Analysis of Metabolites in Complex Biological Samples Using LC/MS and Multivariate Data Analysis : Metabolic Fingerprinting and Detection of Biomarkers

Sammanfattning: To facilitate early diagnosis of diseases and elucidation of the processes involved in their development and progression, various specific compounds or ‘biomarkers’ are often monitored. The first step is to decide which compounds to analyze. An untargeted approach within the field of metabolomics, can provide a unique chemical fingerprint representing the biological state of the studied organism. Fingerprints can be used for classification and thus facilitate the identification of biomarkers and investigation of drug metabolism.In this work a method was developed for the general detection of metabolites using liquid chromatography/mass spectrometry (LC/MS) to obtain data for metabolic fingerprinting. One part of the project focused on generating the data and the other on analyzing the acquired data. First part: Solid phase extraction was applied to rat urine samples and protein precipitation to human plasma samples. Several LC/MS variants were used, i.e. reversed phase LC, hydrophilic interaction LC and ultrahigh pressure LC (UHPLC) coupled to a triple quadrupole MS or time of flight (ToF) MS. Second part: Methods for handling LC/MS data and extracting information, e.g. curve resolution, were evaluated. In addition data fusion methods were investigated. Principal Component Analysis (PCA) and partial least squares (PLS), were applied for pattern recognition. Furthermore, 3-way classification methods, such as parallel factor analysis (PARAFAC) and N-way PLS, were also explored.The developed method was applied to the early detection of phospholipi-dosis in drug development and to search for indicators of successful treatment in breast cancer therapy. Potential biomarkers were suggested, but have not yet been fully evaluated. In addition, new metabolites of an antidepressant drug were discovered and identified using this approach.