Multi-omics Analyses of Facial Skin in Acne Identify Distinct Microbial and Metabolic Features at Lesional and Non-lesional Sites

Abstract

The microbial and biochemical landscape of clinically normal-appearing skin in individuals with acne remains uncharacterized. Here, we performed longitudinal multi-omics profiling of facial skin from 10 individuals with moderate acne and 10 healthy controls, integrating 16S rRNA gene sequencing, shotgun metagenomics, and untargeted metabolomics across lesional and non-lesional sites. Compositional tensor factorization revealed that non-lesional acne skin occupies a distinct intermediate state between healthy and lesional skin in both the microbiome and the metabolome. Machine learning models distinguished healthy from non-lesional acne skin with 70% accuracy, demonstrating that molecular dysbiosis occurs in areas of the skin without visible lesions. Non-lesional sites exhibited reduced microbial diversity, strain-level shifts in Corynebacterium and Lawsonella correlating with disease severity, and metabolic alterations, including elevated lipids and perturbed amino acid and dipeptide profiles. Microbe-metabolite co-occurrence network analyses revealed that healthy skin is enriched for protective metabolites such as urocanic acid, while acne-associated skin shows distinct co-occurrence patterns. These findings establish that acne represents a field effect disorder, with molecular alterations extending beyond visible lesions to encompass the entire facial skin ecosystem. This molecular signature of pre-lesional skin provides potential biomarkers for early intervention and suggests that effective acne treatment may require holistic approaches targeting the broader skin environment rather than individual lesions alone.