Still the less well known relation of the genomic and proteomic disciplines, metabolomics — the study of the metabolome, which consists of the small molecule metabolites found in biological fluids such as blood, saliva and urine — may yet hold some interesting insights into health and illness. As our recent articles on the epigenome (
PJ 2013;290:23) and microbiome (
PJ 2013;290:247) suggest, the various -omics disciplines are starting to make some real waves across lots of different sectors of science and medicine.
Genomics, the study of the genome, and proteomics, the study of proteins produced by an organism, hold the lion’s share of the funding and research conducted so far under the umbrella of bioinformatics. In this article we turn our attention to a slightly less well known branch of the -omics: metabolomics, based on the analysis of low molecular weight metabolites, their changes over time and their potential relevance to health and disease. The emphasis in this particular discipline is on the analysis and the organisation of collected data to yield homogenous, meaningful results.
The sweet taste of urine (to ants)
Although the term “metabolomics” (or “metabonomics”)1 is a relatively new addition to the biochemistry texts, the notion of analysing a biological sample for functional information about the health or ill health of a person has been around for a while. Diabetes, for example, has a history dating back to medicine in the ancient world, as observations on the sugar content of urine being a measure for the appearance of the disease were noted. “Sweet urine disease” was traditionally diagnosed by the attraction of ants to the sugar content of a person’s urine, equivalent to the modern-day urine glucose dipstick test.
Fast forward a few thousand years and things have moved forward. Ants generally have less of a role to play in modern medicine, although we retain the knowledge that biological samples might hold some important functional clues pertinent to the diagnosis and management of lots of different conditions.
When it comes to metabolomics, technology is key. Biological samples, such as blood, urine, saliva and cerebrospinal fluid, can be complicated media to work with, containing literally thousands of compounds, encompassing both primary and secondary metabolites reflective of both endogenous and exogenous sourcing. Separating out such components is therefore an important first stage to analysis, made all the more easier following the introduction of chromatographic and related methods of separation.
High-performance liquid chromatography and the newer ultra-performance liquid chromatography remain the industry standard processes complemented by the use of gas chromatography (GC) and capillary electrophoresis methods. Based on both sample medium and separation as a consequence of differing characteristics of the analyte(s) under examination (polarity, charge), each separative method offers its own set of advantages and disadvantages.
Detection is the next stage. Only a few decades ago, detecting compounds by HPLC was predominantly carried out by ultraviolet or electrochemical detectors. Although such techniques were generally suitable for the analysis of known metabolites in purified or uncomplicated media, such detection methods, however, did little to aid in identifying unknowns or providing structural information of metabolites of interest. By contrast, modern-day detection offers significantly more sophisticated information, as nuclear magnetic resonance spectroscopy and mass spectrometry take centre-stage.
The final stage of the metabolomics process is data analysis or data mining and how, behind every metabolomist, there is a statistician and a good software package. Depending on the type of metabolomic analysis undertaken, various statistical methods are available, most based around the concept of categorisation and grouping (remember all those Venn diagrams at school) as terms like “principal components analysis” and “partial least squares regression” are the norm.
Systems biology, based on metabolite-protein network analysis, for example, also represents an important tool to metabolomics as individual or collective metabolites are modelled back to their source genes with the hope of a more integrated inspection of pathways of interest. The culminations of all these processes, which are often used in tandem with each other, frequently generate a significant amount of information.
Given the potential of metabolomics, a seemingly endless supply of research has been generated on all manner of health conditions. Cancer biology represents a core area of investigation. Various themes have emerged, including applying metabolomics to cancer progress and staging, tumour characterisation and predicting drug response and toxicity.2
When it comes to the potential applications of metabolomics to other areas of science and medicine, none is more tantalising than the possibility of a relatively simple diagnostic test for conditions currently lacking such a luxury. In particular the multitudes of psychiatric and behaviourally presented conditions which are seemingly increasing, and in the most part are idiopathic, are a primary target for the metabolomics field.
Schizophrenia is one of a number of psychiatrically defined conditions that, in both the short and long term, seriously affect the quality of life of patients and those around them. Currently defined solely on the basis of observed symptoms, diagnosis is a time- and resource-intensive operation. The lack of consensus on the underlying genetic and biological causes has hindered the development of a consistent biological test for the condition.
