Untargeted Serum Metabolic Profiling by GC×GC-HRTOF-MS

Whilst many laboratories take appropriate care, there are still cases where the performances of untargeted profiling methods suffer from a lack of design, control and articulation of the various steps involved. This is particularly harmful to modern comprehensive analytical instrumentations that otherwise provide an unprecedented coverage of complex matrices. In this work, we present a global analytical workflow based on comprehensive two-dimensional gas chromatography (GC×GC) coupled to high resolution time-of-flight mass spectrometry (HR-TOF-MS). It was optimized for sample preparation and chromatographic separation, and validated on in-house QC samples and NIST SRM 1950 samples. It also includes a QC procedure, a multi-approaches data (pre)processing workflow and an original bias control procedure. Compounds of interest were identified using mass, retention and biological informations. As a proof of concept, 35 serum samples representing 3 subgroups of Crohn’s disease (with high, low and quiescent endoscopic activity) were analyzed along with 33 healthy controls. This led to the selection of 31 unique candidate biomarkers able to classify Crohn’s disease and healthy samples with OPLS-DA Q2 0.48 and ROC AUC 0.85 (100% sensitivity and 82% specificity in cross validation). 15 of these 33 candidates were reliably annotated (MSI level 2).

尽管诸多实验室已采取规范的管控措施，但仍有不少场景下，非靶向谱分析方法的性能会因相关各环节缺乏系统化设计、严格管控与合理衔接而大打折扣。这一问题对于本可实现复杂基质样本前所未有的覆盖度的现代综合分析仪器而言，负面影响尤为致命。本研究提出一套基于全二维气相色谱（comprehensive two-dimensional gas chromatography, GC×GC）耦合高分辨飞行时间质谱（high resolution time-of-flight mass spectrometry, HR-TOF-MS）的全局分析工作流。该工作流针对样本前处理与色谱分离环节进行了优化，并通过内部质控（quality control, QC）样本与美国国家标准与技术研究院标准参考物质1950（NIST Standard Reference Material 1950, NIST SRM 1950）样本完成验证。该工作流同时涵盖质控流程、多策略数据（预）处理工作流以及原创性偏倚管控流程。本研究基于质谱特征、保留行为与生物信息对目标分析物进行鉴定。作为概念验证，本研究共分析了35份克罗恩病（Crohn’s disease, CD）血清样本，涵盖内镜活动度分为高、低、静止期的3个亚组，同时纳入33份健康对照血清样本。最终筛选出31个独特的候选生物标志物，其经正交偏最小二乘判别分析（orthogonal partial least squares discriminant analysis, OPLS-DA）得到Q²值为0.48，受试者工作特征曲线下面积（receiver operating characteristic curve, ROC AUC）达0.85，交叉验证灵敏度达100%、特异性达82%，可有效区分克罗恩病与健康样本。其中33个候选标志物中有15个可实现可靠注释（质谱注释等级2, MSI level 2）。