(P13-038-25) A Pilot Study to Compare Plant Metabarcoding With Biomarkers and Dietary Recall Data

Objectives: Dietary recall methods rely heavily on participant memory and honesty, which may contribute to inaccuracies and measurement error. We aim to compare plant metabarcoding and non-invasive biomarker methods to determine concordance with dietary recall data and improve dietary assessment.

Methods: A pilot study of 9 participants undergoing medical management of hemorrhoids was conducted. Participants provided a stool sample, three dietary recalls, and VeggieMeter screening at two time points over a 1-month period. Healthy Eating Index (HEI) scores were calculated from telephone-administered recalls using the Nutrition Data System for Research (NDSR). FoodSeq is a method to identify residual plant DNA in stool samples, reducing reliance on self-reported data. The trnL-P6 gene from stool samples was amplified to identify recently consumed plants. Plant metabarcoding richness (pMR) was defined as the number of unique plants identified in a sample, per participant. FoodSeq-detected plants were compared to reported plants in food recall data at each timepoint. The Veggie Meter non-invasively measures skin carotenoids. Correlations between pMR, skin carotenoid scores (SC), and HEI scores were assessed at each timepoint and using the average of the two timepoints.

Results: FoodSeq detected 24% of plants reported by participants via NDSR recall; 13% of FoodSeq-detected plants were unreported by participants, while 63% of reported plants were undetected by FoodSeq. Unreported FoodSeq-detected plants included cannabis (n=1/9), cocoa/chocolate (n=6/9), and grapes/wine (n=3/9). A significant positive correlation between pMR and HEI was observed at the first time point (ρ=0.711; p=0.031), but not at the second time point (ρ=-0.203; p=0.599) or across both time points (ρ=0.283; p=0.254). A non-significant negative correlation between pMR and SC was observed at the second time point (ρ=-0.593; p=0.0921) and across both time points (ρ=-0.163; p=0.517).

Conclusions: This pilot study shows variability in dietary intake measurement across methods, some concordance between HEI and pMR, and little agreement between pMR and SC. Further analysis will assess whether combining methods improves dietary pattern and diversity estimation.

Funding Sources: Funding for this study was provided by the Allen Foundation.