A lipidomics score captures improved dietary fat quality and is related to cardiometabolic risk

Abstract

Abstract Funding Acknowledgements Type of funding sources: Other. Main funding source(s): Grant from the German Federal Ministry of Education and Research (Bundesministerium fuer Bildung und Forschung) to the German Center for Diabetes Research (DZD grant 82DZD00302), Joint Programming Initiative A Healthy Diet for a Healthy Life, as part of the ERA-HDHL cofounded joint call Biomarkers for Nutrition and Health (01EA1704). Background Deep lipidomics might more accurately capture metabolic effects of dietary interventions that aim to modulate the fat quality of the habitual diet compared to standard risk markers. We generated a multi-lipid score (MLS) to summarize the effect of a dietary intervention aiming to substitute saturated with unsaturated fats and assessed its association with incident cardiometabolic diseases. Methods The Dietary Intervention and VAScular function study (DIVAS) was a 16-week, randomized, controlled parallel unsaturated-for-saturated fat substitution (8% total energy, Figure 1A) trial in hundred-thirteen men and women. In regression models, we examined the diet effect on single lipids and combined the significantly affected lipids (P<0.05, multiple testing corrected) in a weighted multi-lipid score (MLS). Then, we used case-cohort designs and Prentice-weighted multi-variable adjusted Cox regression in the European Investigation into Cancer and Nutrition (EPIC)-Potsdam cohort to associate this MLS with primary cardiovascular disease, defined as composite outcome combining stroke and myocardial infarction, and type 2 diabetes (T2D) incidence. Deep lipidomics data (940 molecular lipid species) were available in subsets of both studies (DIVAS: n=113; EPIC-Potsdam: subcohort, n=1262; CVD cases, n=551; T2D cases, n=775). Results The DIVAS unsaturated fat-rich diet reduced lipid metabolites with medium or long-chain fatty acid residuals containing no or few unsaturations in ceramides, cholesterol esters, diglycerides, and phospholipids, which we combine in the MLS (Figure 1B/C). Aside from the MLS, only non-high-density lipoprotein cholesterol (non-HDL-C) was significantly affected in the diet (Figure 1D). In multivariable adjusted models, the MLS, scaled to the observed change in DIVAS (i.e. the theoretical intervention effect), was strongly associated with reduced risk for incident CVD and T2D (Figure 2A). Further adjustment for established risk markers did not meaningfully alter the association, with the exception of the association with T2D and adjustment for plasma triglycerides. Translated to a hypothetical effect on case occurrence, the DIVAS diet intervention-induced MLS difference was associated with markedly fewer CVD (32%, 95%-CI: 21% to 42%) cases and T2D (26%, 95%-CI: 15% to 35%) cases in EPIC-Potsdam (Figure 2B). In contrast, initially observed associations of non-HDL-C, the only other intervention-affected marker, were attenuated when adjusting for the MLS. Conclusion Capturing diet effects using lipidomics profiling, as we show with the generated MLS, might provide superior biomarkers to monitor cardiometabolic benefits of changes to dietary fat quality.