Should ME/CFS genetic research focus on using post-COVID ME/CFS and recovered COVID controls?

forestglip

forestglip

Moderator
Staff member
A goal in experiments is to control for as many confounders as possible to increase the chance that any positive results are actually related to the outcome of interest.

My concern with just taking a bunch of people with ME/CFS and a bunch of controls and doing a GWAS is that you might get genes that are related to infection susceptibility, and you might not be able to tell them apart from genes related to ME/CFS. I suspect that might be part of the signal in the Zhang HEAL2 paper, where genes related to COVID susceptibility are emerging in a study on ME/CFS. And we might see some genes related to susceptibility to infections like EBV in the DecodeME study.

Theoretically, you could meticulously check which infection preceded each person's ME/CFS (not that it's even possible for everyone, and some didn't have an infection they can point to), then find healthy controls who also got these infections.

But we have a better option with long COVID, thanks to the massive amounts of virus testing that have been done: only those who are fairly certain that well-defined ME/CFS began shortly following a confirmed COVID infection as the case group, and those who recovered from a confirmed COVID infection for the control group. Maybe there should also be matching for initial COVID severity.
 
I think there is still a problem here, and maybe a worse version of the same problem @forestglip.

When you say susceptibility to infection X you are really implying susceptibility to ME/CFS specifically as a consequence of X infection. That might correlate with severity of the X infection, or susceptibility to acute symptoms, but it doesn't matter if it does. Most of the infections we are interested in are epidemic or endemic and most people have had them. We don't really have much information on people who were exposed to X but did not get infected. Things like HLA genes as far as we know just influence things like severity and time course, not whether you get the infection.

So your study will pick out genes for susceptibility to ME/CFS specifically as a consequence of Covid 19. Which is what we wanted to avoid? Pretty much everyone has had Covid so the control group is just everyone without ME/CFS.
 
I'm unsure what you're alluding to here. What is EBV susceptibility supposed to mean? As far as I know pretty much everyone gets infected with glandular fever once and that's normally about it.

The Covid problem might be a different one with certain studies picking up risk genes associated with severe Covid and as a consequence PICS and then as a consequence of that LC but I see it harder to see how that would be relevant for something like DecodeME.
 
Jonathan Edwards said:
Most of the infections we are interested in are epidemic or endemic and most people have had them.
Click to expand...
Jonathan Edwards said:
Pretty much everyone has had Covid so the control group is just everyone without ME/CFS.
Click to expand...
EndME said:
As far as I know pretty much everyone gets infected with glandular fever once and that's normally about it.
Click to expand...
Yes, mostly I was thinking about avoiding having controls who have never had an infection at all. So maybe not a big concern. It slipped my mind that 90% of people have EBV in their system.

Still, while the science might not be settled, a lot of studies are finding that initial COVID severity is associated with having long COVID. I don't know about the research on this for other infections and ME/CFS, but it seems plausible that it would apply there as well.

So if you get controls who mostly had mild/asymptomatic infections and didn't go on to get long COVID, and cases who mostly had more severe infections and did get long COVID, then you would expect to see genes related to COVID severity that might have little to do with an ME/CFS physiological process, apart from confirmation that more severe infectious disease makes it more likely.
 
forestglip said:
So if you get controls who mostly had mild/asymptomatic infections and didn't go on to get long COVID, and cases who mostly had more severe infections and did get long COVID, then you would expect to see genes related to COVID severity that might have little to do with an ME/CFS physiological process, apart from confirmation that more severe infectious disease makes it more likely.
Click to expand...

Yes, but you wouldn't want to make that comparison anyway would you?
You would want to match the controls for severity of acute Covid.
But you are still likely to pick out genes that relate specifically to a mistaken immune response to Covid that gets you into ME/CFS. Those genes might not apply to what gets you from EBV to ME/CFS. That would certainly apply if you are using the sort of theory that invokes peptides in grooves and MHC variants.
 
Jonathan Edwards said:
You would want to match the controls for severity of acute Covid.
Click to expand...
Yes, that's mainly what I'm arguing for, and something that's not possible if you are using ME/CFS cases that followed a hodge podge of other infections.
Jonathan Edwards said:
But you are still likely to pick out genes that relate specifically to a mistaken immune response to Covid that gets you into ME/CFS.
Click to expand...
Sure, that may be unavoidable, and may be interesting anyway.
 
forestglip said:
Yes, that's mainly what I'm arguing for, and something that's not possible if you are using ME/CFS cases that followed a hodge podge of other infections.
Click to expand...

