The most detailed genetic analysis of myalgic encephalomyelitis ever conducted

PrecisionLife is a UK-based precision medicine company transforming how complex chronic diseases are understood and managed. Its AI-led combinatorial analytics platform identifies the specific mechanisms that drive disease in heterogeneous patient populations – providing insights that often cannot be detected using traditional GWAS methods.

We work across the drug discovery and clinical pathways, from identifying and validating novel drug targets and finding drug repurposing candidates, to developing novel genetic biomarkers for specific target mechanisms, to building differential triage, prognostic, and therapy selection tools for the clinic.

PrecisionLife has undertaken multiple landmark studies in ME and long COVID, including the first large-scale genetic analyses to reveal reproducible disease mechanisms and clinically actionable patient subgroups. Across these projects, PrecisionLife has worked with partners including Action for ME, the DecodeME team, and leading academic researchers to build the evidence base needed to inform future mechanism-matched diagnostic tests, targeted repurposing trials and precision therapies for people living with ME and long COVID.

For more information on our work in ME and long COVID, visit: www.precisionlife.com/me

To find out about members of our team, visit: www.precisionlife.com/team

The LOCOME (LOng COvid and Myalgic Encephalomyelitis Diagnostics Stratification) project was a two-year research effort launched in December 2023 that has just completed. It was spearheaded by the UK precision medicine company PrecisionLife, working in partnership with Action for ME and Prof. Chris Ponting, principal investigator for the DecodeME project (MRC Human Genetics Unit at the University of Edinburgh). Backed by more than £600,000 in funding from the UK Government’s Innovate UK Advancing Precision Medicine programme, the initiative sought to leverage DecodeME study data alongside PrecisionLife’s advanced combinational analysis techniques.

The project’s goal was to categorize individuals with ME and Long Covid into distinct subgroups defined by mechanisms based on identification of specific DNA variants linked to disease-related genes. These subgroups can correspond to particular symptom patterns and patient experiences, enabling the creation of “disease signatures” and stratification tools. Through this approach, researchers aim to deepen understanding of the biological processes driving each subgroup, identify biological markers that could enhance diagnostic accuracy, and uncover new or existing drugs that target subgroup-specific mechanisms. Ultimately, this work will support the design of clinical trials to evaluate potential novel treatments for ME/CFS and Long Covid.

Yes, if you consented for your DecodeME data to be shared anonymously with other researchers, it could have been included in the LOCOME project analysis.

During the DecodeME enrolment process, the following question was asked of all potential participants, concerning sharing data with other researchers:
“I’m happy for my data to be shared with other researchers in future studies approved by DecodeME. The studies might be about diseases other than ME/CFS. I understand that it won’t be possible to identify me from my data and that my NHS health data won’t be shared.”

86% of DecodeME participants consented to sharing data with other researchers, such as the LOCOME team, therefore amplifying their generous contributions to ME/CFS research as well as many other potential future disease studies.

It is important to note, that no one in the LOCOME team has any way of identifying any of the participants of the DecodeME project. All data is shared anonymously and under strict usage controls.
(see: FAQs | DecodeME | Institute of Genetics and Cancer)

PrecisionLife’s combinatorial analysis platform attempts to identify combinations of features, ‘disease signatures’, that when they occur together are highly associated with a physically identifiable characteristic of a disease (a ‘phenotype’). These features may be genetic, clinical, environmental or other factors, and will occur in many cases and relatively few controls.

When working in genetic data, these disease signatures capture the effects of interactions between genes and biological networks. Compared with methods that identify only single features on their own (e.g., GWAS), we find more disease signatures and they mirror more closely the complex networks of genes and physiological systems that are seen in the biology of complex diseases.

We can cluster disease signatures that have common components (e.g. a shared genetic variant) to create larger sub-groups that are strongly associated with specific subgroups of patients with similar disease features (e.g. speed of progression, severity, response to specific drugs).

‘Monogenic’ genetic diseases, such as sickle-cell disease, cystic fibrosis and some cancers, re disease that are caused by a single DNA mutation, often in the coding region of the gene. Sometimes this changes the way proteins are produced or their 3D structure, which affects their ability to play a normal role in metabolism. This abnormal function may go on to cause the symptoms we observe in disease.

‘Polygenic’ diseases in contrast are caused and affected by multiple genes. In some cases there may be different mechanisms that result in the same clinical symptoms, in others there may be interactions between several genes that together cause the disease. In complex diseases both may be true.

