Passive transfer of fibromyalgia symptoms from patients to mice, 2021, Goebel et al

Of interest for the Canadian crowd, CIHR (Canadian Institutes for health research) has contributed 125,000$ For a fellowship on this project.

https://webapps.cihr-irsc.gc.ca/decisions/p/project_details.html?applId=412619&lang=en

Other than that, CIHR has not funded biomedical research-mainly behavioral, and it may be due to the lack of encouragement in getting in the field, which leads to a lack of grant application.

 

I am a bit puzzled by all this.

On the one hand, we have a study that appears to have impressive results.

The effect sizes are fairly large. They are also seen across a range of behavioural and biological factors. So, for mice injected with IgG from patients (compared with controls) researchers found an increase in paw sensitivity, a reduction in paw grip strength and a reduction in movement.

They also found a reduction in skin innervation and binding to satellite glial cells and neurons in the dorsal root ganglia (though not to sensory neurons). This last finding was backed up by patient IgG binding to human dorsal root ganglia.

As a suite of results, it does look impressive to me.

Also, because there was pooling of IgG, I wonder if the strength of the results might be partly explained by only a minority of patients having this particular problem but at least one of those minority appear in every pooled sample. That would make the overall results look more impressive.

On the other hand are the concerns raised by @Jonathan Edwards @Hutan and @Nightsong . Which all sound like good reasons to doubt the findings.

On top of that, the autoimmune hypothesis to explain fibromyalgia is not new. If it really was a clear-cut autoimmune disease, surely researchers will offend other clear signals by now (notwithstanding the points already made about clinical presentation not fitting)?

And I note that the head of Fibromyalgia Action UK started his comments to the Guardian with a warning that these results needed to be replicated and on a larger sample.

All of which leaves me with this question. How is it possible to get such a comprehensive suite of results if there is really nothing going on?

In some ways, it reminds me of the XMRV Science paper. Here, the evidence appeared strong because it came from several different lines of enquiry, including PCR and antibody testing. Again, there were reasons to doubt the findings because they were so clear-cut and many patients don't have an apparent infectious onset.

So, it can happen, but I'm wondering how it could have happened in this case?

I would dearly like to see a replication in fibromyalgia. If it does replicate, then that becomes very interesting. And at that stage, maybe there is a case for looking for the same in ME/CFS. In any event, the lead author, David Andersson, said he was hoping to run a similar study for ME/CFS.

 

This group looked at gene expression: https://pubmed.ncbi.nlm.nih.gov/32532082/

this 1990 paper looked:

Screening for autoantibodies in patients with primary fibromyalgia syndrome and a matched control group
https://pubmed.ncbi.nlm.nih.gov/2204364/


and a second 1990 paper shut the door:
Absence of autoantibodies in primary fibromyalgia
https://pubmed.ncbi.nlm.nih.gov/2084244/

 

The leeway around clinical criteria does unfortunately reduce the certainty of a diagnosis of fibromyalgia. Ideally, fibromyalgia could be better defined by having one or more characteristic features as for delayed post-exertional malaise in ME/CFS -- even though its expression is highly heterogeneous --, rather than chronic widespread pain which is shared by multiple painful conditions.

The inability to find a consistent auto-antibody, the differences between autoimmune disease and the course and clinical presentations of fibromyalgia, the absence of positive reports of immunotherapy in FM indeed cast doubts about these promising results.

The long delay between the publication of the pre-print (2019) and the publication of the study in a journal is also questionable. The authors seemingly conducted additional experiments, but did they do so based on the feedback from a peer reviewer if they submitted to another journal than JCI and were rejected? I would have expected this study to be published in a rheumatology journal, even though none of the authors are rheumatologists.

Given that the results seem "too good to be true" and that the study has received quite some publicity, it would make sense that independent teams will now attempt to replicate it.

 

It seems to me that it's difficult to assess what the impact of the fairly loose fibromyalgia diagnostic criteria has on identifying patients that all have a single condition. I think it's plausible that once you exlude other conditions, and choose patients that have symptoms characteristic of fibromyalgia, you're quite likely to end up with patients that all have "true fibromyalgia", as I'm not aware of any other conditions that can cause symptoms very consistent with and characteristic of fibromyalgia whilst being normal on other findings.

