Plasma TMAO increase after healthy diets: results from 2 randomized controlled trials with dietary fish, polyphenols, and whole-grain cereals

Abstract

ABSTRACT Background Plasma trimethylamine N-oxide (TMAO) has drawn much attention as a marker of several chronic diseases. Data on the relation between diet and TMAO are discordant and few human intervention studies have assessed causality for this association. Objectives We aimed to evaluate the effects on plasma TMAO of diets based on foods rich in polyphenols (PP) and/or long-chain n–3 fatty acids (LCn3) or whole-grain cereals (WGCs), in individuals at high cardiometabolic risk. Methods An ancillary study was performed within 2 randomized controlled trials, aimed at evaluating the medium-term effects on cardiometabolic risk factors of diets naturally rich in PP and/or LCn3 (Etherpaths Project) or WGCs (HealthGrain Project). Results In the Etherpaths study (n = 78), the changes in TMAO (8-wk minus baseline) were statistically significant for the diets rich in LCn3 (+1.15 ± 11.58 μmol/L) (P = 0.007), whereas they were not for the diets rich in PP (−0.14 ± 9.66 μmol/L) (P = 0.905) or their interaction (P = 0.655) (2-factor ANOVA). In the HealthGrain Study (n = 48), the TMAO change (12-wk minus baseline) in the WGC group (+0.94 ± 3.58 μmol/L) was significantly different from that in the Refined Cereal group (−1.29 ± 3.09 μmol/L) (P = 0.037). Considering the pooled baseline data of the participants in the 2 studies, TMAO concentrations directly correlated with LCn3, EPA (20:5n–3), and protein intake, but not SFAs, fiber, MUFAs, and PP intake. Among food groups, TMAO directly correlated with the intake of fish, vegetables, and whole-grain products, but not meat, processed meat, and dairy products. Conclusions Diets rich in LCn3 of marine origin or WGCs significantly increased plasma TMAO concentration. These changes mirrored the direct associations between TMAO concentrations and intakes of fish and WGCs, suggesting that TMAO reflects intakes of these healthy foods and, therefore, it is not a universally valid biomarker of cardiometabolic risk independent of the background diet. These trials were registered at clinicaltrials.gov as NCT01154478 and NCT00945854.