Microbiome as a prediction of immunotherapy response in lung cancer.

Immune checkpoint inhibitors (ICIs) have revolutionized the treatment of lung cancer (LC), offering durable responses in non-small cell lung cancer (NSCLC) and, to a lesser extent, small cell lung cancer (SCLC). Nevertheless, clinical outcomes remain highly heterogeneous, with many patients experiencing primary or acquired resistance and/or immune-related adverse events (irAEs) that impair their quality of life and treatment adherence. The human microbiome, particularly in the gut and oral compartments, has emerged as a critical modulator of systemic antitumor immunity and a promising noninvasive predictive biomarker for ICI efficacy and toxicity. This narrative review synthesizes the current evidence on microbiome composition, diversity, and function in patients with LC receiving ICIs as monotherapy, dual blockade, or in combination regimens, as well as clinically relevant biomarkers associated with treatment response and toxicity. Higher gut microbial alpha diversity and enrichment of beneficial taxa (e.g., Akkermansia muciniphila, Faecalibacterium prausnitzii, and certain Firmicutes) are consistently linked to improved progression-free survival (PFS) and overall survival (OS), mediated by microbial metabolites such as short-chain fatty acids and inosine, which enhance T-cell priming, tumor microenvironment remodeling, and gut-lung axis communication. Microbiome-disruptive exposures, particularly antibiotics and proton pump inhibitors (PPIs), induce dysbiosis and are strongly associated with poorer survival outcomes. Mechanistic insights from preclinical models and clinical cohorts, alongside clinical confounders, underscore the complementary role of the microbiome relative to established markers such as programmed death-ligand 1 (PD-L1) and tumor mutational burden. Prospective standardization of metagenomic profiling and microbiome-modulating interventions represents a key next step in translating these findings into personalized immunotherapy strategies for LC.