But does that matter? If these are infections that almost everyone gets than why is severity of acute infection relevant? We don't want to deliberately pick controls with mild acute illness and ME/CFS cases with severe acute illness but if acute illness is ignored then you are just looking at susceptibility to ME/CFS in the context of exposure to a range of infections. That will be less dependent on quirks of gene products interacting with any specific microbe.

Let us say a GWAS picks out HLA-C7. That might just be a molecule that fits Spike protein a certain way. Or it might pick out CD57+ T cells, which are crucial to the recovery from EBV infection in a way that does not apply to other infections.
 
Jonathan Edwards said:
If these are infections that almost everyone gets than why is severity of acute infection relevant? We don't want to deliberately pick controls with mild acute illness and ME/CFS cases with severe acute illness
Click to expand...
I'm having a hard time following. If you don't want to deliberately pick cases and controls with different severities, then clearly it's a confounder so why not match severity as closely as possible.

Jonathan Edwards said:
Let us say a GWAS picks out HLA-C7. That might just be a molecule that fits Spike protein a certain way. Or it might pick out CD57+ T cells, which are crucial to the recovery from EBV infection in a way that does not apply to other infections.
Click to expand...
But neither of these sounds like it gets the heart of ME/CFS. It might provide some interesting clues about how the body handles EBV specifically, and it'd be good to test that on its own where cases and controls have different acute severity. But if cases and controls differ in both ME/CFS status as well as acute infection severity, then you'd have a harder time picking out which genes apply to what.
 
Actually, maybe the heterogeneity of different infections preceding ME/CFS achieves a similar goal.

If you only look at post-COVID ME/CFS, then, even if matching for severity, there might be a lot of COVID specific genes that have nothing to do with the common processes underlying ME/CFS after different infections.

But if you have cases with ME/CFS following a wide assortment of infections, then those microbe-specific genes will be muted, compared to ME/CFS-specific genes.

There's still the issue of genes relating to general susceptibility to severe acute illness no matter the microbe, though.
 
forestglip said:
I'm having a hard time following. If you don't want to deliberately pick cases and controls with different severities, then clearly it's a confounder so why not match severity as closely as possible.
Click to expand...

Why is it a confounder? If there is a causal relation between gene and ME/CFS in the context of infection X, the acute illness is surely epiphenomenal to what you are studying? It might be a sign that you are more likely to get ME/CFS, it might not, but it doesn't matter.

Another problem for Covid-19 is that people get virus-specific long term problem like loss of taste or lung problems, and at least the latter might be confused with ME/CFS fatigue.
forestglip said:
But neither of these sounds like it gets the heart of ME/CFS.
Click to expand...

Why not? We are looking for some genetic factor that increases risk of a 'maladaptive' immune response to an infection generating a long term signalling process. The diseases we know that look like that often have HLA or T cell genes giving risk. The heart of ME/CFS is the 'bug' in the immune software that leads to continued problems. These are the sorts of molecule that so far we know get involved in that. Immunologists often assume that these things are to do with handling specific peptides from specific environmental antigens but there are other explanations that may be more likely. Chlamydia, salmonella and shigella all trigger long term arthritis in people with HLA-B27.
 
forestglip said:
Actually, maybe the heterogeneity of different infections preceding ME/CFS achieves a similar goal.
Click to expand...

That is what I am suggesting.
forestglip said:
There's still the issue of genes relating to general susceptibility to severe acute illness no matter the microbe, though.
Click to expand...

So for ME/CFS I see that as likely part of the heart of it, just as B27 is part of the heart ofreactive arthritis.
 
Jonathan Edwards said:
Why is it a confounder? If there is a causal relation between gene and ME/CFS in the context of infection X, the acute illness is surely epiphenomenal to what you are studying? It might be a sign that you are more likely to get ME/CFS, it might not, but it doesn't matter.
Click to expand...
Yes, the processes of the acute illness are important. I think what I'm mainly suggesting is, if you are able to study them separately in order to make it easier to match gene to specific process, why not do that?

Why not have one cohort that outwardly had identical infections, then a week later half got ME/CFS? Now you can be more sure that all the significant genes are related to what happens after the infection, and you aren't trying to tease apart whether the gene just allowed the microbe to replicate more or something else more specific to ME/CFS, like maybe whether it allowed to microbe to subsequently persist asymptomatically or get into the brain, for example.

