基于LASSO回归分析代谢组学指标对儿童重症肺炎支气管肺泡灌洗术预后的预测价值OA
Analysis of predictive value of metabonomics indicators for prognosis of bronchoalveolar lavage in children with severe pneumonia based on LASSO regression
目的 基于 LASSO 回归分析代谢组学指标对儿童重症肺炎支气管肺泡灌洗术(BAL)预后的预测价值,为临床精准评估患儿治疗结局、优化治疗决策提供依据.方法 选取 2023 年 6 月至 2025 年 2 月盐城市妇幼保健院收治的 80 例接受 BAL 治疗的重症肺炎患儿,根据治疗后 72 h 量化标准,将治疗反应良好的 60 例纳入治疗敏感组(预后良好),治疗反应不佳的 20 例纳入治疗不敏感组(预后不佳).收集两组患儿外周血样本,采用超高效液相色谱质谱联用进行非靶向代谢组学分析.先通过单因素分析筛选与预后关联的代谢组学指标,再经 LASSO 回归进一步筛选关键预测指标,以受试者操作特征(ROC)曲线及曲线下面积(AUC)验证代谢组学指标对预后的预测效能.结果 单因素分析结果显示,溶血磷脂酰胆碱(LPC 16:0)、鞘氨醇-1-磷酸(S1P)、脯氨酸、柠檬酸、花生四烯酸、尿苷二磷酸葡萄糖(UDPG)、犬尿氨酸、次黄嘌呤、甘油磷脂酰胆碱、琥珀酸、色氨酸、磷脂酰肌醇、乳酸、鸟苷三磷酸、神经酰胺、葡萄糖-6-磷酸均与儿童重症肺炎 BAL 治疗敏感性相关(P<0.05),溶血磷脂酰乙醇胺、谷氨酸与治疗敏感性无显著关联(P>0.05).LASSO 回归多因素分析显示,LPC 16:0、S1P、脯氨酸、花生四烯酸、犬尿氨酸、柠檬酸、UDPG 为儿童重症肺炎 BAL 预后的影响因素(P<0.05).基于这 7 个关键指标的预测 ROC 分析显示,AUC 为 0.91(95%CI:0.86～0.95),预测效能较高.结论 通过 LASSO 回归筛选出的 LPC 16:0 等 7 种代谢组学指标,对儿童重症肺炎 BAL 预后具有良好的预测价值,可作为临床评估患儿治疗结局、优化治疗决策的重要参考指标.
Objective To analyze the predictive value of metabonomics indicators for prognosis of bronchoalveolar lavage(BAL)in children with severe pneumonia based on LASSO regression,and to provide a basis for clinical accurate evaluation of children's treatment outcomes and and the optimization of treatment decisions.Methods From June 2023 to February 2025,80 children with severe pneumonia who received BAL treatment were selected from Yancheng Maternal and Child Health Hospital.According to the quantitative criteria at 72 hours after treatment,60 children with good treatment response were included in the treatment-sensitive group(good prognosis),and 20 children with poor treatment response were included in the treatment-insensitive group(poor prognosis).Peripheral blood samples were collected from children in both groups,and ultra-high performance liquid chromatography-mass spectrometry was used for untargeted metabonomics analysis.First,univariate analysis was performed to screen metabonomics indicators associated with prognosis,followed by LASSO regression to further screen key predictive indicators.The receiver operator characteristic(ROC)curve and the area under the curve(AUC)were used to verify the prognostic predictive efficacy of the metabonomics indicators.Results The results of the univariate analysis showed that lysophosphatidylcholine(LPC 16:0),sphingosine-1-phosphate(S1P),proline,citric acid,arachidonic acid,uridine diphosphate glucose(UDPG),canine urea,hypoxanthine,glycerophosphocholine,succinic acid,tryptophan,phosphatidylinositol,lactic acid,guanosine triphosphate,ceramides,and glucose-6-phosphate were all associated with treatment response in children with severe pneumonia(P<0.05),while lysophosphatidylethanolamine and glutamic acid showed no significant association with treatment response(P>0.05).Multivariate analysis using LASSO regression revealed that LPC 16:0,S1P,proline,arachidonic acid,canine urea,citric acid,and UDPG were factors influencing the prognosis of BAL in children with severe pneumonia(P<0.05).ROC analysis of the prediction based on these 7 key indicators showed that the AUC was 0.91(95%CI:0.86-0.95),indicating high predictive efficacy.Conclusion The seven metabonomics indicators including LPC 16:0 screened by LASSO regression have good predictive value for the prognosis of BAL in children with severe pneumonia,and can be used as important reference indicators for clinical evaluation of children's treatment outcomes and optimization of treatment decisions.
刘琳玲;祁静;李胜
江苏省盐城市妇幼保健院儿童呼吸科,江苏 盐城 224000江苏省盐城市妇幼保健院儿童呼吸科,江苏 盐城 224000江苏省盐城市妇幼保健院儿童呼吸科,江苏 盐城 224000
医药卫生
儿童重症肺炎代谢组学支气管肺泡灌洗术LASSO回归
Severe pneumonia in childrenMetabonomicsBronchoalveolar lavageLASSO regression
《妇儿健康导刊》 2026 (6)
189-193,5
江苏省儿童早期发展与慢病防治重点实验室开放课题(YLT2024016)江苏省盐城市卫健委医学科研项目(YK2024046).
评论