UK Grand Challenge

[Seanko]

On the back of the UK CMRC 2015 conference, its board approved the plans for the implementation of the Grand Challenge at its November 2015 meeting. It would look at genomics, epigenomics and other omics

This national study will collect more than 10,000 samples of data an focus on phenotyping and subtyping data using statistical analysis, bringing in genomics and other areas. Prof George Davey-Smith (of Bristol University) is very keen to be involved.

Decisions must still be made about:

• whether the Grand Challenge will study adults, or children, or both
• collecting the right type of clinical data
• how to phenotype patients and what objective measurements to use (eg. physiology, known immunology measures, sleep, cognition, psychological, post-exercise fatigue, pain).
• what data is used (new versus previously collected) and how to ensure standard operating procedures.

The level of expertise needed will need to be high and across a range of areas to ensure we collect good clinical phenotypes. Funding must be secured if the Grand Challenge is to become a reality.

The Wellcome Trust has a new collaborative awards scheme, which funds collaborative projects of this nature up to £4m, providing each collaborator plays a proactive part within the project.

A Grand Challenge meeting is being convened in spring 2016 and an outline application could be considered for September. Additional funders could also be considered.
Source: page 37 CMRC 2015 Conference Report

 

[Seanko]

ie It is the next big funding drive for the CMRC in the UK

 

[Cort]

On the back of the UK CMRC 2015 conference, its board approved the plans for the implementation of the Grand Challenge at its November 2015 meeting. It would look at genomics, epigenomics and other omics

This national study will collect more than 10,000 samples of data an focus on phenotyping and subtyping data using statistical analysis, bringing in genomics and other areas. Prof George Davey-Smith (of Bristol University) is very keen to be involved.

Decisions must still be made about:

• whether the Grand Challenge will study adults, or children, or both
• collecting the right type of clinical data
• how to phenotype patients and what objective measurements to use (eg. physiology, known immunology measures, sleep, cognition, psychological, post-exercise fatigue, pain).
• what data is used (new versus previously collected) and how to ensure standard operating procedures.

The level of expertise needed will need to be high and across a range of areas to ensure we collect good clinical phenotypes. Funding must be secured if the Grand Challenge is to become a reality.

The Wellcome Trust has a new collaborative awards scheme, which funds collaborative projects of this nature up to £4m, providing each collaborator plays a proactive part within the project.

A Grand Challenge meeting is being convened in spring 2016 and an outline application could be considered for September. Additional funders could also be considered.
Source: page 37 CMRC 2015 Conference Report

Whoa! We're getting some good research going! 10,000 data points. If they can't dig subsets out of that I don't if they ever will. I recently talked to Ron Davis - he said we are digging too deep not to find something - and soon!

Congrats to the Wellcome Trust. i have a lot of admiration for the very active ME groups over in the UK. (Still gotta do that MERUK blog )

 

[Simon]

Thanks, Cort

There's a little more about the Grand Challenge, and quite a lot more about George Davey Smith and omics approaches in my recent blog The Power and Pitfalls of Omics: George Davey Smith’s storming talk at ME/CFS conference (Pt 1 of 2)
(am I allowed to plug that here?)

I very much hope more news about the Grand Challenge will emerge soon

"Decisions must still be made about:

• whether the Grand Challenge will study adults, or children, or both"

I hadn't considered it might include children.
But perhaps the biggest call of all will be how they define patients - which case definition? Will they wind up recruiting mainly Oxford-criteria patients? I guess that will be one of the main questions for the researchers to consider when they meet up in April

(hey, that's my first post here)

 

[Seanko]

@Simon please feel free to add in the detail. I just copied & posted the source material

My one concern is that Esther Crawley fancies a chunk of the action with psychological phenotypes...

 

[Seanko]

At end of 2015 Crawley & Peter White published a paper in the Journal of Psychosomatic Research (!!!) on psycho-social phenotypes along with Peter White & Dutch researchers.

http://www.jpsychores.com/article/S0022-3999(15)30028-3/abstract

I am sure we would prefer other scientists to have a say in how patients are grouped in terms of physical symptoms such as PEM & orthostatic intolerance and using appropriate patient definition criteria.

