How to select trial participants with ME/CFS?

[jnmaciuch]

That sounds like something that would be very expensive in clinicians' time and home visits, so only able to be done with a very small sample for a student dissertation. How about making use of an ME/CFS biobank that has already collected both information about the pwME and blood samples? You could then presumably do a much larger study as all the work would be lab work and data analysis.

That would be great if possible, it would just be a matter of whether the data collected, on either the sample or patient reported outcome side, is sufficient since we truly don’t know what analytes or details will be predictive of subtypes. It would need to be a pretty wide and comprehensive net at first.

As I have it in my mind right now this would definitely be a small group and the work would be carried out almost entirely by me. However, if a distinct pattern emerges from that small group, I could potentially do some algorithmic feature selection to get a small, predictive list of features that could be used to identify those groups in larger cohorts.

At this point it’s more of a pipe dream since I would need Grant funding and other resources to put together (as well as more effort into study design with my PI), but I just wanted to put the idea out there to get feedback

 

[jnmaciuch]

It does sound complicated, and a lot will depend on what it is you're really trying to get at. It seems to be sort of about subgroups, but also about diagnostic challenges and the issue of bias in both patients and clinicians.

I'm not sure what the simple version of the question is ... but I realise it's only one idea, and you've only sketched it out in rough anyway!

Mostly it’s a matter of finding homogenous study groups in the absence of a clear marker which defines all ME/CFS (if that exists).

It makes it less likely that something which strongly effects a small portion of pwME is obscured by a very heterogeneous study group, and in the absence of a disease biomarker, could prevent accidental inclusion of people who have some other illness that sounds similar on the face but doesn’t present similarly on a broad molecular level

 

[Utsikt]

@jnmaciuch why not wait for the DecodeME results and see if that gives you something to take a closer look at?

 

[Kitty]

Mostly it’s a matter of finding homogenous study groups in the absence of a clear marker which defines all ME/CFS (if that exists).

It makes it less likely that something which strongly effects a small portion of pwME is obscured by a very heterogeneous study group, and in the absence of a disease biomarker, could prevent accidental inclusion of people who have some other illness that sounds similar on the face but doesn’t present similarly on a broad molecular level

[This is only meant as a response to your thoughts, I don't expect answers! ]

Yep, that's clearer. But how does the fact that we already have two distinctive and rather unusual symptoms—PEM and the urgent need to lie down—affect it?

Also, what would be the effect of a drug candidate coming along in the meantime? That is one way of confirming a diagnosis or a subgroup, albeit not the ideal approach.

 

[jnmaciuch]

@jnmaciuch why not wait for the DecodeME results and see if that gives you something to take a closer look at?

It’s likely that DecodeME results will be out before I have a chance to put this together anyways, so I definitely will.

However, it’s very likely that DecodeME won’t provide subtype information if it’s not something that is genetically encoded—the multi-omic data would incorporate things like metabolomics, lipidomics, transcriptomics, etc. that provide a snapshot of what active disease looks like, rather than predisposing factors.

DecodeME results may overlap with multi-omic profiles to some extent, but if you try to define subtypes based on genetic screens, you’re probably going to include a lot of people that have a genetic risk factor for the disease but not the actual disease.

 

[hotblack]

Yes I had this thought—it could be adapted for severe (though possibly not very severe) by being an interview over text or email with one question at a time.

I should have realised you would have That asynchronous nature I think works really well for me and others. And I’m absolutely sure there’s plenty of people here who would be happy to help give input or trial methods.

I’d be interested in if tools like slack or even simpler IM or forums could be used if that interview style with back and forth is needed.

Presumably you really need blood here? (I’m just thinking of the context of workload for those doing the study and comparing when I’ve had to be visited for blood samples to be taken compared to just posting off saliva samples)

 

[jnmaciuch]

[This is only meant as a response to your thoughts, I don't expect answers! ]

Yep, that's clearer. But how does the fact that we already have two distinctive and rather unusual symptoms—PEM and the urgent need to lie down—affect it?

