Science Media Centre:
https://www.sciencemediacentre.org/...blood-based-diagnostic-biomarkers-for-me-cfs/
As well as Chris Ponting's comment there are these two:
Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:
“I think this new research is an interesting development in the search for biomarkers for ME/CFS, and one that may possibly prove important later. But it’s crucial to note that the researchers describe it as a proof-of-concept study, that is, one that is aimed at finding out whether the overall approach of measuring epigenetic markers in blood samples can potentially be useful. Its aim is not to produce a working diagnostic tool, not yet.
“I think the study does establish that the approach is
potentially useful. This matches the quote from the lead researcher in the press release, saying that “Our discovery offers the potential for a simple, accurate blood test to help confirm a diagnosis […]”. He is rightly careful not to claim that the research already offers such a test.
“The research paper points out a number of important limitations about the research and its findings, and mentions other research that will have to be done before this can be taken further. Some of that other research is likely to be time-consuming and expensive.
“I think there are several reasons to be sceptical about the exact figures given in the paper and the press release about how well the new test, developed by the researchers, performs. I’m not saying by any means that the procedure is useless – only that on the basis of what we know so far, it’s just not possible to say how well it would work, if and when it is further developed for real clinical practice.
“The new research is a case-control study. In the simplest terms, this means that the researchers used blood samples from some people who had confirmed diagnoses of ME/CFS, known as the
cases, and blood samples from other people, the
controls, who were as far as possible known
not to have ME/CFS. Then they looked for differences in how the DNA (taken from the blood samples) was folded in these two groups, and they found some differences.
“I can’t personally comment on the details of the approach of looking at DNA folding, because that is beyond my expertise, though I have no reason to doubt that it can be useful. But I can and will comment on points about statistics and machine learning.
“One important limitation is that the blood test wouldn’t, as far as I can see, help in deciding which of the participants in the study had or did not have ME/CFS. That’s because it would already be clear who was a case and who wasn’t. The cases were already diagnosed as having ME/CFS, presumably using something like the four key ME/CFS symptoms listed on the NHS website*. The four are extreme fatigue; serious sleeping problems; problems with thinking, memory and concentration (‘brain fog’); and post-exertional malaise. Also, all of them were classified as having severe CFS and were housebound.
“The controls, however, were chosen so that they all had reasonable exercise tolerance, and had never had any of those four key symptoms in the past or at present.
“So all of these are people who would be reasonably easy to diagnose as having ME/CFS, or not having it, using existing diagnostic approaches. A blood test might well speed up their diagnoses, and possibly overcome the scepticism that still exists in some quarters about whether ME/CFS is a real condition.
“But the new results don’t tell us anything direct about how the test method would work in diagnosing people who don’t have such a severe form of ME/CFS, or whether it can distinguish between people who have ME/CFS and other conditions that involve severe fatigue.
“Also anyone who had any history of chronic illnesses (apart from ME/CFS in the cases), or any cancers (current or previous) or autoimmune conditions, or had had various DNA-altering therapies, was excluded from the research. So there’s no guarantee as to how the test, developed in this research, would work with people who have had some of those conditions.
“It can’t tell us about how the test might perform in telling people with ME/CFS from people with other chronic diseases than involve inflammation, for instance. The researchers point this out, and until we know whether or not the test can make this distinction, it can’t be used to decide what treatment such patients should have.
“Overall, it’s probably not too much of a simplification to say that the researchers have found a blood test that seems to work pretty well on the easier diagnosis decisions, so that the concept has potential, but that we just don’t yet know how well it might work in more difficult diagnostic situations.
“Also, as the researchers clearly point out in their paper, the research can’t directly say anything about
how the differences in DNA folding patterns are caused. That means that the research doesn’t throw any direct light on what actually causes ME/CFS. (That was not its intention anyway.)
“What
is potentially more widely useful is the analysis of the genetic pathways involved in the differences in DNA folding patterns between the cases and the controls, that the researchers carried out. Putting this possibly too simply, this means that they looked at the biological function of the genes that are near the places on chromosomes where the folding differences were. The details of all that are beyond my own expertise, but the findings should give researchers information about where to look for potential causes or treatments. That’s all in the future though.
“As well as all this, there are important reasons to be cautious about the exact numbers quoted for the performance of the diagnostic tool.
“There is actually rather a lot of statistical uncertainty about many of them. (Details are in Table 2 of the paper.) For instance, the sensitivity is quoted in the press release and the paper as 92%. This means that 92% of the people who were known to have ME/CFS had a positive test result.
“But this is a statistical estimate. Yes, 92% is the best estimate if one really insists on a single figure. However, numbers in Table 2 of the paper indicate that it could plausibly be much lower, down to 73%, or rather higher, up to 99%. If it were really somewhere around 73%, that would mean that about a quarter of people who really do have ME/CFS would not in fact have a positive test result.
“There are some details about the way the study was carried out that seem unusual to me, and that in my view may have introduced some biases and limit how far the findings can be generalised “beyond the groups of people directly involved.
“Some of those concerns relate to the complication that all the ME/CFS cases in the study (in the part of the data that was used to construct the diagnostic tool, and also in the validation data that was used to assess the performance of the tool and provided the data in Table 2 of the paper) came from the same source, the LSHTM biobank, whereas some but not all of the controls used in the validation (but not to construct the diagnostic tool) come from a different data set, the ‘OBD repository’.
“The research paper gives no detail or references for either of these data sources, though information about the LSHTM biobank is available online**. Indeed the paper does not say what the OBD repository actually is. I presume it is a data source held by Oxford BioDynamics, who funded this research and whose employees make up most of the research paper’s authors, but no detailed information seems to be available.”
*
https://www.nhs.uk/conditions/chronic-fatigue-syndrome-cfs/symptoms/
**
https://cureme.lshtm.ac.uk/
Dr Alastair Miller, Retired Consultant Physician in Infectious Disease and Internal Medicine, said:
“The authors are claiming a higher rate of sensitivity and specificity than in most other biomedical tests. 96% accuracy is almost unheard of for this sort of test, so this is quite a remarkable claim. My main concern with this study is the lack of appropriate controls. They are using healthy controls rather than those with other chronic conditions such as depression or fibromyalgia or even MS. My worry is that it will prove to be yet another false dawn, launched with a huge amount of hype and will raise patients’ expectations unrealistically.”