目的 分析活动性肺结核（ATB）患者的粪便代谢谱特征，挖掘肺结核新型无创诊断标志物。 方法 以初治ATB患者33例（TB组）和健康人30例（HC组）为研究对象，采集两组新鲜粪便样本，采用超高效液相色谱-质谱联用（UHPLC-MS）非靶向代谢组学技术进行检测。通过偏最小二乘判别分析（PLS-DA）筛选差异代谢物（VIP ≥ 1.0、|log₂ FC| ≥ 0.585且P ≤ 0.05），结合KEGG通路富集与聚类分析确定潜在生物标志物，用ROC曲线评估其诊断效能。 结果 共筛选出515个（ESI-： 312；ESI+： 203）差异代谢物，主要富集于嘌呤代谢、胆汁分泌及烟酸与烟酰胺代谢通路。在ESI-模式下，VIP值排名前20的差异代谢物多数在TB组中呈下调趋势，仅PGE₂、N-｛3-［（3，5-二氟苯基）氧］吡啶-2-基｝-4-戊基苯磺酰胺、N'-2-乙酰吡啶-2-羧酰腙、5-羟基吲哚-2-羧酸上调；而在ESI+模式下，VIP值排名前20的差异代谢物除3-乙酰基-2，5-二甲基呋喃和富马酸在TB组中下调外，其余代谢物均显著上调。代谢通路分析结合功能预测，确定3种潜在生物标志物（PGE₂、富马酸和5-羟基吲哚-2-羧酸），其AUC分别为0.911、0.859和0.824，三者联合诊断的logistic回归模型的AUC为0.957。 结论 活动性肺结核患者差异代谢物主要富集于嘌呤代谢、胆汁分泌及烟酸与烟酰胺代谢通路；PGE₂、富马酸、5-羟基吲哚-2-羧酸的联合诊断模型均对肺结核具有良好的诊断效能。

Objective To analyze the fecal metabolic profile of active tuberculosis（ATB） patients in order to explore novel non-invasive diagnostic biomarkers. Methods A total of 33 treatment-naive ATB patients（TB group） and 30 healthy controls（HC group） were recruited. Fresh fecal specimens from both groups were examined using UHPLC-MS-based untargeted metabolomics. Differential metabolites were selected by partial least squares discriminant analysis（PLS-DA） with the thresholds of variable importance in projection（VIP） ≥ 1.0， |log? FC| ≥ 0.585 （1.5-fold change）， and P ≤ 0.05. Potential biomarkers were recognized through KEGG pathway enrichment and clustering analyses， and their diagnostic effectiveness was assessed via receiver operating characteristic（ROC） curves. Results A total of 515 differential metabolites were identified （312 in ESI-； 203 in ESI+）， which were primarily enriched in purine metabolism， bile secretion， and nicotinate/nicotinamide metabolism. In the ESI- mode， 16 out of the top 20 metabolites were significantly down-regulated in the TB group， whereas PGE?， N-｛3-［（3，5-difluorophenyl）oxy］pyridin-2-yl｝-4-pentylbenzenesulfonamide， N'-2-acetylpyridine-2-carbohydrazone， and 5-hydroxyindole-2-carboxylic acid were up-regulated. In the ESI+ mode， 18 out of the top 20 metabolites were up-regulated， while 3-acetyl-2，5-dimethylfuran and fumaric acid were down-regulated. Metabolic pathway analysis combined with functional prediction identified three potential biomarkers—prostaglandin E₂（PGE₂）， trans-2-Butene-1，4-dicarboxylic Acid， and 5-hydroxyindole-2-carboxylic acid， with areas under the curve（AUC） of 0.911， 0.859， and 0.824， respectively. The logistic regression model integrating these three biomarkers achieved an AUC of 0.957 for combined diagnosis. Conclusions Differential metabolites in patients with active pulmonary tuberculosis were mainly enriched in purine metabolism， bile secretion， and nicotinate and nicotinamide metabolism pathways. Both PGE₂， trans-2-Butene-1，4-dicarboxylic acid， 5-hydroxyindole-2-carboxylic acid， and their combined diagnostic model showed good diagnostic efficacy for pulmonary tuberculosis.