Int Microbiol. 2026 May 30. doi: 10.1007/s10123-026-00845-w. Online ahead of print.
ABSTRACT
Although hypertriglyceridemia (HTG) is a significant contributor to lipid-associated pathologies such as atherosclerotic cardiovascular disease, its regulation by host‒microbiome interactions remain insufficiently characterized. While the gut microbiota (GM) is known to influence cholesterol metabolism, its specific role in systemic triglyceride (TG) homeostasis, particularly in non-Western populations, is poorly defined. This study aimed to identify preliminary robust GM signatures associated with HTG and to assess their translational potential using integrated multiomics and explainable machine learning approaches. In a cross-sectional investigation of 50 well-phenotyped adults from Northwest China, we combined 16S rRNA sequencing, shotgun metagenomics, and ensemble machine learning (LightGBM/XGBoost) to elucidate the associations between the GM and TGs. Microbial features were rigorously linked to serum lipid profiles through dual-algorithm validation and SHAP interpretability analysis, while functional potential was assessed via KEGG pathway mapping. Subjects with HTG exhibited a distinct gut microbial configuration, marked by consistent enrichment of Faecalibacterium and Bacteroides coprocola (positively correlated with serum TG levels) and depletion of Bifidobacterium pseudocatenulatum and Lactobacillus salivarius (inversely correlated). Machine learning converged on five exploratory consensus biomarker taxa, three of which were independently confirmed by LEfSe analysis (Faecalibacterium). Functional profiling further revealed the upregulation of microbial starch and sucrose metabolism pathways in the HTG cohort. Our findings establish a preliminary gut microbial signature for HTG patients and suggest context‑dependent associations of butyrate-producing taxa such as Faecalibacterium. By integrating multiomics with explainable artificial intelligence, this work addresses key challenges in reproducibility and mechanistic inference in microbiome research. These results pave the way for novel microbiota-targeted therapeutic strategies, including precision probiotics and dietary interventions, to modulate lipid metabolism, pending further validation in expanded cohorts and functional studies.
PMID:42215825 | DOI:10.1007/s10123-026-00845-w