Deep phenotyping of post-infectious myalgic encephalomyelitis/chronic fatigue syndrome, 2024, Walitt et al

[bobbler]

Yes, we're in a strange world where the Daily Mail is more sympathetic than The Guardian.



I think they're trying to claim an effect that doesn't exist.

There was little change for 4/6 healthy controls, 1 had a modest reduction, 1 had a large reduction. There was little change for 2/8 patients, 1/8 had a small reduction, 2/8 had a small increase, 1/8 had a modest increase, 2/8 had a large increase.

I'm not seeing a distinct pattern and I think there is too much noise in the method, or the method doesn't consistently invoke the hypothesised effect (reduction in motor cortex excitability)

It has been that way for a long time it seems.

As a side note, I've gone back to the daily mail article and finished reading to the end - which includes a description of the illness. I will paste it on the thread to do with the NHS stuff because this seems pretty 'simple English' as much as that was. But due to including certain aspects seems that bit more appropriate.


"Symptoms of CFS vary per patient and over time.

The most common include an extreme physical and mental tiredness that doesn't go away with rest as well as problems sleeping and with thinking, memory and concentration.

Other symptoms include muscle or joint pain, a sore throat, headaches, flu-like symptoms, dizziness and nausea as well as a fast or irregular heartbeat.

At its mildest, CFS sufferers can perform everyday activities with difficulty but may have to give up hobbies and social activities to rest.

The most severe CFS patients are essentially bed-bound and may full-time care unable to feed or wash themselves or even go to the toilet unassisted.

One of the greatest challenges with CFS is getting a diagnosis due to a lack of a tests that can show a patient has it.

With no such test currently available, patients are forced to undergo diagnosis by process of elimination, with medics progressively ruling out other conditions until CFS is the only one remaining.

There is currently no cure for CFS. Treatment instead revolves around therapy, lifestyle changes and the use of some medications to alleviate symptoms."

 

[Hutan]

Hmmm - does this muddy the waters just enough for Cochrane...

I've wondered the same. I think it's highly likely that Walitt and his friend Shorter are well connected with the BPS people who delayed progress on the new Cochrane exercise therapy review with their complaint. I'm sure there is plenty of communication and that drafts of the paper were shared with them.

It would not surprise me at all if there is a BPS plan to produce papers that prop up the ideology, including CBT and GET, before the analysis for the exercise therapy review starts.

 

[bobbler]

I've wondered the same. I think it's highly likely that Walitt and his friend Shorter are well connected with the BPS people who delayed progress on the new Cochrane exercise therapy review with their complaint. I'm sure there is plenty of communication and that drafts of the paper were shared with them.

It would not surprise me at all if there is a BPS plan to produce papers that prop up the ideology, including CBT and GET, before the analysis of literature for the exercise therapy review starts.

I just remember someone saying how arduous it was to be part of this trial and do hugely feel for the patients who went through all of this

I guess at the moment you can’t stop people riffing on what they want but it didn’t always used to be this way? The data itself is supposed to almost lead to conservative discussions of everything that could mean in the old days? Or am I imagining days the didn’t exist but should?

 

[SNT Gatchaman]

PI-ME/CFS participants had more self-reported total cognitive complaints (66.9 ± 14.7 versus 92.1 ± 17.6, p = 0.00006) and in all five cognitive domains measured: attention, verbal memory, visuoperceptual, language, and visual memory (Fig. 1c).

Why were the healthy controls reporting such high cognitive complaints? These scores are way higher than I would have self-reported prior to 2021 (which would have been zero on PHQ15 and I'm sure zero-equivalent on anything else).

 

[Hutan]

Why were the healthy controls reporting such high cognitive complaints? These scores are way higher than I would have self-reported prior to 2021 (which would have been zero on PHQ15 and I'm sure zero-equivalent on anything else).

I think the answer might lie in the age distributions. With about 45% of the healthy controls in the 50 to 60 year age range, you have a lot of people feeling older, perhaps having some issues with menopause too, and so tending to not feel quite as sharp as they did when they were in their 30s.

It goes to the comments about the poor cohort matching. What was it?, only 23 or 24% of the ME/CFS cohort were in that oldest age range, with 45% in the 18-30 age range. The selection of more older healthy controls may have served to dampen down the differences in self-reported cognitive performance between the cohorts.

(Edited to add: I'm still struggling to understand how a study undertaken by one of the most prestigious and well resourced institutions in the world could be so poor. I mean, cohort matching is the most basic thing to do, and we aren't even talking about anything more complex than sex and age range.

