Preprint Initial findings from the DecodeME genome-wide association study of myalgic encephalomyelitis/chronic fatigue syndrome, 2025, DecodeMe Collaboration

[forestglip]

Just had a quick look at if any of the top 25 significant genes from Table S3 were significant in a few other genetic studies. I only have the genes saved from a few of the genetic studies that have been done, so I hope to check others too at some point, but I found that DCC and CDK5RAP1 were significant in the following study (which had 77 total significant genes):

Neurodevelopment Genes Encoding Olduvai Domains Link Myalgic Encephalomyelitis to Neuropsychiatric Disorders, 2025, Lidbury et al

And that study actually even found that DCC was significant in another cohort:

Comparison with Nevada (USA) ME/CFS sample: We conducted a replication study of our positive Australian associations on the GWAS raw data genotyped in an ME/CFS cohort recruited from Nevada in the United States [22]. Several associations shared by both cohorts were successfully identified, namely [...] (3) other variants anchored in the coding regions of the RAPGEF5, CSMD3, DCC, ALDH18A1, GALNT16, UNC79, NCOA3 genes.

And I've seen DCC in a couple other studies lately:

- Large-scale genome-wide analyses of stuttering, 2025, Polikowsky et al
- Disentangling nature and nurture: Exploring the genetic background of depressive symptoms in the absence of recent stress exposure using a GWAS approach, 2025, Erdelyi-Hamza et al

The DecodeME study says it is associated with chronic pain as well:

However, 9 of the 22 GWAS-1 associations were associated in R-1 [Lifelines and UK Biobank cases] with p-values < 0.05, a larger proportion than expected by chance. These nine included four DecodeME loci (RABGAP1L, FBXL4, OLFM4, CA10), plus LRRC7 – a gene associated by MAGMA gene-based testing (below, Table S4) – and DCC, a gene that has repeatedly been associated with chronic pain (41).

 

[Hutan]

We defined genes as Tier 1 genes if: (i) they are protein-coding genes, (ii) they have GTEx-v10 expression quantitative trait loci (eQTLs) lying within one of the FUMA-defined ME/CFS-associated intervals, and (iii) their expression and ME/CFS risk are predicted to share a single causal variant with a posterior probability for colocalisation (H4) of at least 75%.

That's from the Candidate Genes document. Can someone explain what that means, specifically the ii part? There is a database that says what impact a gene has on specific tissue? Does that mean that there might be more possible Tier 1 genes that don't yet have eQTL data?

And what's a FUMA-defined ME/CFS-associated interval?

Oh, and something about the iii part would be good too.

 

[voner]

I am reading the Preprint. So what does this sentence in the Preprint tell us?

We also tested for associations using only cases who did not report infection at onset (5,841 cases and 259,909 controls; ‘GWAS-No-Infection’). This yielded no significant associations (Table 3).

 

[Hutan]

I am reading the Preprint. So what does this sentence in the Preprint tell us?

Maybe the group of participants not identifying an infection as a trigger was too small to produce associations. Maybe it is a much more mixed group, with some people who don't have ME/CFS. Probably both.

 

[Jaybee00]

I think the author on the PP is

“DecodeME collaboration (2025).

Initial findings from the DecodeME genome-wide association study of myalgic encephalomyelitis/chronic fatigue syndrome.”

 

[JonBiz]

Maybe the group of participants not identifying an infection as a trigger was too small to produce associations. Maybe it is a much more mixed group, with some people who don't have ME/CFS. Probably both.

Maybe you can get ME from an infection too mild to notice?

 

[Robert 1973]

Maybe it is a much more mixed group, with some people who don't have ME/CFS. Probably both.

I think we need to be cautious about suggesting some people don’t have ME/CFS, when we don’t know what ME/CFS is physiologically. If someone meets a particular definition, then they have ME/CFS by that definition, even if it transpires that there are subgroups who have different illnesses.

DecodeME was very careful in selecting process. Assuming that participants were honest with their answers, then they have ME/CFS according to the definition that DecodeME chose to use.

Some participants may have undiagnosed conditions which would explain their symptoms and therefore negate their ME/CFS diagnoses, but I can’t see any other way that a participant could be said to not have ME/CFS.

