Background: A gut-muscle axis through which the microbiome influences skeletal muscle has been hypothesized. However, sex-specific association between the characteristics of gut microbiota and skeletal muscle mass has not yet been reported. Herein, we performed sex-specific analyses of faecal microbiota composition for the skeletal muscle mass in a population-based cohort.

Methods: We collected faecal samples of 1052 middle-aged participants (621 men and 431 women) who attended health screenings, and we analysed the intestinal microbiota using 16S rRNA gene sequencing. Relative muscle mass was calculated using a bioelectrical impedance analysis and presented as the skeletal muscle mass index [SMI (%) = total appendicular muscle mass (kg)/weight (kg) × 100]. We categorized the subjects into four groups by the quartile of the SMI. Association tests between gut microbiota and SMI were conducted according to the microbial diversity, taxonomic profiling and functional inference in a sex-stratified manner.

Results: The mean age and SMI of the total participants were 44.8 years (standard deviation [SD], 8.2) and 41.4% (SD, 3.9), respectively. After adjustments for possible covariates such as age, body mass index and regular physical activity, the highest quartile (Q4) group of SMI had higher alpha diversity than the lowest quartile (Q1) group in male participants (coefficient = 10.79, P < 0.05, linear regression model), whereas there was no difference in diversity among SMI groups in females. At the species level, Haemophilus parainfluenzae (coefficient = 1.910) and Roseburia faecis (coefficient = 1.536) were more abundant in the highest SMI (Q4) group than in the lowest SMI (Q1) group in males. However, no significant taxon was observed along the SMI groups in females. The gut microbiota of the lowest SMI group (Q1) was enriched with genes involved in biosynthesis of amino acids and energy generation compared with that of the highest SMI group (Q4) in both sexes, although the significance of the inferred pathways was weak (P < 0.05 but the false discovery rate q > 0.05).