Elena Chekmeneva1, Gonςalo Correia2, María Gómez-Romero2, Queenie Chan1, Paul Elliott1, Jeremy K. Nicholson2 Elaine Holmes2
(1) Department of Epidemiology and Biostatistics, School of Public Health, Imperial College London; (2)2Computational and Systems Medicine, Department of Surgery and Cancer, Imperial College London
We present a practical comparison of nESI-HRMS and reversed-phase UPLC-MS methods for multiplexed targeted and global metabolic profiling of the urine specimens from the INTERMAP study. The test set consisted of 132 randomly selected samples which included 22 pairs of blinded replicated and different types of quality control samples. Selected metabolites were quantified using the stable isotope labelled internal standards (IS). Both methods were validated according to the FDA guidelines. The quantification results, sensitivity and dynamic ranges were assessed and compared between two methods. The classification ability was examined on the basis of the multivariate analysis of the global profiles.
Large scale metabolic phenotyping and metabolome-wide analysis are increasingly being applied to large scale studies in epidemiological and clinical settings.1 Sustained analysis of epidemiological studies (typically with thousands of participants) requires tools capable of high-throughput analysis such as direct infusion mass spectrometry (DIMS). In contrast, traditionally used hyphenated techniques such as LC-MS (MS/MS) can be time consuming with extremely high associated cost for the analysis of large sample numbers. Thus, the use of DIMS for large-scale epidemiological studies could be an optimal means of accelerating MS analysis whilst keeping its costs affordable. The main disadvantage of DIMS methods is the inability to distinguish between isomeric species in single stage MS mode, which requires chromatographic separation or additional MS/MS experiments.
Recently, we have developed a high-throughput nanoelectrospray high resolution mass spectrometry (nESI-HRMS) method for multiplexed parallel untargeted and targeted quantitative metabolic analysis of the urine samples.2 Automated nESI-HRMS technology presents advantages such as high sensitivity, high-throughput (less than 3 min per samples in both positive and negative ionization mode), wide dynamic range due to low ion suppression, lack of carry-over and ability to unambiguously discriminate isobaric ions. The urine sample preparation is simple, and the injectable sample volume is of the order of nL. Apart from low specimen consumption, it also ensures a stable instrument performance without any contamination. Moreover, nESI method provides additional information in the form of high resolution accurate mass global profile of the samples. It was demonstrated to be suitable for metabolic profiling of large sample numbers and was applied for the analysis of +10,000 24h urine collection samples from the INTERMAP (INTERnational collaborative study of MAcronutrients, micronutrients and blood Pressure) cross-sectional study on relation between diet and blood pressure among 4680 men and women ages 40-59 years in Japan, PRC, UK, and USA.3
There are few studies comparing LC-MS and DIMS approaches for targeted MS/MS assays in proteomics and drug analysis,4,5 and for serum HRMS metabonomic study of kidney cancer.6 In the present analysis we performed a practical comparison of nESI-HRMS and reversed-phase UPLC-MS7 methods for multiplexed parallel targeted quantitative and global metabolic profiling of the urine specimens from one population of the INTERMAP study. The chromatographic separation was performed on a standard C18 column with a 12 min elution gradient, using water/acetonitrile, both with 0.1% formic acid, as mobile phases. Detection in positive and negative ionization mode was carried out in two independent runs. To the best of our knowledge, this is the first comparative evaluation of these two methods for urinary metabolic profiling. The test set consisted of 132 randomly selected samples which included 22 pairs of blinded replicated (“split”) samples to assess the overall technical error. Different types of quality control (QC) samples were included in the analysis by both methods, namely, the pooled urine samples, injected periodically during the run to monitor the system stability; dilution QC samples for biologically relevant feature selection for exploratory profiling; validation QC samples to monitor accuracy and precision of the quantitative analysis.
Selected metabolites (presenting different biochemical pathways, some xenobiotics and diet related metabolites) were quantified using the stable isotope labelled internal standards (IS) by standard addition calibration in a pooled urine sample in nESI-HRMS to account for any matrix effect. In UPLC-MS analysis, the calibration curves were created in blank solution (water) spiked with IS due to the detector saturation if standard addition method was used. Both methods were validated according to the FDA guidelines. Moreover, in validation and pre-validation UPLC-MS assays it was shown that low and high level metabolites required the analysis at different dilution factors of the original urine sample (1/20 and 1/3) which had an impact on the analysis time. The nESI-HRMS analysis was performed in 1/50 sample dilution.
The quantification results, sensitivity and dynamic ranges for different metabolites were assessed and compared between two methods. Moreover, the classification ability was examined on the basis of the multivariate analysis of the global profiles. It is shown that nESI-HRMS covers a wide range of metabolites (polar metabolites, fatty acids, lipids, acylcarnitines) and the information about all ionized components in a sample can be obtained in the infusion mode, while the analysis by UPLC-MS, although more specific, may require the use of different stationary and mobile phases for the adequate separation of metabolites of very different polarities (e.g HILIC is used for very polar compounds and a modified RP for lipids).
1. E. Holmes, et al, Nature, 2008, 453, 396;
2. E. Chekmeneva, et al., Analytical Methods, 2015, accepted manuscript;
3. J. Stamler, et al., J. Hum. Hypertens, 2003, 17, 591;
4. Y. Xiang, et al., Anal. Chem., 84, 1981;
5. E. R. Wickremsinhe, et al., Rapid Commun Mass Sp, 2005, 19, 47;
6. L. Lin, et al., Analyst, 2010, 135, 2970;
7. E. J. Want, et al., Nat Protoc, 2010, 5, 1005.