Preprint Replicated blood-based biomarkers for Myalgic Encephalomyelitis not explicable by inactivity, 2024, Beentjes, Ponting et al

Thank you @Chris Ponting, @Simon M, Sjoerd, Ava, Julia, Amanda, Nima and Gemma :party::party::party:

Some suggestions:

For the purpose of this particular paper I would suggest simply stating that more females than males have an ME/CFS diagnosis. It's not relevant here whether it's 3:1 or 5:1 or whatever
paper said:
To our knowledge, the overall combination of blood marker changes we observed does not present in any other disease. For example, although primary biliary cholangitis is accompanied by elevated ALP and GGT levels (and post-exertional malaise [51]) it is also marked by high circulating levels of bilirubin rather than the lower levels we observe for ME (Fig. 2A).
Ref #51 only describes post-exertional fatigue, also states this gets better with exercise. Can't categorically discount PEM happening here - could be the authors gave a poor description of PEM due to their focus on fatigue, and that the study cited for the improvement with exercise is just as poor as the ME GET studies - but if there is PEM in primary biliary cholangitis this paper doesn't show it.
paper said:
According to UK National Health Service guidance, exercise is “the miracle cure we’ve all been waiting for” [5]. Nevertheless, exercise is not a universal panacea: it is contraindicated among those with cardiovascular disease, anaemia and hyperthyroidism, for example
For anaemia only intensive exercise is contraindicated, other exercise is still recommended. It's probably the same for the other conditions. I wonder if this section is necessary to have at all? Or if it can be made less specific? Really all that can be claimed is that some types of exercise are contraindicated in some health conditions, typically (always?) that's intensive exercise which is cautioned against, and nobody is recommending intensive exercise for ME

In fact, on reflection, much of the introduction section could be shortened significantly as it has little relevance to the study itself. IMO there's really no need for every ME study to repeat all the general background and stats (which are often poorly evidenced), instead just link to a review on all that (sorry, can't think of a good one on top of my head). All that's needed for this particular study are the bits in bold
introduction said:
Physical inactivity accelerates the loss of cardiovascular and strength fitness, shortens healthspan and increases all-cause mortality risk [1, 2, 3]. It lowers insulin sensitivity and elevates the synthesis of triglyceride, ceramide and sphingomyelin in muscle [4]. According to UK National Health Service guidance, exercise is “the miracle cure we’ve all been waiting for” [5]. Nevertheless, exercise is not a universal panacea: it is contraindicated among those with cardiovascular disease, anaemia and hyperthyroidism, for example [6]. A patient might also only accept exercise as treatment if they believe its benefit outweighs its cost [7]. Myalgic encephalomyelitis (ME; also known as chronic fatigue syndrome, CFS) is a [suggestion: an often debilitating] disease of unknown pathogenesis defined by post-exertional malaise (PEM), the [suggestion: a specific pattern of] dramatic worsening of symptoms after even minor mental or physical exertion [8]], which usually lasts at least 24 hours, in contrast to other fatiguing illnesses [9]. ME has no cure and no widely effective therapy [10]. About 10% of people experiencing viral (such as with Epstein-Barr, Ross River virus or SARS-CoV-2 virus) or bacterial (such as with Coxiella burnetii) infection subsequently present ME or ME-like symptoms [11, 12]. In addition, over one-third of people with ME report not experiencing an infectious episode preceding their initial symptoms [13, 14]. Full recovery from ME is rare, at about 5% [15]. It is a female-dominant disease, with females outnumbering males by up to five-to-one; females also report more severe symptoms [13, 14]. In common with many female-biased diseases it has a high burden (e.g., in disability-adjusted life years) and low overall research funding [16]. ME is not rare, as it affects 0.19% − 0.86% of people in western countries [17, 18]. Individuals with ME commonly report PEM, pain, fatigue, sensitivities to noise, and cognitive and autonomic deficits [13] and a health-related quality of life worse than 20 other conditions [19].

There are no clinical biomarkers for ME. A high priority for people with ME is an accurate and reliable diagnostic test [20]. Findings from dozens of biomarker studies have shown limited reproducibility, perhaps due to their typically low sample sizes, their frequent use of inappropriate statistical tests [21] and the known heterogeneity of ME’s symptoms and potentially aetiology [22].

