Development and validation of blood-based diagnostic biomarkers for [ME/CFS] using EpiSwitch®… 2025, Hunter et al. (Oxford Biodynamics)

Some media links on this paper, for what they are worth:

Go to,
https://www.geneonline.com/blood-te...es-chronic-fatigue-syndrome-with-96-accuracy/
https://medicalxpress.com/news/2025-10-blood-chronic-fatigue-accuracy.html
https://www.theguardian.com/society...-test-to-diagnose-chronic-fatigue-syndrome-me
https://www.thetimes.com/uk/healthcare/article/first-blood-test-chronic-fatigue-syndrome-l7shfd09q
https://bioengineer.org/breakthrough-blood-test-for-me-chronic-fatigue-syndrome-unveiled/
https://www.independent.co.uk/news/...d-test-chronic-fatigue-syndrome-b2841122.html
https://www.yahoo.com/news/articles/scientists-develop-first-ever-blood-061051249.html
https://news.sky.com/story/scientis...to-diagnose-chronic-fatigue-syndrome-13446510
https://www.itv.com/news/2025-10-07...ood-test-to-diagnose-chronic-fatigue-syndrome
https://www.mirror.co.uk/news/uk-news/scientists-claim-developed-first-ever-36027263
https://newatlas.com/imaging-diagnostics/chronic-fatigue-accurate-blood-test/
https://www.independent.co.uk/bulletin/news/me-chronic-fatigue-blood-test-b2841425.html
https://www.tipranks.com/news/compa...rough-blood-test-for-chronic-fatigue-syndrome
https://www.telegraph.co.uk/news/20...ood-test-can-detect-chronic-fatigue-syndrome/
https://www.upday.com/uk/uknews/bre...eveloped-for-chronic-fatigue-syndrome/l92vnfq
https://www.proactiveinvestors.co.u...atigue-syndrome-with-96-accuracy-1079911.html
https://www.bgnes.com/health/scientists-develop-first-blood-test-to-detect-chronic-fatigue-syndrome
https://www.thenews.com.pk/latest/1...ic-blood-test-in-chronic-fatigue-breakthrough
 
Last edited:
I mean apart from the fact that if it was real it would be life changing for every single one of us.

I understand your skepticism and I share it, but I think you go a little far here.
Why?

In my mind seeing if they can reproduce the results, albeit while resolving the concerns raised, is exactly what we need.
Because the results are meaningless if they don’t control for confounders. They tell us absolutely nothing about what’s wrong with pwME/CFS.

And even if by luck they have stumbled upon something that’s actually related to the diseases process, we would have no way of knowing because it might all be random noise.

We need a study design that’s capable of falsifying the hypothesis: there are no unique biological differences between pwME/CFS and non-pwME/CFS.
 
I think both @hotblack and @V.R.T. were operating under the assumption that the study design for the “replication” would be altered to better control for confounders
Then it wouldn’t be a replication. Replication is to do the same thing in a different lab by a different team and seeing if it gives the same result.

I have no issue with using the same tests on two much more tightly controlled cohorts. But that’s probably the responsibility of the company that is trying to earn money off of this..
 
The deeper problem is that this team do not understand what a diagnostic test is for. There is no point in developing a diagnostic test that correlated 100% with a syndrome. If it does, you know you already had a 100% method of recognising the syndrome so the test is unnecessary.

You may get a near 100% discrimination with barn door cases against barn door normal people with a test that turns out to be useful, but the usefulness is demonstrated by its performance in uncertain cases. And it is only useful in those uncertain cases if it reflects some real biological feature that has prognostic or treatment implications.

It is time people stopped thinking their job is to find a test that correlates perfectly and realise that the test is only useful if it tells us something about the underlying biology. The disease then gets re-classified by its biology and we move forward.
I disagree strongly on that. You are absolutely correct that in an ideal world we do not need that, that it is not just possible but reliable enough to diagnose using clinical criteria alone, and a test that does not add an understanding of the biology involved is far less useful.

But we live very far from an ideal world, the vast majority of clinicians fail miserably at making this diagnosis, and most won't even bother unless they can send some test returned back to them with a probability rating.

Until we have a test showing us the precise biology, things will remain at a standstill for the most part, but a test that is as reliable as the experienced application of clinical criteria is still very valuable. Its usefulness is much lower because pointing to a specific pathology is a strict requirement for most clinicians, but that's entirely a choice.

There are respected diseases that don't even have that much. The problem isn't with the tests themselves, but a test like this is still valuable for clinicians who still respect the illness, but simply cannot develop the deep expertise needed to apply them reliably. And as we know, lots of clinicians have spent years doing so, and still do far worse than the average patient would.

