Gut microbiota-based prediction of clinical response to sublingual immunotherapy in Artemisia pollen-induced allergic rhinitis: A prospective cohort study

The gut microbiota plays a crucial role in modulating mucosal immunity and allergic responses, yet its predictive value for sublingual immunotherapy (SLIT) outcomes remains underexplored in Artemisia pollen-induced allergic rhinitis (AR). In this single-center prospective cohort study, 204 adults with Artemisia pollen-induced AR underwent baseline stool collection before initiating standardized SLIT. Gut microbiota was analyzed using 16S rRNA sequencing of the V3-V4 region, with prespecified features including Shannon diversity index, composite abundance of butyrate-producing bacteria (Faecalibacterium, Roseburia, Eubacterium rectale group), and Prevotella-to-Bacteroides (P/B) ratio. Clinical response was defined as ≥30% reduction in combined symptom-medication score (CSMS) during the peak pollen season. We developed three prediction models: Model A (clinical variables only), Model B (clinical variables plus microbiota features), and Model C (parsimonious model via L1 regularization). The response rate was 54.41% (111/204). In multivariable analysis, all three microbiota features independently predicted treatment response: butyrate-producing bacteria (OR = 1.59, q = 0.006), P/B ratio (OR = 1.43, q = 0.020), and Shannon diversity (OR = 1.33, q = 0.046). Model B demonstrated superior discrimination compared to Model A (AUC 0.79 vs 0.71, ΔAUC = 0.08, P = 0.021), with improved calibration (intercept α = -0.03, slope β = 0.98) and significant net reclassification improvement (NRI = 0.36, P = 0.002). Decision curve analysis confirmed greater net benefit across clinically relevant threshold probabilities. The parsimonious Model C maintained good performance (optimism-corrected AUC = 0.78) with 77.48% sensitivity and 72.04% specificity. Baseline gut microbiota characteristics, particularly butyrate-producing bacterial abundance, microbial diversity, and Prevotella/Bacteroides community structure, significantly predict SLIT response in Artemisia pollen-induced AR and provide substantial incremental value over conventional clinical parameters. These findings support the integration of gut microbiota assessment into pretreatment stratification algorithms for allergen immunotherapy.