The aetiopathogenesis of fatigue: unpredictable, complex and persistent

[lansbergen]

If I exercised, my external environment would become both stable and predictable. It would consist of my bedroom ceiling and the inside of my eyelids.

I was there but as I could not lay on my back it was not the ceiling I saw when I could open my eyes.

 

[Stewart]

If you google Julia Newton she is attached to a University/NHS website that promotes CBT/GET as treatment for ME.
She also doesn't study ME but fatigue.
Also, I remember quite vividly, reading someone here on PR who said they had an appointment with JN and they where appalled when JN herself proposed some form of CBT/GET to her.

It's hard to parse where some people -- researchers/clinicians stand sometimes. But it increasingly looks to me like JN is no friend to the ME community.

Snowdrop - if you're thinking about the same PR message that I am, then your recollection isn't entirely accurate. It was the patient/poster who specifically raised the issue of GET with Julia Newton (because another poster in the same thread had recounted a bad experience with an OT who worked in the CRESTA fatigue clinic) - and Julia replied that it was one of the only treatments for ME that was recommended by the NHS, and she asked the patient why she hadn't tried it. The patient did say that they were appalled by this - but that when they changed the subject Julia didn't press the issue.

If you work in the NHS I don't think you have much personal freedom to denounce the NICE guidance on CFS/ME (which promotes CBT and GET as 'specialist care' for people with ME) however much you disagree with it. While it's only guidance and not automatically binding, NHS bodies are expected to adopt it unless they have a good reason not to. If the Clinical Commissioning Group/Hospital Trust that employs you has taken the decision to follow the NICE guidance (as the Newcastle Upon Tyne NHS Foundation trust seems to have done) I think your choices boil down to 1) go along with it and work round it as best you can, or 2) resign. The two things that struck me about the interaction detailed above was that it was the patient - not Julia Newton - who raised the subject of GET, and that she dropped it when the patient made it clear she didn't want to discuss the issue any further.

I've tried to read the paper and I'm quietly confident that most of it went completely over my head. However the impression I was left with was that it wasn't really about CFS (even though they kept mentioning it) - it seemed to me it was about the study of fatigue more generally. Julia's area of expertise is autonomic nervous system dysfunction and fatigue rather than ME/CFS specifically, and it looks like the study was written with that broader perspective in mind. That said, I thought it was extremely unhelpful that they kept referring to "fatigue and CFS" as though the two were synonymous when (as Valentjin has already pointed out) they're not. It would have been much better if they simply hadn't mentioned CFS in the paper at all.

The fact that this study refers to the "apparent efficacy" of CBT and GET (my emphasis) could be read as a less than fulsome endorsement of these therapies. But maybe I'm reading too much into their choice of words. Julia Newton's team has received over £1 million of funding from the MRC for their ME/CFS research, so it's probably not politically expedient for them to openly question the findings of the (MRC funded) PACE trial.

I realise this is an excessively detailed defence of Julia Newton, and I promise that I'm not related to her and I'm not a member of her research team. However I was an attendee at the one day conference that the ME Association held with her team in November 2014 (you can see the programme details here) and I was left in no doubt that she and her team are actively hunting for a biological explanation - there was no hint of a suggestion that they thought CBT or GET were the way forward for patients. However I think that the top-down structure of the NHS and the overly-reverential politics of research funding in Britain mean that you have to give the impression that you accept the status quo - even as you're working to undermine it. To do otherwise would be to put your career in jeopardy and ensure that your research never receives the funding it needs.

 

[JaimeS]

If that's the case, why not just avoid mentioning CBT or GET? Then one would not appear to endorse or condemn it, if the point were to avoid stepping on certain toes.

Please, researchers of the world. Ignore it, and maybe it will go away...

 

[Sean]

...and this may explain the apparent efficacy of cognitive behavioural therapy and graded exercise therapy.

"apparent efficacy"

 

[alex3619]

'add up' to fatigue in this manner:

Fatigue = influence of factor A + influence of factor B...

I would like that explained using chemical kinetics models. Really. It could be funny.

 

[Sean]

I would like that explained using chemical kinetics models.

The fatiguocline.

That's where it's all happening, baby!

 

[JaimeS]

I would like that explained using chemical kinetics models. Really. It could be funny.

