What would successful brainstorming about ME/CFS genes look like?

[Kitty]

We would need to be very careful but it may be that we can build a prima facie case for the following having no specific role in ME/CFS:

TNF
IFN alpha beta gamma
IL-6
IL-17
Collagen fibre integrity
Mast cells and tryptase
Specific antigen presentation on Class I and II
Various mitochondrial enzymes
And so on

Might there be ways to drill down further into the data with the objective of trying to exclude things, or would we need to wait for a SequenceME-type project that will look at all the variants including the rarer ones?

 

[Jonathan Edwards]

That's an ill-advised leap to make,

Whatever leap that was, it was yours, @jnmaciuch, not mine. I feel my way through these things and it has done me very well in the past. Everything is up in the air but unless we bounce around off all these issues we end up stuck in the same old theory we had yesterday.

 

[Jonathan Edwards]

Might there be ways to drill down further into the data with the objective of trying to exclude things, or would we need to wait for a SequenceME-type project that will look at all the variants including the rarer ones?

It will be hard to exclude anything for certain but if the ensign is pointing south the wind is usually coming from somewhere near north. I think there is probably a mine of negative information in DecodeME that with care could prevent us from wasting more time of theories that are never going to fly.

 

[SNT Gatchaman]

If symptoms are mediated by interleukins, TNF or interferons then why aren't receptor genes influencing presentation of the disease?

We still have the X chromosome data to come and there are some receptors there: IL-2R and some other IL-nRs.

The X chromosome has other potentially rich pickings eg: neuroligins, various zinc finger proteins, some ubiquitins, TLR-7, synaptophysin, pyruvate dehydrogenase, some NADH dehydrogenase subunits, some HDACs, ABCD1, AIFM1, androgen receptor, ACE2, vasopressin receptors, Ephrin B1.

 

[forestglip]

I think one potential way to find evidence against concepts is a mendelian randomization approach. Not the questionable versions we see pop up here where they just take SNPs associated with eating breakfast or whatever. But the good MR where you take a mutation definitively known to increase or decrease a cytokine.

For example, mendelianrandomisation.com gives the example of mutations in CRP known to cause increased CRP. So if your hypothesis is that more CRP leads to ME/CFS, you check if people with ME/CFS tend to more often have the specific mutation that increases CRP. If not, that's a bit of evidence against that hypothesis.

 

[Jonathan Edwards]

For example, mendelianrandomisation.com gives the example of mutations in CRP known to cause increased CRP. So if your hypothesis is that more CRP leads to ME/CFS, you check if people with ME/CFS tend to more often have the specific mutation that increases CRP. If not, that's a bit of evidence against that hypothesis.

Yes, very much what I was thinking of.

 

[forestglip]

Yes, very much what I was thinking of.

I'd just say not to put to much stock in such mutations not being genome-wide significant hits because that's quite a high bar to clear. Hopefully, someone who knows how to do it properly digs into the specific mutations.

The following is not definitive. I barely know what I'm doing and could be doing something wrong. More just playing with the idea of what might be possible.

That said, I looked up two SNPs known to change the amount of CRP in the DecodeME summary stats:

The observed association of two SNPs, rs3093059 and rs1205, with circulating CRP is also reported by several previous studies22, 33,34,35,36,37.

The p-values aren't anything impressive.

 

[jnmaciuch]

Whatever leap that was, it was yours, @jnmaciuch, not mine. I feel my way through these things and it has done me very well in the past. Everything is up in the air but unless we bounce around off all these issues we end up stuck in the same old theory we had yesterday.

What I'm pointing out is that narrowing down options based on whether the central players came up in a genetics study is no better than asking a magic 8 ball if those pathways are involved in ME/CFS. The lack of primary genetic hits in a pathway does not conclusively [edit: or even reasonably] exclude a pathway any more than a magic 8 ball does, for the reasons I explained. Anyone is welcome to feel things out however they wish, but doing so in a discussion invites others to point out important fallacies in their logic.

 

[jnmaciuch]

I think one potential way to find evidence against concepts is a mendelian randomization approach. Not the questionable versions we see pop up here where they just take SNPs associated with eating breakfast or whatever. But the good MR where you take a mutation definitively known to increase or decrease a cytokine.

For example, mendelianrandomisation.com gives the example of mutations in CRP known to cause increased CRP. So if your hypothesis is that more CRP leads to ME/CFS, you check if people with ME/CFS tend to more often have the specific mutation that increases CRP. If not, that's a bit of evidence against that hypothesis.

Unfortunately I think that falls to the same issues that I already described--a mutation that increases or decreases expression of a cytokine permanently enough to be detected at the point of sample collection for a mendelian randomization is probably going to predispose to other more progressive and inflammatory diseases instead of ME/CFS. But that doesn't mean the cytokine isn't relevant in ME/CFS.

 

[forestglip]

Unfortunately I think that falls to the same issues that I already described--a mutation that increases or decreases expression of a cytokine permanently enough to be detected at the point of sample collection for a mendelian randomization is probably going to predispose to other more progressive and inflammatory diseases before it leads to ME/CFS. But that doesn't mean the cytokine isn't relevant.

I'm not sure I follow. You've got two identical people, except person A that you know is more likely to have a higher level of CRP than person B on average. If you think that CRP causes ME/CFS, then you'd expect person A to be more likely to have ME/CFS.

Person A might also be more likely to have lots of other things like cardiovascular disease too. I don't see why that invalidates the ME/CFS part.

 

[jnmaciuch]

I'm not sure I follow. You've got two identical people, person A that you know is more likely to have a higher level of CRP than person B on average. If you think that CRP causes ME/CFS, then you'd expect person A to be more likely to have ME/CFS.

Person A might also be more likely to have lots of other things like cardiovascular disease too. I don't see why that invalidates the ME/CFS part.

But the logic is not reciprocal for mendelian randomization--if the MR shows that having a higher level of CRP is associated with ME/CFS, that's a good reason to look into it as a causal agent. But it doesn't mean you can exclude CRP if there's no significance, because the relevant role of CRP in ME/CFS might have very little correlation with absolute levels at baseline or whatever proxy measurement is being performed in the study.

The relevance of CRP could be the particular interaction with another cytokine (which would be rate limited by the levels of the other cytokine much more than CRP), an inhibitory feedback loop involving 8 other intermediaries, or it might involve CRP at a point where CRP gets extremely elevated anyways due to a stimulus and the absolute levels at that peak don't vary much between people with and without the mutation.

Perhaps the absolute levels at baseline will correlate with whatever the relevant phenomenon is, but they just as well might not. I definitely understand the instinct to want to leverage genetics data as much as possible to hone in on viable hypotheses, but this would unfortunately be a "throwing the baby out with the bathwater" situation.

 

[forestglip]

if there's no significance, because the relevant role of CRP in ME/CFS might have very little correlation with absolute levels at baseline or whatever proxy measurement is being performed in the study.

Oh I agree with all that you said. I meant if the hypothesis is specifically, "if I increase your CRP, that will cause you to be more likely to have ME/CFS". Which admittedly is not accounting for a lot of CRP related hypotheses.

 

[jnmaciuch]

Oh I agree with all that you said. I meant if the hypothesis is specifically, "if I increase your CRP, that will cause you to be more likely to have ME/CFS". Which admittedly is not accounting for a lot of CRP related hypotheses.

Ah my bad I thought you were speaking more broadly, since the prior discussion was about making the assumption that certain pathways are not involved in ME/CFS. It could definitely be useful for hypotheses about the very thing being measured in the MR, but like you said, it would only be a small fraction of possibilities for the involvement of XYZ.