UK Genome Wide Association Study (GWAS) project - draft website goes live, feedback sought on recruitment plan, and updates

[Adrian]

I also do not think that there is any particular value in this cohort for any other research. I think it would be a mistake to think that once this cohort is collected you can do much more. This is a unique situation where a very large sample is needed for the statistical reasons associated with GWAS. Any other study ought to use a better documented cohort derived on a population basis.

I think there is value in that this cohort could be used to help select smaller sets. This may be particularly the case if there are potential subtypes due to different genes which could lead to researching into different mechanisms.

Or be used for epidemiology studies (such as the one norwegen one that looked at what age people were first ill).

 

[Jonathan Edwards]

I think there is value in that this cohort could be used to help select smaller sets. This may be particularly the case if there are potential subtypes due to different genes which could lead to researching into different mechanisms.

Or be used for epidemiology studies (such as the one norwegen one that looked at what age people were first ill).

The key problem is denominators. I had significant reservations about using any population recruited through the internet for any purpose originally but I was persuaded that for this particular task an internet based sample, if it was big enough, would be of value. For most other studies where you need large numbers I think the denominators need to be known. Certainly for an epidemiological study you need to know the size of population you are sampling from and the weighting in sampling.

It is a common misconception that gathering cohorts for one study will help with all sorts of other studies. The King's Twin Project is maybe an unusual example where it makes sense, since genetic studies of almost anything can use twins.

I agree that it seems a pity not to make use of people who have already volunteered, but privacy and other issues come so high on the agenda that I think it should be a very secondary consideration. It is unlikely that a sample of this size is going to be needed again. Most spin off physiological studies would need no more than 100 individuals. If 5,000 show interest within a fortnight for this then re-finding 100 should not be so hard.

 

[JemPD]

I agree that it seems a pity not to make use of people who have already volunteered, but privacy and other issues come so high on the agenda that I think it should be a very secondary consideration.

Agreed.
What would be truly terrible is if we were unable to recruit sufficient numbers for this study because of privacy concerns introduced due to enthusiasm for the possibility of other, as yet unknown, unspecified & unplanned studies. ISTM that the critical thing is to ensure we get enough numbers to make this study viable. 25,000 people to volunteer is going to be a huge ask & we need to be careful that we don't sacrifice the number one goal for the sake of goals 2,3,4 whatever.

I know BPSers will likely take the finished anonymised data - eg 'X% of PwME have xyz gene', & twist it to mean whatever they want. Cant do much about that & an acceptable risk for the possible gain I think. But if there is any chance of them getting their hands on my saliva and or my name & contact details, then I wouldn't feel safe to participate.

 

[Saz94]

In an ME facebook group that I'm in, somebody has asked "why can't they use the raw data from the DNA tests that some of us have already paid for? The OMF does." How should I answer that? I know why not, but can't think how to articulate it... damn brain fog.

 

[Amw66]

In an ME facebook group that I'm in, somebody has asked "why can't they use the raw data from the DNA tests that some of us have already paid for? The OMF does." How should I answer that? I know why not, but can't think how to articulate it... damn brain fog.

Data varies as chip versions have changed over time and so gene content has changed

May not have particular Gene's as most are selective ( unless you have a full genome) .

OMF do not use 23andme which is probably the widest consumer accessible provider. ( photo posted last year showed gene kits on a desk and i' m sure that they were not 23andme)

Klimas has I think and some of the papers published have been discussed here. Beyond my knowledge, but people don't seem to have been impressed.

 

[TrixieStix]

Data varies as chip versions have changed over time and so gene content has changed

May not have particular Gene's as most are selective ( unless you have a full genome) .

OMF do not use 23andme which is probably the widest consumer accessible provider. ( photo posted last year showed gene kits on a desk and i' m sure that they were not 23andme)

Klimas has I think and some of the papers published have been discussed here. Beyond my knowledge, but people don't seem to have been impressed.

I would imagine the fact that consumer DNA tests are thought to misread & misreport a not insignificant number of SNPs would also be a barrier.

 

[Andy]

A general reply regarding some of the latest issues/concerns raised in this thread - I don't think it is practical reply to each post individually.