This is not what we need, but it would be a positive step.
 
Wouldn’t it possibly be useful if just for legitimising the illness and increasing chances of people getting proper support? (Probably would help with disability insurance).

If a test showed some pattern of biology in ME/CFS that did not occur in other conditions then it would be useful in the same way that DecodeME has been useful but it has nothing to do with being a diagnostic test. DecodeME does not give us a diagnostic test.

If you want to legitimise the illness of an individual then a test that legitimises the illness provides support for the individual if they have the typical features. But you don't use a rheumatoid factor (which is positive in some patients and not others) to legitimise somebody's RA.
 
I disagree strongly on that.

Yes, I can see that but you are missing the key distinction which so many researchers are failing to see, and as a result focusing on the wrong aspects of the data, analysing it the wrong way and very likely missing something really useful.

And this is not about an ideal world. I am making the argument I am making precisely because it is a real world argument. We do not diagnose syndromes with tests. We diagnose processes that we have already understood and have gold standards for. In the real world it is essential that doctors do not use a rheumatoid factor test to diagnose rheumatoid arthritis. Unfortunately, a lot of them do.
 
I can certainly see a scenario where a clinical test could be part of diagnosis.

That occurs as soon as the thing tested for becomes part of the diagnostic concept. And it doesn't have to have particularly high specificity or sensitivity. Rheumatoid factor is a good example. A positive test can tip the balance of probabilities, but that is only really of relevance as a prognostic tool. If the patient has bad arthritis they need treatment whatever you call it.

A genetic analysis of the fibrillin genes will allow you to make a firm diagnosis of Marfan Syndrome but in reality Marfan Syndrome stopped being a syndrome and became a genetic concept decades ago.
 
For example a combination of clinical tests aid diagnosis of Guillain-Barré Syndrome (GBS), but the overriding disease mechanism is unknown.

All you need to know is the aspects of the mechanism that the test tests. The diagnostic concept of Guillain-Barré very much includes the idea of peripheral nerve dysfunction giving a particular type of EMG profile.
 
For ME/CFS we have no idea what we want to identify or whether there is one of them or six. And it is clinically staring us in the face now. It is defined by its clinical presentation. We don't want a test for ME/CFS. We want a tests for the things that might cause ME/CFS.
Yes. We do.
 
Yes. We do.

I am well aware that many people think that way but the more people with ME/CFS and advocacy organisations for ME/CFS call for and fund projects focused on this sort of diagnostic test the longer the disease will remain in limbo with researchers churning out useless analyses of what might otherwise have been valuable data.

At least Robert seems to understand that !

Look at inflammatory rheumatic disease - we went from total ignorance to having highly effective treatments for all of them without ever bothering with a diagnostic test.
 
This on CD4 from the pathway analysis

The comparative analysis of the pathways showed significant overlap in IL10, IL2 and CD4 in all three networks, with more individual overlaps between ME/CFS and each therapy individually (Fig. 4).

HLA-DRA also popped up in Fig 4 although I’m not sure of the significance.

Some quotes on the potential for specific subgroups

From the ME/CFS nucleome profile, we have selected the most significant CCs (765) that are associated with the 50 IL2 STRING genes/proteins. When analysed for these CCs, ME/CFS patients exhibited clear clustering with 18/29 division of the 47 patients analysed.
This clear stratification indicates the potential for IL-2-related epigenetic markers to serve as a basis for patient subgrouping and personalized treatment approaches

A focused analysis of IL-2-associated pathways highlighted its central role in T cell regulation and immune activation. Hierarchical clustering based on IL-2-linked CCs demonstrated distinct patient subgroups, indicating the potential for biomarker-driven stratification and personalised therapeutic approaches in ME/CFS (Figs. 5 and 6).

And

Overlap with therapeutic networks (rituximab, glatiramer acetate/Copaxone) suggests a subset of patients whose epigenetic regulatory architecture reflects pathways targetable by B-cell depletion or T-cell-modulating agents — consistent with prior reports of heterogeneous rituximab responses in ME/CFS and the network overlap we observe
 
The section on diagnostic value has some interesting bits too. I don’t think they’re completely ignoring the limitations here and they do call this a proof-of-concept study

We acknowledge that larger, multi-centre cohorts are needed to confirm generalisability.
Future work with external disease cohorts will be essential to confirm disease specificity and refine biomarker panels capable of discriminating ME/CFS from other systemic inflammatory or autoimmune disorders.
Although we employed independent training and validation subsets, both were derived from overlapping biobank sources. Larger-scale, prospective validation in multi-centre cohorts will be essential to establish external validity.

