Automated quantification of metabolites in blood-derived samples by NMR

NMR is widely applied in the field of metabolomics due to the quantitative nature of the technology and the reproducible data generated. However, because of severe spectral crowding, quantifying individual metabolites in body fluids such as serum and plasma remains a challenge. In this study, a method to automatically annotate and quantify a number of small metabolites in human serum and EDTA plasma is introduced. It combines the superior signal-to-noise ratio of the commonly applied CPMG and NOESY1D pulse sequences with the superior resolution of the 2D JRES experiment to construct a model that extracts the metabolite concentrations directly from the 1D spectra without tedious deconvolution. The performance of the method was assessed by comparing the calculated areas of the various glucose peaks with known clinical values, by comparing several peaks of the same metabolite (extracted versus non-extracted), and by comparing areas obtained from various NMR pulse sequences. Additionally, the models were tested on independent datasets. It was found that for many metabolites peaks could be assigned that show a consistent behavior, indicating a precise quantification. The same method should be applicable to other biofluids with a stable composition and pH, such as CSF fluid, cell extracts, and cell media.