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

[Jonathan Edwards]

A question that remains is what exactly is causing the increased risk in women. Is it genetic variants on the X chromosome, hormonal or immune differences?

I don't think the 'increased risk in women' - i.e. why women have more ME/CFS than men - can be due to genetic variants of the X chromosomes because any variants will be shared out equally to both men and women. Genetic variants on the X chromosome may explain why women with ME/CFS have it and women without ME/CFS don't but that is a separate issue.

And we almost certainly cannot separate hormonal from immune issues because immune differences are very likely dependent on cumulative hormonal environment over many years, with changes at menarche and menopause. There might be a polymorphism in a gene for an immune receptor that is directly influenced in expression level by a single gene on the X chromosome with dosage effect (double for women). It that polymorphism was skewed differently in women with ME/CFS from women without ME/CFS we would have something meaty to work on but this may be wishful thinking.

 

[forestglip]

So the control data QA sweep used GnomAD (v2.1.1) which is an hg19 reference. But the DecodeME paper references hg38 locations?

As far as I understand, that shouldn't really change anything. h19 and hg38 are just different ways to refer to a SNP and can be converted between each other.

 

[wigglethemouse]

As far as I understand, that shouldn't really change anything. h19 and hg38 are just different ways to refer to a SNP and can be converted between each other.

I understand. I just expected to see something in the methods to explain what they did. I could find no mention of the remapping the genomic location from hg19 to hg38 in the text using a search and a quick read of the relevant sections in methods.The methods are very detailed. Perhaps I missed it.

I always get concerned when I spot check a main finding in the paper with available information and there is a mismatch. In this case one of the main findings is OLFM4 variant 13-53194927-GT-G rs35306732 in the paper but it does not exist in publicly available data on geneatlas for the UK Biobank axiom array data from which the control data was taken. LINK

@Chris Ponting would you be kind enough to explain why I can't find the OLFM4 variant from the paper in the data on the geneatlas tool?

 

[Ravn]

There’s considerable confusion in other patient groups about GWAS. Understandably so, it is all rather complicated. I see people interpreting the results as there definitely being something wrong with the 8 highlighted genes. Some who have dna data for themselves are finding they’re not having the snps listed and are wondering if that means they don’t have ME. So lots of misunderstandings

I think labelling the 8 regions with one particular gene each may be contributing to the confusion. I expect it’s standard procedure for this type of study but is there another way of handling this that’s clearer for lay people?

Like talking more prominently about risk-increasing/decreasing regions found, without attaching a specific gene’s name to them, maybe label the regions according to location or even just as risk regions 1 to 8. And then deeper in the guts of the paper explain that in those regions there are all these candidate genes - each of which may or may not increase or decrease risk, more research needed - and of those these 8 look especially interesting based on what we currently know but new knowledge from more research may shift the focus to one or more of the others

 

[Jonathan Edwards]

There’s considerable confusion in other patient groups about GWAS. Understandably so, it is all rather complicated. I see people interpreting the results as there definitely being something wrong with the 8 highlighted genes. Some who have dna data for themselves are finding they’re not having the snps listed and are wondering if that means they don’t have ME. So lots of misunderstandings

To be honest that sounds like an inevitable learning curve that will crop up if people do in for things like getting their DNA tested without understanding what the findings would mean. At least questions are being asked and they may learn.

I don't think it is the job of scientists writing papers using technical methodology to explain the basic biological principles behind how that methodology works. They should give enough detail for scientists not immediately in their field to see how the technique is being applied but not a lecture on the principles of genetic associations with disease.

 

[Chris Ponting]

I understand. I just expected to see something in the methods to explain what they did. I could find no mention of the remapping the genomic location from hg19 to hg38 in the text using a search and a quick read of the relevant sections in methods.The methods are very detailed. Perhaps I missed it.

I always get concerned when I spot check a main finding in the paper with available information and there is a mismatch. In this case one of the main findings is OLFM4 variant 13-53194927-GT-G rs35306732 in the paper but it does not exist in publicly available data on geneatlas for the UK Biobank axiom array data from which the control data was taken. LINK

@Chris Ponting would you be kind enough to explain why I can't find the OLFM4 variant from the paper in the data on the geneatlas tool?

My guess is that this variant is absent from the dbSNP Release used by GeneAtlas at the time, but present in the reference panel that we used for imputation (namely, UK Biobank Whole Genome Sequencing variants). Not all variants are listed in all resources unfortunately.

