(P41-23-26) Bile Acids and Personalized Nutrition: Biomarkers of Metabolic Responses to Dietary Challenges

Objectives: We aim to distinguish metabolic responders from non-responders to a walnut feeding trial using insulin resistance and sensitivity indices and identify which bile acid (BA) features best predict responder status.

Methods: Adults with obesity (BMI≥30 kg/m²) completed a crossover, controlled, complete feeding study with 3-week walnut and corn oil interventions. Glycemic and insulinemic responses were assessed through 6-hour mixed meal tolerance tests (t0-t360).

Fecal samples were collected on days 18-20. Participants with complete data (N=14) were classified as responders or non-responders. Multiple BA feature sets were constructed, including absolute plasma/fecal concentrations, compositional shifts, and diet-induced changes (deltas) for 4-h AUC and individual time points. PCA was applied to handle dimensionality; PC1 from each set served as a predictor in Firth’s penalized logistic regression to classify responder status.

Results: Baseline markers (BMI and fasting glucose) did not predict responder status. The most significant models were derived from absolute plasma BA concentrations during the walnut intervention. PC1 of absolute plasma 4-h AUCs significantly predicted status (OR = 2.50, p = 0.011), driven by total BAs, UDCA, and DCA. Similarly, absolute concentrations across all time points were significant (OR = 1.52, p = 0.011), driven by total BAs and DCA at 60 and 90 minutes. Absolute fecal BAs performed poorly (OR = 0.921, p = 0.633). Univariate analyses confirmed responders had significantly higher postprandial AUCs for total BAs, CA, CDCA, DCA, and UDCA specifically after the walnut intervention.

Conclusions: Standard clinical markers failed to predict metabolic responses to a walnut-enriched diet. Instead, “responder” status was characterized by a distinct postprandial plasma BA signature. Specifically, plasma BA response patterns showed predictive potential for individual metabolic response variability. The overall magnitude and compositional shifts of BAs distinguished responders from non-responders. These findings support BAs as integrative biomarkers of diet-microbiome-host metabolism, offering a potential biomarker for personalized nutrition.

Funding Sources: USDA NIFA (2020-67017-30836/project accession no. 1021932), Personalized Nutrition Initiative External Partners Program, and CONAHCYT of Mexico.