Whilst cardiopulmonary exercise testing does not initially differentiate between people with ME and control individuals, it does so in a follow-up test one day later [23, 24]. This test, however, is not in common use because it risks triggering PEM.

Any clinical biomarker would need to account for individuals’ inactivity relative to the general population. This is because many people with ME do not exercise and often restrict their activity [25] to reduce the risk of subsequent PEM. Some have proposed that it is this avoidance of activity that inhibits recovery by perpetuating ME symptoms following an acute illness [26, 27, 28]. However, therapies based on physical activity or exercise are not effective as a cure [29], implying that ME is instead an ongoing organic illness [30, 31]. It has also been claimed that any physiological abnormalities seen in people with ME might be caused by their inactivity [32].

In this study, we undertake 3 groups of analyses using UK Biobank (UKB) data [33][suggestion: clarify this is not the Cure ME biobank] on (i) 31 blood cell and 30 blood biochemistry phenotypes; (ii) 251 NMR-measured metabolites; and, (iii) 2,923 proteins. Specifically, we quantify which blood traits, Nuclear Magnetic Resonance (NMR) metabolomics, and proteomics features are significantly different between ME cases versus controls, for males or females, or all combined, controlling for age (and sex for male and female combined analyses). The large UK Biobank data sets for ME cases and controls provided substantial statistical power to evaluate hypotheses, also allowing comparison between male-only and female-only analyses, something that had not been previously achievable. We take advantage of three mediators of sedentary lifestyle to determine whether any molecular or cellular trait associated with ME cases is explicable by physical inactivity.[suggestion: rewrite to only mention each point once e.g male-female, 3 mediators]
 
I tend to agree with @Ravn about not needing a lot of background on ME/CFS, and also with @Hutan 's post (on the epidemiology thread) on sex ratio.

Just as a technical point. My main concern about internet recruitment originally was that the GWAS might pick up an allele linked to behaviour that included answering internet calls. The defence would be that we do not know of any such alleles and if they had a major influence they might have been picked up in other studies.

But we do have one genetic difference that is well known to be associated with behaviours relating to health care and that is the XX, XY difference. Rates of attending GPs and outpatients are confounded by female-related health issues in long adults, which is where differences are regularly reported. But we know that women are more likely to deliberately take up healthy diets (regardless of whether they stick to them!). My guess is that a call for volunteers for an ME/CFS study is likely, for a complicated combination of reasons, to be overweighted towards women responders by a factor of 1.5-1.7. I have no data but that means that for me the most likely interpretation of a 5:1 response is that it reflects a 3:1 prevalence.

Maybe, as Rain says, the safest thing, not to raise hares about methodology, is just not to quote a figure. It isn't relevant beyond there being a female predominance.
 
Last edited:
@Hutan is the post now moved on sex ratio.
Not sure if that is a question, but no, no posts were moved but I did another post on that topic here:
ME Epidemiology - prevalence, peak ages of onset and gender ratio


I totally agree with @Ravn about the Introduction and the lack of a need to get into the details of ME/CFS epidemiology. I very much agree with the suggested shorter version.

I agree that these two sentences aren't needed.
Whilst cardiopulmonary exercise testing does not initially differentiate between people with ME and control individuals, it does so in a follow-up test one day later [23, 24]. This test, however, is not in common use because it risks triggering PEM.
I would like to note for future reference though, that I don't think the 2xCPET test is not in common use because it risks PEM. The clinics that offer the test specifically to people with ME/CFS in the US and in Germany don't have any shortage of participants, many of them doing it to support insurance claims. While people at the severe end of the ME/CFS spectrum won't or can't do the test, people with mild ME/CFS can and do. The sentences could be interpreted as meaning that people with ME/CFS choose not do a diagnostic test because of a bit of post-exertional malaise. Actually, I think many people with ME/CFS would walk over hot coals if it meant getting a clear diagnosis. Also, the sentences as written make it sound as though the test is reliably and unquestionably diagnostic.

I think there are three main issues resulting in the 2xCPET not being in common use in many places
1. The uncertainty as to whether it really is diagnostic,
2. The difficulty of finding an exercise physiology facility within a feasible travelling distance that will do the two day testing protocol
3. The lack of awareness of the test among people with ME/CFS and exercise physiologists

Anyway, moving on from the Introduction...
 