The ‘omnigenic’ disease model suggest that some chronic diseases may be linked to very many (e.g., several thousands) of genes. Some of these will be ‘core’ genes that are involved with central biological functions like energy production, metabolism etc. Because of the way biology works, these genes may be influenced by the function of dozens of other ‘peripheral’ genes that affect the absorption and supply of nutrients from the diet, hormone levels and immune system function. Changing the way a peripheral gene functions (faster/slower) may affect the mechanism, but is not likely to be good drug target. Core genes on the other hand are more directly linked to the output of a mechanism and so make better drug targets.

Our study identified 2,311 genes that are associated with ME, including 259 candidate core genes linked to relatively common disease signatures that show the strongest evidence of reproducibility across different sets of patients.

The true number of genes involved is probably greater than this, however, as we had relatively poor coverage of the genome in our datasets and we applied conservative approaches when identifying genes that are reliably associated with ME. Future analyses of additional ME cohorts, especially non-European patients, will likely identify additional genes involved in ME.

Yes, biology is very complex and chronic conditions often have many different genes involved. Some scientists believe that most complex, chronic diseases are at least influenced by thousands of genes. For example, GWAS, which identifies many fewer gene-disease associations than combinatorial analysis, has identified over 300 genes that are significantly associated with increased risk of heart attack, and the most predictive genetic models of coronary artery disease include millions of genetic variants. Similarly, GWAS has identified over 700 genes linked to type 2 diabetes and recently published genetic models for estimating type 2 diabetes risk rely on tens of thousands to millions of genetic variants that are believed to affect disease risk.

No, we don’t believe so. As noted above, coronary artery disease and type 2 diabetes are also highly polygenic – huge numbers of genes directly or indirectly affect peoples’ risk of developing those two conditions – but effective treatment is readily available for both. By targeting ‘core’ genes in the major mechanisms we can design drugs that gain effective control of those processes for many patients.

We believe the way forward for the ME community will be to use the results of this study and other genetic analyses to identify core biological mechanisms and genes linked to ME. These results can then be used to stratify people based on shared ME disease biology and help identify treatments that are most likely to work for each group.

We consider a patient subgroup as a group of patients who have a shared genetic disease risk (i.e, shared disease signatures linked to disease genes). Our analysis found ME patient subgroups that are linked to different genes and mechanisms based on their disease signatures.

Some patients may have more than one disease risk, i.e., have disease signatures that affect multiple genes and often multiple biological mechanisms. Further research is needed to better define these patient subgroups; however, this stratification of patients by their disease mechanism/s is the foundation for developing targeted treatments in ME in the future.

43% of the genes that we had previously associated with long COVID are also linked to ME in our DecodeME combinatorial analysis, suggesting high overlap between the two conditions.

In addition, we found that about a quarter of long COVID genes are also among the very top ranked genes that can differentiate people with and without ME. This is especially true for genes specifically linked to severe and fatigue dominant forms of long COVID that are most similar to ME.

Additional research is needed to better characterize the shared biology and differences between the two diseases, but the top shared features include genes linked to inflammatory and metabolic signalling, brain functioning and cognition, and regulation of the body’s 24-hour internal clock (circadian rhythm). 

We identified the candidate core genes by starting with over 22,000 ME disease signatures linked to higher risk of ME in all 3 patient cohorts. We then looked at the SNPs (genetic variants) to see which genes they are contained in. Just including genes where the SNPs mapped to the gene body (it’s coding region) identified the genes gave us 2,311 genes. We then looked at which of those genes were most closely associated with disease and at high prevalence (across more than 20% of participants). This gave us 259 candidate core genes.

We consider the candidate core genes to be important as they have a higher level of association to disease across the 3 patient populations and may therefore be relevant to the most patients. These core genes provide valuable clues about the biological processes that may be involved in ME as they relate to the disease signatures most strongly linked and most common in people with ME.

Out of more than 22,000 disease signatures linked to ME, we found that some were much more common and showed a stronger link to ME. We identified 259 candidate core genes linked to these signatures that affect biological processes like the immune system, how brain and nerve activity is regulated, inflammation, and how the body’s cells respond to damage and pass signals to regulate their processes via the calcium channels. Many of the genes have other functions as well and further research will be required to establish their exact role in disease.

A ‘drug repurposing candidate’ is an existing drug being evaluated for a new therapeutic use beyond its original disease indication. This usually refers to drugs that are on market, have gone to generic (off-patent) status or which may have been withdrawn from the market for reasons other than safety. It may also include reformulations, or combinations of existing therapies. The related term repositioning may also include drugs still in clinical development, or drugs that were previously discontinued for reasons other than safety, e.g., they failed clinical trials in their original disease indication. By using existing knowledge relating to mechanisms, safety data, or clinical experience, these candidates offer new treatment opportunities without starting from scratch.

