NMR of biofluids and pattern recognition: assessing the impact of NMR parameters on the principal component analysis of urine from rat and mouse.

The ability to interpret metabolic responses to toxic insult as expressed in altered urine composition and measured by NMR spectroscopy is dependent upon a database of proton NMR spectra of urine collected from both control and treated animals. Pattern recognition techniques, such as principal component analysis (PCA), can be used to establish whether the spectral data cluster according to a dose response. However, PCA will be sensitive to other variables that might exist in the data, such as those arising from the NMR instrument itself. Thus, studies were conducted to determine the impact that NMR-related variables might impart on the data, with a view towards understanding and minimizing variables that could interfere with the interpretation of a biological effect. This study has focused on solvent suppression methods, as well as instrument-to-instrument variability, including field strength. The magnitude of the NMR-induced variability was assessed in the presence of an established response to the nephrotoxin bromoethanamine. Changes caused by the model toxin were larger and easily distinguished from those caused by using different solvent suppression methods and field strengths.