Predictive Value of Baseline Gut Microbiome Characteristics for the Response in the Next Pollen Season After Sublingual Immunotherapy in Artemisia Pollen–Induced Allergic Rhinitis: A Single-Center Prospective Cohort Study.

Background Artemisia pollen allergic rhinitis is a major health burden, with sublingual immunotherapy showing variable effectiveness. This study explores the potential of gut microbiota as a biomarker to better predict treatment outcomes. Current predictive methods, such as IgE levels, skin prick tests, and symptom scales, often fail to accurately predict treatment outcomes. Objective To evaluate the predictive value of baseline gut microbiome features for sublingual immunotherapy response and develop a practical clinical score for patient stratification. Methods A single-center prospective cohort study enrolled 204 participants. Pretreatment stool samples were analyzed using 16S rRNA V3–V4 sequencing to assess Shannon diversity, the proportion of butyrate-producing bacteria, and the Prevotella/Bacteroides ratio. Three models were developed, with Model A based on clinical variables, Model B incorporating microbiome features, and Model C using L1 regularization for feature selection. Model performance was evaluated through AUC (DeLong), calibration intercept α and slope β (Bootstrap), NRI/IDI, and decision curve analysis, with Model C validated internally. Results The median improvement in CSMS over the peak 6-week pollen season was 32.83% (95% CI 28.87–36.61), with a response rate of 54.41% (95% CI 47.56–61.10). In Model B, microbiome features significantly predicted response, with ORs of 1.59 for butyrate-producing bacteria, 1.43 for the Prevotella/Bacteroides ratio, and 1.33 for Shannon diversity. Model B increased AUC from 0.71 to 0.79 (P = 0.021) and showed improved calibration (α=−0.03; β = 0.98). Model C, with a threshold of p ≥ 0.62, had a sensitivity of 77.48%, specificity of 72.04%, and AUC of 0.78. Conclusions Baseline gut microbiome features enhance the prediction of sublingual immunotherapy outcomes. The interpretable, low-dimensional score offers a practical tool for patient stratification and decision-making, with potential for further validation and clinical application.