Also, what would be the effect of a drug candidate coming along in the meantime? That is one way of confirming a diagnosis or a subgroup, albeit not the ideal approach.

I appreciate the questions since it helps me flesh out this idea more From seeing discussions on this site I think there actually is an important amount of heterogeneity even in those two symptoms.

E.g. people who get only delayed PEM 24 hours later vs. those who get symptoms within a few hours that last into the next day(s), whether or not someone has orthostatic intolerance, whether they are laying down because they feel sleepy or “fatigued” (and the ambiguity in those words as well), exactly what symptoms are associated with PEM (pain vs. no pain), etc.

In that respect, there’s going to be heterogeneity within pwME and a lot of overlap between what some subset of pwME describe and what some subset of people with another illness describe.

That overlap isn’t a problem for an unbiased clustering approach—it’s just looking for the most parsimonious way to create groups where the members resemble eachother more overall than people from other groups (as one example of an algorithmic approach).

And to your last point, this is something I imagine being complementary to ongoing research—if there is a drug trial, it would be incredibly informative if you observed that all the responders happened to be in (predetermined) subtype A, rather than claiming a subset after the fact with no data to back up what makes responders different from non-responders.

 

[hotblack]

To get back to the original question… wasn’t one of the points of DecodeME to get a cohort that can be used again? Many of us gave permission not only for data sharing but to be contacted by other researchers.

So at least in the UK shouldn’t the answer to the question how do we select participants be… contact DecodeME?

 

[jnmaciuch]

I should have realised you would have

Please keep pointing out problems! I may be better than your average clinician/researcher at spotting problems (it’s a low bar) by virtue of having the illness, but I’m still likely to miss things!

I’d be interested in if tools like slack or even simpler IM or forums could be used if that interview style with back and forth is needed.

That could work potentially! I think the main concern would be data privacy—if it was a series of pre-written prompts, they could be loaded up one at a time in a secure survey portal. Then the responses could be aggregated and given to the expert panel to do at their own time as well.

They could also be presented in a random order that way so if someone is no longer able to participate after a while, we still got a decent selection of answers. And having follow-up questions in this same format would allow someone else to curate them to remove repetitive questions from the experts. But if someone is having trouble writing out whole answers I could definitely chat with them and format their responses later.

Presumably you really need blood here? (I’m just thinking of the context of workload for those doing the study and comparing when I’ve had to be visited for blood samples to be taken compared to just posting off saliva samples)

Yes I think that would be the most feasible, it would allow you to do at least 3-4 multi-omics panels with the least amount of sample collection. Obviously you’re going to miss a lot of tissue-specific signals that way. But from my previous multi-omic experience you actually get more comprehensive information from the blood than you’d expect.

 

[Kitty]

whether they are laying down because they feel sleepy or “fatigued” (and the ambiguity in those words as well)

That's what I meant by unusual: people with what we call ME/CFS aren't driven to lie down for those reasons. They might choose to lie down for them, but that's not the same thing.

They're lying down because they have to, and although a strong need to lie down will occur in other chronic conditions, there's usually an explanation for it. People with ME/CFS describe a drive so urgent that lying down (or at least sitting down as reclined as possible) is non-negotiable.

E.g. people who get only delayed PEM 24 hours later vs. those who get symptoms within a few hours that last into the next day(s)

Yes, this might be important...but it might not. PEM's an interesting phenomenon and the time course is a core feature of it, but we run into problems when we try to define the delay too sharply. It's not necessarily consistent even in the same person (partly because there's rarely a clear beginning and end to whatever triggered it), and some of the differences may be due to illness severity.

I wonder if looking at PEM from the point of view of subgrouping might be a bit of a trap, a bit like trying to define the symptoms in a precise way. The harder you try to pin it down the more elusive it seems to get, yet it's still obstinately there. Obstinately and recognisably there.

 

[jnmaciuch]

Yes, this might be important...but it might not. PEM's an interesting phenomenon and the time course is a core feature of it, but we run into problems when we try to define the delay too sharply. It's not necessarily consistent even in the same person (partly because there's rarely a clear beginning and end to whatever triggered it), and some of the differences may be due to illness severity.