 

[voner]

Maybe the group of participants not identifying an infection as a trigger was too small to produce associations. Maybe it is a much more mixed group, with some people who don't have ME/CFS. Probably both.

So, what does this GWAS report indicate for those people who did not report an infection as a trigger?

 

[forestglip]

I think the author on the PP is

“DecodeME collaboration (2025).

Initial findings from the DecodeME genome-wide association study of myalgic encephalomyelitis/chronic fatigue syndrome.”

The full list of authors is at the end of the paper:

Genetics Delivery Team (alphabetical order):
Thibaud Boutin, Andrew D. Bretherick, Joshua J. Dibble, Esther Ewaoluwagbemiga, Emma Northwood, Gemma L. Samms, Veronique Vitart

Project and Cohort Delivery Team (alphabetical order):
Øyvind Almelid, Tom Baker, Malgorzata Clyde, Anne Connolly, Diana Garcia, Shona M. Kerr, Claire Tripp, Jareth C. Wolfe

Patient and Public Involvement (alphabetical order): Jackie Goold, Gemma Hoyes, Sian Leary, Simon J. McGrath, Julie Milton, Anna Redshaw, Jim M. Wilson

Marketing and Communications Team (alphabetical order):
Helen Baxter, Danielle Boobyer, Claire Dransfield, Daphne Lamirel, Isabel Lewis, Nina Muirhead, Ella Ponting, Anne Redshaw, Charles Shepherd, Alice Turner

University of Edinburgh Team (alphabetical order):
Sumy V. Baby, Sjoerd Beentjes, John Ireland, Ava Khamseh, Ewan McDowall, David Perry, Joshua Slaughter

Genetic Epidemiology of ME/CFS Consortium (alphabetical order):
Erik Abner, Cindy G. Boer, Estonian Biobank Research Team, Sarah Finer, Genes & Health Research Team, Hele Haapaniemi, Hanna M. Ollila, Beth Pollack, Judith Rosmalen, Erika Romppanen, Sirine Saafi, Richa Saxena, Nasa Sinnott-Armstrong, Anniina Tervi, Lea Urpa, Jesse Valliere, David A. van Heel

Management Team (alphabetical order):
Sonya Chowdhury, Andy Devereux-Cooke, Chris P. Ponting

 

[JonBiz]

So, what does this GWAS report indicate for those people who did not report an infection as a trigger?

I could be wrong, but I understood it to mean that whatever process causes CFS after an infection involves the same genes as CFS without an infection.

 

[Jaybee00]

The full list of authors is at the end of the paper:

But they say to cite it as I pasted it above.

 

[jnmaciuch]

That's from the Candidate Genes document. Can someone explain what that means, specifically the ii part? There is a database that says what impact a gene has on specific tissue? Does that mean that there might be more possible Tier 1 genes that don't yet have eQTL data?

And what's a FUMA-defined ME/CFS-associated interval?

Oh, and something about the iii part would be good too.

FUMA is a program for annotating GWAS results, I’m guessing they used SNP2GENE which basically groups together associated SNPs located close together and assigns them all to a relevant gene, part of a gene, or genomic region (if it doesn’t code for a gene).

So basically it’s saying that a gene was called tier 1 if:
1) associated SNPs were assigned to a protein-coding gene
2) there exists information about alleles that change the expression of that gene (from databases that catalog whether certain genetic variants are associated with different expression levels of mRNA, or sometimes protein, for a certain gene in certain sampled tissues)
3) it’s reasonably likely that at least one identified SNP both has a causal association with ME/CFS (which has to be inferred because of confounders such as linkage disequilibrium) and causes expression level differences

[edit: others should feel free to correct me if I’ve made any errors in my explanation, I’m pretty certain this is correct from my general knowledge of GWAS but haven’t done this analysis myself so am not an expert]

 

[ElephantNerd]

I think we need to be cautious about suggesting some people don’t have ME/CFS, when we don’t know what ME/CFS is physiologically. If someone meets a particular definition, then they have ME/CFS by that definition, even if it transpires that there are subgroups who have different illnesses.

Agree so strongly!!

