Physical activity, gut microbiota and steatotic liver disease: a five-year cohort adds a metabolite to the picture

A prospective cohort published in Annals of Hepatology reports that moderate-to-vigorous physical activity (MVPA) is inversely associated with incident steatotic liver disease (SLD), and identifies gut microbial and serum metabolite changes that track with it.

Design. 2,942 free-living adults followed from 2020 to 2025, with 565 incident SLD cases. SLD was assessed by transient elastography rather than by self-report. Physical activity in the full cohort came from a validated questionnaire. A subset of 754 participants was characterised more intensively: faecal samples collected twice one year apart (16S rRNA sequencing, with species-level profiles from shotgun metagenomics), targeted metabolomics on baseline serum, and accelerometer-measured activity.

Findings. Highest versus lowest tertile of MVPA was associated with lower SLD risk (HR 0.78, 95% CI 0.63–0.98). Sedentary hours ran the other way (HR 1.26, 95% CI 1.00–1.58). Twelve genera showed longitudinal change in relation to MVPA (pFDR < 0.05). Proteus was positively associated with SLD risk (OR 1.32, 95% CI 1.01–1.72), supported at species level for Proteus cibarius. Serum L-valine was inversely correlated with MVPA (β = −0.23, pFDR = 0.031) and positively associated with SLD risk (OR 1.78, 95% CI 1.22–2.63); L-valine also correlated with the MVPA-associated genus Synergistes.

What it establishes, and what it does not. This is observational. The authors describe associations between activity, microbiota-related metabolites and a more favourable SLD profile; they do not demonstrate that modifying the microbiota alters SLD risk, and no intervention was administered. Activity in the main cohort was self-reported, a known source of overestimation, though the microbiome and metabolomics analyses rest on the accelerometer-measured subset. The sedentary-behaviour confidence interval reaches 1.00.

Relevance for the category. Two points are worth noting for anyone working on metabolic-health positioning. First, the signal here is a risk-associated genus, not a beneficial strain — the study offers no target for a probiotic formulation and no head-to-head strain data. Second, branched-chain amino acids appear on the risk side of this analysis, which sits awkwardly beside BCAA supplementation marketed to the same middle-aged demographic. Neither observation supports a claim in either direction; both are reasons to read the intervention literature carefully before building messaging on this paper.