Systemic antibody responses against human microbiota flagellins are overrepresented in chronic fatigue syndrome patients, 2022, Vogl et al

Funding Information
Austrian Science Fund (FWF): Erwin Schrödinger fellowship J4256
Solve ME/CFS Initiative (SMCI): Ramsay Award

 

case selection is probably good then

That's a lot of peptides. Presumably there is the possibility that these little bits of protein antigens might join together to make a protein that people with ME/CFS are actually reacting to, but there may be no reaction to the protein fragments. It might be like a key that fits into a lock - if you take the key and smash it up into little pieces, none of the pieces are going to open the lock. I don't know if there are many examples of this technology finding useful things?

So, that's sounding a bit disappointing. Clearly peptides are binding in both healthy controls and the people with ME/CFS. And no single peptides bound at significantly different rates in the two groups.

This figure shows the interesting finding:


ME/CFS patients were found to be a lot more likely to have antibody responses to peptides associated with bacterial flagella (the 'tails' that some bacteria have to help with movement). Also Figure 1E, showing that the ratio of antibody responses in ME/CFS and controls is about 3 to 1 for the flagella-related antigens, but pretty much 1 to 1 for the rest of the antigen groups.

 

So, then they tried to narrow down the particular bacteria that might be eliciting the antibody response.

Some of the detail is given in Table 2a below. Note the ratios on the right (incidence in ME/CFS vs controls). The biggest ratio is for a Legionella flagella with 15% of ME/CFS patients responding and only 2.5% of healthy controls responding. I'm not too sure what to make of this, as the differences don't look too compelling. It's certainly not a situation of overwhelming levels of response in the ME/CFS people versus basically no response in controls.


They then went on to use machine learning to develop a sort of biomarker signature that supposedly would identify ME/CFS patients. Here's the 15 top performing peptides:

The story, if there really is one here, is looking pretty complicated. And I wonder to what extent different diets might be influencing things.

They also looked at parameters in the blood, finding lower creatine kinase in the ME/CFS group (as previously found in UK-MEB samples), but not finding much else of note. We didn't get very excited by that finding when it was first reported, because of the risk that the levels are associated with activity - and there are big differences in activity levels between the severe ME/CFS patients and the controls.

They next tried making machine-learning generated algorithms to identify the ME/CFS individuals from the controls based on the blood test parameters, and found that it sorted the patients from controls better than the antibody models did. But, that might just be largely based on the creatine kinase, with the activity level confounding. And then, of course, they made biomarker models with the antibody results and the blood test results.

 

Discussion section

So, this is interesting - an increased response to Lachnospiraceae flagellins was found in the ME/CFS group, and has been reported in Crohn's disease.

They do discuss the possibility that the overlapping findings in the two diseases might be due to problems in the lab processes, but they don't think so. They also discuss the concern about creatine kinase just being a marker of activity, and the possibility that diet, or long periods of not eating might be contributing to the differences.

All up, the antibodies to some bacteria proteins is interesting, but with no clear differences between patients and controls (there's a lot of overlap), I'm less excited about this than some other recent findings.