【作者】 李吉；

【导师】 刘心昱；

【作者基本信息】 中国医科大学 ， 公共卫生（专业学位）， 2025， 硕士

【摘要】 目的:克莱恩-莱文综合征(Kleine-Levin Syndrome,KLS)是以过度睡眠和认知障碍为特征的罕见神经系统疾病,临床诊断依赖症状学标准,缺乏特异性生物标志物。本研究旨在通过液相色谱-质谱(Liquid Chromatography-Mass Spectrometry,LC-MS)联用技术,系统分析KLS与发作性睡病(Narcolepsy,NC)患者血清及脑脊液代谢谱特征,筛选可用于临床鉴别诊断的生物标志物组,并揭示KLS不同病程时期(发作期与清醒期)的代谢机制,为疾病精准诊疗提供科学依据。研究方法:本研究纳入2004～2021年北京大学人民医院确诊的KLS患者(n=74)、NC患者(n=85)及健康对照(n=64),采集血清及脑脊液样本。采用基于LC-MS的非靶向代谢组学方法获取KLS患者的血清和脑脊液代谢谱,创新性使用基于支持向量机和迭代特征删除相结合的代谢特征筛选方法(Support Vector Machine-Recursive Feature Elimination,SVM-RFE),以确定与KLS和NC两种疾病相关的代谢特征,进而筛选潜在的疾病诊断标志物。代谢通路富集分析基于京都基因与基因组百科全书(Kyoto Encyclopedia of Genes and Genomes,KEGG),通过受试者工作特征曲线(Receiver Operating Characteristic curve,ROC)评估组合标志物诊断效能,并通过Bootstrap重抽样法(1000次)对标志物进行内部验证。针对19例同时具备血清及脑脊液样本的KLS患者,进一步分析KLS病程阶段中血清和脑脊液中代谢物的动态变化及代谢特征。结果:基于液相色谱-质谱联用技术,非靶向血清代谢组学共定性到559种代谢物,非靶向血清脂质组学共定性到334种脂质,数据质量稳定可靠。在血清中确定了由肉碱C10:3(Carnitine C10:3)、3-羟基丁酸(3-Hydroxybutyrate)、癸酸(FFA10:0)、甘氨鹅去氧胆酸(Glycochenodeoxycholic Acid)、苯丙酰苯丙氨酸(Phe-Phe)和鞘磷脂(SM(d18:1＿24:1))组成的KLS诊断标志物组,其AUC值为0.950,灵敏度和特异性分别为0.854和0.923;以及由3-羟基丁酸(3-Hydroxybutyrate)、赖氨酸(Lysine)、溶血磷脂酰胆碱20:5(LPC(20:5))、磷脂酰胆碱16:1＿18:3(PC(16:1＿18:3))、醚连接磷脂酰乙醇胺34:2(PE O-34:2)和甘油三酯18:1＿18:1＿20:4(TG(18:1＿18:1＿20:4))组成的NC诊断标志物组,其AUC值为0.887,灵敏度和特异性分别为0.807和0.875。基于非靶向代谢组学在脑脊液共中定性到168种代谢物,数据质量稳定可靠。针对19例同时具备血清及脑脊液样本的KLS不同病程阶段患者,进行血清和脑脊液关联分析,结果显示KLS和NC患者中可能存在不同程度的主动转运系统功能障碍。结论:通过非靶向代谢组学和脂质组学分析,我们在血清中筛选出由6种代谢物组成的KLS和NC诊断标志物各一组;首次系统揭示了KLS发作期与清醒期脑脊液代谢特征差异,筛选出具有分期诊断价值的差异代谢物;此外,通过整合发作期与清醒期的血清及脑脊液代谢组学数据发现,KLS病程中存在显著的全身-中枢代谢协同紊乱,其机制可能涉及主动转运系统障碍、血脑屏障破坏及中枢神经炎症反应。该研究为KLS的精准诊断和病理机制解析提供了新视角。更多还原

【Abstract】 Objective:Kleine-Levin Syndrome(KLS)is a rare neurological disorder characterized by excessive sleepiness and cognitive impairments.Clinical diagnosis only relies on symptomatic criteria due to the lack of specific biomarkers.This study aims to systematically analyze the metabolic profiles of serum and cerebrospinal fluid(CSF)of KLS and Narcolepsy(NC)patients using liquid chromatography-mass spectrometry(LC-MS)technology.The goal is to screen biomarker panels that can be used for clinical differential diagnosis and to reveal the metabolic mechanisms of KLS during different disease stages(hypersomnia episode and interepisode period),providing a scientific basis for precise diagnosis and treatment of the disease.Methods:In this study,serum and CSF samples were collected from KLS patients(n=74),NC patients(n=85)and healthy controls(n=64)diagnosed at Peking University People’s Hospital from 2004～2021.The serum and CSF metabolic profiles of KLS patients were analyzed using an LC-MS-based untargeted metabolomics approach.Innovatively,a metabolic feature screening method combining Support Vector Machine and Recursive Feature Elimination(SVM-RFE)was applied to identify metabolic features common to both KLS and NC,followed by the screening of potential disease diagnostic markers.The metabolic pathway enrichment analysis was based on the Kyoto Encyclopedia of Genes and Genomes(KEGG).The diagnostic efficacy of the combined markers was assessed by Receiver Operating Characteristic curve(ROC),followed by internal validation via the Bootstrap resampling method(1000 iterations).The dynamic changes and metabolic profiles of metabolites in serum and CSF were further analyzed in19 KLS patients with both serum and CSF samples during the disease stage of KLS.Results:Based on LC-MS,a total of 559 metabolites were characterized by non-targeted serum metabolomics and 334 lipids by non-targeted serum lipidomics with stable and reliable data quality.A diagnostic biomarker panel for KLS was established based on serum differential metabolites,consisting of carnitine C10:3,3-hydroxybutyrate,decanoic acid(FFA 10:0),glycochenodeoxycholic acid(GCDCA),phenylalanylphenylalanine(Phe-Phe),and SM(d18:1＿24:1),with an AUC value of 0.950,and a sensitivity and specificity of0.854 and 0.923,respectively.Additionally,a diagnostic biomarker panel for NC was identified,comprising 3-hydroxybutyrate,lysine,LPC(20:5),PC(16:1＿18:3),PE O-34:2,and TG(18:1＿18:1＿20:4),with an AUC value of 0.887,and a sensitivity and specificity of0.807 and 0.875,respectively.What’s more,168 metabolites were identified in non-targeted metabolomics analysis of CSF,with stable and reliable data quality.For 19patients with KLS at different stages of the disease,who had both serum and CSF samples,an association analysis was conducted on the serum and CSF.The results showed varying degrees of blood-brain barrier disruption in both KLS and NC patients.Conclusions:Through untargeted metabolomics and lipidomics analysis,we identified a set of 6 metabolites serving as diagnostic biomarkers for KLS and NC in serum,respectively.For the first time,we systematically compared the metabolic profiles in CSF of KLS patients during hypersomnia episodes and interepisode periods,identifying differential metabolites that distinguish these phases.Furthermore,integrating serum and CSF metabolomics data from both phases revealed a significant systemic-to-central metabolic dysregulation throughout KLS progression.This dysregulation may involve impairment of the active transport system,blood-brain barrier disruption,and central neuroinflammatory responses.This study provides novel insights for precision diagnosis and elucidating KLS pathological mechanisms.更多还原