Wellll, remember, that's the one the paper rejects as simplistic. The systems biological formula they use is described here, as part of a neat little primer on systems biology as a whole.

In short, yes, they could have used a more modern source for that, but perhaps the 2002 paper is seminal in some way; and yes, systems biology is an important and widespread method used to describe complex biological systems. I'm gaining appreciation for why that idea is in the paper.

Now I just wish they'd chopped the second half of the paper off of the rest and shoved it in the shredder.

-J

 

[anciendaze]

That "simplistic" formula is an explicit statement of additive combination, one of two criteria fundamental to the idea of linearity. In medical circles the other, proportional response, sounds logical for some reason: responses are proportional to causes. Such reasoning regularly leads to overdoses of prescribed medication, a problem most of us have experienced. This kind of reasoning also leaves no room for the prolonged drop in aerobic capacity seen in 2-day CPET tests, where exercise above a threshold has the opposite effect from exercise below that threshold.

Most medical reasoning doesn't even make it to the stage of realizing that additive combination may be on a logarithmic scale, yielding power-law responses. This is where mathematical modelers can use linear differential equations. Talking about ratios of rates brings you closer to the chemical reaction kinetics Alex mentioned, but most such systems are strongly nonlinear. Even simplistic models of individual neurons that produce qualitatively plausible behavior are strongly nonlinear.

The vast majority of medical researchers seem to be fixated on levels and setpoint control while dealing with strongly nonlinear processes more closely resembling what engineers call PID control (Proportional-Integral-Derivative) and differential signalling depending on ratios of rates. None of them have answered the important question of how biological setpoints are ever calibrated. The reason for this omission is simple: biology does not work the way clinical laboratories do.

The distinction between reasoning based on levels and reasoning based on rates becomes overwhelming when you are dealing with clonal expansion that amplifies tiny signals by huge factors. Without better ideas of how to identify causes you will constantly overlook them as too small to be relevant. This is precisely the problem we see in research on a long list of diseases. Most advances in pharmacology have come from random trial and error, often in a different medical context than the one where a drug ends up proving useful.

A second problem is that reasoning is commonly based on straight-line chains of inferences when biological systems which are not heading directly for progressive failure necessarily involve regulatory cycles of causation. This is translated into "holistic" medicine which is, more often than not, simply vague.

If we don't get beyond this confusion, scarcely any advances will be possible, except by sheer luck.

 

[alex3619]

I think a big issue is the practice of creating disease categories based on common symptoms. These are often heterogeneous cohorts. Similar symptoms, wildly varying causality. CFS is one such, and its a big reason why CFS research is having a hard time. A single biomarker is probably not feasible. However this does not mean that a cohort subgroup wont have a single biomarker. Failure to find a single unifying biomarker does not prove that a biomarker does not exist. Most psychiatric research suffers from the same flaw. Depression, for example, is probably a common symptom to a myriad of conditions.

Yet in tighter cohorts we have things like the energy loss after activity, which is measurable. Yet even this could mask a heterogeneous cohort, as shown by the Lights research.

The broader you make the cohort, the less likely it is you will find a single biomarker.

 

[alex3619]

I am a big fan of systems modelling. I think we can learn a lot from such modelling.

However the key point here is model. They are trying to create a model in the hope it can be used to create answers , reduce reality to an equation. I think there is serious risk here. I think a similar issue undermines a lot of BPS theory.

Models are not reality.

Systems models can give us a more structured understanding of interactions, and so help us refine better research questions. The hard research still has to be done though. They can also give us heuristics. Hard definitive answers do not come from systems modelling alone though.

 

[anciendaze]

There is a connection in what you describe with statistical power and problems George Davey Smith mentioned with small cohorts. Big cohorts are desirable from one standpoint, yet almost certain to cloud issues with respect to causation. Big cohorts also correlate with big funding and centralized control, which has been a real problem for us so far.

I've been thinking about some examples of rare diseases which are likely to show up in ME/CFS cohorts unless you are extremely careful. The post on Health Rising about joint hypermobility syndrome reminded me of some odd cases I already knew about.

