Impact of immunoadsorption on autoantibodies and physical performance in post-COVID patients […] exploratory study of 18 participants, 2026, Wossidlo+

[SNT Gatchaman]

Impact of immunoadsorption on autoantibodies and physical performance in post-COVID patients: a prospective exploratory study of 18 participants

Spoiler: Authors

Wossidlo, Christine; Steinbeck, Felix; Geerdes-Fenge, Hilte; Sombetzki, Martina; Mitzner, Steffen; Loebermann, Micha; Reisinger, Emil C; Koball, Sebastian

PURPOSE
Autoantibodies against β-adrenergic and muscarinic acetylcholine receptors are said to play a role in the pathophysiology of post-COVID syndrome. To date, only a limited number of studies have investigated the effectiveness of immunoadsorption. This prospective exploratory study investigates the effectiveness of immunoadsorption by assessing reductions in autoantibody levels and improvements in physical performance.

METHODS
In a prospective exploratory clinical trial, 18 patients with post-COVID syndrome and elevated levels of antibodies against β1/β2-adrenergic receptors and/or M3/M4-muscarinic acetylcholine receptors were treated with a course of five consecutive immunoadsorption sessions. Clinical assessments, objective functional performance tests (e.g., step counters), laboratory parameters, and self-reported performance were performed during the study.

RESULTS
A total of 15 women and 3 men were included in the study. Immunoadsorption was well tolerated. All participants showed a significant reduction in circulating antibodies against β1/β2-adrenergic receptors and M3/M4-muscarinic acetylcholine receptors; however, these levels returned to baseline within the following 3 to 6 months.

OBJECTIVE
determinations of physical performance showed no significant improvement following immunoadsorption. Self-reported evaluation revealed a slight improvement in pain and fatigue. However, the correlation analysis failed to demonstrate a significant association between changes in autoantibody levels and subjectively perceived symptoms.

CONCLUSION
Immunoadsorption was safe and resulted in a short-term, significant reduction in measured autoantibodies. However, there were no improvements in objectively measured performance. Our results indicate that immunoadsorption is not an effective approach for patients with post-COVID syndrome.

Web | DOI | PDF | Infection | Open Access

 

[Utsikt]

This is a shot to the bow of Scheibeinbogen’s hypothesis. They had patients with elevated levels of autoantibodies, managed to reduce them but found no effect even in an open-label study and no correlation with any of the measurements.

The patients had an average step count of 9000 so there might be some ceiling effects, although they also had an average bell score of 40 before the trial.

If you want to construct an autoantibody hypothesis it would involve loops that require suppression of the autoantibodies for a longer time than what was achieved here. Although that might be difficult to reconcile with the sudden onset of symptoms people with post-infectious onset frequently report.

They also mention that some RA and MS patients do report transient improvements with this treatment, further distancing LC from those mechanistically.

In rheumatoid arthritis and multiple sclerosis, a clinical response has been observed in 48% and 88% of cases respectively within a follow-up period of up to 12 weeks after immunoadsorption [43]. In contrast, our data do not show such a pronounced clinical efficacy in post-COVID patients, despite the significant reduction in autoantibodies and immunoglobulins. This suggests that the observed symptom burden may be driven by mechanisms beyond circulating autoantibodies.

For som reason they claim that it would have been unethical to use a control group:

A placebo-controlled design would have required CVC placement in patients with poor peripheral venous access, to simulate blood exchange without immunoadsorption. Moreover, due to the complexity in assessing therapeutic response and the subjective nature of widespread symptoms, a very large sample size would have been required to detect notable improvement in symptoms, making funding somewhat difficult. After careful considerations a placebo group could not be not established in our study for ethical and logistical reasons.

If Fluge and Mella can do a study with 44+22 participants, I doubt an RCT of this would have required a «very large sample size»..

 

[Silip]

This is a shot to the bow of Scheibeinbogen’s hypothesis. They had patients with elevated levels of autoantibodies, managed to reduce them but found no effect even in an open-label study and no correlation with any of the measurements.

The patients had an average step count of 9000 so there might be some ceiling effects, although they also had an average bell score of 40 before the trial.

If you want to construct an autoantibody hypothesis it would involve loops that require suppression of the autoantibodies for a longer time than what was achieved here. Although that might be difficult to reconcile with the sudden onset of symptoms people with post-infectious onset frequently report.

They also mention that some RA and MS patients do report transient improvements with this treatment, further distancing LC from those mechanistically.

For som reason they claim that it would have been unethical to use a control group:

If Fluge and Mella can do a study with 44+22 participants, I doubt an RCT of this would have required a «very large sample size»..

Their average step count was 9000!?
The average step count for a person is between 6000 and 10'000 depending on the country.
(Source: https://pmc.ncbi.nlm.nih.gov/articles/PMC3197470/)
And a the same time they had Bell 40?
If they filled out the Bell-Score correctly they on average had to have a reduction in their functioning and activity by 50-70%, of course it's technically possible but very unlikely if their average step count still is in the norm and even the most affected persons still made above 6000 steps a day.

I'm not a big fan of either immunoadsorption or Dr. Scheibenbogen's theories on autoimmunity in general but to derive much from such a sample for Long Covid let alone M.E. in my opinion wouldn't be serious.

