Can we trust biomarkers identified using different non-targeted metabolomics platforms? Multi-platform, inter-laboratory comparative metabolomics profiling of lettuce cultivars via UPLC-QTOF-MS

Introduction Data analysis during UPLC-MS non-targeted metabolomics introduces variation as different manufacturers use specific algorithms for data treatment and this makes untargeted metabolomics an application for the discovery of new biomarkers with low confidence in the reproducibility of the results under the use of different metabolomics platforms. Objectives This study compared the ability of two platforms (Agilent UPLC-ESI-QTOF-MS and Waters UPLC-IMS-QTOF-MS) to identify biomarkers in butterhead and romaine lettuce cultivars. Methods Two case studies by different metabolomics platforms: (1) Waters and Agilent datasets processed by the same data pre-processing software (Progenesis QI), and (2) Datasets processed by different data pre-processing software. Results A higher number of candidate biomarkers shared between sample groups in case 2 (101) than in case 1 (26) was found. Thirteen metabolites were common to both cases. Romaine lettuce was characterised by phenolic compounds including flavonoids, hydroxycinnamate derivatives, and 9-undecenal, while Butterhead showed sesquiterpene lactones and xanthosine. This study demonstrates that high percentages of the most discriminatory entities can be obtained by using the manufacturers' embedded pre-processing software and following the recommended processing data guidelines using commercial software to normalise the data matrix.