Biomarkers for ME/CFS - discussion thread on the next steps for testing biomarkers, and why we need them

There is no doubt truth in what you say but it may also be overly pessimistic. This is purely anecdotal but I know two young women who are both highly intelligent. With one I had a discussion when she was about 16 about what she was going to do for a career. I told her that with her brain she could make a lot of money in the city if that was what she wanted but she needed to decided if she wanted to sell her soul or do something socially useful. She is currently training to be a surgeon and hoping to combine that with academic research.

The other graduated with a first in physics and successfully applied to do a PhD. When the PhD was postponed for a year she decided to take up an offer to spend 12 months working in the city. I told her mum that she wouldn’t take up her PhD when she realised how much money she could make in finance. She is now very rich.

 

I agree the test needs sufficient validity but think we are where we are and because of the BPS stigmatisation a biomarker is critical for large scale acceptance in wider society (including ill informed GPS who don't get the nuances of diagnostic process) and validation of patients in order the disease and the patients get the respect that is needed and these are valid and important needs. True the BPS garde can try to undermine it but it will be harder, they will be out on a limb with flimsy arguments and get dismissed by the majority.
It might also help with problems like the ongoing research funding panel issue as it becomes significantly harder to reject proposals and could hopefully help in reigniting interest in researching the condition at the much larger scale it needs.

Interestingly the The ME Association has finally picked up the baton on the seahorse nanoneedle stuff and replicated it so at least that initiative is live and kicking again.

https://meassociation.org.uk/2024/0...-properties-of-blood-from-people-with-me-cfs/

https://meassociation.org.uk/2024/08/explaining-electrophysiological-properties-of-cells/

 

They already are. In 40 years of trying they've produced no evidence of a biopsychosiocal basis for ME/CFS.

We need to keep asking the same question of everyone who claims to know how to treat or manage ME, whoever they are and whatever they're saying: where can I read the evidence for that, please?

It'll make no difference to those for whom it's a belief system rather than a science, but they're lost anyway. But it might stimulate thinking in some, and it's more effective than trying to counter non-evidence with still-inconclusive evidence.

I worry that any biomarkers found will also be regarded as somewhat inconclusive. One thing all GPs know is that screening results vary a lot between individuals, and because of this, thresholds put in place to separate normal from not-normal are fairly arbitrary. Unless results are a long way out of range or abnormal across two or more domains, they're rarely viewed as more than a minor part of the clinical picture.

 

i think the cpet more significant than that and if it had been a reliable option then me doing that whilst moderate would have saved me being out through suggestions of going to the gym whilst doing a more than full time role.

and the walls colleagues dragged me on because of their own issues. And all sorts of decisions I was effectively threatened snd coerced into by worse more life threatening options from behaviourism on the other side vs basic things I should have then and now been entitled to (hence it feels very constructed)

I’d have had a very different life if that cpet had existed at any one of the five significant points where I hit coerced through the same awful dystopian terrified gif your own life and freedom loop.

and that’s because it’s so ‘on the nose’ in what it shows and because it’s individual

and because it underlines people aren’t unfit so it isn’t behavioural

and exercise clearly would make us feel worse so isn’t depression and ‘you’ll descend into the depths if we don’t keep you working’ con so gp doesn’t even sign you off for a day. I had less sick leave than people who were never ill despite having serious other illness during that timeframe on top of my M.E. . If I hadn’t had it already chances are their approach would have caused it anyway.

it’s that freedom to allow people to think in the dirty way they instinctively want to due to false superiority being trained in (flase because they aren’t thick, but they aren’t all ‘top of their class’ which is rewriting history so don’t ’know better’ than all their patients) . Add in the suggest of lack of insight combined with delusion and it’s a death warrant fir the individual and their autonomy and human rights to exist . With no evidence and then knowing full well it’s not something to be cured even if it was behavioural the intention is crude behavioural psych aka bullying re-education. Brutal stuff that wouldn’t be signed off to that extent for dogs without protests.

but the idea any physiology does this is sadly untrue . Even in haematology or other areas problem tests that aren’t in the area if eg cancer can end up being called pseudo and inference from correlation be used to infer the cause behavioural even when there is just a massive list of things that are different across msssive cohorts. So it’s cherry picking. Without further investigation and normally without asking for actual history from horses mouth. Just like they get to stick on that we never exercised when we were often more active then forced to do more than most of our colleagues. Until we keeled and our bodies couldn’t.

