Inter-laboratory reproducibility of an untargeted metabolomics GC–MS assay for analysis of human plasma

• Published in: Scientific reports Vol. 10; no. 1; p. 10918
• Main Authors: Lin, Yanping, Caldwell, Gary W., Li, Ying, Lang, Wensheng, Masucci, John
• Format: Journal Article
• Language: English
• Published: London Nature Publishing Group UK 02.07.2020; Nature Publishing Group
• Subjects: 631/1647/320; 692/53; Blood Chemical Analysis - instrumentation; Blood Chemical Analysis - methods; Data processing; Experimental design; Female; Gas Chromatography-Mass Spectrometry - instrumentation; Gas Chromatography-Mass Spectrometry - methods; Humanities and Social Sciences; Humans; Instrumentation; Laboratories; Laboratory Proficiency Testing; Male; Metabolites; Metabolomics; Metabolomics - instrumentation; Metabolomics - methods; multidisciplinary; Plasma - chemistry; Reproducibility; Reproducibility of Results; Sample preparation; Science; Science (multidisciplinary); Software; Standardization
• ISSN: 2045-2322; 2045-2322
• DOI: 10.1038/s41598-020-67939-x

There is a long-standing concern for the lack of reproducibility of the untargeted metabolomic approaches used in pharmaceutical research. Two types of human plasma samples were split into two batches and analyzed in two individual labs for untargeted GC–MS metabolomic profiling. The two labs used the same silylation sample preparation protocols but different instrumentation, data processing software, and database. There were 55 metabolites annotated reproducibly, independent of the labs. The median coefficient variations (CV%) of absolute spectra ion intensities in both labs were less than 30%. However, the comparison of normalized ion intensity among biological groups, were inconsistent across labs. Predicted power based on annotated metabolites was evaluated post various normalization, data transformation and scaling. For the first time our study reveals the numerical details about the variations in metabolomic annotation and relative quantification using plain inter-laboratory GC–MS untargeted metabolomic approaches. Especially we compare several commonly used post-acquisition strategies and found normalization could not strengthen the annotation accuracy or relative quantification precision of untargeted approach, instead it will impact future experimental design. Standardization of untargeted metabolomics protocols, including sample preparation, instrumentation, data processing, etc., is critical for comparison of untargeted data across labs.