 

[Cort]

Thanks, Cort

There's a little more about the Grand Challenge, and quite a lot more about George Davey Smith and omics approaches in my recent blog The Power and Pitfalls of Omics: George Davey Smith’s storming talk at ME/CFS conference (Pt 1 of 2)
(am I allowed to plug that here?)

I very much hope more news about the Grand Challenge will emerge soon

"Decisions must still be made about:

• whether the Grand Challenge will study adults, or children, or both"

I hadn't considered it might include children.
But perhaps the biggest call of all will be how they define patients - which case definition? Will they wind up recruiting mainly Oxford-criteria patients? I guess that will be one of the main questions for the researchers to consider when they meet up in April

(hey, that's my first post here)

Of course you are allowed to plug it here . I had missed your blog....Thanks for posting it here. I hope they finally dump that stupid Oxford criteria - that would say something, huh?

 

[Cort]

At end of 2015 Crawley & Peter White published a paper in the Journal of Psychosomatic Research (!!!) on psycho-social phenotypes along with Peter White & Dutch researchers.

http://www.jpsychores.com/article/S0022-3999(15)30028-3/abstract

I am sure we would prefer other scientists to have a say in how patients are grouped in terms of physical symptoms such as PEM & orthostatic intolerance and using appropriate patient definition criteria.

I think these guys are on the losing end. If Davis right - if we are digging too deeply into the molecular basis of this disease not find significant problems - which he is starting to find - it's just a matter of time before White and his ilk fade away.

 

[Simon]

@Simon please feel free to add in the detail. I just copied & posted the source material

My one concern is that Esther Crawley fancies a chunk of the action with psychological phenotypes...

There will be a lot of researchers from a lot of different disciplines in the Grand Challenge, most biomed. I have no problem collecting psychological data alongside a load of biomedical data and seeing where that leads. I think the main idea is that phenotypes will emerge from the data; when that data is rich in biomedical detail I think the findings could be pretty interersting

 

[IrisRV]

There will be a lot of researchers from a lot of different disciplines in the Grand Challenge, most biomed. I have no problem collecting psychological data alongside a load of biomedical data and seeing where that leads. I think the main idea is that phenotypes will emerge from the data; when that data is rich in biomedical detail I think the findings could be pretty interersting

If they are using the Oxford Definition, there is likely to be one or more purely psychological phenotypes. In the best of all possible worlds , this could lead to proving the Oxford definition pulls in psych-only patients along with ME patients and is therefore a bad definition. That would require that the researchers identify which definitions each patient fits and publishes that data. How likely is that to happen?

Does anyone know the position of these researchers on open data? Do they intend to release their data?

 

[IrisRV]

Having thought about this over the course of the day, I'm becoming more concerned about them collecting psych data if they're using the Oxford definition. Well, I'm concerned about them using Oxford at all. They could easily have 90% psych patients and 10% ME patients in their cohort. They will then "prove" that cfsme patients have predominently psych symptoms and don't show biomedical issues at any statisticlly significant level. This could one giant cluster f**k.

Those sneaky, er, people could be intending this very result. Identify a non-representative cohort, don't find the expected abnormality, and claim loudly through the media that you've just proved ME patients have no biomedical abnormalities. That could be disasterous for patients. Our only hope would be for the NIH to say much more loudly than they have that "they don't have the skillset".

Please, please tell me they're not using Oxford or any other sneaky tricks to gather a non-representative cohort.

 

[Simon]

Most omics researchers seem to be in open data so I hope that will apply to this study too

could easily have 90% psych patients and 10% ME patients in their cohort. They will then "prove" that cfsme patients have predominantly psych symptoms and don't show biomedical issues at any statisticlly significant level

OK, so assume worse case scenario they do have a 90% Oxford vs 10% CCC split. The 10k sample would be big enough to show a clear and different subgroup for the CCC, assuming they are different at the biomedical level (which I do).