I wonder if looking at PEM from the point of view of subgrouping might be a bit of a trap, a bit like trying to define the symptoms in a precise way. The harder you try to pin it down the more elusive it seems to get, yet it's still obstinately there. Obstinately and recognisably there.

Those are all great points, though I wonder if that's something that would be addressed anyways through an in-depth interview. The point of categorization would be to see if people who share details of PEM (i.e. maybe there's a group who all describe PEM with muscle pain that starts a few hours after activity) also share other features of the illness.

"PEM" is how you'd refer to the phenomenon across pwME, but what you're more concerned with re: subtypes is whether the specific details look more similar within a subgroup of people.

And if there truly are no consistent patterns and it's a complete mix-and-match across the board, that would be good to know as well.

 

[EndME]

Those are all great points, though I wonder if that's something that would be addressed anyways through an in-depth interview. The point of categorization would be to see if people who share details of PEM (i.e. maybe there's a group who all describe PEM with muscle pain that starts a few hours after activity) also share other features of the illness.

"PEM" is how you'd refer to the phenomenon across pwME, but what you're more concerned with re: subtypes is whether the specific details look more similar within a subgroup of people.

And if there truly are no consistent patterns and it's a complete mix-and-match across the board, that would be good to know as well.

Perhaps a question for the historians:

Is there a comparable situation (e.g. a condition where by all accounts there are no specific signs amongst different groups pointing to different processes) where anybody was able to identify subgroups of an illness without identifing any pathological clues first?

 

[Kitty]

I wonder if that's something that would be addressed anyways through an in-depth interview. The point of categorization would be to see if people who share details of PEM (i.e. maybe there's a group who all describe PEM with muscle pain that starts a few hours after activity) also share other features of the illness.

Yeah, I can see that it's arguable both ways—are difficult phenomena best approached by dissecting them or looking for the right position to view them from? Maybe that's part of the worry.

Is this for a dissertation or a doctoral thesis, by the way? I don't know anything about the former, but I have worked with a couple of people doing the latter. It left me with all sort of impressions, but one of the clearest was that tracking down ghosts is asking for a lot of grief.

 

[jnmaciuch]

Yeah, I can see that it's arguable both ways—are difficult phenomena best approached by dissecting them or looking for the right position to view them from? Maybe that's part of the worry.

Is this for a dissertation or a doctoral thesis, by the way? I don't know anything about the former, but I have worked with a couple of people doing the latter. It left me with all sort of impressions, but one of the clearest was that tracking down ghosts is asking for a lot of grief.

It's a potential PhD thesis. My work would be focused on the multi-omic integration part, since that is what I've already been working on. The interview part would be, at best, a bonus showing that there's identifiable phenotypic differences associated with the molecular subtypes. That same idea could potentially be addressed with just a battery of regular surveys, but we all know the problems with those surveys.

[Edit: I'm still in the part of my PhD where I'm taking classes and supposed to be "learning the field," so this is me thinking a couple steps in advance]

 

[bobbler]

Perhaps a question for the historians:

Is there a comparable situation (e.g. a condition where by all accounts there are no specific signs amongst different groups pointing to different processes) where anybody was able to identify subgroups of an illness without identifing any pathological clues first?

I think it is quite common to be able to do so depending on what you mean by pathological.

For example a group might have been defined as 'atypical' or another named group at least due to not responding to a medication that the majority not only responds to but does so significantly and/or on an objective measure (eg skin clears up, increased function, increased wellness are complicated in where is objective vs judgment call/assessment?).

 

[EndME]

I think it is quite common to be able to do so depending on what you mean by pathological.

For example a group might have been defined as 'atypical' or another named group at least due to not responding to a medication that the majority not only responds to but does so significantly and/or on an objective measure (eg skin clears up, increased function, increased wellness are complicated in where is objective vs judgment call/assessment?).

I think by the time you have a medication that works (even if not for everybody) I would consider that as there already having been a clue towards pathology, at the very least that situation in any case wouldn't be comparable to ME/CFS (since there is no known drug with efficacy).