Not sure if that is a question, but no, no posts were moved but I did another post on that topic here:
ME Epidemiology - prevalence, peak ages of onset and gender ratio


I totally agree with @Ravn about the Introduction and the lack of a need to get into the details of ME/CFS epidemiology. I very much agree with the suggested shorter version.

I agree that these two sentences aren't needed.

I would like to note for future reference though, that I don't think the 2xCPET test is not in common use because it risks PEM. The clinics that offer the test specifically to people with ME/CFS in the US and in Germany don't have any shortage of participants, many of them doing it to support insurance claims. While people at the severe end of the ME/CFS spectrum won't or can't do the test, people with mild ME/CFS can and do. The sentences could be interpreted as meaning that people with ME/CFS choose not do a diagnostic test because of a bit of post-exertional malaise. Actually, I think many people with ME/CFS would walk over hot coals if it meant getting a clear diagnosis. Also, the sentences as written make it sound as though the test is reliably and unquestionably diagnostic.

I think there are three main issues resulting in the 2xCPET not being in common use in many places
1. The uncertainty as to whether it really is diagnostic,
2. The difficulty of finding an exercise physiology facility within a feasible travelling distance that will do the two day testing protocol
3. The lack of awareness of the test among people with ME/CFS and exercise physiologists

Anyway, moving on from the Introduction...

Can we set up a new thread on this topic? It throws up some interesting questions in the context of 'maybe things moving forward' for the future and with implementation of guideline etc, but I think I got the impression point 1. was a key issue when I last read about it.

Then I realise if it is being commonly used in some big countries then there might by now be more info on that? it would also just be interesting to get an idea how many have been/are being run these days etc. ..
 
Last edited:
it would also just be interesting to get an idea how many have been/are being run these days etc.

I suspect that, because Workwell carries them out privately for individuals needing evidence for insurance or pension claims, many of the results aren't published. We would probably only hear about those run as part of trials.
 
For the purpose of this particular paper I would suggest simply stating that more females than males have an ME/CFS diagnosis. It's not relevant here whether it's 3:1 or 5:1 or whatever
Yes. Statements about ME being a mainly female illness are very worrying, as this risks ME being seen as a 'women's problem' and therefore not to be taken seriously. The immediate thing that struck me, as mentioned by a few other posters, is that it may simply be that women may more actively seek help and to take part in trials, whereas it may be more of a 'guy thing' to not talk about it and not join trials. This may skew people's perceptions.

There's an account on Twitter some folks here may know of, called Royal Free 1955, that posts old news articles about ME. What strikes me in the 1980s-early 1990s articles, is how many feature male sufferers. When I got ME in 1983 I only met 3 people with ME and all were men. Now with social networking I know many men with it. Yet nowadays, news media portray it as more a women's issue. It's a disturbing shift.
 
It looks like the protein with the greatest significance for men only is BCHE (adjusted p=0.0003) which is a cholinesterase - Butyrylcholinesterase.

upload_2024-8-31_17-20-32.png

Its function appears to be as a non specific cholinesterase that is structurally similar to acetylcholinesterase, but that in addition to just acetylcholine it also breaks other chain lengths acylcholines as well as certain drugs. It's produced in the liver and it's blood levels can be indicative of liver function supposedly.

Its cohen's d effect size from the raw data is 0.8, and its Total Effect is also significant in the paper. I'd be curious to see what the raw data looks like as a strip scatter plot. With cohen's d=0.8, normal distributions, and the sample sizes used here I guess it might look something like this simulation I made:

upload_2024-8-31_17-30-51.png

@Jonathan Edwards I know you had some interesting theories around acetylcholinergic neurons - do you have any thoughts on this? Given the possible link of gulf war illness to nerve agents that affect cholinesterase, and to the slight possibility that Wuest's study has picked up something with ACh nerve endings in their muscle biopsy staining, could this be interesting?
 
Last edited:
This looks like a really exciting paper I'm looking forward to properly reading through it.





Here is the same table you showed but for the NMR results rather than blood traits (only showing some of the top ones, there's many more with Cohen's D > 0.2). What's really nice is that is corroborates the lipids/fats results from the blood biochemistry (as measured enzymatically) by measuring it in an orthogonal way (NMR). This to me helps allay concerns about the results being something technical or not clinically relevant.