 

@Jonathan Edwards From reading the other thread I’m not clear if you would disagree with what I’ve written above or not. If you disagree, please can you (or anyone else) explain what I’m misunderstanding?

It seems like an important issue for advocacy as well as research, so I want to be as sure as possible that I understand the arguments properly.

——

Question to moderators: should the name of this thread be changed to reflect more accurately the issues that Jo has raised about biomarkers and diagnostic tests, which have been moved here from the preprint thread?

 

Broadly yes. There are I think two points. One is that although both patients and doctors tend to think assessment consists of 'making a diagnosis' in reality, in modern medicine it is mostly about establishing what is wrong, - where, how much and various other aspects. 500 years ago 'physicians came from far and wide... to name the disease they cannot cure...' Things have changed, or should have. Unfortunately, 'making a diagnosis' is easier if you want to do medicine without any thought process.

Even more unfortunately, in the last twenty years medical training has been all about diagnostic pigeon holes and pathways. I rang my GP to ask if I could see a rheumatologist to discuss my rather puzzling knee pain. She replied that I had to see the physio and go on the knee pain pathway. The ANA test is often considered as a diagnostic test for lupus but it isn't really that at all. It is a test that should influence your vigilance and regularity of monitoring because it has a prognostic value in predicting that there is higher risk of someone with one or more features of lupus having more later. So you are right in that very often these are prognostic tests.

The second point, of relevance to research studies is that you are likely to get rather little joy out of just searching for correlations between tests and an already clinically defined diagnostic group. The popular idea that you want 95% sensitivity and specificity is completely spurious. If 28% of patients had a particular tests above normal range that would be a major finding. If a test result within normal range but above a certain level predicted disease course that would be useful. And so on.

Looking back, the topic arose this time around because of a poorly informed press article giving a misleading view of the significance of the study from Chris Ponting's team. I picked up on this but I was also trying to raise a point about confusion, as I saw it, within the preprint, about what the study was trying to do. The study is important but in the preprint form I think it undersells itself by firing off comments about aims and interpretations in the introduction and discussion that similarly put the focus in the wrong places. There is also a lot of emphasis on mathematical analysis of correlations that I think puts too much faith in an 'ME' diagnostic box and as a result may lead to false conclusions.

It all centres around the argument that if you want 'better' diagnosis that assumes that your diagnosis so far is unreliable - in which case you are not going to find a more reliable way to do it just by searching for correlations with the unreliable method. In the end the Edinburgh team do not make that mistake themselves, because they did not find tight correlations. But the study may reveal crucial information about the nature of the unreliability of defining ME cohorts and using them in any sort of causal analysis - which is what they are hoping to do with the 'mediation' technique.

 

One point though is that the diagnosis often references the exclusion of other causes of fatigue. Of course that is probably unnecessary if an adequate history of defining features, such as PEM, is obtained. In practice though I think nearly all practitioners would submit a new patient to a barrage of clinical tests, all of which come back normal. That's expensive and wasteful, with an implicit opportunity cost (including to the funding of research on the actual mechanisms). There will also be occasional false positives and iatrogenic harm from incidental findings.

If there were some theoretical blood test with 100% accuracy as above, that was similar to something like thyroid function tests, then you could do that first and save all the irrelevant tests.

EDIT: TFTs was a poor choice and diverted the discussion below. I simply meant a clinical test that was easy and cheap, in this case a blood test that doesn't require a particular time of day or starving for example. I wasn't equating to how TFT results are assessed and translated to something clinically meaningful. Instead this is a hypothetical blood test that perfectly comports to "the clinical diagnosis of ME/CFS" as imagined in EndME's quote.

