Pros and cons of genetic studies of families who have multiple members with ME/CFS

[Sasha]

On another thread we've been talking about whether it would be a good idea to do genetic analysis of familes with multiple members who have ME/CFS, particularly if several get it as children (which would indicate a strong genetic 'signal'.

Chris is clearly aware of the option. There may be major practical or statistical difficulties in setting up a good quality study of multi-case families. I would be interested to know what everyone's thoughts on that might be.

I thought the topic deserved its own thread so here it is!

 

[Jonathan Edwards]

My understanding is that GWAS data on families will not tell us more because that is a statistical game with common variants that are not likely to be causing disease in any direct way.

Although members of a family will share 50% of each others DNA my guess is that if rare variants turn up in multicase families they are much more likely to be telling us something interesting than in the studies we have so far. And we already have some clues from DecodeME that could be used for choosing a narrow range of genes.

 

[Utsikt]

How would you determine if the genes are relevant?

 

[Creekside]

Here's my thought process for the ME/genetics question. Imagine that the critical part of ME's mechanism is some structural component, such as astrocyte dendrites. Maybe it's the length or diameter--or the length/diameter ratio--or the number and length of microtubules in those. Those numbers might be affected by genes, but that might be a minor factor compared to your circulating nutrients on day 478 of your life, or whether you had a viral infection then or before or after, or your excitement level while that dendrite was having a growth spurt. So, it wouldn't be likely that the genetic contribution to ME would be found in a genetic study.

I have no reason to believe that ME does involve a structural component like that, but I'm pretty sure no one has measured such things in PWME vs (appropriately unhealthy) controls. Furthermore, it needn't affect all astrocytes (or whichever type of cell); it might be only a few special cells in a small part of the brain causing the ME effects, which would require a large and expensive study to discover. A study that only took a few samples from one part of the brains would conclude "No, that's not the problem."

 

[Utsikt]

So, it wouldn't be likely that the genetic contribution to ME would be found in a genetic study.

I don’t follow. If one gene makes it more likely that whatever ME/CFS is occurs (say fewer microtubules), then that will show up in a large enough genetic study because the people with that gene will be overrepresented in the ME/CFS group.

That’s different from this kind of study, though. I’m not sure how they would work.

 

[Creekside]

If one gene makes it more likely that whatever ME/CFS is occurs (say fewer microtubules), then that will show up in a large enough genetic study because the people with that gene will be overrepresented in the ME/CFS group.

It depends on how large the effect is and how large the effect of other factors is, and how many false positives (multiple PWME who share a gene, but due to a non-genetic common factor). If the study requires a billion PWME to overcome noise, it's going to be a long wait.

Also, if there is a genetic contribution, it might be fairly indirect. Maybe the gene makes people more likely to respond a certain way to a certain microbe or nutrient or hormone at a certain point in their development, which affects something else, which affects something else, which eventually affects the chance of developing ME. I'm not expecting these studies to turn up an ME-gene. I'm not expecting even a combination of genes with a strong correlation with ME. I do expect some correlation with some genes, which doesn't result in any further knowledge, simply because if you search for correlations in a large enough data set, you will find some, even if the data is purely random numbers. Redo with different samples and you'll get a different set of correlations.

People from the same family also share a lot of non-genetic factors, such as diet, environment, activity/sleep habits, etc. You might find some gene correlations between those members who developed ME, but their actual common factor was eating clams from a polluted bay.

If you had multiple families with many members having ME, then I'd agree that genetic studies of them might be worthwhile. I don't get the impression that there are many such families, and they might be victims of other commonalities, such as the previous example.

 

[Jonathan Edwards]

How would you determine if the genes are relevant?

I assume that you would look at their functions and make intelligent guesses much in the way that that has been done for the Zhang study and Liz Worthy's study. The advantage of multicase family studies is just that it seems likely that rare but heavily risk-bearing gene variants would show up more clearly in those.

 

[Utsikt]

It depends on how large the effect is and how large the effect of other factors is, and how many false positives (multiple PWME who share a gene, but due to a non-genetic common factor). If the study requires a billion PWME to overcome noise, it's going to be a long wait.

WGS’s don’t require billions of cases. It seems like you can get by with 10-15 thousand.

Also, if there is a genetic contribution, it might be fairly indirect. Maybe the gene makes people more likely to respond a certain way to a certain microbe or nutrient or hormone at a certain point in their development, which affects something else, which affects something else, which eventually affects the chance of developing ME.

All of those steps would also have genes influencing them, making it more or less likely that you end up with ME/CFS.

But I feel like we’re getting a bit off topic for the thread, which was small genetic studies on families, not large scale WGS. Perhaps the SequenceME threads are more suited.

 

[Amw66]

Perhaps it's a combination of gene expression that's common , not one single gene?

 

[arachel]

Single-family (pedigree) studies have been very useful in genetics historically, but these are typically useful for single-gene disorders. They aren't very useful for complex traits, or traits in which multiple genes cause the disease. Assuming ME/CFS is a complex disorder (which is where the evidence is generally pointing), single-family studies would not be particularly useful. However if it happens to be the case that there is a single gene that can cause ME/CFS, which is totally possible, this could potentially be useful.

 

[Jonathan Edwards]

Assuming ME/CFS is a complex disorder (which is where the evidence is generally pointing), single-family studies would not be particularly useful.

I am thinking of the example of lupus, which is probably as 'complex' as any disease. Identifying rare genes was very useful - particularly complement gene deficiencies. By about 1980 we knew that a tiny proportion of lupus cases could be entirely accounted for by individual complement gene defects - C1q to C3 components - and that C4 aberrations were far more common in lupus. That led to the Peters/Schifferli model of lupus as a process involving failure of intravascular complement-mediated clearance, which still stands.

Most lupus is not due to complement gene defects but that does not matter. If one in a hundred cases of ME/CFS were due to a single gene error then my guess is that that proportion would be enriched in multicase families (where cases are more likely to be monogenic or oligenic), maybe to one in twenty. If you studied 100 families then there would be a real chance of picking up several rare variants of the same gene group. And looking at those in detail might point directly to an interpretation. For the complement genes in lupus the risk is different for different genes along that C1-C9 cascade. C3 is critical and a defect of C3 produces an aberrant form of lupus. C5-9 defects do not produce lupus if I remember rightly. That allowed people to home in on the C1-C3 cascade involved in immune complex clearance but not extravascular membrane attack.