Use of a pre-analysis osmolality normalisation method to correct for variable urine concentrations and for improved...

Metabolomics analyses of urine have the potential to provide new information on the detection and progression of many disease processes. However, urine samples can vary significantly in total solute concentration and this presents a challenge to achieve high quality metabolomic datasets and the detection of biomarkers of disease or environmental exposures. This study investigated the efficacy of pre- and post-analysis normalisation methods to analyse metabolomic datasets obtained from neat and diluted urine samples from five individuals. Urine samples were extracted by solid phase extraction (SPE) prior to metabolomic analyses using a sensitive nanoflow/nanospray LC-MS technique and the data analysed by principal component analyses (PCA). Post-analysis normalisation of the datasets to either creatinine or osmolality concentration, or to mass spectrum total signal (MSTS), revealed that sample discrimination was driven by the dilution factor of urine rather than the individual providing the sample. Normalisation of urine samples to equal osmolality concentration prior to LC-MS analysis resulted in clustering of the PCA scores plot according to sample source and significant improvements in the number of peaks common to samples of all three dilutions from each individual. In addition, the ability to identify discriminating markers, using orthogonal partial least squared-discriminant analysis (OPLS-DA), was greatly improved when pre-analysis normalisation to osmolality was compared with post-analysis normalisation to osmolality and non-normalised datasets. Further improvements for peak area repeatability were observed in some samples when the pre-analysis normalisation to osmolality was combined with a post-analysis mass spectrum total useful signal (MSTUS) or MSTS normalisation. Future adoption of such normalisation methods may reduce the variability in metabolomics analyses due to differing urine concentrations and improve the discovery of discriminating metabolites associated with sample source.