What are the necessary conditions and criteria for a theoretical model of ME/CFS?

It will be interesting to see what is found but I am sceptical about him finding anything very much. Generally speaking if blood brain barrier is altered enough to allow white cells in you either rapidly become comatose or, if it is local, appear to have had a stroke or similar similar. MRI scans can pick up quite subtle inflammatory change in brain. If this was going on there would at least have been a few cases reported of dramatic MRI findings I think.

And I cannot think of any reason why a normal regulatory mechanism that has got disturbed would involve white cells going in to brain.

 

Cytokine assays, at least up until I retired in 2010, found little or no value in clinical medicine because they are extraordinarily poor at telling you what is going on even when there is barn door pathology. In RA the cytokines are going berserk in joints and liver. You can tell that because the C-reactive protein level goes up but cytokine assays were expensive and useless. Most cytokines act very locally. Blood levels can give clues if you are lucky. So there could easily be cytokine activity in brain on that basis. But substantial cytokine production by microglia would be expected to produce very nasty neurological signs that we do not see. And hot MRI scans.

 

Yes, but quite a lot is known about signals from infection and exercise that will impinge on the brain - through hypothalamus mostly. The brain loop idea would be that the response is set wrong. We don't know much that would help explain how that would work but we know that it can happen for certain other signals. Narcolepsy might be an example of the brain's response setting being offset.

 

There was a recent paper, maybe in ScienceDaily about a detailed study of changes in metabolites and organs that occur due to exertion. It seems that the changes are much more complicated and widespread than was assumed before. Figuring out which of those changes--or ones that the study didn't reveal--is involved in triggering PEM might be quite difficult, especially if it's not a direct connection.

 

A few models should help point out what criteria are missing. The feedback will lead to improved models, leading to more feedback. That seems better than trying to create a perfect set of criteria before trying them out on a model.

 

I agree, but with the caveat that I think we need to be careful about the cherrypicking/fishing for that would happen if we just threw data out there, given we know about the latencies and patterns but I don't know what might be manufactured if someone took a few good days vs bad and cart before horsed with exertion vs biofeedback into claiming 'on the days they did more they were better' type thing.

I saw recently someone on social media putting their sleep data on including oxygen info and time in light, deep, REM etc and don't know how accurate that is but if - noting that the more severe you are the more 'all the stuff that seems harder to measure' (noise, thinking, light, time vertical) becomes as relevant - that was included then I have a gut feeling it could be of interest. When I was able to work I used to say that the first thing to be affected was my sleep - but not in the ways the silly sleep hygeiners think (!).

I remember the thread here about high blood pressure where a researcher had an article about the 'what if' of it being seen as a symptom/indicator of something underlying instead of 'a symptom to be treated' and just reducing it with tablets whilst not being curious necessarily of the why.

Over-exertion or other things like being ill, or just nothing we can put our finger on, can 'set off' something and sometimes there seem to be different patterns: hypersomnia or non-refreshing sleep or being exhausted but not able to get rest and so on. What if instead we could describe these in terms that weren't 'sleep' but about what was actually going on, because it feels part of the reaction itself and part of the healing / attempt at getting back to balance ( goodness knows what the right term is when I mean only being able to return back to a pretty debilitated 'norm for your ME severity' and definitely not 'recovery' )

 

To my knowledge nobody has made any claims about proof of causation. On the contrary, my argument was that we don't know either way and that both possibilities are sufficiently plausible to not shut the door on either of them just yet. Until we have technology to provide more conclusive evidence both possibilities are just assumptions

Anyone creating a model or hypothesis remains free to choose which possibility they think is more plausible and put forward their arguments for that choice

A checklist to measure models and hypotheses against, on the other hand, requires a higher level of evidence than we don't know either way (note I wrote 'higher', not '100% absolutely certain', which doesn't exist)

I'm not arguing for or against any particular theory of onset, I'm agnostic about that. I'm arguing for a high threshold of evidence before including an item on a checklist intended to direct people's thinking

So yes, a model does need to account for the observation that many pwME didn't have symptoms of infectious onset. But a model postulating asymptomatic infections does this just as much as a model postulating a range of other triggers

A model proposing all ME is postinfectious - and assumes asymptomatic infections for those who don't report non-infectious onset - is just fine even though we can't test the assumption with current technology. If what the model proposes happens after the infection gives us testable lines of inquiry that's still useful. Even if the asymptomatic assumption is wrong it may still turn out the model is valid for ME with clear infectious onset and that would still be a win for a large proportion of patients. I wouldn't want any such model dismissed upfront

A model proposing ME occurs only after severe infection, on the other hand, would be invalid because it doesn't match the observed data. But it would still be valid to argue there may be two types of ME. This is also not testable right now but plausible enough to not be discounted upfront and, as in the previous case, if what the model proposes happens after the infection gives us testable lines of inquiry that's still useful for that particular group. The model just can't make any broader claims.

