Invest in ME/Prof Jonathan Edwards statement on UK Rituximab trial, 30 July

I understand your concerns Kati, but fortunately there is no need to worry here. The protocol Drs F,M and K use is the one I devised for my patients in 1999 and which the drug company subsequently took on as standard. The two week gap is actually an artefact of the way I was treating patients then and as far as we know can be one week. Oncologists are more prone to use the metre square rule for technical reasons relating to their cancer work. The topic came up because of the question of cost price (without profit) to a public healthcare system like NHS. It is interesting to know that the charge for infusion may be less than £1000 because I think the charging within the NHS 'internal market' may be a little creative. On the other hand the cost of $6000 for 1gm in the US would be a 100% mark up and that might be used to cover the overheads that otherwise would need to be covered by infusion fees. What is of interest here is the reasonable cost in the context of trials or compassionate use where profit is not factored in - i.e. for the immediate future. It is good to note that Genentech have been helping with drug costs but Genentech only hold the franchise for North America.

This is really fascinating, Ecolimber. I looked at the abstracts and will download the full papers later. What I am thinking is that this is telling us that things are connected in some ways, but not all ways. One aspect of the 'way of modelling' that Dr Jo Cambridge and I have found useful is the idea that autoimmune diseases are loops based on random events but that for every autoantigen (self protein) the loop will behave in a slightly different way. Coeliac disease is already an instance where an environmental trigger does seem to creep in (gliadin) but the randomness should still come in with the antibody to transglutaminase arising in only a few people. Multiple sclerosis is again a bit 'outside of our outside box' because its epidemiology suggests an environmental factor. Hugh Willison has shown that post-infective polyneuropathy or Guillain Barre syndrome may after all fit the 'molecular mimicry' idea that we think is overrated in other situations. The association between narcolepsy and DR1 has been around for decades but I had not realised just how useful narcolepsy could be as a pointer to new variants on how loops might form.

The Mignot review is very helpful. It seems that there are other susceptibility genes related to T cell function so this confirms that immune signalling is almost certainly involved. The DR1 association pointed to that but for years nobody could quite believe the implications. Also the occurrence of narcolepsy in children after flu jabs looks very hard data. The time course of development over a year but not more is also very interesting. That could be the time course of a short lived plasma cell response. There is also the fact that most people with DR1 do not get narcolepsy so there is still a random factor involved. Finally, he admits that no autoantibodies or T cell responses to hypocretin have been found. That to me confirms the suspicion that molecular mimicry is too simple a theory. Something more subtle is going on, just as coeliac is a bit more subtle.

But the idea that is forming in my head now is that all these illnesses that seem to be a bit outside our outside box are neurological. Maybe the old theory of 'hidden antigens' in the brain is nearly right, although the original version has proved to be dubious. (A lot of antigens that might seem to be hidden are made available on the surface of blood vessel cells.) One of the ideas we had about MS was that if the brain is designed to work free of lymphocytes, very particular problems may arise if lymphocytes do manage to work their way in there.

So I am starting to think that ME might be a cluster of these neurological illnesses that are sort of in the autoimmune box, but like other neurological problems may be half out of the outside box. Some ME seems to have a trigger episode so it might work like narcolepsy. But what makes me rather optimistic is not only that the Norwegian study seems to show improvement with removal of B cells but the improvement may drop off when B cells come back and reappear with further treatment. This might not sound so good but to me the key point is that it suggests that at least a good proportion of ME is dependent on continual renewal of plasma cells and antibodies, not just a single post-trigger episode of short lived plasma cells. Paradoxically, that to me makes it more likely that ME can be treated succesfully. The problem with some post-infective neurological problems is that the damage may be done in the first few weeks and after that it makes no difference whether you alter antibodies or not. What is also encouraging about ME is that there as rarely any evidence of structural damage. In narcolepsy it is known that cells disappear. Narcolepsy may still be treatable by replacing the signalling molecules but if the cells have gone it seems likely that replacement would need to continue. In contrast, Fluge and Mella seem to have shown that in some cases of ME/CFS removing B cells can lead to continued remission without regular treatment.

Maybe another way of putting it is that the narcolepsy story is showing just how subtle and unexpected these mechanisms can be and that if narcolepsy can work this way there is every reason to think ME might work in a similar but slightly different way, or maybe more likely several ways with a similar outcome. We need to get rid of the old one size fits all genetic/environmental mimicry model and tease out the subtleties.

I don't think this is what you're saying, but just to check - you're not saying that one possible indication of Rituximab responders vs non-responders could be acute viral onset vs insidious? Or even acute 'anything' onset vs insidious?

Good point , Sasha. I am aware that thinking aloud will inevitably lead people to wonder what the implications are for them. But, no, I think what I am saying is similar to what I said to Firestormm: that these rules of how each disease works may turn out to give the opposite prediction for different illnesses. In narcolepsy it looks as if an autoimmune reaction (if it is really in that box) can occur either with an acute trigger or without. Our modelling approach can handle that. The upshot I think is that we are really not going to know in advance what predicts rituximab response. We can test some ideas but we do not yet know enough to bet our life savings on the answer! I do think, nevertheless, that it is worth investing a considerable amount in finding out what the answer is. Not only would that mean rituximab is given to the right people (assuming the results are confirmed) but also that we would be nearer working out the right treatment for the other people.

