(P29-006-23) Developing and Evaluating the Construct Validity of a Diet Pattern Predictive of Plasma TMAO and Choline

Objectives: The metabolism of choline (highly present in animal products) can produce trimethylamine N-oxide (TMAO), a metabolite with atherosclerotic effects; however, dietary fiber may suppress this metabolic pathway. This study aimed to develop a diet pattern predictive of plasma TMAO and choline concentrations using reduced rank regression (RRR) and to evaluate its construct validity.

Methods: This analysis included data from 1724 women who participated in an ancillary study within the Women's Health Initiative Observational Study, a prospective cohort study of post-menopausal women (ages 50-79 at enrollment). Baseline (1993-1998) dietary data were collected from a food frequency questionnaire. Baseline plasma choline (μmol/L) and TMAO (μmol/L) concentrations were assessed using liquid chromatography-tandem mass spectrometry. The diet pattern was developed using RRR in half of the sample (Development Sample) and applied to the other half of the sample (Validation Sample). RRR is a method that considers both the metabolic process of interest and the covariance structure of dietary data. Energy-adjusted (residual method) MyPyramid equivalents food groups were the predictor variables and plasma choline and TMAO, the response variables. The first RRR factor was retained and used to calculate the diet pattern score for each participant by summing the intake of each food group weighted by its respective retained factor loading value. In each sample, ANOVA was used to test if the means for each biomarker statistically differed across the medians of the quartiles of the diet pattern score.

Results: Discretionary fat, white potatoes, red meat, and eggs were positively associated with the diet pattern with factor loadings >0.15, while yogurt, fruits, added sugar, and starchy vegetables were inversely associated with factor loadings < -0.15.Mean TMAO and choline concentrations increased across increasing quartiles of the diet pattern in the Development (p-trends < 0.0001) and Validation Samples (p-trends were 0.014 for TMAO and 0.0007 for choline).

Conclusions: We identified a diet pattern reflecting the potential of the diet to contribute to plasma concentrations of TMAO and choline.

Funding Sources: National Heart, Lung, and Blood Institute and National Cancer Institute, National Institutes of Health