I think that the issues raised (including concerns about consent/data sharing/re-contacting, validity of the planned large re-contactable cohort, and the question of using commercial genetic testing) are probably best addressed by the actual scientists. However, there will be a delay on this, as a lot of work, mainly by them, is currently going into writing, amending, re-writing, correcting and re-re-writing, the application for funding, so I would ask for everybodies patience at this time - good questions have been asked, valid points have been raised, and we will address them, but we need to focus on the application first, at this time.

 

[JemPD]

Happy to see them prioritising @Andy happy to be patient.

please confer gratitude & encouragement to them

 

[rvallee]

In an ME facebook group that I'm in, somebody has asked "why can't they use the raw data from the DNA tests that some of us have already paid for? The OMF does." How should I answer that? I know why not, but can't think how to articulate it... damn brain fog.

Those are only partial tests, which is why they are cheaper than a full genome reading. They only look at genes known to be significant in some diseases, when the purpose here is to find among all the genes which ones are significant to ME.

 

[InitialConditions]

Those are only partial tests, which is why they are cheaper than a full genome reading. They only look at genes known to be significant in some diseases, when the purpose here is to find among all the genes which ones are significant to ME.

The whole process needs to be standardized too - you can't just bring together data from different tests.

 

[Andy]

Update: the FAQ has been revised and reorganised, https://mebiomed.org.uk/faqs/

As always, if you spot anything that you believe is incorrect and/or have suggestions on how it could be improved then please let us know.

 

[Robert 1973]

From the FAQs:

Can people who have improved or recovered participate?

Those who pass the CCC or IOM criteria can participate. Following these criteria to the letter (as we will do) unfortunately means that those who are fully recovered will not be able to participate. This is an unfortunate but inevitable consequence of the need for us to fully comply with these criteria.

I don’t understand why people who have recovered aren’t being included. Understanding why some people seem to recover and others do not is hugely important. Presumably, it could be that some people have genes which increase the probability of recovery, or it could be that the people who recover have a different illness or a different type of the same illness. I would have thought that these are exactly the sort of questions that a GWAS could answer.

I understand the need to recruit a minimum number of participants who fulfil the CCC or IOM diagnostic criteria, but I don’t understand why an additional group of recovered patients couldn’t be included. Is it a cost issue, or is there another reason for the decision? Is it too late to change the proposed design?

It would be frustrating if the GWAS managed to identify distinct sub-groups but was unable to tell us whether any of those sub-groups were more or less likely to recover or significantly improve in time, without treatment.


NB I’ve not managed to follow all the discussions, so apologies if this has already been discussed. If it has, please can someone post a link? Thanks.

 

[Hoopoe]

Maybe there are concerns about the reliability of retrospectively applied diagnoses.

 

[Robert 1973]

Maybe there are concerns about the reliability of retrospectively applied diagnoses.

I guess there may be but presumably there will be concerns about the reliability of current diagnoses too – or rather, the reliability of the answers to questions which are used to determine whether or not someone fulfils the diagnostic criteria.

I assume that answering questions about historic symptoms is likely to be a bit less reliable than answering questions about current symptoms but I don’t know how much that would matter. My limited understanding is that for a GWAS it doesn’t particularly matter how many unreliable diagnoses there are, provided there are sufficient numbers of reliable diagnoses (however a reliable diagnosis is defined).

Also, let’s assume that a separate, additional cohort of recovered patients is recruited – people who have previously been diagnosed with ME/CFS, whose answers to questions about their symptoms prior to recovery meet the study’s criteria for a diagnosis of ME/CFS. Now let’s assume that the GWAS identifies a number of SNPs that are significantly more common in people who currently meet the ME/CFS diagnostic criteria than in the healthy controls. Whether or not those SNPs are equally common in the recovered cohort would tell us something useful about people who report that they have recovered from ME/CFS, regardless of how reliable their answers are. And that is something I would be very interested to know.

Having said that, looking at the number of likes that my comment has received compared to strategist’s reply I’m wondering if I’m misunderstanding something.

 

[ringding]

I wouldn't read too much into numbers of likes.
Just over a year ago I'd have been in the "almost recovered" category. Now I'm most definitely ill. I'm also not sure of why the "phasing" of symptoms, if remitting/relapsing, should impact on suitability for inclusion. Is anyone ever really recovered?
Surely if someone has had ME of any severity their genetic information would be of use?