And, seriously, only data from eight participants for that chart of (one day) CPET outcomes? Contrast that with a study in a university that almost none of you will have ever heard of, by a team that was completely and utterly under-resourced, in a country with a total population that is smaller than the whole state of Maryland - they managed to get more people with ME/CFS through a two-day CPET investigation than the NIH could manage. There's no conclusion possible other than decision-makers in NIH really didn't give a toss.)

 

[Jaybee00]

From Brian Vastag Mastodon

https://sciencemastodon.com/@brianvastag/111978289834885835


Avi Nath at NIH tells me UCSF is trying to launch a clinical trial of a class of drugs called checkpoint inhibitors in ME/CFS.

The paper he published yesterday mentions these drugs as a target for testing in the disorder, on the theory that they can help clear persistent antigens.

They're cancer drugs & I don't know much about them.

 

[Hutan]

Childhood Trauma and Sexual and Physical Abuse
It's mentioned in the Methodology that the participants completed the Childhood Trauma Questionnaire Short Form and the Sexual and Physical Abuse Questionnaire.

In the paper, this is all I can find about this, in the discussion:

There was also no difference in psychiatric history or reporting of traumatic events between the two groups. Thus, psychiatric disorders were not a major feature in this cohort and did not account for the severity of their symptoms.

By only reporting about it in the Discussion, the omission of a link to the data is less obvious.

I found a line for the Childhood Trauma Questionnaire in Supplementary Data 7. Higher scores indicate more childhood trauma. There was no significant difference between the cohorts, and both cohort means indicated a low level of trauma. However, the healthy controls had a higher mean, so slightly more trauma, (39.10) compared to the ME/CFS cohort (35.8)

I could not find any reporting at all for the results of the Sexual and Physical Abuse Questionnaire. I don't know why that would be. I can't see that privacy would be the reason for not reporting means, especially as the Childhood Trauma Questionnaire covers sexual and physical abuse during childhood. Why report the results of one survey and not the other?

 

[Andy]

Datasets from this study have been added to the mapMECFS portal, https://www.mapmecfs.org/dataset/?organization=nih-intramural

 

[Amw66]

Declines in submaximal CPET measures corresponding with PEM symptoms and signs on the second day of a two day CPET have nothing to do with effort preference in people meeting accepted physiological criteria for maximal exertion. You can’t choose or fake your anaerobic threshold.

 

[Amw66]

Effort Preference is a way to test for anhedonia, the core symptom of Major Depressive Disorder. The unsupported introduction of Effort Preference to the ME world may well be laying the groundwork for more destructive studies, characterizing ME as a somatoform illness. FN 15

 

[Sid]

Hmmm - does this muddy the waters just enough for Cochrane...

No doubt it doesn’t help. Likely it will damage us in all sorts of ways. Can’t eat/swallow? You can prefer your way out of that.

 

[Hutan]

Gene expression profiles

See Fig. 9: Male and female cohorts have distinct differential gene expression profiles in the muscle.
The title claims that male and female cohorts are different when it comes to gene expression in muscle. I could believe that. But look at the PCA charts:



Figure a
There's no overall difference between gene expression in the muscle between the healthy volunteers and ME/CFS group. They've used Principal Component 1 and Principal Component 2 - which explain 16.1% and 10.6% of the variation, which is not great. There's no clear difference between the two groups. Also, note the numbers of data points - I make that 8 healthy controls, 14 ME/CFS - also not great.

Figure b
And then, they give us Chart B, where all the males (healthy + ME/CFS = approx 11) are grouped together, and all the females from the two cohorts (approx 12) are grouped together. They claim that this shows that gene expression in muscle is different in males and females. But, first off, look at the separation of the male and female data points - yeah, there isn't anything like two clear groups. And then look at the axes labels. They are showing PC2 and PC3, explaining 10.6% and 9.7% of the variation. It's a pretty weak result to warrant that Figure title claiming differential gene expression in the muscle of males and females, particularly given that small sample size.
(BTW: Watch out for the use of PCAs that don't explain much variance elsewhere in the paper)



Figure c is a Venn diagram, I think showing that males had 593 differentially expressed genes (i.e. there is different expression between ME/CFS males and healthy males) in muscles, females had 328 differentially expressed genes in muscles, and, of those genes, only 15 genes were differentially expressed in both males and females. Given the tiny sample size (e.g. about 6 ME/CFS males and 6 healthy males) and the very poor replication of findings in the male and female subsamples, surely there are a lot of false positives there. While there may be some sex differences, I'm always dubious when a difference can only found by stratifying small samples into miniscule ones.