TLDR: my key points (which apply beyond the specific example above of infectious vs non-infectious onset)

A checklist to measure models and hypotheses against has the potential to influence people's thinking in positive ways. E.g. we really don't need yet another deconditioning hypothesis but we do need everyone thinking about ME to be thinking deeply about abnormal response to exertion.

The same checklist also has the potential to limit people's thinking in a negative way. This is why we should require a high level of certainty that items included on the checklist are based on solid evidence.

Where in doubt, I prefer to err on the side of keeping an open mind rather than shutting down a line of enquiry prematurely. That's why in an earlier post I suggested a second list of items for things that look important but don't currently have solid evidence for them. The aim is to encourage people to think about them really hard but without being required to force-fit everything into a single model

And now I'm going to 'unwatch this thread' - abnormal response to exertion and all that

 

That may be true, but the same thing applies to virus trigger: there is no conclusive evidence, and we simply assume that the virus triggered the syndrome. All we know is that there is a pattern of Y following X with no direct link. There is no basis for leaving other possibilities open for X=stress while accepting X=virus as a fact. There are more post-virus cases than post-stress cases of course, but there still are enough number of cases that are clearly traceable to stress.

Well, here is a problem: a theory that assume all ME is post-infectious can end up proposing infectious agent is the cause, not just a trigger, of ME/CFS. And you can't reject that theory if you don't have non-infectious cases as a test for the hypothesis.

Good idea! I myself have to remind me time to time that my health, not anything else, is the highest priority. Have a good rest and I hope you recover soon!

 

Tad concerned that this is black box stuff i.e. difficult to measure things that elucidate pathology/mechanism. Perhaps GWAS would provide clues e.g. genes related to infection &/or genes involved in signaling?
Also, metabolomics could potentially turn up clues - apparently coverage is low in ME/CFS - Dr Li (Jackson Laboratory) NIH metabolomics webinar.

 

So good thanks @Kitty!

I wondered if we might refer to late/r recovery’s that appear to be rare, and early recovery’s that appear to be common, mostly getting better in 18months or before about 2-3 years, it’s hard to know exactly where the cut off might be but people meeting the ME criteria which was supposed to be given after some months do recover in pretty high numbers, hence all the “I cured my ME by….and you can too, pay up and I’ll show you how!”

I know that for many who consider themselves recovered they aren’t necessarily recovered, but rather moved out of the severe and moderate state, and managed to stay there, or they were fully recovered but ME comes back for them later.

I don’t what that would all mean but I guess some pathological pathways are more easily turned off or downgraded than others and I guess there could be some clues in there somewhere.





I’m not counting LC with ME presentation because it’s early years on that one, who knows where that will go.

 

Sharing this here as I do not know if Jeff Lubell is a member and his comments are relevant to this thread.

“If I have one plea for medical researchers on #MEAwarenessDay: please construct a model for #MECFS that centers relapses in the pathophysiology of this illness. Static models that do not consider this longitudinal dimension are missing something fundamental.
Relapses are not incidental. They are a core feature of #MECFS. My hypothesis is that the relapses are caused by systemic inflammatory events that cause vascular damage. Happy to be proven wrong! But please, researchers, study this issue!“

https://twitter.com/user/status/1789694835336573188

 

I don't know whether they qualify as criteria, but I had some thoughts on questions for the model.

Some PWME have their ME symptoms increased while fighting a viral infection, and others have their symptoms decreased. I'm not sure of the percentage with no effect. That variable response should say something about the importance--or lack thereof--of immune response to the mechanism of ME. I think it means that immune response can affect ME severity, but isn't the cause of it. I can't recall any people reporting that their ME switches on or off during a cold or flu. Has anyone suffered PEM or avoided PEM only during viral infections?

The same holds true for toxins, such as heavy metals. Such toxins affect many body functions, so affecting ME is reasonable via various mechanisms, but removing all toxins doesn't reliably cure ME. I can't recall reports of people being ME-free after heavy metal treatment and then have it return when environmental exposure raises their levels again.

The all-too-common experience of finding a treatment that works well for a short time, maybe just for a couple of doses, but then stops working and never works again, should mean something, so it's worth testing the model for that.

 

With my current thinking ME is a dynamic process. Its not damage. Its dysregulation, but that could include epigenetic changes. That does not mean that some critical tissues are not damaged, only that ME can get worse or better dynamically, regardless of damage, in at least some cases. I have experienced fantastic responses to treatments that only happened once or a few times, and like @Creekside I have heard this many times from other patients. Then there are my many spontaneous remissions, and relapses a few hours later.

I am much less interested in the genetics of ME than I am the epigenetics of ME. A suite of defective genes might be at the core, that is not ruled out, but a really good model should explain the large array of deficient or excess substrates we find in metabolism, and any other dysregulations the experimental data shows.