 

[Chris Ponting]

Hi. I've posted this elsewhere on Science for ME, but think it's important to leave here too.
About the "only a 1% difference between [the frequencies of] those with ME and controls". This small "effect size" does not matter if you're focused on drug discovery. This is because the success rate from clinical development to approval "is largely unaffected by genetic effect size", see https://www.nature.com/articles/s41586-024-07316-0. The reason for this is subtle but important. Even when DNA variation tweaks biology only slightly, this variation highlights what biological aspects need changing via drugs. And these drugs can be made to alter biology to a far, far greater extent than the genes can.

 

[Hoopoe]

Is it possible that we struggle to understand ME/CFS because we don't understand much about how the body returns to normal after an infection?

In the sense that we know a lot more about infection, inflammation, autoimmunity than about the restoration of homeostasis that follows that fighting phase. We try to build hypotheses with the knowledge that we have and the knowledge about the restoration phase is very limited. The research has failed to find a persistent pathogen, significant inflammation, autoantibodies, tissue damage. In ME/CFS it seems the body has returned to normal in many but not all ways... there must be some specific aspect that has not returned to normal which is causing a lot of problems in the long term.

Without a complete return to homeostasis, the body ends up being unable to tolerate the stress of ordinary daily activities and you get various responses like unrefreshing sleep, fatigue and postexertional malaise, avoidance and pacing behaviours that are psychologized.

 

[Kiristar]

One thing that worries me is how differently different countries (let alone individual physicians) appear to be diagnosing ME/CFS, not least with the muddying of the water from the CFS label. This makes me wonder if it even can be replicated or if that of itself is too much of a confounder.

It makes me think getting to a diagnostic test is more important than first thought for future research reasons.

 

[Andy]

One thing that worries me is how differently different countries (let alone individual physicians) appear to be diagnosing ME/CFS, not least with the muddying of the water from the CFS label. This makes me wonder if it even can be replicated or if that of itself is too much of a confounder.

It makes me think getting to a diagnostic test is more important than first thought for future research reasons.

Until then our screening questionnaire made sure that our DNA donor participants met CCC and/or IOM criteria and is available to be used in any future research.

 

[Simon M]

was anyone else besides me surprised that 86.1% of the cohort reported "muscle pain"? many years of following ME/CFS forums, webinars, conferences led me to believe the number would be much lower. I’m not saying this number is good or bad or inaccurate, just that it surprised me.

Interesting. This came out in 2023, and I was unsurprised then, and I don't think it drew any comment from PwME. I'm sure I've seen similarly high figures in other large patient surveys. Can anyone point to other large symptom surveys?

Assuming this is right, the surveys are more likely to be representative of the broader population than either forums or webinars (I think there are only 400-odd members here, for instance).

Pain is a major feature of my PEM. I had assumed it was for many (though I don't think I have see PEM-pain covered in surveys).

 

[Simon M]

I would think that the genotyped SNPs are the ones with an INFO score of exactly 1 but there are only 408,031 of those in the dataset while the array (Thermo Fisher UKB AxiomTM array) should have around 820.000.

I know there was extensive QC, and your and @forestglip 's Manhattan plots showed how important they were (there would have been lots more exciting results without QC, no doubt lots of blind alleys). Maybe @Andy can comment?

 

[chillier]

A question I have is if there is any way to infer whether we are looking at one common pathological process or more from the GWAS data. A related question is if the risk from each risk variant stacks additively. If you have all 8 for instance would you just combine the increased risk from each other together in a linear way, or do they act together synergistically for example to increase risk even further?

 

[Andy]

I know there was extensive QC, and your and @forestglip 's Manhattan plots showed how important they were (there would have been lots more exciting results without QC, no doubt lots of blind alleys). Maybe @Andy can comment?

Not a chance, far too technical for me I'm afraid. That would have to be asked of Chris or someone else from the Genetics team.

 

[Hoopoe]

One thing that worries me is how differently different countries (let alone individual physicians) appear to be diagnosing ME/CFS, not least with the muddying of the water from the CFS label. This makes me wonder if it even can be replicated or if that of itself is too much of a confounder.

It makes me think getting to a diagnostic test is more important than first thought for future research reasons.

I suspect that it has more to do with how ME/CFS is viewed in each country among healthcare professionals than case definitions.

In Italy nobody uses the ME label, and Fukuda criteria are used and my impression is that there's a lot less toxicity towards patients and the illness. It's widely neglected, more than in the UK it seems, but not widely psychologized, unlike in the UK.

In Italy there's probably very significant underdiagnosis of ME/CFS, and instead some of the patients are given a fibromyalgia diagnosis. Fibromyalgia seems to function as a "we don't know what's wrong and you have fatigue and pain" diagnosis.