No signs of neuroinflammation in women with [CFS] or Q fever fatigue syndrome using the TSPO ligand [11C]-PK11195, 2021, Raijmakers, Knoop et al

Abstract
The pathophysiology of chronic fatigue syndrome (CFS) and Q fever fatigue syndrome (QFS) remains elusive. Recent data suggest a role for neuroinflammation as defined by increased expression of translocator protein (TSPO). In the present study we investigated neuroinflammation in female CFS and QFS patients compared with healthy women, using Positron Emission Tomography (PET) with the TSPO ligand [ 11 C]-PK11195.

Methods
The study population consisted of CFS patients (n = 9), QFS patients (n = 10), and healthy controls (n = 9). All subjects were women, matched for age (± 5 years) and neighbourhood, between 18 and 59 years of age, who did not use any medication other than paracetamol or oral contraceptives, and were not vaccinated in the last six months. None of the subjects reported substance abuse in the past 3 months or reported signs of underlying psychiatric disease on the Mini-International Neuropsychiatric Interview (MINI). All subjects underwent a [ 11 C]-PK11195 PET scan and the [ 11 C]-PK11195 binding potential (BP ND ) was calculated.

Results
No statistically significant differences in BP ND were found for CFS patients or QFS patients when compared to healthy controls. BP ND of [ 11 C]-PK11195 positively correlated with symptom severity scores in QFS patients, but a negative correlation was found in CFS patients.

Conclusions
In contrast to what was previously reported for CFS, we found no significant difference in BP ND of [ 11 C]-PK11195 when comparing CFS or QFS patients to healthy neighbourhood controls. In this small series we were unable to find signs of neuroinflammation in patients with CFS and QFS.

Preprint https://europepmc.org/article/ppr/ppr306486
Post with link to final version here

 

Is this the same methodology as the Japanese PET study?

 

As far as I can tell it seems to be the same or similar.

 

While the authors do not discuss this difference, they only used the Fukuda criteria while participants in Nakatomi et al.'s study met both the International consensus criteria and Fukuda. They do not mention if the patients in their cohort experienced PEM.

The differences in methodology are discussed below (spacing mine):

Although the set-up of this study was similar to that of Nakatomi et al., using the same TSPO ligand ([11C]-PK11195) (5), a number of important differences can be discerned.

First of all, for reasons of homogeneity, our study only included women. Around 75% of CFS patients are female and, although the percentage of women in QFS is lower (52%) (11, 31), we felt that we should avoid a gender effect in a study with such a small sample size. Nakatomi et al. included 30–40% males without presenting separate data for men and women (31). This is important as inflammatory responses are generally higher in males (32). Also, in experimental mouse studies of traumatic brain injury, male mice are more likely to exhibit neuroinflammation compared to female mice (33). One could argue that neuroinflammation is more likely to occur, and perhaps even persist, in males compared to females. However, if neuroinflammation is indeed present, the high percentage of female CFS patients contradicts with this hypothesis.

A second difference between our study and that of Nakatomi et al. is that we distinguished CFS patients, with often heterogenic aetiologies (31), from post-infectious fatigue syndrome patients, i.e., QFS patients.

Thirdly, we used a neighbourhood control group with healthy women that were matched with CFS and QFS patients in terms of age and geographical area in order to accomplish optimal matching and avoid bias due to confounding.

Also, patients, especially those with CFS, that were included in our study had a longer duration of illness than those included in the study by Nakatomi et al (reported mean of 62.4 months). When using small numbers of included patients, as is the case in both studies, subtle differences like these might contribute to the different outcomes that are seen.

This brings us to a fifth and final difference, i.e., the method used for determining the binding of [11C]-PK11195. We used pharmacokinetic binding with an arterial input function whereas Nakatomi et al. used the cerebellum as a reference region in reference tissue modelling. We feel that the latter is methodologically less sound as no brain region is devoid of TSPO, meaning that the cerebellum is not an objective reference region, and the cerebellum may actually be involved in the disease process. Whether binding of the [11C]-PK11195 ligand is considered enhanced, normal or even lowered, may be explained by this difference in methodology.
​

Though in Nakatomi et al.'s study "no arterial blood sampling was performed", they tried to correct for the last point:

Because there is no brain region devoid of TSPO (although a cerebellum reference may add stability to quantitative analyses (18)), and because there was no difference between mean time–activity curves of standardized uptake value in CFS/ME patients and healthy controls (Fig. 1C), we generated parametric images of regional 11C-(R)-PK11195 nondisplaceable binding potential (BPND) using linear graphical analysis according to Logan (19), with the cerebellar cortex as a reference region and corresponding to the linear part of the plot covering the last 40–60 min of measurement (20).
​

But van Elzakker et al. have previously argued that arterial line sampling, as was done here, is more robust:

 

I had not realised that Nakatomi had used cerebellum as a reference region. That seems troublesome because it means the study might be picking up differences in brain region usage within individuals rather than any absolute measure of microglial activation.

 

Interesting study. Unfortunate that the Japanese findings could not be confirmed.

 

Did they use the same "older" ligand in both studies?

 

Maureen Hanson's team have published MRS data showing low levels of glutathione (a very important antioxidant). Would that suggest that neuroinflammation is a relevant research area & free radicals/oxidate stress?

 

This negative result seems more insightful than it appears at first glance. The sounder methodology in this study incites us to give it precedence over that of Nakatomi et al -- so if we suppose that neuroinflammation is truly absent in ME/CFS, does this suggest that autonomic and/or vascular dysfunction may mostly be responsible for cognitive impairments (and possibly partly so for fatigue)?

