Peanut protein allergy is a life-threatening condition affecting 2% of the population in industrialized nations. As a leading cause of food-induced anaphylaxis, it was not treatable until the FDA approval of peanut oral immunotherapy (POIT) in 2020. While POIT has demonstrated efficacy in desensitizing patients to allergenic peanut proteins, remission—defined as the prolonged absence of clinical reactivity after treatment cessation—is observed in a smaller subset of approximately 20–30% of treated patients. Since the gut microbiome is linked to food allergy, we sought to identify fecal microbial predictors of POIT efficacy and develop mechanistic insights into treatment response.

We collected stool samples before, during, and at the end of treatment from children (n = 91)

undergoing POIT in the first double-blind, placebo-controlled, multicenter POIT clinical trial, and

performed 16S rRNA sequencing (n = 327), shotgun metagenomics (n = 184), and untargeted

metabolomics (n = 174). Our integrative multi-omics and machine learning (logistic regression)

analyses identified five microbial-derived secondary bile acid metabolites that were enriched at

baseline and predicted treatment efficacy (AUC = 0.71). Failure to induce disease remission was associated with a distinct fecal microbiome characterized by enhanced bile acid deconjugation, altered amino acid metabolism, and increased in vitro peanut peptide utilization. Thus, microbiome mechanisms of POIT failure may include depletion of immunomodulatory secondary bile and amino acids, as well as antigenic peanut peptides necessary to promote desensitization and remission.