Funny that this question should crop up on the morning I just happened to be reading an old (1997) paper that the late much maligned Bill Reeves (CDC) co-authored.

This is described as an exploratory case control study (patients' n=26) which used a fairly comprehensive array of tests to explore differences in immune parameters between CFS patients and controls. All group comparisons showed no significant differences but when sub-grouped - type of onset, how well they were currently feeling and length of illness - there were some interesting findings (the authors admit the study was underpowered to allow for correction for multiple comparisons so these results could be a statistical artefact). However they did think these results were worth following up. I wonder if they ever did?

(bolding added) :

http://jid.oxfordjournals.org/content/175/1/136.full.pdf

It is interesting to me that the neurotransmitter involved in narcolepsy is a 33 amino acid peptide. This might be significant because there will be only very few ways a T cell receptor or an antibody could 'see' it. T cell receptors pick up short peptide segments - maybe 8 amino acids or so, if I remember rightly and the cutting up into segments tends to be certain standard ways. Antibodies can recognise very small molecules but to get signalling you tend to need two antibodies binding - so you need room for two side by side, which might need about 33 amino acids. With bigger peptides or proteins regulation may be more reliable because there will be lots of ways of binding and they won't all go wrong. So maybe in ME some other short peptides are involved?

Yes, that looks interesting. CD5 is a tricky marker because, unlike many others, it does not seem to divide cells into two groups (+ and -) but gives a continuous spectrum. The idea that CD5 is a marker for autoantibody producing cells is also something that not everyone finds. Nevertheless, I think it may well be a clue to how to find a more practical marker. I think there is another paper that finds high CD5, but the trouble is that you can set the gate for 'high CD5' wherever you like so getting consistent results may not be to be expected.

For many years, Don Staines has been proposing an autoimmune process involving vasoactive neuropeptides (including vasoactive intestinal peptide) as playing a role in ME/CFS and other 'fatiguing' conditions.

http://www.ahmf.org/05access/05staines.html

VIP appears to be a 28 aa neuropeptide but this paper suggests that just a mere 4 aa segment may retain the full neuroprotective potency as the full neuropeptide and will also bind to VIP receptors. It also suggests that physiologically cleaved sequences may naturally occur.

This is all way beyond me but could such a 'naturally cleaved' sequence provide a suitable small molecule?

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC22434/

@Jonathan Edwards
I'd be intrigued to hear your opinions of new and developing therapies in autoimmune diseases such as T-cell vaccination. This seems to be a very new field and wondered how it could be tied into other areas such as the Plasma cells, B-cells and auto-antibodies. I believe it has been trialed in a small number of diseases such as MS and rheumatoid arthritis with varying degrees of effectiveness.
http://www.ncbi.nlm.nih.gov/pubmed/12901569

prof. Edwards,

I'm sure this is no surprise to you, here's a recent comment from Dr. Peterson (One of the most experienced clinicians/researchers in ME) about the use of Rituximab:

Q: Are there any other immune boosters you are interested in?

A (dr. Peterson): IV gammaglobulin. A new area I’m excited about is the cytokine blockers and the immune modulator rituximab, which has gotten lot of press. Hopefully, they will do a much larger rituximab study in the near future. My concern with rituximab is that I don’t know how to predict who will respond. It would be nice to have guidelines, for entrance criteria, etc. where we could give people an idea of whether or not they will respond. The side effects can be very rough.

http://www.prohealth.com/me-cfs/library/showarticle.cfm?libid=18258

sounds like in RA you still have this issue of "whom will benefit from this treatment"??? How much emphasis do you put into trying to figure out which patients will benefit versus not? I guess that's where some of the biomarker information comes in?

If thinking ME through, using the approach I have found useful before, ends me up where Dr Staines has already got to I would be very reassured.

Most peptides and proteins have fairly small 'business ends' with one or more 'handles' attached. Autoantibodies could bind either to the business end, to block its function, or to a handle, which might change the situations in which the business end functions, and how, in all sorts of ways. The selection of an autoantibody producing B cell for bulk antibody production depends on B cells and T cells recognising either different bits of the antigen or maybe bits of other proteins stuck to the antigen, as in coeliac disease. So it all gets very complicated. Quite often bits get cleaved off peptides during use but the enzymes that do this may or may not cut the peptide at different places from those that cut peptides up for the purpose of presenting fragments to T cells. The whole thing has so many possibilities that I am not sure anyone has ever pinned down the exact story for any specific disease yet. I am just thinking that a peptide of about 30 amino acids that lived almost entirely in the brain might fall foul of a particular sort of immune mistake that relates both to it being in a certain size range and being hidden away. There are always risks in drawing analogies between one disease and another but I do wonder whether narcolepsy might be a particularly useful model for thinking about ME. In other words, it's beyond everyone, I think, but going through these issues can still lead to useful ideas for practical scientific work that can take things a step forward.

