Combining -omics in the search for mechanisms in complex trait diseases

Rapid advances in technology have allowed for the collective characterization and quantification of pools of biological molecules, the so-called -omics, that translate into the structure, function, and dynamics of an organism or organisms. Moreover, combining different techniques aims to identify the functions of as many genes as possible, the functional genomics. These new approaches, when applied to complex diseases, try to better understand mechanisms and uncover new biomarkers for helping in the prediction, diagnosis, and monitoring of the given disease.

Using this approach, Menni et al.^[1] combined metabolomics and transcriptomics to look for markers (pathways) that could be related to increased arterial stiffness, assessed as pulse wave velocity (PWV), in an article of the current issue of the Journal of Hypertension. The Menni's study was carried out in a large and well phenotyped cohort of female twins ^[2] and although based on metabolomics, it was completed with a gene expression study.

The trait selected for the study, PWV, is gaining clinical relevance. Considered gold standard for the assessment of arterial stiffness, it has shown to be an independent marker of cardiovascular disease both in the high risk and in the general populations ^[3–6]. Increasing evidence has revealed that measurement of PWV improves cardiovascular risk stratification and has a promising potential as a surrogate endpoint of cardiovascular disease ^[7–9]. In Menni's article, PWV has been assessed by using the Sphygmocor device (AtCor Medical, West Ryde, New South Wales, Australia), with the recommended methodology ^[6]. Potential associations with the trait can result in relevant information for better understanding the mechanisms involved in the development of arterial stiffness. No grounded information exists on the issue up to now, just one article was published, which demonstrated that circulating lipid-related intermediate metabolites are closely associated with arterial stiffness and inflammation and oxidative stress in diabetes ^[10]. In addition, the potential predictive value of the metabolomic markers over or beyond one of the most used cardiovascular risk charts, the Framingham, has also been assessed ^[11].

Metabolomics, which involves the identification and measurement of metabolites in biological samples, is achieved mainly by either high-throughput mass spectrometry (MS) or NMR spectroscopy. In clinical applications, metabolites associated to a particular disease are evaluated to aid diagnosis, build predictive models, discover and confirm pathogenesis mechanisms, and monitor treatment outcomes. Many studies in experimental models suggest a relevant role for metabolomics in the characterization of disease, although the influence of interindividual variability in the metabolomic profile poses a large limitation to the translation into clinical models. The variability of the metabolomic profile depends not only on the genotype but also on age, lifestyle, environmental factors, nutritional status, drugs, and gut microbiota metabolism ^[12]. To overcome these dependencies, it is mandatory to analyse samples from large cohorts, well characterized, and stratified by known risk factors. In this context, NMR spectroscopy and mass spectrometry provide metabolomic solutions for different problems and can be complementary. Despite its relative insensitivity, NMR spectroscopy is ideally suitable for global exploratory metabolomics in large cohorts because of the virtually nonexistent sample processing, the reproducibility of the results, the low cost of the measurement and the speed of data acquisition. MS, although more costly in terms of budget and sample processing, seems the best option for the study of selected metabolites and restricted groups of patients because of the high sensitivity, the spectral resolution, and the existence of large metabolite spectral databases for signal assignment. Therefore, MS metabolomics seems essential for biomarker validation, pathway analysis and pathophysiological mechanistic studies. Menni et al.^[1] performed MS metabolomics for identifying associations between a wide spectrum of metabolites and PWV in women.

The combination of metabolomics and gene-based studies represents a powerful approach for both identifying new patient subgroups with a predisposition or resistance to disease ^[13] and for detecting novel pathways and mechanisms that deserve further exploration. However, combining -omics is not exempt of caveats ^[14]. This combined analysis requires, to be successful, complex statistical methodologies and careful selection of target metabolites. Menni et al. performed the combination of a transcriptomic and metabonomic study, looking at gene expression and metabolites associated to PWV and to Framingham risk scores. This allowed them to overcome the influence of potential confounders and explore mechanistic implications of their metabolic discoveries. Despite many authors defending the power of chemometric analysis in metabolomics ^[15,16], Menni et al. used simple and robust statistical tools for a metabolite-by-metabolite data analysis. For pathophysiology studies, this seems a more adequate approach as the goal is to identify molecules associated to the pathogenesis more than building diagnostic or predictive mathematical models. Probably, more sophisticated tools, such as partial least-square discriminant analysis, hierarchical cluster analysis, or soft independent modelling of class analogy, could provide additional associations. Similarly, using false discovery rate calculations instead of Bonferroni multiple test correction could increase the number of associations. However, despite the obvious limitations, the results seem very solid both because of the statistical significance and more importantly the associations to the expression of biologically related genes.

Metabolic associations are difficult to understand in complex trait diseases ^[17]. The use of different -omics technologies helps to decipher the biological meaning of otherwise obscure statistical associations. By adding gene expression data, Menni et al. further support the biological relevance of their metabolomic results. The authors suggest a potential link between insulin resistance and gut microbiota in blood pressure regulation. The role of gut microbiota in cardiovascular risk and cardiometabolic disease is becoming more and more evident ^[18,19]. However, most of the evidence comes from experimental models, and it is not clear whether the link is causative or not. Due to the large potential interaction among genes, proteins and metabolites in different regulatory pathways of cellular and organ functions, the scenarios can be multiple and the real impact on one phenotype is difficult to assure, mainly when the PWV phenotype integrates different components such as blood pressure values, heart rate, and arterial stiffness itself.

In summary, the work of Menni et al. represents a good example of how the combination of metabolomics and gene-based information in clinical samples may help to increase the knowledge of pathways and mechanisms of disease. There is a need for clinical studies that confirm these associations beyond traditional cardiovascular risk factors. The results provide a novel link to human blood pressure regulation, which deserves further investigation.

ACKNOWLEDGEMENTS

Conflicts of interest

There are no conflicts of interest.