But at that point the study would have a big problem, because with 'only' a thousand "ME" patients they would have limited power to subgroup the CCC group.

 

[Seanko]

@Simon believe there is a meeting IN Bristol about the Grand Challenge inviting interested parties in April?

Do you have any info & what institutions might be interested in being involved? (if this is not subject to embargo)

 

[Simon]

My understanding is that information will start to emerge once the April meeting has been held and the make up of the group has been finalised. But I understand they will be attracting mostly researchers new to mecfs, with big reputations in their own fields.

 

[IrisRV]

Most omics researchers seem to be in open data so I hope that will apply to this study too

I hope so to. I wish we had more than hope to rely on. Fingers crossed, I guess.

OK, so assume worse case scenario they do have a 90% Oxford vs 10% CCC split. The 10k sample would be big enough to show a clear and different subgroup for the CCC, assuming they are different at the biomedical level (which I do).

True, but if they have 90% non-patients and 10% patients, the conclusions of the papers are going to be based on the 90% -- CF from psych issues, CF from stress, etc. The 10% results will likely be considered just noise.

But at that point the study would have a big problem, because with 'only' a thousand "ME" patients they would have limited power to subgroup the CCC group.

Indeed.

I wish this group was as open to our concernes as the NIH team was. Some obvious and easy adjustments (don't use Oxford at all, for example) could make a huge difference in the quality of the research.

At this point I don't know if there's anything we can do but wait and hope for the best -- and I am hoping for the best, because otherwise the consequences could be disasterous.

If this research goes well (from our perspective, not theirs), it could gather huge amounts of valuable data. That's something we certainly need.

 

[Simon]

True, but if they have 90% non-patients and 10% patients, the conclusions of the papers are going to be based on the 90% -- CF from psych issues, CF from stress, etc. The 10% results will likely be considered just noise.

I don't think that would happen, not least because the Grand Challenge is being set up specifically too look for sub-groups. And a 1k subgroup would rise well beyond statistical noise: that's a big reason for planning such huge samples. On the other hand, if it were 90% Oxford-not-other, I would see that as wasting a lot of money

I wish this group was as open to our concernes as the NIH team was. Some obvious and easy adjustments (don't use Oxford at all, for example) could make a huge difference in the quality of the research.

I hope there will be a respectable Oxford-only group in the study: I think the Oxford-only vs other mecfs groups could throw a lot of light on what causes mecfs - though I suspect even a CCC group won't be homogenous. If breast cancer, asthma and diabetes type 1 are all mixes of quite different illnesses, I would expect something as broad as mecfs to be too.

If this research goes well (from our perspective, not theirs), it could gather huge amounts of valuable data. That's something we certainly need.

Yes, it could be very good indeed. There's nothing else on this scale in the pipeline, as far as I know

Added: yes, shame no patient respresentation: I think that would add a great deal to the study

 

[IrisRV]

Do we know if they intend to subgroup their their patients according to definitions -- Oxford, Fukuda, CCC, ICC, and publish the differences in results between these groups? That would be fantastic information. That would make it easier for researchers to rule in and rule out historic studies when doing background research to develop their hypotheses. For example, if the Oxford group doesn't have significant (or specific) immune abnormalities but CCC patients do, then researchers intending to study immune issues in ME would know to disregard research based on Oxford criteria. The same could be true (theoretically) for policy-making agencies. Immune treatments could be deemed suitable for patients diagnosed by CCC, but not by those diagnosed by Oxford.

OTOH, if they are not subgrouping by definition, their data could just look muddled.

How many patients are they studying, 1000? While 10,000 data points is a fair amount of data, if it's only 10 data points per patients and some of those data points are subjective psych information or even subjective reports of symptoms, then it really isn't enough data to draw substantial conclusions.

how to phenotype patients and what objective measurements to use (eg. physiology, known immunology measures, sleep, cognition, psychological, post-exercise fatigue, pain).

If they use all 7 of these measures (I'd like to hear how they are going to manage objective psychological measures), then that's less than 2 data points per area of interest. And where are the -omics in this list?

