Abstract Body: Background: Plant-based diets are recommended to lower the risk of type 2 diabetes (T2D) and coronary heart disease (CHD). Proteomic responses to plant-based diets may reveal biological pathways underlying the associations between plant-based diets and cardiometabolic disease risk.
Hypothesis: Proteomic profiles reflecting adherence and biological response to an overall plant-based diet index (PDI) and a healthy PDI (hPDI) will associate with lower risks of T2D and CHD, but to an unhealthy PDI (uPDI) will associate with higher risks.
Methods: We analyzed baseline plasma Olink antibody-based proteomic profiling and food frequency questionnaire (FFQ) data from 1,660 participants in the Nurses’ Health Study (NHS), NHSII, and Health Professional Follow-up Study. First, proteome-wide association analyses were conducted among 532 proteins for each PDI, adjusting for demographics, lifestyle, and BMI. For proteins with p<0.05, we applied elastic net regression with 10-fold cross-validation to develop plasma proteomic profiles related to baseline self-reported PDIs in a training set (n=1,162). We validated these profiles in a testing set (n=498). Multivariable Cox regression models examined associations between proteomic profiles of PDIs and incident T2D and CHD, adjusting for respective FFQ-derived PDI scores and many potential confounders.
Results: We documented 173 T2D cases and 85 CHD cases during median follow-up periods of 23 and 24 years. We identified proteomic profiles comprised of 35, 54, and 40 proteins that correlated with (Spearman r) the FFQ-derived PDI (0.23), hPDI (0.29), and uPDI (0.23) scores, respectively. Each SD increment in the proteomic profiles of the PDI and hPDI was associated with a lower incident T2D risk (PDI: HR=0.59 [0.50, 0.69]; hPDI: HR=0.75 [0.64, 0.88]), whereas the proteomic profile of the uPDI was associated with a higher incident T2D risk (HR=1.37 [1.17, 1.60]). Each SD increment in the PDI proteomic profile, but not the hPDI proteomic profile, was associated with a lower incident CHD risk (HR=0.76 [0.60, 0.96]), whereas the uPDI proteomic profile was associated with a higher incident CHD risk (HR=1.28 [1.02, 1.60]). All the associations remained statistically significant after further adjustment for BMI, except for the association between the hPDI proteomic profile and T2D risk (HR=0.87 [0.73, 1.03]).
Conclusions: Proteomic profiles of PDIs were associated with risks of T2D and CHD, independent of self-reported diet measures.