Résumé : The role of the humoral immunity in the pathogenesis of alcohol-related liver disease (ALD) has been only marginally explored. This study presents the first integrated analysis of systems serology and machine learning (ML) to elucidate the contribution of humoral immunity to ALD, in a well-defined cohort of patients at different stages of the disease. Our findings revealed a progressive increase in circulating IgG levels directed against multiple gut microbiota antigens, correlating with the severity of ALD. The subclasses of IgG and their binding to Fc receptors (FcR) exhibited an amplified signal in parallel with ALD severity, suggesting enhanced IgG effector functions and potential contribution to ALD inflammation and tissue injury. In vitro assays exploring the capacity of the Fc fragment to induce complement activation, natural killer cell-mediated cytotoxicity, neutrophil or macrophage phagocytosis demonstrated complex profiles that did not strictly align with IgG levels and FcR binding, indicating regulatory mechanisms. Despite this functional heterogeneity, unsupervised ML using principal component analysis (PCA) identified distinct humoral patterns associated with ALD progression, further supporting a role for humoral immunity in the pathogenesis of ALD and as a biomarker of disease severity. Moreover, density-based spatial clustering of applications with noise (DBSCAN) analysis on the PCA biplot of subjects at the most severe ALD stages revealed two clusters of humoral response with similar clinical characteristics but different risk of complications (septic shock and mortality at 90 days), independently of the Model for End-stage Liver Disease (MELD) score. These results underscore the significance of gut microbiota-specific humoral immunity as a key contributor to multiple facets of ALD chronic inflammation, suggesting its potential as a biomarker for disease progression and prognosis, as well as a therapeutic target for improving outcomes in vulnerable patients.