Also, for those who don't recall an infection as a trigger, could some people have had one and never realized? Either an asymptomatic infection or they just didn't connect their MECFS onset with a cold they had when they were younger??

But even if there truly wasn't an infection as a trigger, having a different subtype of MECFS doesn't negate the diagnosis at this time. (And it certainly doesn't negate the reality of the illness, which is what it might FEEL like to have a diagnosis removed without an alternative one given.)

If the researchers end up narrowing the definition of MECFS and leaving certain subgroups out (such as people without particular biomarkers or people who don't recall a viral onset), I sincerely hope they will provide an alternative diagnostic name for the subgroup who doesn't meet the new criteria. That way, they aren't "orphaning" patients who have a different subtype by removing their diagnosis.

In my experience, being ill without a diagnosis makes a patient even more susceptible to the BPS attitudes, and self blame and shame for not managing to think ourselves positively to better health and physical resilience. This attitude of trying to be strong and push through since "it can be overcome by ignoring the symptoms like everyone else who is just tired" can lead to severe decline in function (I speak from experience).

So I would never want the medical community to remove someone's diagnosis without providing an alternative.

Accuracy is important of course; i am not suggesting slapping a random diagnosis on. It would have to be well thought out and match the patient's presentation!

My initial thought is that it might be best to keep MECFS diagnostic criteria as is (based on presentation rather than a biomarker test). Then, as subgroups are identified, each subgroup of MECFS can be specifically named such as "such-and-such subtype of MECFS" vs "MECFS unspecified subtype" for the rest. My instinct is that this will be the best for the patients in the unspecified group....??

But I haven't thought it through, especially on the researcher and clinician side of things, which is a bit above my pay grade. Others much more educated may have better ways of thinking of this?????

Edit/amendment: I haven't found the right words to express my gratitude to the DecodeME team and participants. Thank you all for providing hope and direction!

 

[Sean]

Stronger associations would have been nice, and there is the issue of inconsistency across studies to be addressed. But otherwise a very welcome and important step forward, that should help focus research efforts on more productive possibilities.

Thank you to all involved.

-----------

​

Link to disease: Mutations in HFE can lead to hereditary haemochromatosis, an excessive absorption and accumulation of iron (41).​

...this is from the Supplement on Candidate Genes (which covers more than the 8 highlighted in the paper). Haemochromatosis and issues with transferrin have a number of mentions on the forum. I know some of our members have diagnosed issues with iron, and infections can change iron homeostasis.

Interesting. As previously mentioned on the forum, I have what would seem to be the opposite problem. Persistently low iron (as ferritin) and difficulty getting ferritin level up and keeping it there. Other iron measures seem normal. All a bit confusing, and am still under investigation for it.

Maybe the group of participants not identifying an infection as a trigger was too small to produce associations. Maybe it is a much more mixed group, with some people who don't have ME/CFS. Probably both.

Maybe you can get ME from an infection too mild to notice?

That was my first thought – (more or less) asymptomatic infections would explain it.

 

[Hutan]

think we need to be cautious about suggesting some people don’t have ME/CFS, when we don’t know what ME/CFS is physiologically. If someone meets a particular definition, then they have ME/CFS by that definition, even if it transpires that there are subgroups who have different illnesses.

DecodeME was very careful in selecting process. Assuming that participants were honest with their answers, then they have ME/CFS according to the definition that DecodeME chose to use.

Some participants may have undiagnosed conditions which would explain their symptoms and therefore negate their ME/CFS diagnoses, but I can’t see any other way that a participant could be said to not have ME/CFS.

We should be cautious, yes. But we should also be realistic that it's a possibility. Even the NIH study that appeared to try very hard to characterise the participants found people who, although meeting the symptom criteria, had other reasons for their symptoms. There will be some people in the DecodeME ME/CFS cohort who have another good reason for their symptoms. It's possible that people who can't identify a clear infection immediately prior to the onset of their symptoms may be more likely to be in that group.

Criteria for ME./CFS generally have that out clause of there not being another reason for having the symptoms.

So, what does this GWAS report indicate for those people who did not report an infection as a trigger?