View attachment 22913

What's also cool is that findings relating to high levels of triglycerides, unsaturated fatty acids and low levels of phosphatidylcholines are what are reported lipkin and fiehn's metabolomics study - and in general these findings are probably the most well replicated of the metabolomic data done at baseline (also seen in Naviaux and I think possibly Tronstadt and Hanson's work).

To compare some of the lipid findings to previous studies that have done lipidomics:

This thread's paper describes low levels of phosphatidylcholine, sphingolipids (types of membrane lipid) and high levels of triglycerides.

Low levels of phospholipids and sphingolipids are also seen in Che & Lipkin et al 2022, but triglyceride levels are more conflicting as there are high unsaturated but low saturated levels - this is also a female only effect:
upload_2024-8-31_18-2-50.png

Naviaux et al 2016 replicate low phospholipids and sphingolipids - but do not report on triglycerides so it's unclear if it's not significant or not measured.
upload_2024-8-31_18-3-31.png

Hoel et al 2021 replicate low phospholipids (PC,PE,LPC) and high triglycerides (TAG). However it reports no difference in sphingolipids (SM, CER)/ if anything slightly increased sphingolipids.
upload_2024-8-31_18-4-1.png


Most other metabolomics studies do not measure lipid levels, though there are difference in free fatty acids in the 2 day CPET paper by Hanson's group. I'm sure there are other lipidomics papers I have missed.

EDIT: Also one of the significant proteins in the combined analysis in this thread's paper is the maybe relevant seeming 'Phosphatidylcholine-sterol acyltransferase' (adjusted p = 0.023).
 
Last edited:
Interesting. I don't know if this is addressed in the paper - but are these biomarkers all seen together in the same individual, or do the traits present together in different patient clusters? Eg one cluster with high triglyceride and and a separate cluster with elevated liver markers, and maybe a big cluster with no trait differences at all?

I think this is a key point. I don't think it is addressed in the paper and maybe some analysis should be made.

I am travelling but hope to get some detailed comments down for Chris and co within a week or so. I think there are some major issues to sort out in relation to confounding co-morbidities and the uncertainties around ascertainment of an ME/CFS diagnosis.

I think the identification of superoxide dismutase 3 and complement proteins may be an important finding but I suspect that many of the other findings reflect confounding factors that increase the chance of a CFS diagnosis through all sorts of routes. (They are probably the expected markers of other predisposers to persistent fatigue in this age group.) The study may turn out to be a useful dry run for DecodeME in terms of assessing the risks of selection biases. Fortunately, I think these are going to be greater in this cohort than the Decode cohort. Comparing the two may give some very useful information about essentials for recruitment methodology.
 
To compare some of the lipid findings to previous studies that have done lipidomics:

This thread's paper describes low levels of phosphatidylcholine, sphingolipids (types of membrane lipid) and high levels of triglycerides.

Low levels of phospholipids and sphingolipids are also seen in Che & Lipkin et al 2022, but triglyceride levels are more conflicting as there are high unsaturated but low saturated levels - this is also a female only effect:
View attachment 22977

Naviaux et al 2016 replicate low phospholipids and sphingolipids - but do not report on triglycerides so it's unclear if it's not significant or not measured.
View attachment 22978

Hoel et al 2021 replicate low phospholipids (PC,PE,LPC) and high triglycerides (TAG). However it reports no difference in sphingolipids (SM, CER)/ if anything slightly increased sphingolipids.
View attachment 22979


Most other metabolomics studies do not measure lipid levels, though there are difference in free fatty acids in the 2 day CPET paper by Hanson's group. I'm sure there are other lipidomics papers I have missed.

EDIT: Also one of the significant proteins in the combined analysis in this thread's paper is the maybe relevant seeming 'Phosphatidylcholine-sterol acyltransferase' (adjusted p = 0.023).

Germain et al 2020 reported higher sphingolipids than controls. It seems phosphatidylcholine wasn't significant.
upload_2024-8-31_15-25-32.png
Pathway abnormalities have already been described for sphingolipids in ME/CFS patients, but the results are inconsistent. In a similarly sized cohort, Naviaux and colleagues found ceramides and sphingomyelins to be decreased in the patient cohort, regardless of the gender [16]. This result was not reproduced by Nagy-Szakal and colleagues, where they did not see a consistent decrease in ceramide levels but reported on decreased levels of longer ceramides in patients without IBS [18]. Our dataset instead shows an increase in ceramide levels for our patient cohort compared to controls. When splitting our patient cohort based on their gut symptoms, ceramide levels for patients without gut symptoms tend to be even further increased for CER (18:0), CER (18:1), and CER (20:0).