So I don't think the possible diagnostic uncertainty is a reason to write off this study at all. However it is a reason to have some doubt, and independent replication is definitely something that's wanted here.

 

That’s the part I’m struggling with. I automatically assume that any paper in this field has a 99.9% chance of not being replicable. But unless this is another XMRV type situation, it’s hard to explain away the fact that the disease was induced in animals.

 

So what is 'characteristic of fibromyalgia'? It all seems pretty vague to me.

If it really was characteristic and easy to discriminate then why do physicians find it so hard to agree on who has it?

I have looked after people with painful peripheral neuropathy and it is immediately obvious that they have a painful peripheral neuropathy because that really is characteristic - feet and hands, diffuse, burning, unremitting, associated with swelling and vascular changes.

If these antibodies are targeting peripheral nerves why is the pattern specific and typical?

 

Sorry if this has been covered, but do you think fibromyalgia is a distinct condition?
Do the people diagnosing it think this?

 

I am not sure I know what that would mean. I suspect the people diagnosing it might say yes but I couldn't be sure. presumably it would imply a uniform mechanism that is not present otherwise. I find it hard to see that as likely since physicians cannot agree how many people to include.

 

Fibromyalgia is usually triggered by something, such as physical trauma, and the combination and pattern and frequency of symptoms in fibromyalgia helps distinguish it - the score on the widespread pain index, light sensitivity, exacerbation of pain after exertion or relatively light pressure, "fibro fog". And normal pain killers generally don't work for fibromyalgia

I doubt someone is easy to diagnose with fibromyalgia. But if you exclude other conditions and are left with characteristic symptoms like those above, that someone with appropriate experience would be used to identifying, that may eliminate a fair bit of diagnostic uncertainty.

 

In this video, https://www.s4me.info/threads/iimer...ence-week-london-2019.5907/page-4#post-184672 , it appears to show the first 3 individuals results tested show the same pattern before they pooled. Starts around 4:35.

If it was replicated, perhaps it's not a stricly auto-immune situation.

 

Something I had wondered about but not investigated is whether or not the mouse studies were done in a blinded and systematic way, which would be essential for a reliable result.

Listening to the video by David Anderson it seems not. They tested one patient and one control, I suspect knowing which was which, then did another patient and control and so on. This is the best way to get spurious results. I cannot be sure that the findings are spurious but I just find it very difficult to think otherwise.

 

That does not quite fit with the way things are presented in the video.

 

What's an effect size?

 

In statistics, an effect size is a number measuring the strength of the relationship between two variables in a population

https://en.wikipedia.org/wiki/Effect_size

 

I've looked at the article already but couldn't make sense of it. Can you give examples of effect sizes and how they were derived, assuming I know nothing about statistics?

 

Is it possible that what is causing problems is not IgG but something else?

Some non-antibody factor in the blood that isolation techiques for IgG are not separating from IgG.

This is what seems to be occurring in ME/CFS.

The general idea here is that there exists an as of yet undiscovered class of molecules, probably a product of the immune system, which can cause chronic illness, and problems with this kind of immune response could explain a variety of chronic illnesses that currently can't be explained.

 

Sure, let’s say you have two groups, patients and controls, and you're interested in whether they differ on some outcome you're measuring in your study (e.g. antibody, score on a fatigue questionnaire etc.). The usual approach to this problem (you'll see this in every published paper) is to do a statistical test of the null hypothesis that there is no difference between the groups. So you run your t test and get a 'statistically significant' result (those p-values you see in papers, e.g. p<0.05). Ok great, but this only tells you that getting this result would be quite unlikely if the true difference between the groups was 0. But you don't really care about that. What you generally want to know is: how big is this difference? Is it clinically meaningful? This is where effect size comes in. It helps you to figure out if the difference between these two groups on your outcome of interest is small, medium or large. To do this (in this example), you'd divide the mean difference by pooled standard deviation (standard deviation is just a measure of dispersion) and you'd get a number which you could then interpret as small (0.2), medium (0.5) or large (0.8). (These estimates are rule-of-thumb, not an exact science.) If an effect size is large it means that the scores vary a lot between the two groups (if you plotted them, there would be little overlap).