Few people are terribly surprised when a case labeled CFS turns out to be a case of MS or lupus, even if Dr. House might be. ("It's never lupus.") They would likely never imagine that a patient labeled with CFS could have an autoimmune disease hitting IgG4. Such a disease is unlikely to ever reach a TV series, it is simply too rare. If the disease had not been progressive, eventually providing samples for pathologists, I doubt the marker would have been found. There are a lot of things for which we generate antibodies, and in most autoimmune diseases the question of antibody levels is problematic. (I can tell you that the last disease mentioned is now called "IgG4-related disease" because the antibodies are not always reliable biomarkers.)

What I noticed in the JHS case mentioned above was an association with connective tissue disorders. This also happened to turn up in the periodic paralysis patient I know personally, even requiring complicated surgery to correct deformities causing life-threatening sleep apnea. (I'm told most common surgery for sleep apnea introduces problems patients are not warned about in advance, like having food come out your nose. It paid this patient to do research.) He also had joint hypermobility in childhood.

The problem we run into in his case is that the incidence of periodic paralysis (PP) is about 1:100,000. The population of the U.S. is not large enough to put together a cohort of 10,000. What is more, there are forms of the disease which could be mistaken for CFS. If you do the right tests you will have little doubt that you have found a patient with PP. (In the case of my friend, it took 5 hours after a provocative test before he could leave the clinic. There were episodes of doctors screaming at nurses involved when another doctor realized how close they had come to killing him.) The problem is that you will not do the right tests if the doctors involved have never heard of the disease, and have trouble believing in it. ("That's a classic case of somatization leading to catatonia.")

In reading about the splash in the press caused by a paper on suicide in CFS I first thought of the point James Coyne later made, that there was no way to control for depression with the Oxford definition. In that case you could prevent suicide if you treated patients for depression without even knowing about a diagnosis of CFS. A second thought was that all 5 actual suicides might have been included in the cohort as a result of misdiagnosis. Diagnostic error rates don't come close to ruling out 5 out of 2,147.

This is only the tip of an iceberg, because the NHS tends to funnel patients who are diagnostic problems into CFS specialization. Considering the size of the catchment of that clinic we might be looking at a substantial fraction of all the rare chronic diseases without convenient clinical signs in the U.K. Under these circumstances you can't simply rule out diseases with an incidence of 1:100,000 as too improbable to worry about. There are more such diseases than most doctors imagine.

Now I feel a need to address the point you made about models. It is no accident that the convenient markers based on levels of certain substances work best when applied to progressive diseases. If you miss the marker at one time, it will become more prominent. You may find it at autopsy, but it will make itself known. This is not necessarily true in chronic diseases.

What I'm getting at is that even when we look for markers we regularly place the wrong significance on what we see because there are far more biological processes where ratios of rates are important than those strictly responding to concentrations of molecules. Babies are not calibrated like clinical instruments, their organs must be responding to a different kind of signalling. There are plenty to chose from, if anyone in medicine looks at the literature on biological signalling.

Here I want to describe part of my personal history which had dramatic examples of defective mental models far from medicine, some of which nearly killed me in the military.
(Advice: never ask a helicopter pilot what happens if the engine quits while you are flying, unless you enjoy a thrilling demonstration of an autorotation landing. Fortunately, this was tame compared to my regular work -- it didn't even involve explosives.)

Fast forward now to a civilian job creating simulation models for use in training people to perform aircraft maintenance. (Even way back then computers were a lot cheaper than complete jet aircraft, and much less likely to produce fatal consequences.) The first part of designing a simulation is to understand what you are simulating. This means you read manuals, watch operation and go talk to the people who teach the subject. Guess what? Not only do students have defective mental models, you can even find instructors with the problem. You can't code a hand-waving explanation for a computer, except perhaps for amusement. I found a misunderstanding which had actually resulted in what is euphemized to "equipment loss" was still not understood on the flight line. The embarrassing problem was not officially classified, but nobody wanted to talk about it, (another insight.)

We can skip a long series of anecdotes about discrepancies between mental models and reality discovered in a broad range of simulations which reach all the way to the Space Shuttle, and include side trips to subjects very different from aerospace engineering, like the Strategic Petroleum Reserve.

With all this for background behind me, I ran into an explanation from a doctor of how a medication worked. This didn't make sense to me, but then I do not have appropriate training. The Internet was an infant at the time, but I had access to a good academic library. I checked textbooks and reference material, and got different versions of the same thing. My problem was that I could see no way to turn those explanations into an actual model of the process. If it worked at all, the biochemical behavior would be considerably different from what was being described.

