Comparing capillary electrophoresis-mass spectrometry fingerprints of urine samples obtained after intake of coffee, tea, or water.

Metabolomic fingerprinting is a growing strategy for characterizing complex biological samples without detailed prior knowledge about the metabolic system. A two-way analysis system with liquid separation and mass spectrometric detection provides detail-rich data suitable for such fingerprints. As a model study, human urine samples, obtained after intake of coffee, tea, or water, were analyzed with capillary electrophoresis electrospray ionization time-of-flight mass spectrometry (CE-ESI-TOF-MS). In-house-developed software (in Matlab) was utilized to manage and explore the large amount of data acquired (230 CE-MS runs, each with 50-100 million nonzero data points). After baseline and noise reduction, followed by suitable binning in time and m/z, the data sets comprised 9 and 14 million data points in negative and positive ESI mode, respectively. Finally, a signal threshold was applied, further reducing the number to about 100 000 data points per data set. A set of interactive exploratory tools, utilizing principal component analysis (PCA) and analysis of variance (ANOVA) results based on a general linear model, facilitated visual interpretation with score plots (for group assessment) and differential fingerprints (for "hot spot" detection). In the model study highly significant differences due to beverage intake were obtained among the 10 first principal components (p < 10(-6) for two of the components in both ESI modes). Especially, the contrasts between "coffee" and "tea or water" indicated several "hot spots" with highly elevated intensities (e.g., for uncharged masses 93, 94, 109, 119, 123, 132, 148, 169, 178, 187, 190, and 193) suitable for further analysis, for example, with tandem MS.