Trial Report Plasma cell targeting with the anti-CD38 antibody daratumumab in ME/CFS -a clinical pilot study, 2025, Fluge et al

[V.R.T.]

we cannot discount the possibility that some people with a diagnosis of ME/CFS have other diseases that may be critically different in this respect. You are assuming they all had the ME/CFS you are familiar with. I don't think you can know that.

Agree with what you're saying, but I think the fact it is Fluge and Mella who ran this trial as opposed to say, the US recover program makes it more likely that they carefully selected patients representative of most pwME as we are familiar with them.

 

[Jonathan Edwards]

Agree with what you're saying, but I think the fact it is Fluge and Mella who ran this trial as opposed to say, the US recover program makes it more likely that they carefully selected patients representative of most pwME as we are familiar with them.

One would hope so but who knows whether the criteria we use differentiate? Haemophilia turned out to be two almost identical diseases. ME/CFS is a bit different but not because it is easier to pin down!

 

[ryanc97]

The trouble is that although your intuition may be right that a person with the disease that most members here probably have may well be unlikely to triple a step count for months, we cannot discount the possibility that some people with a diagnosis of ME/CFS have other diseases that may be critically different in this respect. You are assuming they all had the ME/CFS you are familiar with. I don't think you can know that.

I would like to believe Dara works but I have seen enough trial results to know that you cannot take these findings as proof of 'response'.

I think the chance of it being another disease is impossibly small, because F/M chose the patients.

I think the chance of placebo is very low, because it’s not a survey, it is an observed outcome, and if it were placebo, step counts would not be sustained. The patients will crash and go back after a month of increased steps.

Another Possibility is that the data was faked or manipulated (eg inflating step counts or faking it) but given what I understand of F/M, this is unlikely. This is my worst fear.

So as I mentioned the main question is why it didn’t work in non responders, and if LLPC is the culprit

 

[ryanc97]

On a side note, I am very curious on why the response was minimum two months.

If it were antibodies, surely the response would be linear in time?

One can argue for the long live of IGG and saturation of what it binds to, but assuming whatever they bind to is saturated, that does not account for different severity levels (ie more severe = more AABs)

 

[Jonathan Edwards]

I think the chance of it being another disease is impossibly small, because F/M chose the patients.

I know Oystein and Olav well. I have sat talking with them for hours. They are very honest and humble folk and I think would agree that they cannot possibly know how many diseases their patients fall under. What I am sure of is that they would not fiddle data. But they cannot know if there are two illnesses that look the same, one of which is the one you are familiar with and one of which is not. It may have a lot in common but may fluctuate very differently.

 

[EndME]

Well believe it or not nobody in ME/CFS trial history has bothered to do a pre recording of step counts and step counts.... insanity.

Plently of trials have recorded pre trial step count and post trial step count, even if that data is often not commented on. I think Fluge and Mella have put in quite a bit of additional effort on trying to recognise step count trends including a longer pre recording phase, but even PACE had objective (null) data. Even CBS clinicans sometimes use actimeter data as part of their monitoring process.

To me, the question is not whether it is placebo. Clearly it works. For some. The question is why it works for responders, why it doesn't for non responders and also what does this reveal about the underlying mechanism of ME/CFS.

You don't triple your step count consistently for months due to placebo.

The Rituximab trial included actometer data, including step count (pre trial for a week and then later at 17-21 months during follow-up), and I'm sure you will find people in the placebo group that had the same improvements as the people here in the Daratumab study (you will probably even find some individuals in the PACE study who had similar sustained improvements in step count in both arms despite the treatment having no efficacy). "Placebo" in this case includes "natural full recovery" and in a recent high profile study of ME/CFS patients that were probably selected at least as rigourously as here, that rate was not 0. This happens. I'm sure you can find hundreds of unblinded studies with objective outcome measures that looked drastic but where it turned out that things do not do anything. I don't see how one can draw any conclusions before the placebo-controlled study is completed.

I agree that it is hard to explain how exactly the recovery rate could be so large in this study if the drug were to have no efficacy but then again you have the possibility of all sorts of other biases contributing to the seen data. I’d be surprised if Fluge and Mella thought that this data looked categorically different from their Rituximab data, which turned out to have no efficacy.

 

[ryanc97]

I know Oystein and Olav well. I have sat talking with them for hours. They are very honest and humble folk and I think would agree that they cannot possibly know how many diseases their patients fall under. What I am sure of is that they would not fiddle data. But they cannot know if there are two illnesses that look the same, one of which is the one you are familiar with and one of which is not. It may have a lot in common but may fluctuate very differently.

Yeah, they are good guys for sure, so it's definitely real.

Plently of trials have recorded pre trial step count and post trial step count, even if that data is often not commented on. I think Fluge and Mella have put in quite a bit of additional effort on trying to recognise step count trends including a longer pre recording phase, but even PACE had objective (null) data. Even CBS clinicans sometimes use actimeter data as part of their monitoring process.


