Lovejoy, J. | bioRxiv Science

bioRxiv2021

Weight loss response following lifestyle intervention associated with baseline gut metagenomic signature in humans

We report a weight-loss response analysis on a small cohort of individuals (N=25) selected from a larger population (N[~]5,000) enrolled in a commercial scientific wellness program, which included healthy lifestyle coaching. Each individual had baseline data on blood metabolomics, blood proteomics, clinical labs, lifestyle questionnaires, and stool metagenomes. A subset of these participants (N=15) lost at least 10% of their body weight within a 6-12 month period and saw significant improvement in metabolic health markers ( weight loss group), while another subset of individuals (N=10) undergoing the same lifestyle intervention showed no change in BMI over the same timeframe ( no weight loss group). Only a single baseline blood analyte, a metabolite linked to fried food consumption, was (negatively) associated with weight loss, but a large number of baseline stool metagenomic features, including complex polysaccharide and protein degradation genes, stress-response genes, respiration-related genes, cell wall synthesis genes, and gut bacterial replication rates, were significantly associated with weight loss after explicitly controlling for baseline BMI. Together, these results provide a set of baseline gut microbiome functional features that are associated with weight loss outcomes.

microbiology↗

bioRxiv2019

Blood metabolome signature predicts gut microbiome α-diversity in health and disease

Defining a healthy gut microbiome has been a challenge in the absence of detailed information on both host health and microbiome composition. Here, we analyzed a multi-omics dataset from hundreds of individuals (discovery n=399, validation n=540) enrolled in a consumer scientific wellness program to identify robust associations between host physiology and gut microbiome structure. We attempted to predict gut microbiome -diversity from nearly 1000 analytes measured from blood, including clinical laboratory tests, proteomics and metabolomics. While a broad panel of 77 standard clinical laboratory tests and a set of 263 proteins from blood could not accurately predict gut microbial -diversity, we found that 45% of the variance in microbial community diversity was explained by a subset of 40 blood metabolites, many of microbial origin. This relationship between the host metabolome and gut microbiome -diversity was very robust, persisting across disease conditions and antibiotics use. Several of these novel metabolic biomarkers of gut microbial diversity were previously associated with host health (e.g. cardiovascular disease risk, diabetes, and kidney function). A subset of 11 metabolites classified participants with potentially problematic low -diversity (ROC AUC=0.88, Precision-Recall AUC=0.76). Relationships between host metabolites and -diversity remained consistent across most of the Body Mass Index (BMI) spectrum, but were modified in extreme obesity (class II/III, but not class I), suggesting a significant metabolic shift. Out-of-sample prediction accuracy of -diversity from the 40 identified blood metabolites in a validation cohort, whose microbiome samples were analyzed by a different vendor, confirmed the robust correspondence between gut microbiome structure and host physiology. Collectively, our results reveal a strong coupling between the human blood metabolome and gut microbial diversity, with implications for human health.

bioinformatics↗