Valid measures of diet are essential for monitoring and improving human health. One simple measure is dietary diversity (i.e., the number of different foods eaten over a period of time), which can serve as a marker of nutritional inadequacies. Surveys of dietary diversity, however, are limited by current assessment techniques that rely on dietary self-report. Another approach to surveying dietary diversity would be to use biomarkers or biochemical indicators of diet. However, a dietary biomarker that specifically captures overall dietary diversity has yet to be developed. Here, we will develop and validate a new technique, known as DNA metabarcoding, for enumerating the number of dietary plant and animal species individuals consume (dietary species richness). Our approach builds on a conceptual insight made by ecologists studying complex feeding practices in wild animal populations, which is that dietary DNA survives digestion and can be detected in stool using high-throughput DNA sequencing. Our preliminary studies support the promise of this approach: we have successfully amplified and sequenced more than a hundred dietary species from over a thousand human stool samples collected across multiple countries. Our pilot work has also shown significant correlations between the number of dietary species captured by DNA metabarcoding and survey-based indices of dietary diversity and quality. To further establish DNA metabarcoding as a reliable and useful marker of dietary diversity, our team of experts in fecal genomics, nutritional epidemiology, and biostatistics will pursue two Specific Aims. First, we will optimize DNA metabarcoding for assessing intake of dietary animal species. This Aim will build on our existing protocols for metabarcoding analysis of dietary plants. We will perform bench-top experiments under well- controlled lab settings using mixtures of intact and processed foods. We will then test our most promising protocols using a repository of stool samples collected from human cohorts undergoing controlled feeding. Second, we will test the validity and utility of measuring dietary diversity using DNA metabarcoding. This Aim will apply the technique to: 1) a cohort of primarily African-American/Hispanic youth from low-income families enrolled in a study of obesity treatment; and, 2) a cohort of 1,000 individuals of African descent from five countries with varying dietary habits and cardiometabolic disease risk. We will use these studies to validate that metabarcoding species richness reflects existing measures of dietary diversity measured by recall-based surveys of dietary intake. These real-world cohorts will further allow us to integrate metabarcoding data into models of metabolic disease risk, examine temporal trends in metabarcoding results, and identify potential geographic and socioeconomic determinants of dietary species richness.

(Funded by NIH 1R01DK128611-01A1)