The inconsistencies reported between studies of ceramide metabolism, and more generally sphingolipids, points to a complicated landscape in ME/CFS. Their prevalence in cell membranes, with their involvement in many cell processes, as well as the numerous unrelated pathologies linked to an imbalance in their levels [30], suggest that changes in sphingolipids are not directly linked to ME/CFS, but are rather a consequence of the disease and the life style that it inflicts on patients such as disturbed diets, lack of physical activity, and medication regimes.
 
In fact, on reflection, much of the introduction section could be shortened significantly as it has little relevance to the study itself. IMO there's really no need for every ME study to repeat all the general background and stats (which are often poorly evidenced), instead just link to a review on all that (sorry, can't think of a good one on top of my head).
So as not to take this thread too far off topic I've made a separate thread for it. Hopefully somebody there will be able to suggest some good general reviews suitable for linking to.
https://www.s4me.info/threads/how-to-write-a-good-introduction-for-an-me-cfs-paper.40017/
 
BCHE
Its cohen's d effect size from the raw data is 0.8, and its Total Effect is also significant in the paper. I'd be curious to see what the raw data looks like as a strip scatter plot. With cohen's d=0.8, normal distributions, and the sample sizes used here I guess it might look something like this simulation I made:

The z score for BCHE in ME/CFS men is over 5 (for women it's heading towards 2). Given the z score of 5, I'm not understanding your hypothetical scatter plot. Shouldn't the ME/CFS data points be a lot higher? I could easily be misunderstanding something there.


I'm quite keen on the idea of problems being created by toxic cholinergic substances - as seems to be the case in Gulf War Illness, and also in Ciguatera (which has a Long Ciguatera phenomenon). Here, BCHE seems to be elevated relative to the mean in controls. And it's a cholinesterase

which is a cholinesterase - Butyrylcholinesterase.
so presumably there is plenty to break down some toxic cholinergic substances.

And it is low levels, not high levels that indicate inflammation and liver problems:
Butyrylcholinesterase as a prognostic marker: a review of the literature
Butyrylcholinesterase (BChE) is an α-glycoprotein synthesized in the liver. Its serum level decreases in many clinical conditions such as acute and chronic liver damage, inflammation, injury and infections, and malnutrition.
BChE levels are strongly influenced by inflammation, sensitively decreasing in the acute inflammatory phase and promptly increasing when inflammation improves
High levels seem to be associated with high protein intake. It also seems to be higher in people with higher body weight and hyperlipidemia. There is a suggestion that BChE can be high when there is mild inflammation and metabolic syndrome.
Conversely to what happens in case of malnutrition, during metabolic syndrome (diabetes, hypertension, hyperlipidemia) and/or hepatic steatosis, BChE positively correlates with the presence of mild inflammation
So, it seems a bit complicated.

This UK Biobank study suggested that there were markers of insulin resistance (elevated triglycerides-to-HDL cholesterol)
Could the higher levels just be the result of a high protein diet and/or high triglycerides?

I think it would have been good to have some information on the BMI of participants.
 
And it is low levels, not high levels that indicate inflammation and liver problems:
Butyrylcholinesterase as a prognostic marker: a review of the literature


High levels seem to be associated with high protein intake. It also seems to be higher in people with higher body weight and hyperlipidemia. There is a suggestion that BChE can be high when there is mild inflammation and metabolic syndrome.

So, it seems a bit complicated.

This UK Biobank study suggested that there were markers of insulin resistance (elevated triglycerides-to-HDL cholesterol)
Could the higher levels just be the result of a high protein diet and/or high triglycerides?

I think it would have been good to have some information on the BMI of participants.


At least in my case it doesnt fit together.
The first 8 years I had a mild case of ME/CFS. I couldnt do sport anymore but I tried to walk a lot to keep my condition even if I was often very exhausted.
My cholesterol level was always perfect but I had high level of triglycerides. Also my BMI was good. My metabolic doctor was always surprised. She proposed to me to eat more vegetarian food. The more my food was vegetarian the worse my triglycerids were. Later I had stomach problems and I tried low carb diet and I was very surprised, my triglycerides vere perfect.
 
Back
Top Bottom