Science Media Centre's "Expert reaction to biomarkers for CFS/ME":
A paper published in the journal Science Advances has reported the presence of a specific biomarker signature in patients early in the course of CFS/ME, which was not seen in patients with a longer duration of the illness or in healthy individuals. The biomarkers relate to immune signalling messengers.
Dr Diana Prata, Group Leader, Institute of Molecular Medicine (IMM), University of Lisbon and Visiting Lecturer, Centre for Neuroimaging Sciences, King’s College London, said:
“This study advances the field in terms of understanding the pathophysiology of ME/CFS, but regardless of how statistically strong a correlation between two variables (presence of disease and presence of a given substance in the blood) a given test cannot be considered a real diagnostic biomarker, without showing high specificity or high sensitivity – ideally both. That is, it has to get most of the positive cases right or most of the negative cases right, in order to be clinically useful. Furthermore, even then, that doesn’t make it so. Prevalence of both the disease, and the substance, in the population has to be taken into account, to then deduct the so called positive and negative ‘predictive values’: the proportion of the obtained positive and negative test results that are indeed true. For example, if the proportion of patients with the illness or with the specific plasma immune signature is very low, the clinical use decreases substantially.
“These authors did no assessment of these values and thus, the idea from a statement in the press release that that ‘there is unequivocal evidence for a diagnostic biomarker’ is unsubstantiated.
“This specific illness is not in my area of expertise, but I have published a recent review on biomarkers where we had to address similar issues: http://www.ncbi.nlm.nih.gov/pubmed/24877683.”
Prof. Michael Sharpe, Professor of Psychological Medicine, University of Oxford, said:
“Whilst this finding that some patients with CFS/ME have an immune abnormality is potentially interesting, we should treat it with great caution.
“This type of study (a case-control study) is notorious for producing findings that other researchers subsequently fail to replicate.
“Everyone who has worked clinically with patients with CFS/ME knows this is a real illness; this study neither proves nor disproves that observation.”
Dr Esther Crawley, Reader in Child Health, School of Social and Community Medicine, University of Bristol, said:
“Studies of biomarkers are important to help us develop hypotheses about this important illness. In this paper, the differences between those with ‘short’ illness duration of 3 years compared to those with longer illness duration are fascinating and will help researchers develop new treatments and explore factors that predict outcome in adults.
“Whilst these results are important for researchers, it is important for patients to understand that these studies are not currently suitable as a test to diagnose ME/CFS. Although the authors say ‘integration of these immune markers with clinical findings will provide clinicians with a stronger framework for establishing an ME/CFS diagnosis, and, possibly, make it easier to rule out other conditions at an earlier time point’, this paper does not compare these cytokine perturbations with controls who have other illnesses.
“Further work is needed before it will be possible to use these results to rule out other conditions. In addition, patients need to accept that the authors do not claim that they are suitable in the early stage of the illness to predict outcome.”
Prof. Derek Hill, CEO of IXICO plc and Professor of Medical Imaging Science, UCL, said:
“The authors have analyzed stored blood samples and clinical data from several hundred ME sufferers that participated in completed research studies, along with controls subjects who didn’t have ME. The paper reports a number of molecular markers in the blood that are associated with the ME in its various stages. The results that the paper emphasise are from small proteins in the blood called cytokines, but it isn’t clear how wide a range of blood markers they initially looked for: the more molecules you search for in this sort of experiment, the higher the chance that associations will be found by pure chance.
“It is notable that when the authors compared all the patients with ME with all the controls, “no substantive differences were found” and the authors then seem to have further broken down their analysis to look at patients in early and later stages of the disease – and it is in this further analysis of smaller groups that they found significant associations. Sub-group analyses of this type do further risk associations that arise from pure chance. To be confident in the results in this paper, it is important that these or other researchers replicate these findings in the same or preferably other patient with ME. Furthermore, in order to demonstrate that these molecules could be used to help develop new treatments or to diagnose patients, it would be important to work out how the biomarker signals change as the patient gets better or worse, and also how accurate the biomarker separates patients with ME from patients who don’t have ME but have similar clinical symptoms (rather than just comparing with normal controls).
“Discovering a biomarker and turning it into a diagnostic test that is used on patients is comparable to the challenge of discovering and developing a new drug: it can take well over a decade and has a high chance of failure. This paper is an important step forward in identifying candidate biomarkers for ME. But much more work is now needed to demonstrate these findings can be replicated in prospective studies and in demonstrating that the biomarker test performs reliability before regulators would consider the biomarker qualified for the purposes of developing new drugs. And even more work would be requited before it would be approved as a diagnostic test. ”
Prof. Naveed Sattar, Professor of Metabolic Medicine, University of Glasgow, said:
“This study suggests a pattern of inflammatory markers may differentiate early ME/CFS from healthy controls, an intriguing result on the face of it. Whilst the findings are of some interest, the results should be considered preliminary. The study would have benefited from better control of age and also consideration of other simple characteristics which influence body inflammation levels, including, in particular, individual’s body mass index and their smoking habits, both factors which can alter inflammatory status.
“Indeed, we know that simple markers of inflammation can differ by several fold between lean non-smokers and obese smokers, as can an individual’s social status – inflammation levels are higher in less affluent populations due to factors not well described. Hence, the current results should be considered only hypothesis generating rather than definitive and need to be confirmed in better controlled studies in future with very careful matching.”
Prof. Stephen Lawrie, Head of the Division of Psychiatry, University of Edinburgh, said:
“This field – the biology and especially immunology of ME/CFS – has been bedeviled by false dawns for at least 20 years. This is a small study and the fact that the ‘biomarker’ does not internally replicate is a concern. Yes, it could be that there is a different immune profile in acute and chronic ME/CFS but it is at least as likely that the finding in acute patients is down to chance and hence a false positive signal.”
Prof. Peter White, Professor of Psychological Medicine, Queen Mary University of London (QMUL), said:
“This study is impressive when considering the numbers studied and the care taken by the renowned scientists undertaking it. But I think it is premature to conclude that they have found ‘a diagnostic biomarker for disease’, something that would make diagnosis much easier. Only one out of the 51 immune proteins studied was elevated in all cases compared with controls, something that could happen by chance alone. Finding more elevated immune proteins in those with a short duration of illness is less convincing when this was found to be more important than the association with the severity of illness: one of the fundamental tests of a biomarker. I hope the authors will go on to re-examine their data after stratifying their samples by other factors that determine the different sub-groups that most scientists now accept make up this illness. Finally, as the authors themselves suggest, we need to see these results replicated independently. ”
‘Distinct plasma immune signatures in ME/CFS are present early in the course of illness’ by Hornig et al. published in Science Advances on Friday 27th February.
Prof. Derek Hill: IXICO has no commercial interest in ME or ME biomarkers
Prof. Peter White: Does voluntary work for UK departments of Health and Work and Pensions and Ministry of Defence. Also does paid work for a re-insurance company.
No other interests declared