Introduction: We applied multi-omics to study individual responses to dietary interventions and the effects of dietary interventions on the gut microbiome in a long-term randomized controlled trial.
Methods: We profiled host genetics using Global Screening Array and measured the gut microbiome via shotgun metagenomics and fecal and plasma metabolomics using liquid chromatography-mass spectrometry at baseline and 18 months in the 18-month DIRECT-PLUS Trial of 294 men and women (mean age: 51 y) who were randomized into three groups: healthy dietary guidelines, Mediterranean diet (MedDiet) and MedDiet + polyphenols.
Results: We observed considerable variability in individuals’ responses to dietary interventions, as evidenced by mean coefficients of variation of post-intervention changes in multiple outcomes: BMI at 92, HOMA-IR at 108, hs-CRP at 103, and LDL/HDL at 96. Age, sex, and baseline diet, smoking status, weight, waist circumference, pulse, diastolic and systolic blood pressure, and multi-omics collectively accounted for the following proportions of the variability in individuals’ post-intervention changes BMI (4.4%), fasting plasma glucose (11.0%), HOMA-IR (3.7%), and hs-CRP (13.8%). Furthermore, the integration of baseline multi-omics significantly improved the prediction of individual responsiveness to diet (AUC from 0.60 to 0.90), jointly defined by post-intervention changes in BMI, HOMA-IR, LDL/HDL, and hs-CRP. Our longitudinal analysis revealed MedDiet-induced alterations in metabolites and the gut microbial taxonomy, including an increase in butyrate producer Blautia glucerasea and a decrease in Bifidobacterium angulatum.
Conclusion: Our study demonstrates the application of multi-omics in predicting personalized responsiveness to dietary intervention. These findings set the foundation for precision nutrition for more effective prevention of metabolic disease.

The BPRHS is a prospective cohort that enrolled 1500 participants at baseline. Data was collected across four waves, over thirteen years of follow-up. Data from the fourth wave was used in the present study. At wave 4, 309 BPRHS participants self-reported PPI use and self-collected, metagenomically profiled, stool samples. PPI use was classified as any past use. Cross-sectional associations between gut microbial taxa, functional pathways, and PPI use were analyzed using omnibus tests, multivariate linear modeling in MaAsLin2, and random forest classifier in feature-wise analyses. We further compared our results with the non-Hispanic Health Professionals Follow-Up Study to validate key findings and examine ethnicity-related differences.

Among 309 participants (mean age 68.8 years; female 74.6%), 112 (36%) self-reported PPI use. After adjusting for relevant covariates, we observed an enrichment of Streptococcus parasanguinis (=3.16, FDR P=0.01), S. anginosus (=2.89, FDR P<0.01), S. salivarius (=2.56, FDR P=0.01), S. gordonii (=1.98, FDR P=0.15), and Rothia mucilaginosa (=1.54, FDR P=0.06), among PPI users compared to non-users. Streptococci and Lactobacilli predominantly contributed to the functional pathways associated with PPI use. The observed enrichment of oral-typical taxa, such as Streptococci, among PPI users in the BPRHS, suggests the potential of PPIs to alter gut microbial composition. More studies are needed to understand the impact of PPI use on the gut microbiome in different ethnicities.