I think Figures d, e, f and g are mostly the product of the false positives and can be ignored. The authors take the expression values for each participant for only the genes that were differentially expressed between the ME/CFS and healthy cohorts. And then they do a Principal Component Analysis on them. And, amazing! the ME/CFS and healthy cohorts form discrete groups. Well, not so amazing, because they only used the data points that were different between the two groups, and didn't use the datapoints that didn't show a difference. So, those PCAs aren't really telling us anything new.

( Just further on that - the paper says "In the male PCA of 593 DE genes, clustering was observed based on the disease status (Fig. 9d, e)." and
"In the females, PCA of the 328 DE genes showed distinct clusters of HV and PI-ME/CFS samples (Fig. 9f, g)."
I think that's highly misleading and circular for the reason I explained in the preceding paragraph.)


Figures h,i, j, k - h and i are for males; j and k are for females.
The authors try to group the differentially expressed genes into pathways. The two separate figures for each sex are for pathways that are claimed to be up-regulated and down-regulated.
So, that's adding quite a layer of interpretation that may not be well founded. @DMissa had some concerns about this approach - I've copied his comment here:

I wish there was more detail in the gene expression sections but it's limited to the typical surface level bioinformatic pathway analysis that too often neglects to take into account the actual function of the particular gene products. I really hate this modern trend of "well 5/9 of these genes "involved" in X process are elevated in expression so X process is probably upregulated". Okay and what if one of those 5 inhibits the pathway? What if a rate-limiting enzyme of the pathway is meanwhile downregulated? Well, I guess it is more taxing to read about the function of a few dozen genes of interest that the pathway analysis turned up, rather than only giving the data to a bioinformatician and asking them to press a button and calling it a day. One could call this issue a matter of "effort preference".

I couldn't see much overlap in the pathways listed for males and females, and find that worrying. Yes, there probably are sex differences, but males and females with ME/CFS both report that dead, heavy feeling in muscles in response to exertion, so there should be some commonality.

So...
there could be some gene expression differences in there that are of interest. In particular, I'd like to look at the 15 gene expressions that were different between ME/CFS and healthy volunteers in both males and females.
But real gene expression differences are surely swamped by the false positives created by the small sample sizes and mis-matched cohorts.

(I think there is a stray sentence at the end of the Figure 9 caption that is actually talking about a supplementary figure.)

I'm done for the night. Feel free to point out errors or disagree with me.

 

[Simon M]

Consideredtogether, these data suggest that there is an alteration in autonomictone, implying central nervous system regulatory change.

Autonomic dysfunction in ME/CFS

Tilt table testing found no POTS or OI differences with controls. However, they also found reduced baroslope, which I think means that regulation of blood pressure when it changes is slower than optimal.

A decrease in baroslope (Fig. 2h) and longer blood pressure recovery times after the Valsalva maneuver (3.0 ± 0.2 versus 4.1 ± 0.4 sec, p = 0.014) in PI-ME/CFS reflect decreased baroreflex-cardiovagal function

A wearable 24-hour ECG showed reduced heart rate variability (HRV). This is an indicator of ill-health and might well be normal in chronic illness. There were some specific HRV differences that I don't understand:

PI-ME/CFS participants also had altered frequency domain differences, with decreased high frequency (HF) and low frequency (LF) power (Fig. 2e, f). Non-linear analyses adjusted by hour of the day further demonstrated group differences in (p = 1.1E−12), SD2 (p = 2.1E−05), and SD1/SD2 (p = 3.9E−07).

They also found increased daytime heart rate suggesting increased sympathetic activity and a reduced drop in nighttime heart ratee in PwME, suggesting reduced parasympathetic activity [consistent with observed
HRV changes in rMSSD, HF power, pNN50, and SD1].

Considered together, these data suggest that there is an alteration in autonomictone, implying central nervous system regulatory change.

Note that for the Hr & HRV studies they only had data on 13 or 14 subjects. I'm not sure how they failed to get 24-hour ECG data (which is easy to set up and easy to wear) on all subjects. Apart form 2 HRV measures (lnHF&LF), the p values are modest and not corrected for mulitple comparisons.

Do you think these findings are significant, @Jonathan Edwards?

