ME/CFS Science Blog
Senior Member (Voting Rights)
Yes not sure if that statement holds up, hadn't seen it. Thanks for highlighting it.This is the sentence:
Yes not sure if that statement holds up, hadn't seen it. Thanks for highlighting it.This is the sentence:
I'm not sure what you are trying to say here, because doesn't the same seem to apply to DecodeME (https://www.s4me.info/threads/initi...2025-decodeme-collaboration.45490/post-637039 )?Don’t they just show sample overlaps?
If a substantial portion of the hEDS group had ME/CFS, isn’t a correlation just as expected?
What does that mean for hEDS?
I'm not sure what you are trying to say here, because doesn't the same seem to apply to DecodeME
As I understand it, in GWAS terms this is small scale and potentially subject to inaccurate results, in the same way that previous small scale GWAS on ME/CFS didn't turn up the same results as DecodeME. I think it would be worth taking the results with at least a pinch of salt as well.Abstract said:1,815 cases and 5,008 ancestry-matched controls.
2- DecodeME data strongly suggests that self-reported symptoms and self-reported diagnosis of ME/CFS (verified by self-reported symptoms) identify a group of people suffering from a distinct pathology
Quite possibly, but the authors seem to suggest that heterogeneity of their different cohorts seems to not have driven the results here but rather the opposite. Maybe they picked the studies for their meta-analysis in such a way that would make things work or maybe its sufficient for a low percentage of people with hEDS to have EDS to deliver such results (but I guess the second argument applies similarly to ME/CFS).Moreover, I think you have to consider the context of trawling for cohorts. DecodeME would hopefully have appealed to almost anyone with ME/CFS who had heard about it. They manage to get maybe as many as 20% of all UK active cases recruited. That still leaves room for bias but maybe as mitigated as possible.
Trawling 'hEDS' is a very different ball park, I think. Unlike ME/CFS with a cardinal feature of PEM, hEDS seems to be just about whatever you think makes you a bit stretchy. The variation in estimated prevalence is something like 0.02%- 10%, i.e. 500 fold. At least for ME/CFS it is probably no worse than 5 fold. If this is the study put up on another thread that I looked at it recruited through an advocacy group. EDS advocacy looks to me like the biggest can of worms out (bar MCAS).
Quite possibly, but the authors seem to suggest that heterogeneity of their different cohorts seems to not have driven the results here but rather the opposite. Maybe they picked the studies for their meta-analysis in such a way that would make things work or maybe its sufficient for a low percentage of people with hEDS to have EDS to deliver such results
But people in "All of US" are not answering specific hEDS online questionnaires (at least from what I've understood) and from what I've seen the authors seem to suggest that their results seem to be quite homogeneous amongst the different cohorts.Not sure what heterogeneity of cohorts driving results means here? Maybe the tendency to answer online questionnaires is very homogeneous? As far as I can see none of these patients have EDS. Whatever concept they think they are building for a polygenic multisystem problem it isn't EDS by definition.
But people in "All of US" are not answering specific hEDS online questionnaires and from what I've seen the authors seem to suggest that their results seem to be quite homogeneous amongst the different cohorts.
@Jonathan Edwards how would you explain SLC39A13 showing up in this meta-analysis? Doesn't it seem to suggest a relationship between hEDS and EDS?One of the more interesting findings, however, was the gene-based analysis where SLC39A13 showed up. It encodes a zinc transporter critical for connective tissue development. Mutations in this gene cause the recessive spondylocheirodysplastic form of EDS. Perhaps this points to (undiagnosed) EDS cases in the hEDS group.
It's a good question. Ideally, we'd see if there's still a correlation even if all the people with depression were excluded in decodeme. (Or ME/CFS in this study).In DecodeME, 1/3 of the participants had depression. So wouldn’t you expect a certain correlation between the DecodeME GWAS and a depression GWAS then?
The data that they used for ME/CFS is indeed the DecodeME data (see table 7 of supplementary data). Looks like ME/CFS status was verified in the hEDS population by ticking a ME/CFS box in a questionnaire and for the ALL of us set is was simply using EHR data.
In this study 3 cohorts were used. The authors seem to suggest that their results are somewhat consistent amongst these cohorts. One of these cohorts is the "All of US" cohort, which is a cohort that is not specific to one illness and I thought illness status is confirmed view EHR records (I'll have to dig deeper whether that is really the case). If the results are indeed consistent amongst the 3 different cohorts, I find it very hard to paint a picture fitting with online behaviour.I am not sure who 'All of US' are but maybe they are the people who tend to answer online questionnaires of any type. Sorry, I haven't had a chance to see exactly what is being compared with what here but the route to answering yes on an hEDS questionnaire is a strange and complicated one for sure. In the last study I saw 90% were women, yet EDS is if anything male predominant. It is way more muddled than ME/CFS.
@Jonathan Edwards how would you explain SLC39A13 showing up in this meta-analysis? Doesn't it seem to suggest a relationship between hEDS and EDS?
In DecodeME, 1/3 of the participants had depression. So wouldn’t you expect a certain correlation between the DecodeME GWAS and a depression GWAS then?
Fair enough if they did it correctly.I don't think necessarily. When recruiting for a depression GWAS people who are 'depressed' or even suffer depression, in the context of some other debilitating illness that would make any normal person depressed, may well be excluded. I have a neighbour who is depressed because her husband has Alzheimer's but I doubt she would get put in to a depression study.
That's a very fair point. But I still think it's very curious that they here use 3 seemingly quite different cohorts (in terms of recruitment) and in all 3 of them the OR for the 2 risk loci is much higher than any of the risk loci identified in DecodeME (the lowest OR for any of the 2 loci in any of the cohorts is still 1.25). I don't think we've seen something comparable for ME/CFS, in fact I think we've seen the opposite. Maybe they chose their cohorts "wisely" or something else is off but I think that requires some more sophisticated argumentation.As I understand it, in GWAS terms this is small scale and potentially subject to inaccurate results, in the same way that previous small scale GWAS on ME/CFS didn't turn up the same results as DecodeME. I think it would be worth taking the results with at least a pinch of salt as well.
I think that the effect sizes seen in DecodeME are the norm in GWAS. There's also an evolutionary theory behind this stating that if a common SNP was more strongly associated with disease, it would have been deleted.Maybe they chose their cohorts "wisely" or something else is off but I think that requires some more sophisticated argumentation.
I struggle to make sense of such an argument in the context of hEDS. Maybe it never affected reproduction rate much and it was partially advantageous to be bendy to shoot an arrow?There's also an evolutionary theory behind this stating that if a common SNP was more strongly associated with disease, it would have been deleted.
The fact that this loci had an OR of 1.8 in "MUSC1" and 1.4 in the "All of us" could suggest that maybe you can pick up something that isn't necessarily genuine, but I still find it hard to construct a story about a somewhat systemic confounder when at least one cohort seems to be rather different (in age, sex, recruitment handling, socioeconomic background, race and ethnicity).So the strong effect (OR = 1.66) found here is a bit curious and unusual: either it points to a very important gene in that region or the findings are not very robust.