A lipid-related metabolomic pattern of diet quality

Abstract

Adherence to a healthy diet has been associated with reduced risk of chronic diseases. Identifying nutritional biomarkers of diet quality may be complementary to traditional questionnaire-based methods and may provide insights concerning disease mechanisms and prevention. To identify metabolites associated with diet quality assessed via the Alternate Healthy Eating Index (AHEI) and its components. This cross-sectional study used FFQ data and plasma metabolomic profiles, mostly lipid related, from the Nurses' Health Study (NHS, n=1460) and Health Professionals Follow-up Study (HPFS, n=1051). Linear regression models assessed associations of the AHEI and its components with individual metabolites. Canonical correspondence analyses (CCAs) investigated overlapping patterns between AHEI components and metabolites. Principal component analysis (PCA) and explanatory factor analysis were used to consolidate correlated metabolites into uncorrelated factors. We used stepwise multivariable regression to create a metabolomic score that is an indicator of diet quality. The AHEI was associated with 83 metabolites in the NHS and 96 metabolites in the HPFS after false discovery rate adjustment. Sixty-three of these significant metabolites overlapped between the 2 cohorts. CCA identified "healthy" AHEI components (e.g., nuts, whole grains) and metabolites (n=27 in the NHS and 33 in the HPFS) and "unhealthy" AHEI components (e.g., red meat, trans fat) and metabolites (n=56 in the NHS and 63 in the HPFS). PCA-derived factors composed of highly saturated triglycerides, plasmalogens, and acylcarnitines were associated with unhealthy AHEI components while factors composed of highly unsaturated triglycerides were linked to healthy AHEI components. The stepwise regression analysis contributed to a metabolomics score as a predictor of diet quality. We identified metabolites associated with healthy and unhealthy eating behaviors. The observed associations were largely similar between men and women, suggesting that metabolomics can be a complementary approach to self-reported diet in studies of diet and chronic disease.