 

[siobhanfirestone]

I thought the same.

I think the T cell populations they flagged up were those with more innate function. Nothing in what they or others have found would suggest a need to clear foreign antigen. This looks to me like pretty low level immunological analysis-by-meme.

apparently "Avi Nath at NIH tells me UCSF is trying to launch a clinical trial of a class of drugs called checkpoint inhibitors in ME/CFS. The paper he published yesterday mentions these drugs as a target for testing in the disorder, on the theory that they can help clear persistent antigens. They're cancer drugs & I don't know much about them"
I am quite worried about this as, whilst we both disagree about how to interpret the exact data regarding the specific immune manifestations and cells that might be driving ME, wether its innate or adaptive these drugs will FUCK patients up, like they literally can CAUSE autoimmunity at worse case: https://pubmed.ncbi.nlm.nih.gov/32390537/
We know that there are likely sub-sets within ME, maybe in general between a sort of active virus VS immune system hyperactivity/autoimmunity-esque line. If the trials do not make a serious effort to work out which sub-set someone is in i strongly encourage people to stay the fuck away for their own safety frankly

 

[FMMM1]

Decided to post this on Twitter - basically @JonathanEdwards ideas - unattributed!

 

[Jonathan Edwards]

It would not surprise me at all if there is a BPS plan to produce papers that prop up the ideology, including CBT and GET, before the analysis for the exercise therapy review starts.

I am not sure that propping up ideology can have any impact on a review of evidence for efficacy of treatments.

And in as much as this study says anything specific it says that CBT and GET wouldn't address the problem itself.

 

[Jonathan Edwards]

Do you think these findings are significant, @Jonathan Edwards?

I doubt it. My main concern is that PWME are bound to be in a different autonomic state because they are the ones being tested for an illness - the ones with a threat.

 

[NelliePledge]

And in as much as this study says anything specific it says that CBT and GET wouldn't address the problem itself.

The wording was not as clear cut as yours.
May not be particularly effective (or something on those lines) was what registered with me which unfortunately still leaves the door slightly ajar.

 

[Jonathan Edwards]

Avi Nath at NIH tells me UCSF is trying to launch a clinical trial of a class of drugs called checkpoint inhibitors in ME/CFS.

The paper he published yesterday mentions these drugs as a target for testing in the disorder, on the theory that they can help clear persistent antigens.

I find this very strange. As far as I know nobody as found any evidence of residual antigen and the weight of evidence in post-infective syndromes is that persistent antigen is not the problem. So there is no reason to use checkpoint inhibition on the basis of the current study.

Moreover, if ME is driven by continued immune activation checkpoint inhibitors are most likely to take it worse. I don't buy this story of immune exhaustion. If the immune system is exhausted you may be a risk of catching new infections but in inactive system is not going to be producing ME symptoms.

 

[chillier]

Gene expression profiles

See Fig. 9: Male and female cohorts have distinct differential gene expression profiles in the muscle.
The title claims that male and female cohorts are different when it comes to gene expression in muscle. I could believe that. But look at the PCA charts:

View attachment 21163

Figure a
There's no overall difference between gene expression in the muscle between the healthy volunteers and ME/CFS group. They've used Principal Component 1 and Principal Component 2 - which explain 16.1% and 10.6% of the variation, which is not great. There's no clear difference between the two groups. Also, note the numbers of data points - I make that 8 healthy controls, 14 ME/CFS - also not great.

Figure b
And then, they give us Chart B, where all the males (healthy + ME/CFS = approx 11) are grouped together, and all the females from the two cohorts (approx 12) are grouped together. They claim that this shows that gene expression in muscle is different in males and females. But, first off, look at the separation of the male and female data points - yeah, there isn't anything like two clear groups. And then look at the axes labels. They are showing PC2 and PC3, explaining 10.6% and 9.7% of the variation. It's a pretty weak result to warrant that Figure title claiming differential gene expression in the muscle of males and females, particularly given that small sample size.
(BTW: Watch out for the use of PCAs that don't explain much variance elsewhere in the paper)

View attachment 21164

Figure c is a Venn diagram, I think showing that males had 593 differentially expressed genes (i.e. there is different expression between ME/CFS males and healthy males) in muscles, females had 328 differentially expressed genes in muscles, and, of those genes, only 15 genes were differentially expressed in both males and females. Given the tiny sample size (e.g. about 6 ME/CFS males and 6 healthy males) and the very poor replication of findings in the male and female subsamples, surely there are a lot of false positives there. While there may be some sex differences, I'm always dubious when a difference can only found by stratifying small samples into miniscule ones.