Repurposing drug candidates offer major advantages over traditional drug development. Because many already have established safety profiles, development can proceed faster, with lower cost and reduced risk of failure (a new drug takes approximately 10 years and $2.6B to develop from scratch).

Repurposing/repositioning also gives researchers access to proprietary compounds, existing technical and formulation data, and well-characterised patient groups. With strong disease knowledge guiding new applications, repurposed drug candidates can accelerate therapeutic innovation and bring potential treatments to patients more quickly.

We found several genes strongly linked to the most common ME signatures (which we refer to as ‘candidate core genes’) that represent potential drug targets for ME. These genes and associated biological processes provide a foundation for future research and could be used to guide testing of new treatments for ME patients.

We have identified many genes among the 2,300 genes or the candidate core gene subset linked to long COVID in our earlier research. We think that the common genes reflect shared disease pathology between ME and long COVID and may also benefit subgroups of long COVID patients. However, it is also important to note that there are also many long COVID genes that are not biologically associated with ME and vice versa, so we consider ME and long COVID to be partially overlapping but different diseases.

Yes. We found several genes among the 2,300 genes linked to ME signatures that have been linked to drugs currently being tested as potential treatments for ME. For example, among the 259 candidate core genes that seem to play the largest roles, the gene TLR3 is linked to treatments currently under investigation in ME clinical trials.

TLR3 plays a key part of the body’s immune system that helps protect against viruses and also plays a role in managing how the body responds to inflammation. Rintatolimod, a drug which works by selecting binding to the protein product of TLR3 gene, has been tested in Phase II and Phase III clinical trials with ME patients and has shown statistically significant improvements in primary endpoints. 

At a very basic level the more disease risk genes a person has and the bigger the potential effect of those disease risk genes, the higher the chance of developing disease that person will have. This is not however the whole story as the person will also have to encounter a triggering event (e.g. a viral or bacterial infection), and even then they may have protective or compensating genetic mechanisms that will make them more resilient to the disease.

We showed using a very simplistic method that people with the most signatures (top 10% signature count score) have approximately 68% greater odds of having ME compared to people with the fewest signatures (bottom 10% signature count score). This demonstrates a clear genetic component to ME, but the signature count presented in this paper is not on its own intended to be a strong predictor of who will develop ME.

We believe that the true genetic contribution is likely greater than reflected in the signature count score, which is highly simplistic and does not capture important genetic contributions such as protective genetic features, interactions between signatures, or the importance of mechanisms and subtypes to ME biology.

Yes. Although we showed that people with many ME disease signatures are significantly more likely to have ME than people with few disease signatures, we also observed DecodeME participants who have few signatures.

The signature count score was included in the manuscript only to test that the full set of signatures is associated with disease in aggregate, and it represents an overly simplistic method for estimating peoples’ individual risk of ME. For example, some people may develop ME because they possess a combination of interrelated signatures associated with one biological mechanism even though they possess relatively few signatures from the total set.

We’re continuing to run some additional studies using the remainder of the DecodeME dataset that we haven’t yet had chance to analyze. We expect to do further analyses looking at whether there are differences between women and men, different ages and groups with different symptoms. We intend to write up our findings and make them available to the community to provide a foundation for future research.

However, now that the LOCOME and MELO projects have been completed, we do not have any further funding to support this work, so this will not be as intensive as it has been during LOCOME. We hope to be able to support ME and long COVID projects that have funding, and we’re working with several academic and clinical groups around the world to support their grant applications. We are discussing opportunities for new research with foundations and research funding bodies.

PrecisionLife works in many different disease areas including ALS, endometriosis, rheumatological, respiratory, neurodegenerative and cardiovascular diseases. We have a number of clinical collaborations and trials ongoing in these diseases, aiming to bring new clinical tests to market and to find and validate drug repurposing candidates.

Not at the moment. We are working with various groups in the UK, USA, the Netherlands, and Australia to support the design and recruitment of targeted clinical trials using repurposed drugs, but those will require separate funding and will be run by the clinical researchers rather than PrecisionLife.

People with ME and with long COVID shared their views and lived experiences to help shape the delivery and direction of the LOCOME project from start to finish including input from a dedicated Patient and Public Involvement (PPI) Group throughout.

Yes, the LOCOME team alongside our publication, are releasing our results as ‘supplementary materials’ including lists of all the genetic variants (SNPs) and the genes they are related to, as well as a list of the candidate core genes, that we have found to be associated with ME by this study.

This will help to maximize the utility of our findings to the community by allowing other researchers to look closely at the results for further interpretation and plan additional research based on these findings.