基于肠道微生物群预测蒿属花粉诱导的变应性鼻炎对舌下免疫治疗的临床反应：一项前瞻性队列研究

摘要(中文译文)

肠道微生物群在调节粘膜免疫和过敏反应中起着至关重要的作用，但在蒿属花粉诱导的过敏性鼻炎（AR）中，其对舌下免疫治疗（SLIT）结果的预测价值仍未得到充分探索。在这项单中心前瞻性队列研究中，204名蒿属花粉诱导的AR成年人在开始标准化SLIT之前接受了基线粪便采集。使用V3-V4区的16S rRNA测序分析肠道微生物群，具有预先指定的特征，包括Shannon多样性指数、产丁酸细菌的复合丰度（Faecalibacterium、Roseburia、Eubacterium rectale组）和Prevotella与拟杆菌（P/B）的比率。临床反应被定义为花粉高峰期联合症状药物评分（CSMS）降低≥30%。我们开发了三种预测模型：模型A（仅限临床变量）、模型B（临床变量加微生物群特征）和模型C（通过L1正则化的简约模型）。反应率为54.41%（111/204）。在多变量分析中，所有三种微生物群特征都独立预测了治疗反应：丁酸产生菌（OR=1.59，q=0.006）、P/B比（OR=1.43，q=0.020）和香农多样性（OR=1.33，q=0.046）。与模型A相比，模型B显示出更优的区分能力（AUC 0.79 vs 0.71，ΔAUC=0.08，P=0.021），校准得到改善（截距α=-0.03，斜率β=0.98），净重新分类得到显著改善（NRI=0.36，P=0.002）。决策曲线分析证实，跨临床相关阈值概率的净收益更大。简约模型C保持了良好的性能（乐观校正AUC=0.78），敏感性为77.48%，特异性为72.04%。基线肠道微生物群特征，特别是产生丁酸盐的细菌丰度、微生物多样性和普氏菌/拟杆菌群落结构，显著预测了蒿属花粉诱导的AR中的SLIT反应，并提供了比传统临床参数更大的增量价值。这些发现支持将肠道微生物群评估整合到过敏原免疫治疗的预处理分层算法中。

Abstract

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.

如何引用