What I'm trying to get at is whether there is any historical precedence for people to subgroup ME/CFS further without any pathological clues. An argument can be made if there is a group with specific signs that points to something distinct, but as far as I'm aware that doesn't really exist but I'm happy to admit that it possible that either someone hasn't asked the right questions or that something like OI vs no-OI, certain cognitive problems vs none could be such a sign.

 

[bobbler]

I think it is quite common to be able to do so depending on what you mean by pathological.

For example a group might have been defined as 'atypical' or another named group at least due to not responding to a medication that the majority not only responds to but does so significantly and/or on an objective measure (eg skin clears up, increased function, increased wellness are complicated in where is objective vs judgment call/assessment?).

Is only going with those who have definable PEM going to mean that those who don't yet/aren't currently in that position (either because they can't discern it from the rest of their fatigue, or it's rolling PEM and they aren't going to be given a break soon etc or they don't get it as often, or even don't have it etc) aren't going to benefit from results/if there is a move forward?

Is PEM a better and safer 'proxy' from putting people through over-exertion for a long period of time vs deterioration (or retrospectively measuring for that) ?

If we wanted to take that analogy to 'treatment' and it were actually really a treatment then effectively GET, and any variations of it, or I think extrapolations of that paradigm/parody would be the same thing.

The difficulty or situation that we've had is due to that objective vs judgement call/assessment and who gets to do it and what their aim has been, or more precisely attitude to the null or alternative hypothesis has been (in allowing that to be 'made provable' if it is only true if the system sees it but the system won't see it) for ME/CFS.


PEM has sort of ended up in the middle of this assumed to be a/the(?) mediating factor by some when different types of exertion/over-exertion/insufficient recovery rest and the cumulative aspects contributing to both is perhaps the closest we are at currently.

Yet as a thought experiment has anyone who gets PEM ever benefitted from, or got more well despite, GET? we just don't know whether not triggering PEM (impossible for most) is either possible or avoids the deterioration of continual over-exertion ie 'riding limits' over a more extended period of time, and from my experience of riding PEM at least then the answer is no.

It is going to cause my brain to fuse out trying to think through the different groups and ifs and buts here (eg non-PEM but deterioration, PEM and deterioration and so on...)

and then it cycles round to this issue of the (deliberately in some circles) ambiguous definition of fatigue, not very rigorous descriptions of patterns of it and different types with different symptoms across all sorts of illnesses and dumping pots.

I mention these because one problem eg with PACE type things from the old days I remember rightly coming up was that if someone who had only depression was seen as having fatigue, and depression actually is a condition where people feel better and function increases with more 'doing'/exercise, then if someone filled a trial with subjects whom noone knew if they had one, the other or both you might actually just be re-testing 'does GET improve depression' rather than cfs or me/cfs and making claims based on extrapolation that is inaccurate, as well as in those with both not understanding how something can be both good for one and bad for the other (and then that eventually being bad for the first one too as people's health gets worse).

BUt as others have said there even in the cfs days was generally no actual benefit found so the claims of there being some magic, undefined group of people with cfs/me (in the term used by eg clinics who don't want to change away from GET and claims it's OK because they 'treat CFS/ME not ME/CFS') that can still be treated with something that is at best useless because maybe it doesn't harm them isn't something based on anything.

 

[bobbler]

I think by the time you have a medication that works (even if not for everybody) I would consider that as there already having been a clue towards pathology, at the very least that situation in any case wouldn't be comparable to ME/CFS (since there is no known drug with efficacy).

What I'm trying to get at is whether there is any historical precedence for people to subgroup ME/CFS further without any pathological clues. An argument can be made if there is a group with specific signs that points to something distinct, but as far as I'm aware that doesn't really exist but I'm happy to admit that it possible that either someone hasn't asked the right questions or that something like OI vs no-OI, certain cognitive problems vs none could be such a sign.

I thought quite often it operated in reverse where a treatment that helped came from observing coincidence eg didn't things like quinine for lupus come from where it was taken for other things like malaria, viagra was from a heart trial as I think were the claims about anti-depressants (that I think they still don't know really how they work).