 

[Jonathan Edwards]

I don't understand why the questionnaire cannot ask whether either now or in the past the person has fulfilled the various criteria.

CCC and IOM criteria do not have to be applied in the present tense. They remain the same criteria applied in the past tense.

The past might actually be more reliable than the present since PWME may well not know exactly how things are in the present if they are pacing effectively.

 

[Sisyphus]

I don't understand why the questionnaire cannot ask whether either now or in the past the person has fulfilled the various criteria.

CCC and IOM criteria do not have to be applied in the present tense. They remain the same criteria applied in the past tense.

The past might actually be more reliable than the present since PWME may well not know exactly how things are in the present if they are pacing effectively.

“PWME may well not know exactly how things are in the present if they are pacing effectively.”

There’s a substantial amount of significance in this one sentence. i) I have learned, at considerable cost, that feeling better for a month does not mean I am either free of or safe from the disease. ii) not only pacing of activity, but unknown factors which may relax for a time may give one a partial respite. That improvement comes to a crashing end not only from over-activity but from some trivial stressor (heat, a common cold, sleep disruption, etc) or for no observable reason. There are certainly unknown body processes involved, there could be stochastic processes triggering a relapse.

 

[Snow Leopard]

From the FAQs:
I don’t understand why people who have recovered aren’t being included. Understanding why some people seem to recover and others do not is hugely important. Presumably, it could be that some people have genes which increase the probability of recovery, or it could be that the people who recover have a different illness or a different type of the same illness. I would have thought that these are exactly the sort of questions that a GWAS could answer.

Agreed, this seems like a serious oversight. Those in remission are a notable subset.

 

[Amw66]

I guess there may be but presumably there will be concerns about the reliability of current diagnoses too – or rather, the reliability of the answers to questions which are used to determine whether or not someone fulfils the diagnostic criteria.

I assume that answering questions about historic symptoms is likely to be a bit less reliable than answering questions about current symptoms but I don’t know how much that would matter. My limited understanding is that for a GWAS it doesn’t particularly matter how many unreliable diagnoses there are, provided there are sufficient numbers of reliable diagnoses (however a reliable diagnosis is defined).

Also, let’s assume that a separate, additional cohort of recovered patients is recruited – people who have previously been diagnosed with ME/CFS, whose answers to questions about their symptoms prior to recovery meet the study’s criteria for a diagnosis of ME/CFS. Now let’s assume that the GWAS identifies a number of SNPs that are significantly more common in people who currently meet the ME/CFS diagnostic criteria than in the healthy controls. Whether or not those SNPs are equally common in the recovered cohort would tell us something useful about people who report that they have recovered from ME/CFS, regardless of how reliable their answers are. And that is something I would be very interested to know.

Having said that, looking at the number of likes that my comment has received compared to strategist’s reply I’m wondering if I’m misunderstanding something.

If this is, as multiple reports suggest, a relapsing and remitting condition, then " recovered" genetic info may play an important role.

 

[MSEsperanza]

I think it could be very helpful if someone were able to answer the concerns expressed in the comments on the Virology Blog article, in particular provide more details about the diagnostic / selection criteria.

If these details aren't clear yet (according to the FAQ IOM or CCC or Fukuda (*) will be used), perhaps it could be added to the FAQ (explicitly at each question regarding diagnosis) that he project team is still working on this, incorporating feedback, also on how to ensure that PEM isn't understood as just feeling bad after exertion.

(*) [edit: this has now been answered, they just forgot to remove Fukuda in the FAQ update (Q: "What case definitions will be used?") see Andy's reply.]


Q: "How will the study ensure...participants really do have ME/CFS?"

• "We take diagnosis very seriously. You will be diagnosed using the CureME patient questionnaire that has been in use for the UK ME/CFS Biobank for several years."

It seems the questionnaire isn't available on the CureME website. Is this intended?

• "a GWAS does not need 100% diagnostic accuracy to produce valid results."

This seems a very relevant info. I think it would be good to link to a source that explains why.

• "as part of the study, some patients will have an assessment by a clinician with knowledge of ME/CFS to confirm the diagnosis is accurate".

Could be re-assuring to know what percentage of participants is intended to be assessed by a clinician and how this will be done (selection, home visits?).

(And don't know but perhaps better to write "participants" instead of "patients"?)