Optimized breath analysis: customized analytical methods and enhanced workflow for broader detection of VOCs

作者：Wisenave Arulvasan, Julia Greenwood, Madeleine L. Ball, Hsuan Chou, Simon Coplowe, Owen Birch, Patrick Gordon, Andreea Ratiu, Elizabeth Lam, Matteo Tardelli, Monika Szkatulska, Shane Swann, Steven Levett, Ella Mead, Frederik‐Jan van Schooten, Agnieszka Smolinska, Billy Boyle, Max Allsworth · 发表于：Metabolomics · 年份：2025 · DOI：10.1007/s11306-024-02218-8 · 被引用次数：9 · 研究领域：Advanced Chemical Sensor Technologies、Analytical Chemistry and Chromatography、Biochemical Analysis and Sensing Techniques

INTRODUCTION: Breath Volatile organic compounds (VOCs) are promising biomarkers for clinical purposes due to their unique properties. Translation of VOC biomarkers into the clinic depends on identification and validation: a challenge requiring collaboration, well-established protocols, and cross-comparison of data. Previously, we developed a breath collection and analysis method, resulting in 148 breath-borne VOCs identified. OBJECTIVES: To develop a complementary analytical method for the detection and identification of additional VOCs from breath. To develop and implement upgrades to the methodology for identifying features determined to be "on-breath" by comparing breath samples against paired background samples applying three metrics: standard deviation, paired t-test, and receiver-operating-characteristic (ROC) curve. METHODS: A thermal desorption (TD)-gas chromatography (GC)-mass spectrometry (MS)-based analytical method utilizing a PEG phase GC column was developed for the detection of biologically relevant VOCs. The multi-step VOC identification methodology was upgraded through several developments: candidate VOC grouping schema, ion abundance correlation based spectral library creation approach, hybrid alkane-FAMES retention indexing, relative retention time matching, along with additional quality checks. In combination, these updates enable highly accurate identification of breath-borne VOCs, both on spectral and retention axes. RESULTS: A total of 621 features were...