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

The Times:
Scientists have found biological signatures in the blood of people with myalgic encephalomyelitis (ME), a breakthrough that could lead to the first reliable test for the debilitating condition.
Click to expand...

I think it’s a shame that The Times article puts the emphasis on the potential for these results to lead to a blood test for ME/CFS. That would be great but I think it’s somewhat misleading to suggest that these results are likely to lead directly to a blood test. As far as I understand, these results point to small but significant differences at a population level, which gives us valuable clues as to what may be going wrong in ME/CFS, but doing the same tests on an individual would be of no use in confirming a diagnosis.
 
Andy said:
Coverage on Radio 4 starts at approx. 1:22:40
Click to expand...
Chris did very well, some great responses to questions that weren't particularly helpful from the reviewer.


Some quick notes I made on the list @forestglip posted above of the blood measures found in both cohorts. Not complete by any means.

Significantly increased in both
Alkaline phosphatase - ALP, high in liver and bone diseases, infections, vitamin D deficiency
glucose - high in diabetes, inactivity, eating too many carbs, infections, kidney and liver disease
hba1c - (indicator of longer term levels of blood glucose) as for glucose, also anemia
leukocyte count - infections, inflammation
neutrophil count - infections, inflammation
triglyceride/HDL-C ratio
triglycerides - eating a lot of carbs, being overweight and inactivity, diabetes, kidney and liver disease, hypothyroidism, sleep deprivation
triglyceride glucose index

Significantly decreased in both
HDL-C - the 'good' cholesterol, can be reduced by being overweight and inactivity
 
Robert 1973 said:
The Times:


I think it’s a shame that The Times article puts the emphasis on the potential for these results to lead to a blood test for ME/CFS. That would be great but I think it’s somewhat misleading to suggest that these results are likely to lead directly to a blood test. As far as I understand, these results point to small but significant differences at a population level, which gives us valuable clues as to what may be going wrong in ME/CFS, but doing the same tests on an individual would be of no use in confirming a diagnosis.
Click to expand...
It might be an exaggeration of this sentence in the abstract:
Nevertheless, these results keep alive the future ambition of a blood-based biomarker panel for accurate ME/CFS diagnosis.
Click to expand...

Agree that the times framing is not only exaggerating but maybe misleading.
 
InitialConditions said:
What is the stats/maths behind this statement, please, i.e., where does the 150 come from?
View attachment 26666
Click to expand...

If you do a statistical test with a significance threshold of p = 0.05, by definition there is a 5% of you getting a positive result by chance. Therefore if you do ~3000 tests as the authors have done in this paper, you expect to see 5% of those tests to be positive completely spuriously = 150 false positive results. He's using this to argue all the results are false positives.

You fix this problem by doing multiple test correction which is what the authors did, making his argument complete nonsense.
 
The no-a-priori-hypothesis claim is silly - there's a clear research question & rationale for the methodology; it's not some fishing expedition. I've only skimmed the paper but think he's conflating the cross-sex replication set (116) with the total (511) & the 150-by-chance/116 observed is wrong b/c @ q=0.05 there would be expected to be ~26 false+ve results in the total and ~6 in the sex-replicated set. He's assuming a simple per-test alpha=0.05, but they use Benjamini-Hochberg. The results are far above chance. Carson also grouses that the authors did not find any "signature" - well, they state openly that no single trait cleanly separates cases from controls - and also goes on to claim that they "could not replicate their findings" - but they present multiple layers of replication - sex stratification (166 replicated in males and females) and the AoU cohort (9/14 in the same direction). The self-reported diagnosis point is intrinsic to the UK Biobank study population.
 
Nightsong said:
He's assuming a simple per-test alpha=0.05, but they use Benjamini-Hochberg
Click to expand...
This method controls the false discovery rate (FDR) at 0.05, which is mentioned multiple times in the paper and graphs. It means that among the significant findings, the expected proportion of false positives (the false discoveries) is only 5%.

It's not that complicated and standard in large studies, so quite strange that Carson got this wrong.
 
chillier said:
Therefore if you do ~3000 tests as the authors have done in this paper, you expect to see 5% of those tests to be positive completely spuriously = 150 false positive results. He's using this to argue all the results are false positives.
Click to expand...
Because they controlled the FDR at 0.05, the expected proportions of false positives among the 116 significant features is approximately 6 (not 150).
 
The Carson comment is just blatantly wrong, so I sent an email to SMC to ask they correct it:
Carson appears either to have not read the full paper to see that they used false discovery rate correction, or to not understand what FDR is. With an FDR of q<.05, fewer than 5% of positive findings are expected to be false positives.

The study found 116 markers that replicated between males and females, which means fewer than 6 are expected to be false positives, not 150.

I ask that this comment be revised to correct this misrepresentation of the study in question.
Click to expand...
 
JemPD said:
I',m a bit bewildered by all this... arent these tests done as standard blood work for each of us before diagnosis? FBC, LFT & glucose must be, surely?
Click to expand...
I haven't read all this thread or the research yet, but Chris Ponting in the BBC interview explains that they are seeing patterns or trends rather than things that can specifically be pinned down as diagnostic at the individual level. More research needs to be done to see if they can work out specific things that will be diagnostic of ME/CFS for individuals.
 
I also don't see any significant overlaps with the results in Pu et al. - has Carson even skimmed through either paper, or is he just winging it?
Pu et al said:
Our findings signaled that the levels of glutamic acid and phosphatidylcholine (32:0) were consistently elevated in the blood of patients with depression, while those of tryptophan, kynurenic acid, kynurenine, acetylcarnitine, serotonin, creatinine, inosine, phenylalanine, and valine were lower. In urine samples, the concentrations of isobutyric acid, alanine, and nicotinic acid were increased, whereas those of N-methylnicotinamide and tyrosine were decreased. A thorough examination of 23 proteomic studies unveiled that only one protein, ceruloplasmin, was consistently altered in the blood of patients with depression. Besides, a convergence comparison was also performed to prioritize circulating metabolites. The top-ranked metabolite was tryptophan, followed by kynurenic acid, acetylcarnitine, creatinine, serotonin, and valine
Click to expand...
 
Yann04 said:
What stimgatising and derogatory language does he accuse the authors of using?
Click to expand...
I also don't get this. The only statement I see is:
​
"Evidence that there is a large number of replicated and diverse blood biomarkers that differentiate between ME/CFS cases and controls should now dispel any lingering perception that ME/CFSis caused by deconditioning and exercise intolerance (Wessely et al, 1989; Moss-Morris et al, 2013; Sharpe, 1995; White et al,2011)."​

I doubt that this can be seen as stigmatising or derogatory.
 
You must log in or register to reply here.
Back
Top Bottom