Abstract

Background: Diet and the gut microbiota are each associated with hepatic steatosis, but their joint variation and shared explanatory contribution are rarely quantified in older Asian community-dwelling populations. Methods: In 701 non-heavy-drinking Kyotango longevity cohort adults, habitual diet (BDHQ; 31 food groups) and stool 16S rRNA microbiome (47 genera; CLR-transformed) were related to the fatty liver index (FLI) by canonical correlation analysis (CCA), reduced-rank regression (RRR), and bootstrap mediation; FIB-4 was a secondary exploratory outcome. Results: Four dietary patterns and four microbial clusters emerged. CCA revealed multivariate diet-microbiota co-variation (four significant canonical correlations; r = 0.40–0.46; all p < 0.05). Combined RRR (n = 697 with complete FLI data) explained 11.1% of FLI variance in-sample (permutation p = 0.006), although cross-validated R2 was negative, reframing the model as hypothesis-generating rather than predictive. Bootstrap mediation suggested that 12.6% of the diet-on-FLI effect was carried by the microbiota (95% bootstrap CI excluding zero). Of 1457 FDR-corrected food-genus pairs, one was significant (fruits × Eubacterium eligens; r = +0.202, q = 1.0 × 10−4). Conclusions: In this cross-sectional, hypothesis-generating analysis with no individual-level predictive utility, dietary patterns and gut microbial composition co-vary and jointly relate to FLI. The findings describe population-level covariance patterns for future prospective diet-microbiome intervention testing; external validation in independent cohorts is essential.

IPC Classification

Keywords