• Welcome to Phoenix Rising!

    Created in 2008, Phoenix Rising is the largest and oldest forum dedicated to furthering the understanding of and finding treatments for complex chronic illnesses such as chronic fatigue syndrome (ME/CFS), fibromyalgia (FM), long COVID, postural orthostatic tachycardia syndrome (POTS), mast cell activation syndrome (MCAS), and allied diseases.

    To become a member, simply click the Register button at the top right.

ME Research UK: An immunosignature assay for rituximab therapy? [Funding announcement]

Another of his group’s specialities, however, is the development of immunosignatures. An immunosignature uses an array of chemical compounds called peptides to give information about the antibodies present in an individual’s blood.

Prof. Patrick acquired some samples from the Norwegian team with which to develop an immunosignature capable of distinguishing ME/CFS patients likely to respond to rituximab treatment from those who will not. This issue is important because rituximab is associated with potentially serious side-effects and requires clinical monitoring, and it is also an expensive drug.

The preliminary results were promising (200 peptides differentiated responders from non-responders 92% of the time), so ME Research UK has awarded Prof. Patrick a grant to see if the results can be confirmed in a blinded study using a larger number of samples from all 152 patients taking part in the Norwegian randomized trial.

If the immunosignature pattern for the clinical response to rituximab is found to be sufficiently sensitive and specific, it may represent a useful biomarker for ME/CFS patients’ responses, helping to predict those who will and those who will not benefit from rituximab treatment.

http://www.meresearch.org.uk/our-research/ongoing-studies/immunosignature-for-rituximab/
 

Jonathan Edwards

"Gibberish"
Messages
5,256
This is an interesting approach but I think the maths of immunochemistry are very heavily stacked against it being reproducible. If a score has to be derived from responses to 200 peptides it is very unlikely to give a black and white prediction. It is much more likely to be a grey scale - may be distinguishing people with 70%, 50% or 30% chance of responding. I am not sure that is going to make a huge difference to whether rituximab is tried.
 

Invisible Woman

Senior Member
Messages
1,267
To be honest it would be good to know if I was likely to be a responder or not.

However, as long as the doctor prescribing it was experienced and trusted and they thought it was worth a go I would try it.

Interestingly I know someone else who has a debilitating progressive disease and has been offered it but they are too scared to give it a go. Mind you, functionally they are so much more able than I.
 

alex3619

Senior Member
Messages
13,810
Location
Logan, Queensland, Australia
I do like the concept that such testing might lead to a scale of chances of improvement. I am also cautioned by the finding that duration of treatment alters the response rate, some patients need to be treated much longer than others. It is also conceivable that simply using peptides might not be enough. We need to be looking at cytokines, history, age, BMI, comorbidities, sex, duration of illness, etc. etc. etc. Until we understand the issue we need to be open to all the possibilities.
 

Jonathan Edwards

"Gibberish"
Messages
5,256
@Jonathan Edwards

My understanding was that at least one patient came to significant harm in the first trial. Do you think by doing a study of this sort we could work out not who it will work for, but who it could harm?

I am not sure anyone came to harm although I know one had a period of worsening of symptoms. I think it very unlikely that a peptide array would predict adverse events.

Amongst various things that worry me about this approach is that it does not seem to have been tested on any standard disease where one might hope it might give a clear result - rheumatoid or lupus or any other autoimmune disease. It mostly seems to have been used to test vaccine response, where it might make some sense, and cancer immunity. Almost all the papers are by one group.
 

user9876

Senior Member
Messages
4,556
He measured a very lot (125-130k) of peptides and I'm assuming that around 200 separated the groups to a reasonable degree. I don't know the variability in peptides but I would have thought that there would be a fair chance just at random that such groupings could occur.

An interesting thing to do would be to take other groupings of patients at random and see if similar separations could be got through his data set.

More generally I wonder if there is a different analysis which would be more interesting which is a before and after one.
 

Jonathan Edwards

"Gibberish"
Messages
5,256
He measured a very lot (125-130k) of peptides and I'm assuming that around 200 separated the groups to a reasonable degree. I don't know the variability in peptides but I would have thought that there would be a fair chance just at random that such groupings could occur.

An interesting thing to do would be to take other groupings of patients at random and see if similar separations could be got through his data set.

More generally I wonder if there is a different analysis which would be more interesting which is a before and after one.

Agreed on the first bit.

The last bit may be a moderately brilliant inspiration!! It might tell us nothing about responses to rituximab but it might tell us something fundamental about autoantibody production.

We have the puzzle that certain autoantibodies are made by short lived plasma cells that die off after rituximab and others are made by long lived plasma cells that don't (for a long time). Anti-DNA antibodies and rheumatoid factors tend to be short lived. Anti nucleoprotein antibodies tend to be long lived. The most likely explanation is that the flaw in the control mechanism is different for different antigens, leading to b cells going down different maturation pathways. But we do not understand what it is about each antigen that drives this. It may not be that likely but it is not unreasonable to suggest that some fairly simple stereochemical feature of antigenic epitopes is involved that might pop up very easily on a anti-peptide antibody assay. For instance epitopes recognised by short lived plasma cells might tend to have prominent arginine residues (arginine sticks out like a charged sore thumb so tends to crop up in this sort of analysis). If there was a rule as simple as that it would have huge implications for plasma cell targeting strategies. It might also provide a handle for testing how these autoimmune loops work.

So maybe to help things along I should suggest to Jo Cambridge that we get Dr Patrick to do immunosignaturing on bog standard RA and lupus sera and then after rituximab. If that gives no signal it may be time to forget immune signatures but if it comes up with something the implications might be considerable.

