Lead: Mika Ala-Korpela
Key collaborators: John C. Chambers and Jaspal Kooner (Imperial College London, UK), Marjo-Riitta Järvelin (University of Oulu, Finland), Michael Hill (University of Oxford, UK) and George Davey Smith (University of Bristol, UK)
Metabolomics provides a snapshot of an individual’s physiological state, influenced by genetic and lifestyle factors. Urine is produced from blood by the kidneys and contains both endogenous and exogenous compounds. Among the biofluids commonly used in epidemiology, urine has several advantages: it is abundant, sterile, and easy to collect. Urine reflects the function of kidneys, including multiple metabolites from several key biochemical pathways in relation to (patho)physiology and cardiometabolic conditions, gut microbial metabolic activities and short-term food consumption. Urine samples therefore contain abundant and underutilised information for epidemiology and for potential translational applications.
NMR spectroscopy provides a comprehensive quantitative approach for urine analysis and has the potential to offer fully automated high-throughput experimentation in a cost-effective manner, which would be essential for large-scale systems epidemiology (1–3). NMR spectroscopy is highly reproducible and requires only minimal sample preparation; NMR may well currently be the most comprehensive and certainly the most quantitative approach for urine characterisation. However, the signal assignments and quantifications from urine spectra are complicated by signal overlap, as well as considerable variations in signal positions between spectra due to differences in the chemical properties of the samples, such as pH, ionic strength and concentration of multivalent cations. Some software applications exist which have been used in the analyses of urine NMR data, but currently none of them provide comprehensive automated quantification of the metabolic information.
We have recently introduced a detailed experimental set-up, including all the key attributes of sample preparation and NMR experimentation, for quantitative high-throughput urinary analyses (4). We also initially demonstrated how fully automated quantitative analyses perform in the case of urine NMR spectra and proposed an open access quantitative pipeline of urine NMR metabolomics to facilitate large-scale studies. We presented extensive analytical data on intra-assay, intra-individual and inter-individual variation in urinary metabolites. In addition, we detailed the characteristics of quantitative urine metabolite data in epidemiology and presented novel analyses regarding how the urine metabolites associate with circulating metabolites and lipids. Confirmative genome-wide analyses were also presented. All data domains in this recent work (4) substantiate the potential usefulness of quantitative molecular data on urine samples in systems epidemiology.
Our quantitative analytical experimentation indicates high robustness and accuracy of the urine NMR spectroscopy methodology per se. The extensive epidemiological data illustrate clear inherent differences in the intra-fluid metabolic associations based on physiological and metabolic functions: the urine metabolites are in general only weakly interrelated, in contradistinction to highly correlated metabolic pathways represented by the quantitative serum data. The metabolic associations between serum and urine are weak, suggesting combining serum and urine metabolomics would increase the amount of independent metabolic information. While the intra-individual variation in urine metabolites is high, the even higher population-based inter-individual variation does provide a sound base for epidemiological and genetic applications. However, appropriate large-scale studies and replication data are crucial to enable statistically robust findings of biological relevance. The known genome-wide associations detected with a very small number of individuals are reassuring for both the analytical process of the presented urine NMR metabolomics set-up and the intriguing potential of quantitative urine metabolite data in systems epidemiology (4).
We are now in the process of measuring tens of thousands of urine samples for multiple large-scale epidemiological cohorts. Once the automated urine NMR spectral analyses will be available for some 40 urinary metabolites we will start multiple epidemiological and genetic analyses with these extensive novel data.
We anticipate this quantitative methodology to eventually offer a multitude of unique opportunities to study the role of urine metabolites, for example, in cardiometabolic health and diseases and as potential markers of kidney function. To the best of our knowledge, this project is novel both in the open access aspects and in the integrated large-scale systems epidemiology perspective that are likely to result in important epidemiological findings with high translational potential.
The automated quantification of urinary creatinine and glucose from the NMR spectra.
On the left: Building and assessment of the final automated regression models for the absolute signal areas for creatinine and glucose in the NMR spectra (n=999).
On the right: The distribution of absolute urinary concentration (in µm/mM creatinine) in 4548 urine samples in a population cohort. The absolute signal areas for the urinary creatinine and glucose used to calculate the distribution are based on fully automated NMR spectral analyses using the final models illustrated on the left.
- High-throughput serum NMR metabonomics for cost-effective holistic studies on systemic metabolism
Analyst 2009 Sep;134(9):1781–5
- Quantitative serum nuclear magnetic resonance metabolomics in cardiovascular epidemiology and genetics
Circ Cardiovasc Genet 2015 Feb;8(1):192–206
- Quantitative serum nuclear magnetic resonance metabolomics in large-scale epidemiology: a primer on -omic technologies
Am J Epidemiol 2017 Nov 1;186(9):1084–1096
- Proof of concept for quantitative urine NMR metabolomics pipeline for large-scale epidemiology and genetics