There was research material available on pharmacology, so I went to that, and kept finding various versions of hand-waving. Finally, going all the way down to reaction kinetics of the molecules involved, I ran across a paper which said the biological significance of these chemical signals was determined by the ratio of uptake rates at two receptors. That was something I could turn into a model, and would even yield the behavior being described. It might or might not be true of the actual biology, but it was at least feasible to create a model.

More than simply being feasible, this model added considerable sense which had been lacking. It was based on rates rather than levels, and the two receptors were like separate communication channels because molecules which fit one would not fit the other. Since docking of molecules in receptors is a discrete event you could describe this as differential signalling based on pulse repetition rate. Nature was being quite clever because this meant changes in concentration which affected both molecules would have no effect on the biological signal, a kind of noise immunity. The high specificity of receptors also greatly reduced "noise". This kind of signalling also bypassed a problem which had bothered me: how were biological level-sensing mechanisms calibrated? There is nothing naive about biological signalling, the conceptual problem lies on the human end. I have since learned that correctly functioning biological signalling mechanisms are generally much more sophisticated than most people in medicine imagine. This ought to have an impact on endocrinology, immunology and neurology. Many years later I'm getting quite tired of waiting.

The bibliography of that paper told me the concept had already been around for years. Since I was looking at research several years old, I could immediately judge the effect of this breakthrough on the field. The vast bulk of papers afterward went on dealing with the same defective models, tacking on special cases whenever an example which falsified the model appeared. I already knew from other experience that an explosion of special cases was a sign of poor understanding.

Now, it is entirely possible, even likely, for biology to violate human preconceptions about how it ought to work, but in this case there was a model which matched the behavior researchers were describing, as far as I could tell, and it was being ignored. Either the behavior being described was also wrong, and the whole corpus of research nearly worthless, or people were ignoring an internal inconsistency which prevented them from constructing useful models. This continues to this day.

Levels of biomarkers are convenient when you can trust them, but there is no law of nature requiring them. What we have already is anecdotal, but still significant, information about dynamics. It took many years for anyone to actually test patient reports about PEM and reduced capacity. This is a dynamic response to an exercise challenge relating to aerobic thresholds. A great deal of learned nonsense could have been dispensed with if this had been done 30 years ago.

Quite a number of us also show strange responses to a glucose tolerance test. I suspect we have a lot of people with increased sensitivity to insulin. Another weird response concerns steroids, and I am far from alone when I report that I didn't sleep for 48 hours after a shot. Both these relate to the dynamics of glucocorticoid regulation.

We have similar reports of poor regulation of fluids and electrolytes, requiring conscious measures to maintain them. All of these things fall in the category of dysregulation of the HPA axis, but you can plot points along multiple axes for cluster analysis.

Orthostatic intolerance is still another axis, generally indicating poor autonomic control, if there are not fairly specific defects in muscles and joints.

By the time you get through locating patients in the space defined by all these axes I don't think there will be much overlap with the healthy population. The problem is that medical doctors are fairly consistent in avoiding each of these measurements of variation, preferring to think of the changes as "random".

Added: Edited to correct 2,400 to 2,147

 

[JaimeS]

the incidence of periodic paralysis (PP) is about 1:100,000. The population of the U.S. is not large enough to put together a cohort of 10,000. What is more, there are forms of the disease which could be mistaken for CFS.

In that light, it's pretty hilarious that PP was my initial diagnosis. "It's not that I'm certain you really have that," the physician at the time explained. "It's just... that best fits your symptoms, and we need a diagnostic code to run these sorts of tests..."

The vast bulk of papers afterward went on dealing with the same defective models, tacking on special cases whenever an example which falsified the model appeared.

As a teacher, I was always horrified when I discovered that what was in my students' textbook was simply wrong, wrong, wrong. An outdated model with little practical application, meant to serve as a bookmark for the kids to return to 'when they were old enough'. ATOMS. ELECTRONS. One of the worst offenders is electron configuration (3d before 4s) the reasoning for which is still taught incorrectly at the college level across this entire country. We are remarkably poor at letting go of outdated modalities, and we have the unsettling habit of taking a few facts and fudging the rest until it's 'good enough'.

Quite a number of us also show strange responses to a glucose tolerance test. I suspect we have a lot of people with increased sensitivity to insulin. Another weird response concerns steroids, and I am far from alone when I report that I didn't sleep for 48 hours after a shot. Both these relate to the dynamics of glucocorticoid regulation.

