In the age of large-scale omics data, we often find ourselves discussing the merits of genomics, proteomics and transcriptomics. But lesser discussed is metabolomics – the study of small molecules, metabolites, within cells. The overarching goal of metabolomics is to characterise these molecules and their interactions within their biological environment. Despite its potential, metabolomics is considered a relatively novel part of the multiomics toolbox.
Metabolomics research is now on the rise and technology is becoming more advanced. That said, challenges still remain. In this feature, we assess the state of metabolomics and look at the past, present and future to give you a comprehensive overview of this research landscape.
The basic concept of a metabolome can be traced back thousands of years, but the first modern instance of the concept (at the time referred to as a metabolic profile) arose in the 1940s, focusing on the presence of molecular compounds within bodily fluids such as urine. Early methods to assess the metabolome and its links to human disease typically involved paper chromatography. As technology in the life sciences field advanced, eventually gas chromatography-mass spectrometry and nuclear magnetic resonance (NMR) spectroscopy emerged as crucial tools for the profiling of human metabolites from the 1970s onwards.
In the 1990s, metabolic profiling had a new lease of life. The use of liquid chromatography-mass spectrometry came to the forefront of the field and the term ‘metabolomics’ was coined in 1998 by Stephen Oliver, who was working on yeast functional genetics. Continued technological advancements led to improved sensitivity, resolution and speed of metabolite analysis. Plus, the development of high-throughput platforms allowed for the simultaneous analysis of a large number of metabolites.
In the early 2000s, great strides were made in the metabolomics field. In 2004, the first web-based metabolomics database was created, allowing for the analysis of thousands of human metabolites based on mass spec data. Three years later, The Human Metabolome Database was launched, containing the most comprehensive collection of metabolites available at the time. Over the last two decades, the field has continued to grow. Techniques to detect and characterise metabolites have improved over time, and software has advanced to a level that allows for high-throughput analysis. One of the most famous pieces of metabolomics software is MetaboAnalyst, which was developed in 2009 and has been continuously updated since. The platform and tools allow you to process, analyse and annotate a variety of experimental data from a number of sources.
Despite vast progress, the field is still fraught with challenges. Given the vast scale of the human metabolome, perhaps the biggest hurdle is the difficulty of assessing the whole metabolome using only one analysis method, due to the overwhelming diversity of the molecules. Additionally, the most advanced technology is often inaccessible to researchers and clinicians, meaning only subsets of metabolites can be assessed and characterised in one experiment or test.
In spite of these challenges, metabolomics shows great promise as a tool to improve precision medicine. This is because the metabolome represents the closest thing to the raw phenotype of a cell or tissue at a given time or state. Additionally, spatial technology is paving the way for more effective metabolomics analysis, with methods such as mass spectrometry imaging (MSI) allowing researchers to gain regional information regarding metabolites. Annotation of the potentially thousands of metabolites in a sample remains a challenge, although a number of tools are currently in the pipeline to overcome this issue.
Metabolomics can be split into two distinct subgroups: targeted and untargeted analysis, both of which have their own merits in research and healthcare. Targeted metabolomics assesses the presence of predetermined molecules, such as for diagnostics, whereas untargeted metabolomics is a more comprehensive profiling method of the molecules in a sample. Both subtypes are currently being investigated for use in clinical settings.
Metabolomics has played a role in our understanding of cancer, inflammatory bowel disease and a number of other human conditions, contributing to the identification of clinical biomarkers for diagnosis and treatment. One prominent example of this is in cardiovascular disease. The metabolite Trimethylamine N-oxide (TMAO) is a gut microbiota metabolite that is indicative of adverse cardiovascular events and has the potential to be used in diagnostic and preventative settings. Likewise, metabolomics played a role in our understanding of COVID-19 infection and immune response, showing diagnostic and prognostic potential. However, there are still very few examples of metabolomics applications that have made it to the clinic.
The metabolomics field is vast, and its potential to improve human health, and other fields, is immense. Let’s take a look at some interesting papers from the last 12 months.
The potential role of early life feeding patterns in shaping the infant fecal metabolome: implications for neurodevelopmental outcomes: This study, published in December 2023, explored the impact of breast feeding and formula feeding on the infant fecal metabolome and, consequently, neurodevelopmental outcomes. The researchers discovered that feeding patterns in early life can in fact impact the metabolome (Chalifour et al., 2023).
