Dietary Analysis with Lipdomics: First Steps Toward Objective Dietary Analysis of Macronutrient Intake Using Lipdomics

Abstract

Dietary macronutrients are an important and controllable factor in health. Current research best practices use memory-based dietary assessments (MBDA) to estimate nutrient intake. However, there is evidence of bias in information obtained from MBDA as subjects over or under report energy and macronutrient intake. Metabolomics is a powerful tool to used identify molecular biomarkers of disease risk, not only can provide mechanistic insights into macronutrient metabolism but also can inform our understanding of accuracy and precision of MBDA. Metabolomics uses high throughput profiling to identify small metabolites. Lipidomics is a subset of metabolomics that primarily identifies lipids. Here, I describe the use of plasma lipidomic profiling to determine the temporal changes in the plasma lipidome of healthy adult participants provided high fat or high carbohydrate diets and objectively identify dietary macronutrient content. The PUFA study provided a high polyunsaturated fat diet (40-50% fat, 80% polyunsaturated fats) for 3 weeks followed by a high carbohydrate diet (75% carbohydrates - CHO) to 12 individuals. After 2 days of PUFA, 16% of 480 lipids showed significant changes and 27% changed after 21 days. After switching to CHO, 27% percent of lipids changed after 2 days and 30% after 21 days. This demonstrated that a high PUFA diet produced a rapid turnover of the plasma lipidome. Next, we provided a standard diet for 3 days to 24 individuals then randomized to a high fat diet (60% fat, HF) or a high carbohydrate diet (75% carbohydrates, HC) for 3 weeks. Fewer lipidomic changes were apparent within group over 21 days, but significant differences between experimental groups were apparent. We identified a set of ‘sentinel’ lipids comprising mostly plasmalogens and phosphatidylcholines that classified the participants in both studies with 87% predictive value. A significant correlation between macronutrient intake and predictions in lipidomics data from a study of individuals with known diets highlighted the potential use of lipidomics to identify dietary intake in free-living population. In parallel with the MEAL study, we compared known diets of subjects with reported diets during the study. Participants reported kCal intake accurately but over-reported protein intake in all diets. On the HF diets, fat was under-reported by 6.3% and carbohydrate over-reported by 19.8%. On the HC diet, carbohydrate under-reported by 10.8% and fat over-reported by 66%. These results suggest 24HR estimate energy well in controlled feeding studies. However, estimation of extreme macronutrient intakes in the context of a feeding study was poor. To assess dynamic changes in the metabolome from acute intravenous glucose and insulin, we collected plasma from a cohort of obese, metabolically healthy and lean individuals during hyperinsulinemic-euglycemic clamps. Targeted profiles of metabolites demonstrated changes associated with glucose infusion rate (GIR), rather than obesity. S with low GIR showed greater suppression of plasma fatty acids and increased levels of branched-chain amino acids throughout the clamp. Paradoxically, insulin suppressed long chain fatty acids while their cognate acylcarnitines were unchanged but not shorter chained species. This novel finding suggests a clearance difference of acylcarnitines by chain length. These studies collectively demonstrate the utility of metabolomics and lipidomics to identify macronutrient intake. The development of objective biomarkers of intake could produce complementary diagnostic tools for nutritional studies. The identification of differential dynamic changes due to macronutrient exposures and insulin sensitivity has the potential to identify unknown physiological effects of diet.