At this stage in your thinking Professor Edwards, with regard to a likely response to Rutuximab if one has antibodies to the thyroid, would it make any difference if one has peroxidase or anti thyroglobulin antibodies? Is one of these antibodies more likely to indicate a problem with the B cells?

(I have to admit I am not understanding an awful lot of what has been written here so far, just some of the basic information).

Thanks

Pam

To be honest Andrew, I do not know quite what T cell vaccination is about and I am not sure the people working on it do. In the forty years I have been learning about immunology I have never come across evidence for an autoreactive T cell being involved in autoimmune disease. Virtually all my colleagues assume that autoimmunity requires autoreactive T cells but they seem to forget that nobody has found them and that the theories that propose them do not really add up. T cell vaccination seems to be based on the concept of autoreactive T cells and as far as I know there is no evidence it does any good. Autoreactive T cells may well occur in mice that have been deliberately engineered to have autoreactive T cells but that proves nothing. The only human autoimmune disease for which we know what the T cells react to is coeliac disease and the T cells are recognising wheat, not self. The basis of my own approach is that autoimmunity is a situation in which autoreactive B cells (or plasma cells) make antibodies because of a loop that makes use of T cells that are reacting quite normally to foreign antigens. Most of the known autoimmune diseases fit with that quite neatly. (There is a rare human genetic disorder affecting the AIRE protein that may be an exception to this story, with genuine autoreactive T cells, but I am not sure that the story is fully understood.)

Immunologists have for the last thirty years at least assumed that T cells are in charge of immune responses, but the thinking is sloppy, to my mind. Breaking out of that mindset allowed us to develop the only cell-specific therapy that really works for autoimmune diseases (and lots of them) which is rituximab. So even if I am wrong, it has proved useful to think it is the B cells that are misbehaving, not the T cells.

Alongside all this there are a group of diseases in which it seems clear that T cells are hyperactive - like psoriasis and maybe ankylosing spondylitis. (Crohn's disease looks to be a rather odd hybrid mechanism that may also involve antibodies.) However, there does not seem to be any clear evidence that these are autoimmune T cells. They may react to foreign antigens or may be activated through pathways that do not involve specific interaction with antigen. I think we should probably stop thinking of these as autoimmune diseases.

Yes, although I am not sure why so many people think we need large studies, except maybe for safety. A scientific study only needs to be big enough to give a firm answer. I will come back to this on another thread I think.

We do still have the issue of who will benefit in RA, but we know enough to select people who have at least a 60% chance of good benefit and in many cases that is more or less complete suppression of active disease. That seems good enough. The real problem with ME is that the criteria even for the diagnosis are all subjective. At least in RA we can judge likelihood of response on seeing joints swollen to twice their size on ultrasound, a CRP up in the sky and a positive rheumatoid factor antibody. Fluge and Mella's study suggests that 67% of patients on Fukuda criteria respond and that might sound good enough to treat all patients fulfilling Fukuda criteria. But what I think is in the back of many people's minds is the fact that recognising ME is so difficult that estimates of how many people suffer from it vary by a factor of about ten - is it 1% of people or 0.1%? Uncertainty about prevalence is not unique to ME but it seems to be a particularly big problem, maybe for all sorts of reasons. And then there are reasons for thinking that ME is several different problems. It may be possible to press ahead with rituximab without any further ways of separating patients out but I can see that being difficult.

No, I don't think it would be much different. Both would be a sign of a B cell problem.

I think that's using the CCC/ICC (0.2%) vs the extremely broad criteria (Oxford? The broader CDC ones?) that have come in for heavy criticism for many years (2-4%) and aren't regarded as identifying ME (with the cardinal symptom of post-exertional malaise), just CFS (unexplained chronic fatigue with knobs on).

Others will correct me if I'm wrong on that (but let's not derail this thread with a gigantic discussion of ME vs CFS and all the different diagnostic criteria - please start up a new thread or add to an old one if you want to do that and provide a link to it, and let's keep this thread on Rituximab).

http://www.ukrituximabtrial.org/Rituximab news-Aug13 02.htm

Perhaps a more important point than the diagnostic system is the differential diagnosis looking for other causes. There was a paper by Julia Newton reporting that 40% of people referred to the Newcastle clinic received an alternative diagnosis.

http://www.rcpe.ac.uk/journal/issue/journal_40_4/newton.pdf

I also seem to remember a paper looking at people with MS who had been diagnosed with ME early in their illness. But I can't find it so I might be remembering wrongly.

This piece discusses how the two can be mixed up, or alternatively, distinguished: http://www.cfids.org/archives/2000rr/2000-rr4-article03.asp

Yes, it may be hard to pin down exactly why, but it does look as if there is going to be a big uncertainty problem around the corner. Everything in science rests on probabilities or uncertainties. Sometimes you can minimise these but often only if you make a whole lot of unstated assumptions. With criteria that allow the assessor to choose from about fifty symptom options, all of which are subjective, I think uncertainty is bound to be an issue. Anything objective, however low on specificity or sensitivity, would help.