Single-cell spatial metabolomics with cell-type specific protein profiling for tissue systems biology: This paper describes a framework for the use of untargeted spatial metabolomics and targeted multiplexed protein imaging to jointly profile proteins and metabolites in single immune and cancer cells, increasing our understanding of systems-level tissue biology (Hu et al., 2023).
Reverse metabolomics for the discovery of chemical structures from humans: This study used a reverse metabolomics approach, acquiring MS/MS spectra from newly synthesised compounds and compared them to databases filled with metabolomic and phenotypic data. The team uncovered associations to human inflammatory bowel diseases and validated the effectiveness of the approach (Gentry et al., 2023).
Analysis of the effect of different withering methods on tea quality based on transcriptomics and metabolomics: Stepping away briefly from human health, metabolomics can also be used for several other unique purposes. In this study, metabolomics (alongside transcriptomics) was used to assess the impact of different ‘withering’ methods on the quality of tea (Jia et al., 2023).
Machine learning facilitates the application of mass spectrometry-based metabolomics to clinical analysis: A review of early diagnosis of high mortality rate cancers: This review from late 2023 evaluates the use of machine learning to improve the effectiveness of mass spectrometry-based clinical metabolomics in the early detection and diagnosis of cancer (Ngan et al., 2023).
Recent advances in mass spectrometry-based computational metabolomics: Another 2023 review, this article explores the advances in metabolomics technology – specifically, the computational progress that has allowed for better analysis, visualisation and integration of large metabolomic datasets – and how these advances address common challenges (Ebbels et al., 2023).
Role of metabolomics in the delivery of precision nutrition: Diet is a key regulator of human health, and in this article, the authors highlight the potential of metabolomics to inform personalised nutrition programs (Brennan and de Roos, 2023).
Large‐Scale Metabolomics and the Incidence of Cardiovascular Disease: This study, published in early 2023, assessed almost 800 circulating metabolites in over 2000 individuals for associations with cardiovascular disease. They identified 37 significant associations, of which 35 were classed as subclinical markers of disease (Lind et al., 2023).
Probiotic-fermented tomato alleviates high-fat diet-induced obesity in mice: Insights from microbiome and metabolomics: This study explored the impact of eating probiotic fermented foods, specifically tomatoes, which are associated with weight-loss. Using metabolomics and assessing the microbiome, the researchers concluded that fermented tomato was more strongly associated with reduced weight gain and fat accumulation than unfermented tomato in mice (Wei et al., 2023).
Integrated metabolomics and machine learning approach to predict hypertensive disorders of pregnancy: In this study, the authors used metabolomics and machine learning to identify candidate metabolites in urine that could be indicative of hypertensive disorders in pregnancy. These findings could play a role in decreasing maternal-foetal morbidity (Verghese et al., 2023).
In the spotlight
Metabolomic Biomarker Signatures for Bipolar and Unipolar Depression – Tomasik et al., 2023.
Bipolar disorder can be hard to diagnose and is often misdiagnosed as major depressive disorder due to overlapping symptoms. This can be detrimental to the patient, as they will not receive the right treatment and support. In this study, the authors investigated whether metabolomics could be used as a diagnostic tool to differentiate between the two conditions. Using patient samples from the Delta study, the researchers aimed to identify those who were suffering from bipolar disorder that had a misdiagnosis of major depressive disorder. Using a targeted mass spectrometry platform, they analysed metabolites in dried blood spots and identified biomarkers – notably, ceramide – which could differentiate between the two conditions in both the discovery and validation cohorts. The findings highlight the potential of metabolomics technology in the clinic.
Figure 1: Graph showing the importance of each identified metabolite to the classification model. Adapted from Tomasik et al., 2023.
In one of our recent interviews in the ‘Big Challenge’ series, we had the pleasure of chatting to Theo Alexandrov (Group Leader at the European Molecular Biology Laboratory (EMBL)) about the big challenge in metabolomics, particularly regarding the latest single-cell and spatial methodologies. Below is a snippet from Theo’s interview.
‘What would you say is the big challenge in your field?