If I would want to advocate desperately FOR an effect immunoadsorption I even could argue that those people showed improvements 2-3 months after treatment, Bell Score goes from average 40 to 54, step count from 9000 to 9470, pain decreased by 40% etc...
As this isn't much and not placeco controlled this most possibly isn't meaningful at all, but just to mention that it would be possible to argue like that without sounding completely irrational.

 

[EndME]

For some reason they claim that it would have been unethical to use a control group:

That seems a bit peculiar considering that others in Germany have conducted RCTs with immunadsorption, for example the group in Hannover whose trial seemed to have negative results (still unpublished) as well as the trial in Mainz which had negative results.

The patients had an average step count of 9000 so there might be some ceiling effects, although they also had an average bell score of 40 before the trial.

That is very high. It was measured with the Fitbit Inspire 2 smartwatch, which at least isn't known for overestimating step count in general.

However, roughly half (11/18) of the participants are unable to work which suggests some higher level of disability (together with the Bell score), but unfortunately it isn't clarified whether the inability to work is related to their newly onset symptoms or what exactly in general.

I find the symptom characterisation slightly lacking but that may have to do with difficulty in recruitment in general. Something that stands out is the difficulty in recruitment:

Despite the approximately 400 patients seen annually in our post-COVID outpatient clinic, recruitment efforts had limited success. The main reason was that only a subset of patients received a confirmed diagnosis of post-COVID syndrome [2]. Furthermore, many patients did not meet inclusion criteria, primarily due to relevant comorbidities. Among the 133 screened patients, antibodies against β1/β2aR and M3/M4mAcR were not detectable in 29%. Additionally, several patients resigned at short notice for personal reasons, due to safety concerns, or because of spontaneous clinical improvement.

 

[Utsikt]

Their average step count was 9000!?
The average step count for a person is between 6000 and 10'000 depending on the country.

We know from other data that step counters are not reliable for comparisons between different models. They might not even be very comparable between different people using the same model.

But I agree that it seems like it’s within the range of what healthy people report.

However, roughly half (11/18) of the participants are unable to work which suggests some higher level of disability (together with the Bell score), but unfortunately it isn't clarified whether the inability to work is related to their newly onset symptoms or what exactly in general.

The SD for Bell was 15 if I remember correctly, so they might have had a relatively large range of patients.

 

[Felixm99]

I think the only thing you can infer from this study is that IA is not a miracle cure for somewhat loosely defined Long Covid. Which we knew.

On the one hand: No control group, 9000 steps baseline (?!) plus repeated home exercise. I also don‘t think the ELISA test for these AABs is very useful

On the other: an actual increase in Bell score by 12 points which is not nothing, some subjective improvements

All in all I don‘t think this tells us much

 

[V.R.T.]

I'm not a big fan of either immunoadsorption or Dr. Scheibenbogen's theories on autoimmunity in general but to derive much from such a sample for Long Covid let alone M.E. in my opinion wouldn't be serious.

I agree, 9000 steps is such a high baseline it's conceivable people could feel a lot better and not increase steps much.

This study doesn't look like the slam dunk negative finding we thought it was, even if I remain unconvinced by CS' hypothesis.

 

[Utsikt]

I think the only thing you can infer from this study is that IA is not a miracle cure for somewhat loosely defined Long Covid. Which we knew.

On the one hand: No control group, 9000 steps baseline (?!) plus repeated home exercise. I also don‘t think the ELISA test for these AABs is very useful

On the other: an actual increase in Bell score by 12 points which is not nothing, some subjective improvements

All in all I don‘t think this tells us much

I agree, 9000 steps is such a high baseline it's conceivable people could feel a lot better and not increase steps much.

This study doesn't look like the slam dunk negative finding we thought it was, even if I remain unconvinced by CS' hypothesis.

I think we need to consider what it would have looked like if the treatment actually worked, which is why I shared the bit about RA and MS where it was very obvious that it works for some.

Sure, you can construct a version of reality where every study has had some fatal flaw that caused them to be false negatives. But how likely is that at this point?

 

[Silip]

I think we need to consider what it would have looked like if the treatment actually worked, which is why I shared the bit about RA and MS where it was very obvious that it works for some.

Sure, you can construct a version of reality where every study has had some fatal flaw that caused them to be false negatives. But how likely is that at this point?

Not that likely in general.

Regarding this study you still can't seriously include people with a stepcount in the average norm, then give them a treatment and then conclude the treatment wasn't successful because their stepcount hasn't improved enough.

If, right at the start of your study, your patients already have values very similar to those of healthy people for your primary outcomes, why would you expect them to improve?

 

[Utsikt]

Not that likely in general.

Regarding this study you still can't seriously include people with a stepcount in the average norm, then give them a treatment and then conclude the treatment wasn't successful because their stepcount hasn't improved enough.

If, right at the start of your study, your patients already have values very similar to those of healthy people for your primary outcomes, why would you expect them to improve?

There was room for improvement in the subjective outcomes. And they found no correlation between autoantibody levels and changes in any of the outcomes.

I agree that there is a ceiling effect when the average step count was within the healthy range.