However our state of knowledge is so poor, with so very much data we do not understand, that a great many options in models should not be prematurely ignored either. Its fine to prioritize, its not fine to exclude things until we have a better understanding of the physiology and biochemistry of ME. I suspect we absolutely need biomarkers, and as I have said elsewhere there may not be any biomarkers to find, at least not in fully developed ME. The biomarkers may only exist when you are coming down with ME the first time. After that its all dysregulation. I cannot think of a simple and inexpensive way to investigate this.

So I am less sure we should be considering how to rule out models with our limited data, than about how much of ME they explain, their potential explanatory power. Then the researchers can make predictions. Then we can falsify whatever ideas do not work, based on experimental data.

Lastly we need to consider ME might be a number of different diseases, or as I have said before it might be a spectrum disorder. How do we decide on models in these cases?

 

I've suggested one simple, inexpensive way: check existing data for correlations of sets of factors. I expect that most if not all studies have checked their data only for single factors that are abnormal. ME might have several factors that are slightly low and some that are slightly high, and that pattern might be consistent for a high percentage of PWME. It seems a simple, fast check of existing data. Someone might write a program for that and offer it to anyone who wants it, in a language and format that's easy to modify for individual data formats. Alternatively, ask an AI to do it.

 

This might be the case but my concern is that its looking like its mass dysregulation, and everyone might be slightly different, leaving no consistent biomarkers. We could do it with a large and well funded prospective study, grabbing blood samples from tens of thousands, doing a new one every time they get sick, and even more over the months if they appear to have a post viral syndrome, and even more if they fulfil ME, FM or LC diagnostic criteria. The issue here is $$$$$.

 

This is the overwhelming presentation in clinic. Patients repeatedly try and do x or y. Endlessly. Doom loop of try, fail. Rest up repeat.

It's so at odds to the 'fear of activity' model.

pwME tend to have the opposite. Almost a stubborn refusal to give up trying 'this time it will be OK....' and are shocked and profoundly saddened when what they would like (elimination of symptoms) does not occur. One reason why the fear/avoidance/deconditioning model is seductive to patients. At least for a time it can make sense. And then it doesn't.

Humans are loss averse and will often keep chipping away at an unhelpful strategy, especially if it has been conditioned and helpful pre illness, way beyond its usefulness.

 

I just want to clarify two points. First, a large study of healthy people to be followed, at least initially healthy. Second, the point is to capture the onset of ME, LC etc. at the very start. Not years later.

 

You could theoretically do this with early LC for ME/CFS.

Sample
500 post-viral COVID (1-3 months)
----
Sample
200 go meet a CCC definition at 6 months
150 have LC but don't meet CCC
150 recovered


I think this is possible but from healthy would be really tough unless part of an extremely large scale biobank.


The value of this design is that you could develop a diagnostic to enhance current diagnostic process. I mean you're building data on individuals that you would want to diagnose (early ok disease). You could see signatures that separate ME/CFS from non-ME/CFS PVFS.

You would be deeply characterising 500 post-viral fatigue patients with 200 eventually meeting ME/CFS..

But would this design be more fruitful than deeply characterising 1000 ME/CFS patients to look for pathologies and signatures to cluster? That 1000 ME/CFS patients can give a longer history of diseases and treatments they tried to some success or the opposite.

 

There are a couple of groups who have sort of gone down this route. Programms such as LIINC have been setup to track people right from their Covid infection until whatever outcome they arrive at and the recent ME/CFS biobank in Germany is also setup to do something similar.

However, some of those studies have shown that even people that meet the CCC definition at 6 months or at least look very similar to ME/CFS often turn out to look rather different at a later time, as for example seen in Long-term symptom severity and clinical biomarkers in post-COVID-19/chronic fatigue syndrome:results from a prospective observational cohort. This German group surrounding Scheibenbogen is also tracking ME/CFS post EBV, but those studies are probably extremely underpowered to reveal much from an epidemiological stand-point as seen in One-Year Follow-up of Young People with ME/CFS Following Infectious Mononucleosis by Epstein-Barr Virus.

If the studies are anyways severly underpowered from an epidemiological stand-point, for example to figure out how large of a risk factor EBV infection is for ME/CFS, I do agree that it might just be sensible too just look at 1000 ME/CFS patients with a longer disease history and do a deep phenotyping study of those patients.

I do wonder though why there is so much focus on trying to look at signatures that seperate somewhere around 90% of ME/CFS patients from healthy controls and then often do so via something like a random forest classifier that just ends up looking rather random, finding marginal differences in a larger set of people without telling us anything fruitful.

I'm far more interested in a result that shows stark differences between two groups and that tells us something meaningful about pathology even if that result only applies to say 50% of ME/CFS patients. But then I wonder how large the sample set of patients really has to be. Depending on what is done it might sometimes be more important to have someone that look at a problem under a novel angle. There are certainly things wrong with the intramural study by the NIH, however what it did show us that a study expecting somewhere around 90% accuracy from a signature is destined to fail from the get go if it is really supposed to distinguish ME/CFS from HC.