And on use of their technology

Examples of translation include the EpiSwitch PSE test for prostate cancer, which achieved 94% accuracy and reduced unnecessary biopsies by 79% in prospective and real-world studies, and the CiRT test for immunotherapy response, which predicts PD-1/PD-L1 response with 85% accuracy, outperforming standard PD-L1 IHC. Both tests are now commercially available in the US (CPT PLA codes 0433U and 0332U) and UK, with established reimbursement and widespread clinical uptake. On this foundation, the current ME/CFS biomarkers can follow a clear path to clinical translation, supported by further multi-center validation.
Here are links to the papers referenced in this section for people who want to read them

 
Something that strikes me again and again is how many scientists there are who don’t seem to understand – or chose to ignore – some very basic principles which can be perfectly understood by people like me with limited scientific knowledge and no scientific training.

If this company were to start selling tests before they have been validated that would be a red flag.
Mmm.

Incurious enough not to bother laying down a foundational understanding, a flourishing academic career never demanded such. Or maybe even required a leap frog approach.

Or, accidentally on purpose? A highly incentivised approach.


Moving on from such knowledge. Selling something poorly constructed is just as, if not, more profitable than the well made product after all. An expedient approach since the basics can be an obstacle to a certain kind of progress.

We can only guess.

I don’t care which it is.

The end result is garbage.


How to respond to a professional (something) who’d hawk their shoddy merchandise with such overblown proclamations?

Bag it and bin it, I think.


Edit:Mashed quotes, edit ⬇️

@Robert 1973, our priorities require we understand. The calculus for saving life, rather different to that of carving out a career under capitalist logic.
 
Last edited:
This study doesn't warrant all this media attention. Wonder if the SMC is to blame for pushing this?
https://www.geneonline.com/blood-te...es-chronic-fatigue-syndrome-with-96-accuracy/
https://medicalxpress.com/news/2025-10-blood-chronic-fatigue-accuracy.html
https://www.theguardian.com/society...-test-to-diagnose-chronic-fatigue-syndrome-me
https://www.thetimes.com/uk/healthcare/article/first-blood-test-chronic-fatigue-syndrome-l7shfd09q
https://bioengineer.org/breakthrough-blood-test-for-me-chronic-fatigue-syndrome-unveiled/
https://www.independent.co.uk/news/...d-test-chronic-fatigue-syndrome-b2841122.html
https://www.yahoo.com/news/articles/scientists-develop-first-ever-blood-061051249.html
https://news.sky.com/story/scientis...to-diagnose-chronic-fatigue-syndrome-13446510
https://www.itv.com/news/2025-10-07...ood-test-to-diagnose-chronic-fatigue-syndrome
https://www.mirror.co.uk/news/uk-news/scientists-claim-developed-first-ever-36027263
https://newatlas.com/imaging-diagnostics/chronic-fatigue-accurate-blood-test/
https://www.independent.co.uk/bulletin/news/me-chronic-fatigue-blood-test-b2841425.html
https://www.tipranks.com/news/compa...rough-blood-test-for-chronic-fatigue-syndrome
https://www.telegraph.co.uk/news/20...ood-test-can-detect-chronic-fatigue-syndrome/
https://www.upday.com/uk/uknews/bre...eveloped-for-chronic-fatigue-syndrome/l92vnfq
https://www.proactiveinvestors.co.u...atigue-syndrome-with-96-accuracy-1079911.html
https://www.bgnes.com/health/scientists-develop-first-blood-test-to-detect-chronic-fatigue-syndrome
https://www.thenews.com.pk/latest/1...ic-blood-test-in-chronic-fatigue-breakthrough
 
I think both @hotblack and @V.R.T. were operating under the assumption that the study design for the “replication” would be altered to better control for confounders
Yeah, I think some of as are roughly saying the same thing. Maybe with some differences in how we dress it up. Others seem to think it’s not worth pursuing at all.

It seems clear from the factors raised by Chris Ponting and Kevin McConway that this shouldn’t be being reported as a test with the accuracy (or sensitivity and specificity) reported.

There are risks what they found is just plain wrong. But it may not be. If the same results are found in a study performed with the issues raised addressed then that would prove that something has been found here. So that’s what is needed. And what you’d expect after a proof of concept like this.

It would probably be better if they skipped the narrative side too. But it’s hardly as if they’re the only team that does this. Let’s hope they read some of the constructive feedback, continue with the work and can put together a more convincing case with a followup study.
 
Back
Top Bottom