Although many hydrophilic interaction liquid chromatography-high resolution mass spectrometry (HILIC-HRMS) methods have been developed and applied for untargeted metabolite profiling in clinical metabolomics, according to the literature, the suitability of these HILIC-HRMS methods has not been fully evaluated with respect to their performance when they are subjected to statistical analysis. In this study, using a series of human urine samples we investigated the effect of technical variations on multivariate and univariate analysis of the data collected using a previously developed HILIC-HRMS method for untargeted urinary metabolite profiling in clinical metabolomics. The technical variation introduced by sample preparation was more significant than that produced by the HILIC-HRMS method. By using an orthogonal partial least squares (OPLS) model, subtle fold-changes were accurately measured in the urine samples spiked with (13)C and (15)N isotope labelled amino acids at different concentrations. The robustness of this HILIC method was also evaluated by analysing the obtained data from a single urine sample following manipulation of several primary LC parameters. High reproducibility in the chromatographic performance of three ZIC-pHILIC columns with different batch numbers indicated the reliability of the polymer based zwitterionic stationary phase allowing column replacement without compromising the performance of the method.
Keywords: Clinical metabolomics; Isotope labelled amino acids; Technical variation; Untargeted metabolite profiling; hydrophilic interaction liquid chromatography-high resolution mass spectrometry (HILIC-HRMS).
Copyright © 2016 Elsevier B.V. All rights reserved.