And we have the further comparator of lymphoma patients given rituximab despite having normal antibody dynamics. Some of their antibodies will go down and not others too, so we can see what the rules are for epitopes and their plasma cells across the board.
 

Demepivo

Dolores Abernathy
Messages
411
These days Rituximab is out of patent. I've seen postss that biosimilars are available in Europe on this site

Are the biosimilars available at a reasonable price to researchers?
 

Marky90

Science breeds knowledge, opinion breeds ignorance
Messages
1,253
I do like the concept that such testing might lead to a scale of chances of improvement. I am also cautioned by the finding that duration of treatment alters the response rate, some patients need to be treated much longer than others. It is also conceivable that simply using peptides might not be enough. We need to be looking at cytokines, history, age, BMI, comorbidities, sex, duration of illness, etc. etc. etc. Until we understand the issue we need to be open to all the possibilities.

I wouldn`t say the data suggests that some needs longer treatment, rather the contrary..
 

M Paine

Senior Member
Messages
341
Location
Auckland, New Zealand
It would be nice if this peptide screening method could lead to the identification of the actual peptides effected by auto-immunity in ME/CFS, but as already discussed in other papers from this group, that sequence data alone is likely not enough to go on when trying to find similar proteins in the human genome to those identified in the peptide screens (unfortunately :()

We've heard that Fluge & Mella were trying to identify proteins using a large collection of human peptides, which seems like a good bet. I hope that someone has success in the area of antibody target identification.
 

Marky90

Science breeds knowledge, opinion breeds ignorance
Messages
1,253
Isn't the response time variable from 8 weeks to 18 months?

Yeah, but nothing suggests that your chance of recovery increases if your a late responder who takes treatment longer. That`s what you said, or? I might have misunderstood u.
 

Murph

:)
Messages
1,799
He measured a very lot (125-130k) of peptides and I'm assuming that around 200 separated the groups to a reasonable degree. I don't know the variability in peptides but I would have thought that there would be a fair chance just at random that such groupings could occur..

I agree. If you measure a lot of things then by random chance you find some things that look well correlated.

I don't know peptides but I do know a little about making regression models. You can always get more explanatory power by adding more variables. When you get to 200 it is a sign you don't really have a handle on cause and effect.

Maybe this is the wrong mental model for thinking about what these guys are doing. If not, I'd expect the model to flunk when applied to a new patient group.
 
Last edited:

user9876

Senior Member
Messages
4,556
. Anti-DNA antibodies and rheumatoid factors tend to be short lived.

One thing I have wondered about the time delay associated with Rituximab is does that say something about the antibodies. Here you are suggesting that some antibodies are short lived and I assume that indicates something around the speed that Rituximab will work? Then the question becomes one of does the timescale suggest something about the antibody by its lifetime. The complicating factor, however, is that there may be a threshold type function or a sigmoidal effect vs levels function.


So maybe to help things along I should suggest to Jo Cambridge that we get Dr Patrick to do immunosignaturing on bog standard RA and lupus sera and then after rituximab. If that gives no signal it may be time to forget immune signatures but if it comes up with something the implications might be considerable.

I understand what you are saying from an general antibody perspective. There is another case though which is if the peptides are related to the ME disease process (I though someone was suggesting this?) in which case the change with the Rituximab patients may explain improvements? Would this give a hint at disease process?

And we have the further comparator of lymphoma patients given rituximab despite having normal antibody dynamics. Some of their antibodies will go down and not others too, so we can see what the rules are for epitopes and their plasma cells across the board.

As another comparison Rituximab is used to treat EBV in patients who have had a bone marrow transplants (to avoid post transplant lymphoma) although that small population would have a weird blood dynamics.
 

user9876

Senior Member
Messages
4,556
I don't know peptides but I do know a little about making regression models. You can always get more explanatory power by adding more variables. When you get to 200 it is a sign you don't really have a handle on cause and effect.

You need to be really careful about adding more model variables when fitting a model. The problem is the more variables in the model the more likely you are to fit to noise rather that the general trends. Also I've always thought that as you add a variable you want to double the number of data samples.

But I think you were talking about the number of variables in the input. A good supervised classifier will learn (given sufficient data) what is important and what isn't but there is not really enough data and then the model needs to be tested on fresh data - which I think is what they have funding to do.
 

RogerBlack

Senior Member
Messages
902
Exactly my thoughts. If this is successful, and providing we don't see any other treatment discovered, I'd definitely want to find out if I should respond well to Ritux.

I have not found the original Patrick paper.
8% of people being told they would not benefit falsely, plus presumably >0% being told they would benefit falsely starts to become a significant number.

There is also the non-medical medical aspects that are all too real.
'Oh, he's been given rituximab in a trial, he must be really ill'. Is a thing that actually works on doctors too.
I cannot find the paper at the moment, it was on anchoring bias in attitudes by doctors to case studies.

Whatever is the case, any Rituximab treatment absolutely must take samples from amongst the routine bloods I understand are taken with the treatment, and stick them in a biobank somewhere.
 
I have not found the original Patrick paper.
8% of people being told they would not benefit falsely, plus presumably >0% being told they would benefit falsely starts to become a significant number.

There is also the non-medical medical aspects that are all too real.
'Oh, he's been given rituximab in a trial, he must be really ill'. Is a thing that actually works on doctors too.
I cannot find the paper at the moment, it was on anchoring bias in attitudes by doctors to case studies.

Whatever is the case, any Rituximab treatment absolutely must take samples from amongst the routine bloods I understand are taken with the treatment, and stick them in a biobank somewhere.
8% is better than the 33% that the stage 2 trials show don't benefit from Ritux. And instead of 67% benefiting, 92% should benefit instead - those are definitely significant numbers, IF this test is proven to be accurate.