I think Figures d, e, f and g are mostly the product of the false positives and can be ignored. The authors take the expression values for each participant for only the genes that were differentially expressed between the ME/CFS and healthy cohorts. And then they do a Principal Component Analysis on them. And, amazing! the ME/CFS and healthy cohorts form discrete groups. Well, not so amazing, because they only used the data points that were different between the two groups, and didn't use the datapoints that didn't show a difference. So, those PCAs aren't really telling us anything new.

( Just further on that - the paper says "In the male PCA of 593 DE genes, clustering was observed based on the disease status (Fig. 9d, e)." and
"In the females, PCA of the 328 DE genes showed distinct clusters of HV and PI-ME/CFS samples (Fig. 9f, g)."
I think that's highly misleading and circular for the reason I explained in the preceding paragraph.)


Figures h,i, j, k - h and i are for males; j and k are for females.
The authors try to group the differentially expressed genes into pathways. The two separate figures for each sex are for pathways that are claimed to be up-regulated and down-regulated.
So, that's adding quite a layer of interpretation that may not be well founded. @DMissa had some concerns about this approach - I've copied his comment here:


I couldn't see much overlap in the pathways listed for males and females, and find that worrying. Yes, there probably are sex differences, but males and females with ME/CFS both report that dead, heavy feeling in muscles in response to exertion, so there should be some commonality.

So...
there could be some gene expression differences in there that are of interest. In particular, I'd like to look at the 15 gene expressions that were different between ME/CFS and healthy volunteers in both males and females.
But real gene expression differences are surely swamped by the false positives created by the small sample sizes and mis-matched cohorts.

(I think there is a stray sentence at the end of the Figure 9 caption that is actually talking about a supplementary figure.)

I'm done for the night. Feel free to point out errors or disagree with me.

All of this is at p=0.05 and unadjusted, so it's possible every one of those genes in that venn diagram is a false positive.

To be honest I'm not too concerned that those first two PCAs (a, b) look the way they do, they're looking at global differences in gene expression and it would make sense that the vast majority of the genes would be irrelevant.

Agree that it's not informative to do those PCAs (d, f) based on the differentially expressed (DE) genes - something more relevant might have been to take the highly variable genes across everybody, then do a PCA based on that. Then you could see for example whether more variance is explained by men/woman or ME/control or BMI or whatever (maybe they tried something like that and it wasn't interesting). Really though there are better easier ways to explore those relationships. If they had a much bigger sample size maybe it would be a way to look for subgroups/types of ME.

What would have been informative instead of a PCA of the DE genes would have been to show them as a heatmap, and I think it's telling they didn't do that in this figure when they did do it in the previous figures with the same sort of data structures. It probably looked uninteresting. So we don't get a mention of any individual genes then.

The pathways analyses actually do meet multiple testing correction criteria by the looks of it (by pooling genes into groups by related annotated function and then checking if you see that pathways come up more or less than you would expect it to by chance). Problems with this kind of analysis is 1) you need to trust the ontology annotations which may actually be extremely dodgy (eg may be automated annotation based on protein sequence, instead of being described through actual experiments), and 2) it's vague.

Still, besides the concerns about mismatched groups, I don't think the fact there are big differences in the pathways between men and women is worrying in itself, and may actually be potentially interesting. After all theres a maybe 4:1 sex bias in this disease and maybe that reflects the biology. The data suggested females have downregulated fatty acid metabolism related gene pathways (which is seen in the male data only to a much lesser extent) and males have increased neutrophil related immunity pathways (not seen in women at all). In many metabolomics studies there are much greater differences found in women and very little differences men, so this could be consistent with that.

Agree with @DMissa that we want to see actual genes and their mechanisms studied, but it doesn't look like they found any genes here you could really trust is worth following. The sample size is just miles too small (maybe WASF3 actually did meet q@0.05 and they haven't mentioned it?).

It's not really much to go on but for the sake of argument here are some experiment ideas. They mention in the text they see nominal differences in growth hormone receptor signalling and ubiquitin-transferases. These genes tag proteins for destruction so could be worth investigating cell growth, protein synthesis and degradation rates. Could look at the master growth regulator mTOR and its signalling pathways for changes in phosphorylation state. If hormone receptor signalling is altered maybe that's a clue to a possible 'factor in the blood.'