Can cognition be adequately measured with 1 or 2 data points? There are several indentified cognitive dysfunctions already noted -- short term memory, processing speed, focus and attention problems, word-finding, etc).

It's probably even worse where immunology is concerned. My specialist takes far more immune data points than that to get the full picture -- more like 20-30 just in immunology -- proinflammatory cytokines, anti-inflammatory cytokines, NK cell number, NK cell function, CD8/CD4/CD3 cell numbers and more. How can you get any decent picture of immune function in patients with only one or two data points?

Let's not even talk about physiology, which is a very broad field. Even your standard CBC test has more than 2 data points.

So do they have fewer than 1000 patients (in which case the possible 10% ME patients could be two or three people -- easily dismissed as outliers) or do they have a reasonable number of patients, given their lousy definition, but take so little data from each that they could miss almost everything? Neither sounds good to me. Or am I completely misunderstanding their plan? Are they talking about taking 10,000 data points on each patient? Now that would be something!

This would be better if they were using CCC-defined patients, assuming they know how to distinguish PEM from other forms of fatigue/symptom exacerbation (which I'm not confident they do). At least then a much larger number of their cohort would be people with ME, so the data would be a bit clearer. They could learn something from the NIH study which proposes (if I understand correctly) to use patients who fit all the criteria (except Oxford which is a worthless definition as the IOM pointed out).

It all sounds fine until you start looking at the details.

So, there are still a lot of unanswered questions. Hopefully a lot of this will get straightened out before the research starts.

 

[Cort]

I don't think that would happen, not least because the Grand Challenge is being set up specifically too look for sub-groups. And a 1k subgroup would rise well beyond statistical noise: that's a big reason for planning such huge samples. On the other hand, if it were 90% Oxford-not-other, I would see that as wasting a lot of money

I hope there will be a respectable Oxford-only group in the study: I think the Oxford-only vs other mecfs groups could throw a lot of light on what causes mecfs - though I suspect even a CCC group won't be homogenous. If breast cancer, asthma and diabetes type 1 are all mixes of quite different illnesses, I would expect something as broad as mecfs to be too.


Yes, it could be very good indeed. There's nothing else on this scale in the pipeline, as far as I know

Added: yes, shame no patient respresentation: I think that would add a great deal to the study

I agree Simon. Let's get an Oxford only subgroup in there and get some clarity on that definition and who fits it. That would be fantastic...That could throw a new light on so many CBT/GET findings over the years.

 

[Cort]

My understanding is that information will start to emerge once the April meeting has been held and the make up of the group has been finalised. But I understand they will be attracting mostly researchers new to mecfs, with big reputations in their own fields.

Good news! Let's get those researchers into the field and keep them in the field with good study results. Use good criteria to separate out the patients properly and then let prominent researchers have at the data.

My guess is that those types of researchers are drawn to big sample sets with a lot of data. If you gather it - they will come...

 

[Cort]

Having thought about this over the course of the day, I'm becoming more concerned about them collecting psych data if they're using the Oxford definition. Well, I'm concerned about them using Oxford at all. They could easily have 90% psych patients and 10% ME patients in their cohort. They will then "prove" that cfsme patients have predominently psych symptoms and don't show biomedical issues at any statisticlly significant level. This could one giant cluster f**k.

Those sneaky, er, people could be intending this very result. Identify a non-representative cohort, don't find the expected abnormality, and claim loudly through the media that you've just proved ME patients have no biomedical abnormalities. That could be disasterous for patients. Our only hope would be for the NIH to say much more loudly than they have that "they don't have the skillset".

Please, please tell me they're not using Oxford or any other sneaky tricks to gather a non-representative cohort.

I cannot believe that this group would countenance using only Oxford patients. I can see, however if it did, an ME/CFS cohort springing up in the data. I would assume that depressed patients and people with ME/CFS would separate out. There's little to suggest that they wouldn't. I can't remember a single study in which the presence or absence of depression had a significant on the study results.