The Rituximab trial included actometer data, including step count (pre trial for a week and then later at 17-21 months during follow-up), and I'm sure you will find people in the placebo group that had the same improvements as the people here in the Daratumab study (you will probably even find some individuals in the PACE study who had similar sustained improvements in step count in both arms despite the treatment having no efficacy). "Placebo" in this case includes "natural full recovery" and in a recent high profile study of ME/CFS patients that were probably selected at least as rigourously as here, that rate was not 0. This happens. I'm sure you can find hundreds of unblinded studies with objective outcome measures that looked drastic but where it turned out that things do not do anything. I don't see how one can draw any conclusions before the placebo-controlled study is completed.

I agree that it is hard to explain how exactly the recovery rate could be so large in this study if the drug were to have no efficacy but then again you have the possibility of all sorts of other biases contributing to the seen data. I’d be surprised if Fluge and Mella thought that this data looked categorically different from their Rituximab data, which turned out to have no efficacy.

Are you able to post the step count data here Rituximab trial? Which phase was it? I cannot find it. Would be interested. I feel like if there were step counts for Ritux trials I would have known about this already, and I don't.

 

[ryanc97]

This is NOT the Rituximab study data.

This source comes from a no intervention observational study in 2022: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0274472

So F/M did this to try and observe natural variation in ME/CFS patients.

The middle responders, aka a 30-45 point increase in PF SF-36 score, median step count increased from 4.8k to 6k.

The super responders, aka a >45 point increase in PF SF36 score, median step count from 5k to 5k. No increase.

That is a far cry from 2k to 9k steps in Patient 1 in Dara, the super responder. A 20% increase versus a 450% increase. The effect size is 22 times as large.

As we can see, it means the SF-36 is not reliable. Survey data of feeling better is not associated with corresponding increase in step count.

 

[EndME]

Are you able to post the step count data here Rituximab trial? Which phase was it? I cannot find it. Would be interested. I feel like if there were step counts for Ritux trials I would have known about this already, and I don't.

They used step count as part of the Phase 3 trial https://clinicaltrials.gov/study/NCT02229942. In their published paper they write

Each patient used an electronic SenseWear arm- band at home continuously for 5 to 7 days to record baseline level of physical activity (number of steps) (16, 17).

Recording of physical activity with SenseWear armband registration was repeated between 17 and 21 months.

But as the story goes the results in the treatment group were not statistically significant in these measures either.

Beyond the outcome measures described in the article, none of the following secondary end points defined in the study protocol differed significantly between intervention groups: repeated measurements of SF-36-mean5, changes from baseline to 18-month follow-up in maximum number of steps per 24 hours, mean and maximum duration of physical activity of at least 3.5 METs per 24 hours, or longest duration of lasting clinical response (data not shown).

We know the average data won't look like it does here, but my point was that there are individuals for whom it looks will look no different to what has been presented.

 

[EndME]

This is NOT the Rituximab study data.

This source comes from a no intervention observational study in 2022: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0274472

So F/M did this to try and observe natural variation in ME/CFS patients.

The middle responders, aka a 30-45 point increase in PF SF-36 score, median step count increased from 4.8k to 6k.

The super responders, aka a >45 point increase in PF SF36 score, median step count from 5k to 5k. No increase.

That is a far cry from 2k to 9k steps in Patient 1 in Dara, the super responder. A 20% increase versus a 450% increase. The effect size is 22 times as large.

As we can see, it means the SF-36 is not reliable. Survey data of feeling better is not associated with corresponding increase in step count.




View attachment 30441

Thread on that study here.

 

[ryanc97]

Alright, I found the Ritux step count data. It is at the last few pages of the PDF of phase 3 study.

Across treatment centres, there is an approx mean step count increase of 1k across everyone from 3k to 4k. Since we know ritux does not work, we can account any increase to be from placebo. Then:

In Dara, mean step count increase in responders went from 3k to 7k, an increase of 4k.

My point being, the 1k step count increase is a placebo effect. A 4k step count increase, is not a placebo effect. Done. @EndME I think my point is proven.

 

[EndME]

My point being, the 1k step count increase is a placebo effect. A 4k step count increase, is not a placebo effect. Done.

You have forgotten the fact that there (likely) will be people with a 4K step count increase in the placebo group in the Rituximab trial. That was the point.

 

[ryanc97]

You have forgotten the fact that there (likely) will be people with a 4K step count increase in the placebo group in the Rituximab trial. Done.

But the difference is, I am comparing means. You are cherry picking an individual data point. You have to see the means.

First, we do not know if there is a data point with a sustained 4k step count increase in the rituximab study.

Assuming there is for your sake, you cannot just cherry pick it. You have to see the mean, because that aggregates everyone. If you want to cherry pick, note Patient 1 in Dara went from 2k to 8k, so a 6k increase.

If placebo effect is so powerful, why do we not see a 4k step count increase in aggregate in the Rituximab study? We see a 1k increase.

So we know a placebo or a drug that doesn't work can increase step counts by 1k. So what are the odds in Dara, the placebo effect was 4x as powerful?