 

OK now I'm getting into theoretical land but a test that is somewhat diagnostic would be interesting because we could look at what other diseases are actually common comorbidities without that 'maybe the fatigue is caused by x so y doesn't exist' type issue. Which I think is also potentially useful as a clue as to cause and vulnerability - whether these additional diagnoses take place before ME or after.

For me CFS was used as a dumping pot I suspect for my GP to avoid looking into anything else even though the directive had been sent back that something else I had 'was caused by something else' and another GP had immediately suggested a number of likely things (at least one of which I was finally treated for 12+ years later) and to return for the next appointment, upon which the surgery switched me to a partner who told me I had 'CFS' and that they'd send me to a consultant for diagnosis but not anything else. The GP did no more tests. Including eg iron/ferritin, which was picked up 1yr later as under 10. That then also wasn't properly treated or investigated.

So I really wouldn't assume that it hasn't been very much used to do the opposite and cause what I would call iatrogenic harm by deliberately not treating or investigating and using that 'hypochondria' version of the guidelins as a cover for it. I'm all too aware of the way in the UK the what costs GPs and what pays them from gut feeling over a number of years created strong nudges for certain places.

The irony is the diagnosis if it were ME or ME/CFS wasn't wrong - once the guideline and description changed to PEM and to me the CDC suggestions rang bells for me. But that's just how bad the CFS old guideline was in describing what something was, the treatment and their suggestions on investigations. I don't think we can take any of it forward, including the bit about 'excluding other causes' in the same way - it's important to find other things that are treatable and treat them (opposite to old guideline) to reduce load on someone's sytem (and maybe give a chance of reducing debility and a downward spiral that could be avoided), but given PEM is it really the case that you can't have multiple things at once?

 

But the point is that a thyroxine level is 100% accurate (both sensitivity and specificity) as a judge of thyroxine level.

It is no way 100% accurate as a judge of clinically assessed myxoedema.

When the time comes, hopefully soon, when we have a test that is 100% accurate for a causal factor in people with ME/CFS it will be a 100% accurate test for that causal factor, not ME/CFS.

The point is not that we are not wanting tests that tell us 'aha it's that'. We do. The point is that if we assume that a diagnostic test will correlate perfectly with a seriously vague clinical syndrome concept we will come up with a lemon. You start going down all sorts of mathematical rabbit holes that get us nowhere.

Apart from anything, there is probably little point in chasing combinations of test values that are in the normal range but just show up as a skewed combination on some machine learning analysis. In clinical medicine we make use of tests that are consistently and clearly abnormal to explain ill health. There are a very few exceptions where ratios are key, but I cannot actually think of one for the minute.

 

AST / ALT ratio in some liver diseases?

 

Nevertheless, I suspect this may be what happens, only I doubt it's likely to be a single test, not in the near term at least. It is more likely to be a combination of tests, an algorithm e.g. low vasopressin, low NK cell function, low VEGF etc). Say three out of four or five values.

Why a diagnostic amalgam? In part because singularly all tests I'm familiar with have come up short. That might be because we are a heterogeneous group. We simply share a common core of symptoms. It might not be the neatest diagnostic, but I'd happily take it should it ever happen.

It's only when you've no diagnostic and are in many ways on the outside looking in, that you are keenly aware that you need a diagnostic to a) possibly get appropriate treatment research, and b) have medical biases suspended, and c) be allowed back into society.

The amalgam idea might change if they get their act together about tissue diagnostics, but I'm not holding my breath.

 

My understanding of the whole discussion is that is precisely not how things work. My understanding is that the diagnostic process is more complicated than that and always includes signs of exclusions and inclusions. Additionally we have once again seen in the intramural study why exclusion is important (my understanding is that both the patient that had cancer and MS were seen as having an adequate history of defining features). From what I gather signs of exclusions and inclusions are part of every medical diagnosis so there is no specific reason to why it should be any different for ME/CFS. And similar to how there is no “biomarker” for MS, people still get diagnosed with it and receive treatment for it and of course there are also wrong diagnoses there, with some post-mortem studies suggesting that at least 1/20 MS patients actually don’t have MS but some other rarer forms of demyelinating illnesses.