They are part of the overall ME/CFS cohort to which the findings apply to. It could just be that taking a smaller sample reduced the statistical power so much that the signals were too weak. It could be a bit like just analysing the subset of people from Wales. Because the number of participants will be smaller, maybe no genes are significant, but it doesn't mean that the findings for the whole group don't apply to ME/CFS participants from Wales.

 

[ElephantNerd]

should be cautious, yes. But we should also be realistic that it's a possibility. Even the NIH study that appeared to try very hard to characterise the participants found people who, although meeting the symptom criteria, had other reasons for their symptoms. There will be some people in the DecodeME ME/CFS cohort who have another good reason for their symptoms. It's possible that people who can't identify a clear infection immediately prior to the onset of their symptoms may be more likely to be in that group.

Criteria for ME./CFS generally have that out clause of there not being another reason for having the symptoms.

So hopefully those who don't have MECFS but rather have something else will receive a more accurate diagnosis in those cases! more accurate diagnoses is the goal

 

[DHagen]

The limited replication of the findings when tested against other databases is disappointing. There's been some public commentary about that. It's suggested that this indicates that there are a lot of people with CFS and ME/CFS labels in many database who don't actually have ME/CFS, and is a reflection of the care taken to characterise DecodeME participants well. I think that is likely to be true.

If so, then it means that research trying to find biomarkers, or talking about risk factors is going to fail unless there is really careful selection of participants. It's probably even worse with Long Covid.

This really concerns me and, though it is still very early days (hours), I am actually surprised at how little I have been seeing it discussed so far. Perhaps I am once again looking at it incorrectly.

When attempting to replicate these findings in previously existing databases, given that the genetic differences identified "are also often found in people without ME/CFS" (p.2), wouldn't one still expect to see some reflection of this predominance even in a data set feature many unidentified individuals without ME? That is, wouldn't the fact that a fair number of the pw/oME also have these genetic traits mean that even if only 50% of the "ME cohort" actually had ME we'd still be likely to see this in the analysis?

I dearly hope it is true that this reflects "the care taken to characterise DecodeME participants well," which would seem to be the best case scenario - but even this makes me worry that we're going to see this fact used both to disregard and/or minimise this study, or that the suggestion that "there are a lot of people with CFS and ME/CFS labels in many databases who don't actually have ME/CFS" will be used against patients who will accused of not having "real" ME.

Not a knock against the researchers, of course, who are to be commended for (among many, many other things) highlighting this issue.

 

[Sean]

So is an infection triggered neuro-immune condition a broadly correct characterisation of ME/CFS, based on these findings?

Given that the genetic differences identified "are also often found in people without ME/CFS" (p.2), wouldn't one still expect to see some reflection of this predominance even in a data set feature many unidentified individuals without ME? That is, wouldn't the fact that a fair number of the pw/oME also have these genetic traits mean that even if only 50% of the "ME cohort" actually had ME we'd still be likely to see this in the analysis?

I dearly hope it is true that this reflects "the care taken to characterise DecodeME participants well," which would seem to be the best case scenario - but even this makes me worry that we're going to see this fact used both to disregard and/or minimise this study, or that the suggestion that "there are a lot of people with CFS and ME/CFS labels in many databases who don't actually have ME/CFS" will be used against patients who will accused of not having "real" ME.

Maybe it is identifying currently healthy people who have an increased genetic risk of getting ME/CFS.

Would be interesting to follow that group for a few decades and see if they do get it at a higher rate than those without that risk.

 

[DMissa]

My understanding is that the main mitochondria linked gene wouldn't explain "feeble mitochondria". If anything maybe the reverse, but I wonder if it may show that the metabolic clues we have had make sense in an unexpected way.

As the resident expert I will make sure that I find some time to look into this and clarify it for everyone in this thread

 

[jnmaciuch]

So is an infection triggered neuro-immune condition a broadly correct characterisation of ME/CFS, based on these findings?

It would be jumping the gun based on these results alone. The immune or nervous system may well be involved, but these genes aren’t exclusive to those systems and have other functions, and we also can’t differentiate between genes involved in maintaining disease process and genes that are involved in triggering it.

We can say that genetic results point to the potential involvement of genes that have previously been associated with neuronal and immune function (among other things).