Insulin tolerance test, too. Twisk recently wrote this review article; it has a very exhaustive list of irregularities that seem to be common in our patient cohort.

preferring to think of the changes as "random".

If they don't recognize it as belonging to something from a textbook, it MUST be random. Or weren't you aware?

I agree if they just combined the findings / symptoms you mentioned, there would be no healthy people left. Perhaps the CCC wasn't immediately adopted and embraced everywhere because it seemingly had too many variables, and therefore was too 'complex' and 'multivariable' to not apply to other conditions? That's not a perspective I'd considered, before.

-J

 

[anciendaze]

I'm still working on the idea that cases of Ehlers-Danlos sydrome (EDS) can be mistaken for hypochondria, depression or CFS. This is not nearly as rare, and has half a dozen forms about which ordinary MDs don't have a clue. Borderline cases of MS and lupus are even more common.

What does that tell you about the claim that the 5 suicides in a cohort of over 2,147 must have killed themselves because of CFS? Is there any psychiatric diagnosis with an error rate that low?

Even this assumes doctors with psychological presuppositions are competent and alert. Ever hear about patients with Wilson's disease turning up in psychiatric facilities? That one has a pathognomonic sign that would be hard to make up, a copper circle around the iris. If you can't tell this is staring you in the face, you aren't going to catch anything subtle.

Added: corrected number of cases in suicide study.

 

[JaimeS]

Ever hear about patients with Wilson's disease turning up in psychiatric facilities?

I learned it from House, M.D. Fiction = useful.

I'm still working on the idea that cases of Ehlers-Danlos sydrome (EDS)

Oh, look! Another one of the things I was diagnosed with early. A quick pubmed search shows that conflating depression, anxiety, and EDS are not a thing of the past. Out of the top ten studies that popped up on pubmed, two were describing the link between EDS and psychiatric conditions.

-J

 

[anciendaze]

I learned it from House, M.D. Fiction = useful. ...

Since I don't own a TV, and try to avoid them, I couldn't have learned it there. I have exchanged messages with a patient who was unreasonably impressed with my knowledge and asked about her own eye. When I pointed her toward a medical reference she said she did have a copper ring around the iris, but it didn't look like the one in the reference. I told her that depositing copper was not normal in anyone, and she should be examined by a doctor who knew what Wilson's disease was.

 

[Mij]

If you google Julia Newton she is attached to a University/NHS website that promotes CBT/GET as treatment for ME.
She also doesn't study ME but fatigue.

My understanding is that she studies the ANS dysfunction and recommends treatments for OI, so perhaps GET could be useful for that.

 

[anciendaze]

My understanding is that she studies the ANS dysfunction and recommends treatments for OI, so perhaps GET could be useful for that.

You're assuming that orthostatic intolerance and dysautonomia are unrelated to exercise. Unfortunately, they are very much involved. Normal autonomic function reduces blood flow to the gut during exercise. Exaggerated autonomic response can cause serious problems with gut motility as well as OI. We are just beginning to investigate possible bacterial translocation during exercise as a cause of possible autoimmune response.

Anecdotal reports are that patients with OI have more trouble with digestion and exercise than healthy people.

 

[Jonathan Edwards]

This appears to be an invited review in British Medical Bulletin - which is sort of the weekend colour supplement to BMJ I think. The first author is a first year medical student who is probably very motivated and may well turn out to be a great asset to the ME/CFS community but at the moment seems to be distracted by vague issues of metaphysics and mathematical equations (not to mention quotes about quantum mechanics) that would be better considered after a little experience in the field! It is disappointing that the senior authors have not steered this into something more useful. That diagram looks pretty meaningless.

 

[Mij]

@anciendaze I guess it would depend on what type of exercise that is recommended. Mild physical activities in a supine position like swimming or pilates can be helpful for improving/increasing blood volume and reducing blood pooling.

 

[rosie26]

Since I don't own a TV

I haven't had a TV since we went digital in December 2013. I thought at the time that I had everything worked out for the switch over but apparently not. I didn't bother getting it right. And I haven't missed the TV. No more having to listen to lots of bad news everyday and also so nice not to have to listen to endless advertising. I will perhaps get one in the future but I am in no hurry whatsoever.