April 29, 2019

expert reaction to study investigating a potential biomarker for chronic fatigue syndrome / ME

Research published in PNAS suggests a potential biomarker for myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS)

Prof Chris Ponting, Chair of Medical Bioinformatics and Principal Investigator at the MRC Human Genetics Unit, University of Edinburgh, said:

“Esfandyarpour and colleagues show that people with ME/CFS are different from healthy controls with regards to their blood samples’ electrical properties.  Excitingly, they appear to have discovered a distinguishing feature of ME/CFS, and one that can be measured simply and cheaply.  Before this approach is tested clinically, however, three things will be needed.  First, results should be replicated in a second cohort of individuals.  Second, the device should be tested whether it sets apart ME/CFS not just from general health but also from other disorders.  Finally, early indications that the device can distinguish severely- from moderately-affected people with ME/CFS need to be tested thoroughly.  These results also now narrow down the possible molecular and cellular causes of this devastating set of conditions.”

Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:

“This research is interesting, and if it can be the start of developing a blood-based diagnostic marker for ME/CFS, that would be great.  But it’s important to understand that this is only a pilot study.  There’s a long way to go, with many possible pitfalls that could still derail the development, before it gets to something that’s useful in practice.

“So far the method has been tested on only 20 people with diagnosed ME/CFS, and another 20 who did not have a diagnosis of ME/CFS or any related disease, and who also had no blood relatives who had been diagnosed with ME/CFS.  The proposed test did distinguish between these two groups of people quite clearly.  The researchers report the results of using a quite sophisticated machine learning algorithm to distinguish the two groups, but actually the differences were so clear that this was hardly necessary.  But something like that algorithm may well be necessary for future, more demanding, tests of the diagnostic procedure.  Also, it’s a general principle of using machine learning algorithms that one develops the classification method in one set of data, and then validates it by seeing how it works with other, independent data.  That hasn’t been done yet – data from more people is required.  Also, so far there appears to be no data on how well the approach might distinguish between people with diagnosed ME/CFS and, say, people who have a diagnosis of some similar condition that isn’t ME, or people who are ill with some unrelated illness, or people who do have a relative with ME/CFS but do not have the condition themselves.  Until this kind of investigation has been done, with a lot more participants than the 40 used here, we just can’t know whether this new method can be used for diagnosis in the real world.

“An important question to ask about any potential diagnostic test, that finds a difference between people with a disease and those who do not have it, is whether there’s some other difference between the two groups of people apart from whether they have the disease or not.  If there is, perhaps that’s what the test is picking up, and not the disease at all.  I’m not sure that this has been yet considered carefully enough in this case.  We’re told that five of the healthy people were matched to five of the people with ME/CFS for age and gender, but the remaining 15 health people seem not to have been matched in this way.  Perhaps there’s a difference between the two groups in age, gender, or indeed something else, that is contributing to the observed differences at least in part.  Also, the researchers report that they handled the samples from the ME/CFS patients and the controls in exactly the same way, but they do not say whether the people handling the samples and recording the results knew whether each sample was from a patient or a control at the time the samples were being prepared and the measurements made.  Ideally they should not know, in order to eliminate the possibility that the researchers unintentionally do things in a slightly different way for the two groups.  I’m certainly not accusing the researchers of any kind of deliberate cheating here, but with such sensitive equipment, perhaps some tiny, entirely unintentional and unnoticeable change in what the experimenters do might conceivable have an effect on the results.  Again, this can be dealt with in further research.

“The researchers report that it might be possible to use their technology for screening potential drugs for ME/CFS.  That is indeed an exciting possibility, but again I think we’ve got to be careful to consider the limitations of the findings so far.  The idea is that a potential drug can be added to blood samples from a ME/CFS patient, and then if the blood sample starts reacting more like a sample from someone who isn’t a patient, the drug could be useful.  Well, that’s a possibility.  But suppose that the debilitating symptoms of ME/CFS turn out not to be caused by the sort of stress response in blood cells that is being picked up by the new method, but that instead the symptoms and the effect in blood cells are both caused independently by some other problem in the body.  Then a drug that deals with the response in the blood cells might turn out have no effect on the symptoms.  Likewise a drug that seems not to affect the stress response in the blood cells might turn out to be effective against whatever is causing the symptoms.  So the use of the new method for screening potential drugs might turn out to be very useful, or it might not.  I think we just don’t know yet, but further research will throw more light on this.

“In short, this new approach is certainly potentially useful in diagnosis and drug screening, but with a big emphasis on ‘potentially’.  There’s a very long way to go before we know whether the potential can be realised.”

Prof John Martin, Professor of Cardiovascular Medicine, UCL, said:

“If a test is to have meaning it has to be able to be applied to a population of patients who can be defined clinically.  The patients described had a variety of symptoms that could have arisen from a variety of causes.  The population was not clinically defined in a way that could be related to a test.  Further the authors do not relate the cellular finding in the test to a possible cause of the disease.  CFS/ME is probably not a disease but a syndrome.

“It is interesting that all the patients tested had the same response in the test even though they had different severities of complaints.  Could the test be picking up something in the 20 patients not present in the controls such as anxiety?  The biological mechanistic meaning of the test needs exploring and the clinical description of CFS/ME needs specific definition.  The authors should consider whether their test is related to an affect of symptoms and not related to the cause.”

Prof Sir Simon Wessely, Regius Chair of Psychiatry, Institute of Psychiatry Psychology & Neuroscience, King’s College London (IoPPN), and President, Royal Society of Medicine, said:

“There have been many previous attempts to find a specific biomarker for CFS.  The problem is not differentiating patients with CFS from healthy controls.  The issue is can any biomarker distinguish CFS patients from those with other fatiguing illnesses?  And second, is it measuring the cause, and not the consequence, of illness?  This study does not provide any evidence that either has finally been achieved.  It is also regrettable that it is claimed that such a test would give “scientific proof” of the existence of the condition, and prove it is “not imaginary”.  You don’t need a blood test to prove that an illness exists, and nor does the absence of such a test mean that it is “all in the mind”.  Any sub who runs a headline that says ‘new test proves CFS is real and not psychiatric’ should be ashamed of themselves.”

‘A nanoelectronics-blood-based diagnostic biomarker for myalgic encephalomyelitis/chronic fatigue syndrome (ME/CFS)’ by Rahim Esfandyarpour et al. was published in PNAS at 20:00 UK time on Monday 29 April 2019.

Declared interests

Prof Chris Ponting: “I do not believe that I have a conflict-of-interest.”

Prof Kevin McConway: “I am a trustee of the Science Media Centre, but I am writing these comments in my capacity as a professional statistician.”

Prof John Martin: “I have no conflict of interest in the matter of this article.”

Prof Sir Simon Wessely: “SW sees CFS patients regularly on the NHS and has published over 150 papers on the topic.  He is a Trustee of the SMC.”