(P19-048-25) Metagenomic Estimation of Dietary Intake From Human Stool Using MEDI

Objectives: Nutrition intervention studies rely on participant dietary adherence, which is frequently a self-reported measure. Biological samples, including stool, provide objective biomarkers of dietary intake that can complement self-reported data. Herein, we evaluated the ability of the Metagenomic Estimation of Dietary Intake (MEDI) pipeline to detect foods consumed as part of a complete feeding trial.

Methods: Dietary and fecal metagenomic data from a complete feeding study with a 7d menu cycle were utilized. Healthy adults (n=28) participated in a randomized, controlled, crossover trial with three 21-day dietary conditions that were identical except for the inclusion of walnuts, walnut oil, or corn oil in the diet. Participants provided 2 fecal samples between days 18–20 of each dietary condition. Following fecal DNA extraction, shotgun genomic DNA sequencing was conducted using a NovaSeq X Plus. Metagenomic shotgun sequence data were analyzed using the MEDI pipelines for recovering food abundances, including mapping items to FOODB and quantifying food DNA. Binary and continuous read counts of detected foods were analyzed.

Results: MEDI detected varying proportions of consumed foods across the 7d menu cycle: 72% on Day 1 (13/18 foods), 67% on Day 2 (12/18 foods) and Day 3 (10/15 foods), 89% on Day 4 (8/9), 86% on Day 5 (12/14 foods), 69% on Day 6 (9/13 foods), and 73% (11/15 foods) on Day 7. Commonly detected foods included turkey, lettuce, onion, peas, and celery. Walnut DNA was detected in 55% of samples during the walnut condition. Walnut oil and corn oil were not among the foods identified in the dataset.

Conclusions: MEDI effectively profiled dietary intake from human fecal microbiome data, demonstrating robust detection of whole foods but limited sensitivity to highly processed foods like oils. The frequent detection of whole plant-based foods may be attributed to their raw consumption, which supports DNA durability throughout digestion, resulting in food-based DNA detectability within the stool. MEDI can be used to generate objective dietary biomarkers that can complement self-reported dietary data, thereby supporting diet-microbiome research.

Funding Sources: USDA NIFA grant no. 2020-67017-30836/project accession no. 1021932 and USDA Hatch project accession no. 1026591