemokine 25 for PDI and 8 proteins for hPDI modestly but significantly improved the prediction of these indices individually and collectively (P value for difference in C-statistics<0.05 for all tests) ConclusionsUsing large-scale proteomics, we identified potential candidate biomarkers of plant-based diets, and pathways that may partially explain the associations between plant-based diets and chronic conditions.