The next challenge for metabolomics is in data interpretation. In metabolomics, in terms of detection, we have amazing technologies already, in terms of hardware and instruments. We can do a lot of things, but in terms of interpretation, metabolism and metabolic pathways are inherently very complex. Metabolites are building blocks and energy sources, but the same molecules can play signalling roles. There are also many unknown roles of metabolites. Lately, the Rutter lab has published a paper in Nature Cell Biology showing how metabolites have unstudied roles, in particular in controlling the activity of enzymes in their pathways and also enzymes throughout the whole metabolism. This kicked off the discussion of ‘metabo-verse’, a universe of all the small molecules with diverse functions and roles.’
‘Why should people care about this challenge?
We do not have the databases of roles and functions for metabolites, similar to how we have them for genes and for proteins. There is work to be done to create these catalogues of their functions and associations with cell types. I think these databases will eventually explain the roles and contain all the biochemical molecular functions of the molecules. That will be a big breakthrough, but this is our current challenge.’
‘What is being done to tackle the issue, or what should be done to tackle the issue?
First of all, we lack robust protocols, particularly in single-cell and spatial, that are accessible for scientists who are not from analytical chemistry labs. We’ll have impact with this technology only when it will be accessible by biologists and by clinical scientists. Over the last years, it started happening with spatial metabolomics when a number of biology labs installed their first imaging mass spectrometer. However, we’re not there yet. Once this happens, they will create bigger opportunities. For this to work, we need to have better instrumentation, we need to have robust protocols, which are relatively easy to execute, we need to have user friendly software, which is not for geniuses in mass spectrometry.
On top of this, we need to have databases that this software can tap into to enhance data interpretation. And, in particular, we need to link to other omics, because metabolomics is not an ultimate tool, it is orthogonal and complementary to other omics. Obviously, to have a biological or medical conclusion or a decision for drug development, one needs to use all the tools, and have metabolomics in your portfolio. There should be more to be done to link metabolomics with other omics tools that will help achieve much bigger impact.’
Read Theo’s full interview here.
Whilst the concept of the metabolome has existed in some form for thousands of years, its merits within the healthcare sphere have only been relatively recently considered. Due to the challenges associated with the sheer scale of the metabolome, it is unsurprising that the field has grown more slowly in comparison to other omics such as proteomics and transcriptomics.
But this does not mean that metabolomics should not be in the spotlight. The concept is widely touted as the future of precision medicine, allowing for the best possible insight into cellular phenotypes both as a result of genetic and environmental factors, and is highly regarded as a breakthrough diagnostic tool. Work is ongoing to implement metabolic biomarkers into the clinic, allowing for detection and tailored treatment of various conditions, notably cardiovascular disease. However, individual metabolites may not be able to accurately predict disease, and the concept of fingerprint panels has been discussed as a possible alternative.
Metabolomics also has potential for preventative medicine. For example, by employing precision nutrition techniques, individuals could engage in seemingly healthier lifestyles, leading to lower disease burden. Additionally, metabolomics could be used to track treatment response. Despite the potential of the concepts, there are still no FDA-approved metabolomics tests available in these contexts.
There are still a number of challenges with integrating the use of metabolomics in the clinic. Accessibility is still limited, and current tools are mostly insufficient for assessing the metabolome at scale. That said, these problems are now being addressed. New techniques such as lipidomics and fluxomics, alongside the integration of deep learning models and improved computational workflows for data analysis, will all play a role in the journey of metabolomics from bench to clinic. In addition, improved integration of different omics will enhance the clinical utility of all kinds of datasets. For a more complete overview of this topic, please refer to our Multi-Omics Playbook, due to be released late January 2024.
References and further reading
Kell, D.B., Oliver, S.G. The metabolome 18 years on: a concept comes of age. Metabolomics 12, 148 (2016). https://doi.org/10.1007/s11306-016-1108-4
Johnson CH, Gonzalez FJ. Challenges and opportunities of metabolomics. Journal of Cell Physiology. 277, 8 (2012). doi:10.1002/jcp.24002.
Johnson, CH et al. Metabolomics: beyond biomarkers and towards mechanisms. Nature reviews. Molecular Cell Biology 17,7 (2016). doi:10.1038/nrm.2016.25
Gertsman, I, and Barshop, BA. Promises and pitfalls of untargeted metabolomics. Journal of Inherited Metabolic Disease. 41,3 (2018). doi:10.1007/s10545-017-0130-7