This would be in line with Shan's recent systematic review of neuroimaging studies in ME/CFS (bolding mine):

Results
63 full-text articles were included in the synthesis of results from 291 identified papers. Additional brain area recruitment for cognitive tasks and abnormalities in the brain stem are frequent observations in 11 and 9 studies using different modalities from different research teams respectively. Also, sluggish blood oxygenation level-dependent (BOLD) signal responses to tasks, reduced serotonin transporters, and regional hypometabolism are consistent observations by more than two research teams. Single observations include abnormal brain tissue properties, regional metabolic abnormalities, and association of brain measures with ME/CFS symptoms. Reduced resting cerebral blood flow and volumetric brain changes are inconsistent observations across different studies.

Conclusion
Neuroimaging studies of ME/CFS have frequently observed additional brain area recruitment during cognitive tasks and abnormalities in the brain stem. The frequent observation of additional brain area recruitment and consistent observation of sluggish fMRI signal response suggest abnormal neurovascular coupling in ME/CFS.​

But not with Younger's results from whole-brain magnetic resonance spectroscopy (MRS):

Significant between-group differences were detected in several regions, most notably elevated CHO/CR in the left anterior cingulate (p < 0.001). Metabolite ratios in seven regions were correlated with fatigue (p < 0.05). ME/CFS patients had increased temperature in the right insula, putamen, frontal cortex, thalamus, and the cerebellum (all p < 0.05), which was not attributable to increased body temperature or differences in cerebral perfusion. Brain temperature increases converged with elevated LAC/CR in the right insula, right thalamus, and cerebellum (all p < 0.05). We report metabolite and temperature abnormalities in ME/CFS patients in widely distributed regions. Our findings may indicate that ME/CFS involves neuroinflammation.
​

Hopefully this Dutch team can conduct a larger PET study, while Younger continues working on his NIH R01 grant for MRS in ME/CFS. I believe Zack Shan has received funding from the Australian NHMRC too (@Simone may be able to confirm this).

Michael Van Elzakker and Kenneth Kwong (co-inventor of fMRI) also received a Ramsay award in 2019 to study brain perfusion pre- and post-exercise, which could provide helpful hints.

 

It's still all suggestive evidence, but I'd suggest something along the lines of "abnormal neurovascular coupling" too. This is also the most likely explanation for the frequent experience of headaches.

 

What does this tell us?

Hold on there! This is a second study with n=9 and using an old-generation "noisy" tracer as a proxy of inflammation. We can surely safely conclude nothing either way.

Pleny of neuoimagers (such as Chris Chambers) think the whole field struggles because there is so much inherent variation between people's brains, and so many assumptions in identifying specific parts of the brain accurately, that you need large studies. Or pooling of results. I think 50+ is a starting point. Even then, Chambers and others has pointed out that there is a huge amount of flexibility in analysis because it is such a complex, non-standard process. And we know where flexibility of anlysis leads.

That said, this new study seems a bit better than the original. Given the Japanese team have failed to follow up on their original finding after six years (which is not a good sign) I think the safest assumption is that we still have no evidence of neuroinflammation, but mostly we have no good evidence either way.

However, Maureen Hansen and colleagues are doing exactly this kind of PET study before/after exercise. I have a feeling the sample size was N=90. Michael VanElzakker also said he was trying to do this (I'm not sure if he got funding). Ron Davies also said there were going to do this a few years ago but I've not read anything since. For reasons I don't understand, Hansen and I think Michael V were talking about using the old tracer/ligand for TSPO.

So, I think we know nothing but hopefully we should get answers in the next few years.

 

I wonder whether anyone has considered making all medical test results available as anonymous 'big data'. Imagine if this research group could access several hundred or thousand scans to provide some baselines or variabilities. Obviously there would be problems with variables in how the tests were done, maybe missing data on whether the patient had other health problems, etc. Still, there's probably some value to be found there. Some opportunities for abuse too, such as medical insurance companies using it to identify types of people to refuse insurance.

 

https://twitter.com/user/status/1381937543109869571


5 tweet thread

 

I was going to say the exact same thing as Mark Vink. When BPS-researchers do "biomedical" research it's most often with the intention of not finding anything. And because they're trying not to find anything, which isn't that hard to be honest, they often succeed. Vegard Bruun Wyller (in Norway) has been wasting millions and millions of taxpayer money doing "biomedical" research trying not to find anything. And he uses this as "proof" that immunological factors aren't as important as psychosocial factors, which is of course ridiculous, but if anybody calls him out on it they are labelled "ME-activists" and accused of harrasment.

Also a smaller sample size makes it more difficult to find statistically significant results, so when you choose to only include 9(!) participants, that's pretty much a recipe for disaster, unless it's a really big difference there. Most of the researchers who think neuroinflammation is part of the picture emphasize that it's low grade - so looking for it with small sample sizes seems doomed from the beginning.

This paper looks very weak to me, to be honest.

 

Tongue in cheek response. Perhaps we should start referring to BPS activists with one-track minds who can never think outside the box they've barricaded themselves into.

 

If men have a higher inflammatory response than women it means that there is not a simple relationship between damage and inflammation so less inflammation does not always mean more healthy tissue.

If producing inflammation is a dynamic response, then looking for it in ME may be like looking at walking in ME, it changes from hour to hour so it is not a good marker of what is going on.

 

Also, at least some parts of the immune system are non-linear (think of the result a a tiny amount of allergen to some people), so the measure of one or more inflammatory markers may not say much about the severity of ME symptoms.

Another complication: maybe it's not an inflammatory cytokine that triggers the ME symptoms, but rather the body's followup to that cytokine. A cytokine triggers a response from some cells, but the cells have other mechanisms to control that response, and to 'clean up' afterwards. ME could involve those secondary functions, which are probably less well understood.

 

Final version of this paper has now been published:

https://nn.neurology.org/content/9/1/e1113