With the application of metabolomics to large research sets, there are, however, some promising data emerging pertinent to a diagnostic test. Oresic and colleagues3 reported initial results delineating schizophrenia from other psychoses and asymptomatic controls based on serum profiles indicating issues with glucoregulatory processes and amino acid chemistry to be central. Yang and colleagues4 using GC-ToF (time of flight mass spectrometry) also presented results based on the identification of patients with schizophrenia over controls.
This trial involved not just the identification of the biomarkers (via the training patient set) but also the subsequent independent testing and confirmation of the identified biomarkers. Based on the identification of five serum and one urinary marker, the authors report perfect separation of the diagnosed and asymptomatic groups. Further replication is awaited. Similar applications have been reported in other conditions, such as depression,5 and autism spectrum disorders,6 realising the interfering issues of heterogeneity and comorbidity are likely to represent important confounding issues.
Wang-Sattler and colleagues7 reported results for various novel prodrome biomarkers related to prediabetes (impaired glucose tolerance) in their cohort. Altered levels of three metabolites (glycine, lysophosphatidylcholine 18:2 and acetylcarnitine) were identified in the prediabetic group. Work also continues in other important areas such as predicting Alzheimer’s disease progress8 and also in Huntingdon disease.9
Predicting potential responder characteristics to pharmacotherapies and the influence of medicines on intermediary metabolism represent other important areas of investigation for metabolomics. Kaddurah-Doauk and colleagues10 reported initial results based on responder and non-responder “metabotype” characteristics to sertraline for depression. This same group also reported that differing responses to simvastatin treatment may be governed by differences in the gut microbiome, which may also be detectable via metabolomic analyses.11
The analysis of biofluids for functional measures of disease state and potential for disease is of great medical interest. Metabolomics incorporating facets of biology, chemistry, physics and mathematics is a growth area in this brave new world, as technological advances drive ever more sophisticated analyses at ever more reduced costs.
1 Nicholson JK, Lindon CJ. Systems biology: metabonomics. Nature 2008;455:1054–6.
2 Claudino WM, Goncalve PH, di Leo A et al. Metabolomics in cancer: a bench-to-bedside intersection. Critical Reviews in Oncology/Hematology 2012;84:1–7.
3 OreÅ¡ic M, Tang J, Seppanen-Laakso T et al. Metabolome in schizophrenia and other psychotic disorders: a general population-based study. Genome Medicine 2011;3:19.
4 Yang J, Chen T, Sun L et al. Potential metabolite markers of schizophrenia. Molecular Psychiatry 2013;18:67–8.
5 Zheng P, Gao HC, Li Q et al. Plasma metabonomics as a novel diagnostic approach for major depressive disorder. Journal of Proteome Research 2012;11:1741–8.
6 Yap IKS, Angley M, Veselkov KA et al. Urinary metabolic phenotyping differentiates children with autism from their unaffected siblings and age-matched controls. Journal of Proteome Research 2010;9:2996–3004.
7 Wang-Sattler R, Yu Z, Herder C et al. Novel biomarkers for pre-diabetes identified by metabolomics. Molecular Systems Biology 2012;8:615–25.
8 IbÃ¡Ã±ez C, Simo C, Martin-Alvarez PJ et al. Toward a predictive model of Alzheimer’s Disease progression using capillary electrophoresis-mass spectrometry metabolomics. Analytical Chemistry 2012;84:8532–40.
9 Verwaest KA, Vu TN, Laukens K et al. (1)H NMR based metabolomics of CSF and blood serum: a metabolic profile for a transgenic rat model of Huntington disease. Biochimica et Biophysica Acta 2011;1812:1371–9.
10 Kaddurah-Daouk R, Boyle SH, Matson W et al. Pretreatment metabotype as a predictor of response to sertraline or placebo in depressed outpatients: a proof of concept. Translational Psychiatry 2011;1pii:e26.
11 Kaddurah-Daouk R, Baillie RA, Zhu H et al. Enteric microbiome metabolites correlate with response to simvastatin treatment. PLoS One 2011;6:e25482.