So I assume in the days before things like MRI and other tools then certain conditions that presented in certain ways did use those as clues. Maybe it isn't a technical term but I've know a few people who have autoimmune conditions of different types refer to theirs as the atypical type because it didn't respond to the treatments that the majority did

I guess it also depends what we mean by groupings, does that grouping have to be one where the pathology is what causes that group or one where the lack of responsiveness of the pathology does etc. ?

 

[bobbler]

Mostly it’s a matter of finding homogenous study groups in the absence of a clear marker which defines all ME/CFS (if that exists).

It makes it less likely that something which strongly effects a small portion of pwME is obscured by a very heterogeneous study group, and in the absence of a disease biomarker, could prevent accidental inclusion of people who have some other illness that sounds similar on the face but doesn’t present similarly on a broad molecular level

The thing is that there are quite rigorous ways of checking if someone has ME/CFS (particularly as they get more severe from manner of questioning and watching the exhaustion, but also from the way the history is taken) vs ones which are less rigorous, even if just a clinician's assessment.

The added expense has sadly been lumbered onto the ME/CFS research by/where there has been a dumping bucket approach or other concerns from clinics in comparison to other illnesses that might also be diagnosed by a consultant physician using a history and a work-up. And of course the fact those clinics are not a lifetime responsibility, but often a short course and then people's severity changes over the years and so on.

It is tricky to say but of course there is also theoretically always a risk of entryism, if that is the right term for anyone could theoretically join deliberately to mess up the results or whatever other aim, if anything is just online self-declare. But that has to be balanced with access for a group who is so debilitated and probably needs to be studying the most debilitated and inclusion of that has been a real weakness over the years, and there is no infrastructure whatsoever.

And given how dangerous clinics have been for so long, particularly to those more severe, there is almost more likelihood those who did/could attend clinics didn't have ME/CFS or it was less severe than those who couldn't simply based on it being a Krypton Factor challenge in itself to both attend and risk attending (a term I use when they use the 'treatment' itself as a filter as only the least ill could get to the end: The Krypton Factor - Wikipedia)


On top of that there is a complication that if people have gone to clinics or HCPs recently and are in their early stages of ME/CFS quite a lot get told or primed with stories and replacing their own words about their symptoms that are untrue/rewording of their own descriptions to something that is less accurate - but it takes quite a few years to stop using that term eg 'boom and bust' or 'mustn't nap as that's why I can't sleep' and the rest that any phrasing of questions needs to be able to get past.

This is an interesting conundrum by the way but I think we can get there with enough very slow bandying about the issues and catches to something that is better and at least we are aware of the limitations of etc.

 

[Kitty]

It's a potential PhD thesis. My work would be focused on the multi-omic integration part, since that is what I've already been working on. The interview part would be, at best, a bonus showing that there's identifiable phenotypic differences associated with the molecular subtypes. That same idea could potentially be addressed with just a battery of regular surveys, but we all know the problems with those surveys.

Even if the interviews are only thought of as a bonus, I still wonder if PEM might be too difficult a focus. You could get as many different descriptions as there are individuals, and have to commit a lot of time to trying to make sense of them vs your multi-omics results—or explaining why you couldn't. A doctoral programme's tough enough as it is without the risk of wandering into quicksand.

There are other questions which are just as interesting and potentially more amenable to answers. For instance, even if DecodeME provides real insights into the genetics of ME/CFS, it might not offer us much about male-female (as assigned at birth) differences. There have been hints of this in the research, including the findings on amino acid metabolism/ utilisation made by Fluge and Mella in about 2016, but they haven't been chased down. Could multi-omits tell us more?

Then there are the differences in results for mild/moderate and severely affected patients in recent papers. It doesn't appear to have been anticipated, yet the findings were there. Maybe it could be approached from the other end, actually starting with "Can I show differences between cohorts of mild/moderate and severe participants?".

They're important questions, and might have the potential to make more of a contribution than another look at PEM.