 

[ryanc97]

I will go a step further. Here is a table of step count increase, n=46, in responders of Ritux phase 3. Be it placebo or Ritux.

Mean steps went from 2.2k to 3.2k.

My point is empirically, a placebo effect can give you 1k step increase, not 4k.

I mean this makes sense. The 4k likely comes from doing things like commuting to work, walking after work, going out on the weekend, etc.

 

[EndME]

But the difference is, I am comparing means. You are cherry picking an individual data point. You have to see the means.

First, we do not know if there is a sustained 4k step count increase in the rituximab study.

Assuming there is for your sake, you cannot just cherry pick it. You have to see the mean, because that aggregates everyone.

If placebo effect is so powerful, why do we not see a 4k step count increase in aggregate in the Rituximab study? We see a 1k increase.

So we know a placebo or a drug that doesn't work can increase step counts by 1k. So what are the odds in Dara, the placebo effect was 4x as powerful?

But that is an apples to oranges comparison. Obviously the mean step count increase wasn't high, because the drug has no efficacy! If this study had the same sample size as the Rituximab study or was placebo-controlled and there was a mean step-count increase of 4k steps the situation would be completely different. There would be nothing to discuss. But it is not the case.

Currently you just have a handful of individuals that had a large step count increase in the Daratumab study. And we know from numerous studies in ME/CFS that this can also happen when people are given a placebo or nothing at all (as seen elsewhere) so you cannot conclude from such data alone that it has to be due to the treatment having an efficacy even if this effect is more consistent across the group.

So we know a placebo or a drug that doesn't work can increase step counts by 1k.

We know that in a limited amount of individuals the same effects as witness in the Daratumab study has been witnessed. The question would be why the effect would be much more consistent in the Daratumab study without there being treatment efficacy, but that has already been commented on.

Let's see how the objective data of the current study will look like once it will get published. I'd be more than happy if there is response to the drug including an increase in steps.

 

[ryanc97]

All I can say is from a probabilistic perspective, we have observed a placebo effect giving a 1k aggregate step count increase.

Therefore, the probability of a 4k step count increase being placebo is way, way too low, even if the sample size is small.

Even if there were only 6 patients vs 46 patients.

Assume placebo step count increase follows a distribution with 1k at the mean. Say it is normally distributed. Say the stdev is 1k. So the probability of drawing an increase of 4k, say that is 2 stdevs away, is 5%.

Now these samples are independent, so you can multiply 5% by 6. That is like.... 1%.

I know these are all made up figures, but the probability of a 4k effect being placebo is really too small.

I mean, lets say you have one placebo super responder with a 4k increase. What are the odds of ALL 6 your placebo responders having a 4k increase? And not only that, what are the odds they all started increasing their step counts at the same time? At 2 months. Those odds are like winning a lottery.

 

[ryanc97]

Currently you just have a handful of individuals that had a large step count increase in the Daratumab study. And we know from numerous studies in ME/CFS that this can also happen when people are given a placebo or nothing at all (as seen elsewhere) so you cannot conclude from such data alone that it has to be due to the treatment having an efficacy even if this effect is more consistent across the group.

If you can find me a study in ME/CFS that has a sustained doubling or tripling or even a 4x of step counts, sustained, over a year from placebo, please show me.

I would change my mind instantly about Dara. Point is, I don't think such a study exists, because it is not possible for placebo to do that.

 

[Jonathan Edwards]

Assume placebo step count increase follows a distribution with 1k at the mean. Say it is normally distributed.

We don't have the slightest reason to think 'placebo' effects in the broad sense are parametric. It seems extremely unlikely to me. By their very nature we know almost nothing about their quantitative characteristics.

I can understand your enthusiasm but nobody with any experience of trials is going to agree with you I am afraid. We don't know what these results mean. Oystein and Olav would not want people to base decisions on your analysis.

 

[Utsikt]

The NIH effort preference study had a few complete remissions without any intervention.

If you look at lp-fortellinger.no you’ll find many stories of people doing far more than they used to for a long time before eventually crashing.

I know a person that abruptly recovered from 2-3 years of ME/CFS-like LC after a reinfection with covid.

 

[EndME]

If you can find me a study in ME/CFS that has a sustained doubling or tripling or even a 4x of step counts, sustained, over a year from placebo, please show me.

I would change my mind instantly about Dara. Point is, I don't think such a study exists, because it is not possible for placebo to do that.

There is no study where there is a mean tripling of step count that is sustained in a placebo-controlled trial because there is currently no effective treatment for ME/CFS.

What we know is that there are individuals with ME/CFS for whom this however applies. Depending on study context that rate might be as high as 20%. I'm sure you would acknowledge this as well.

The question you're wondering about is: How can this effect be this consistent in the trial if it was to be placebo. You made a assessment that you find it to be unlikely that say something that usually might only happen in 10% all of a sudden happens 70% of the time and under assumptions of normality and so forth and then indeed it is entirely unlikely that this is the case. But of course there is absolutely no reason to assume this as others have already mentioned many times.