But if your ME/CFS test is 100% accurate that just means it is 100% accurate when tested against the CCC (or against a different criteria), rather than anything else. That still means that a patient with MS can in rare cases meet the CCC and not having been discovered as having MS or a comparable situation can occur. That means a 100% accurate test will have to diagnose such a person that has MS with ME/CFS. Is such a test still 100% accurate in the true meaning? The point as I understand it is that those questions don’t matter, because it is not the accuracy of diagnosis that is crucial rather than something that signals a pathology (and does so accurately).

This is obviously the case for other illnesses or diseases as well, including those mentioned by others that don’t need a “biomarker”, neither to be diagnosed nor to be treated. I view signs of exclusion and inclusions signs of any diagnosis, naturally also as part of ME/CFS.

I think thyroid function test is a good example of this. It accurately measures a value that can be tied to pathology but it is not a biomarker of a certain illness in itself. If a patient visits a GP he will never only get a thyroid function test, he will always have a whole barrage of tests done. Testing TSH is of value precisely because it can gives clues to specific pathologies. Once a TSH test returns an abnormal value one can then be sent to a specialist where additional tests are then run to help untangle said pathology and hopefully come up with a treatment. It is of use precisely because it helps untangle pathology. My understanding is that if TSH was just a test that would divide the population into higher and lower values without clues to pathology it would not be used in the first place at all. I don’t see why it would be any different for ME/CFS.

If your “test” just tells you this person has ME/CFS it would not be different to an TSH test that separates the population into high and low values without providing the crucial information that it actually carries. This value seems to largely come from it giving pathological clues. That is to say in a patient with sympoms that match ME/CFS you'll be running your complex machinery of diagnostics just like before, but at some point are now also looking at whether this newly discovered clue for pathology is occurring. This will certainly in many cases shorten the barrage of tests a patient has to undergo, but you don’t need a “diagnostic test” for this. Of course once pathology is discovered you don't test for ME/CFS anymore as it likely stops existing in its current form.

What happens in practice for patients in ME/CFS I think can be mixed, at least in terms of patients picked up in studies used for an accuracy debate. There are those, including those in BPS land, that use ME/CFS (in that case usually CFS) as a dumping ground without performing a single test and then there are the majority of patients that have gone through a whole barrage of tests. The large studies unfortunately often tend to include a lot of patients from the first category, with smaller pathological studies tending to be looking more at the second set of patients. In both cases you will have to define your accuracy against a certain predetermined set of people which creates the whole circular argument of accuracy.

 

I strongly suspect that is because they are not measuring what is wrong - so would not be expected to come up longer.

A combination of tests is very plausible - maybe as many as three or four, but in at least a good proportion of cases at least one of them should be abnormal. What we are seeing with research studies is chasing shifts of tests within the normal range. In hyperparathyroidism the alkaline phosphatase is raise, the calcium raised and if I remember rightly the phosphate low. In myeloma the phosphate is not low. In osteomalacia AP is high and calcium a bit low. And so on. But one or other test is outdside the normal range.

I think the same applies to liver tests. Rations of enzymes like ALT and AST are relevant but at least one of them is abnormal.

 

So if it turns out we need to look for signalling molecules that aren't easy to measure, researchers have at least two problems: working out how to measure them, then working out what the ranges are in healthy people?

(There are probably a lot more problems, specially if a minor illness could completely skew the 'healthy' profile, but don't want to be too pessimistic!)

 

maybe not quite so bad. `Establishing a reliable method for measurement is all about calibration and checking useable ranges in terms of noise and slopes and interference from o there molecules and population scatter and suchlike. By the time you have a replicable marker you are likely to have a normal range.

What I think we should be wary of are studies that find that the mean IL-48 level in ME/CFS of 23.9 is statistically significantly lower than in normals - 19.6 with a range of 9 to 30. (Maybe with similar differences in three other factor with a certain combination seeming to have 'high specificity and sensitivity'.) It may be a real difference but if it is related to causal factors that would really define ('diagnose') an ME/CFS process it is likely to be a loose relation. It is also quite likely to be biased by unknown factors relating to cohort selection.