That glass of wine? Your skin microbes remember

A large-scale integrated analysis of public and local cohort data has revealed that alcohol intake is strongly associated with the composition of the human skin microbiome, and that machine learning can accurately predict drinking status based on microbial profiles.

The human skin harbors a diverse bacterial community, but how lifestyle factors shape this ecosystem remains poorly understood. To address this gap, researchers compiled 16S rRNA sequencing data from multiple population-based studies published between 2005 and 2024, encompassing skin microbiomes from 3,306 individuals across different ages, geographic locations, and skin sites. After standardizing and quality‑controlling the data, the team characterized microbial diversity, co‑occurrence networks, and the influence of 29 covariates.

Among all factors examined, geographic location and age showed the strongest associations with skin microbiome variation. Notably, alcohol intake frequency—ranging from rare to daily consumption—emerged as the third most influential covariate, with distinct microbial patterns observed across drinking groups. The association was further validated in an independent local cohort of 168 participants, where age and alcohol intake again ranked as the top factors driving microbiome differences, consistent with the public data.

Using co‑occurrence‑based modular analysis, the researchers identified five major bacterial modules. Alcohol intake correlated positively with genera such as Streptococcus and Staphylococcus, and negatively with Lawsonella and Micrococcus. Linear discriminant analysis effect size (LEfSe) confirmed Streptococcus as a marker of high alcohol intake and Lawsonella as a marker of low intake.

To test the predictive potential, the team built a random forest model using species‑level microbiome data from the public datasets and validated it on the local cohort. The model achieved a high accuracy in distinguishing alcohol drinking status. Top predictive features included Dermacoccus nishinomiyaensis, Lactococcus lactis, and Brevibacterium aurantiacum, aligning with the LEfSe results.

These findings suggest that the skin microbiome reflects an individual’s alcohol consumption patterns and could serve as a biomarker for lifestyle assessment. While limited by cohort representativeness and incomplete species‑level coverage, this work offers a parallel comparison of factors influencing the skin microbiome and suggests that the skin microbial community not only reflects cutaneous health but also serves as a mirror of systemic physiological states shaped by lifestyle habits. Future studies with broader populations and deeper sequencing are warranted to further elucidate these relationships.