What are we losing with that approach?

I mean, I guess on a population level, it's good to know what genes make people more susceptible to ME/CFS, no matter which step in the process it relates to. But then that implies you'd be interested in, for example, a gene that makes you more inclined to go to large social gatherings where EBV is spreading wildly. It's interesting and potentially useful, sure, but it adds more work in figuring out, is this a "social gathering" gene, a "severe infection" gene, or a "post-infectious sequelae" gene.

Jonathan Edwards said:
Another problem for Covid-19 is that people get virus-specific long term problem like loss of taste or lung problems, and at least the latter might be confused with ME/CFS fatigue.
Click to expand...
I'm not sure how certain we can be that this doesn't apply to many other infections. Without a pandemic of one specific virus that was being tested for by everyone, maybe just as many people have been getting long lasting lung problems from infections but can't tie it to anything specific, or those that can aren't really being focused on in the media or research.

Jonathan Edwards said:
Chlamydia, salmonella and shigella all trigger long term arthritis in people with HLA-B27.
Click to expand...
So yes, things like that are interesting. But what insights can we gain from that if we don't know whether it just allows chlamydia to replicate more and make a person more sick or whether it does something more unique to arthritis. My point is, if we can control for that to narrow down where in the casual pathway a gene fits, why not?
 
Practically, if you're just using any and all ME/CFS cases, there's not really any way to control for severity since everyone has had lots of different infections.

So this doesn't really apply, and maybe it would still just be better to use the heterogenous cohort to not get microbe-specific genes.

But if already doing long COVID-specific ME/CFS, controlling for severity seems valuable.
 
Utsikt said:
Wouldn’t that depend on the question you’re trying to answer?

To my mind, if it’s «what gives rise to ME/CFS?», severity might not matter, because we’d still have to explain the ME/CFS cases in the other group(s).
Click to expand...
Yeah, there are various questions, and why the best approach would just be to record that data on severity, but not perfectly match groups. You can do an uncontrolled analysis, but also do an analysis with severity as a covariate.
 
forestglip said:
Yeah, there are various questions, and why the best approach would just be to record that data on severity, but not perfectly match groups. You can do an uncontrolled analysis, but also do an analysis with severity as a covariate.
Click to expand...
That would probably be the best, gather the data and analyse it from all angles.

Would that affect how the corrections for multiple testing is done, and potentially obfuscate leads?
 
I think it might be useful if it is possible to create a big cohort that meets those criteria, but ultimately it would only address one or two small facets of "susceptibility." That term would include several possible factors, such as:
1) frequency of contracting any infection (or a specific infection)
2) clinical severity of acute illness (which, for like 90% of studies is simply defined by hospitalization for acute respiratory distress)
3) severity as defined by any other metric (viral titres, duration of illness, presence/intensity of other symptoms, all of which might involve vastly different pathways)
4) in the case of EBV, susceptibility to latent infection, separate from acute infection
5) susceptibility of certain cell types/organs to infection (i.e. we know that SARS-CoV-2 can infect the muscle and brain, but don't fully know the extent to which it does so between individuals)
6) relatedly, whether the infection spreads beyond the respiratory tract (which could be more of a function of the robustness of local immune response at the site of infection rather than susceptibility of other cell types)

So you might be able to account for some of these in some limited capacity via cohort selection, but are still only indirectly addressing part of the story. Probably the more feasible route is to compare top GWAS hits to other studies which did look at more specific phenomena, or if possible, functional studies to see whether the relevance of a particular mutation can be explained by one of these mechanisms.
 
forestglip said:
So yes, things like that are interesting. But what insights can we gain from that if we don't know whether it just allows chlamydia to replicate more and make a person more sick or whether it does something more unique to arthritis.
Click to expand...

The insight is that the problem is almost certainly mediated by non-B lymphocytes - either CD8 or NK. If you add in the fact that the people with the same illness have a risk gene encoding for a common cytokine receptor used by T cells you are pretty sure he problem is a general overactivity of a T cell subset. Which gives confidence in using things like anti-IL17, which I am fairly sure turns out to work very well. If you know enough to abolish symptoms you might want more out of curisity but you have solved the problem.

I think there is a high chance that something like B27 will be altering a threshold in ME/CFS patients so that they are more susceptible to getting the illness with a range of triggers. That is all we need to know to home in an a solution
 
You must log in or register to reply here.
Back
Top Bottom