Friday, September 13, 2013

Results from a Phase-II Trial of Teplizumab in Honeymooners

This blog posting covers the results from a phase-II clinical trial of a drug called Teplizumab.  This trial ran from 2005-2011 and was published on line in July 2013:
Abstract: http://www.ncbi.nlm.nih.gov/pubmed/23835333
(The full paper is not available for free on line, but one of the authors sent me a PDF, so I do have the published details: Thanks very much!)

I have written several blogs about this drug in the past, which you can read here:
http://cureresearch4type1diabetes.blogspot.com/search/label/Teplizumab

The sound track for this blog posting is "Wait" by Wang Chung from the "To Live and Die in LA" soundtrack:   http://grooveshark.com/#!/s/Wait/3IYAu1?src=5

Results

The key results are summarized in this graph, which I will explain below:
This graph shows a person's ability to generate their own insulin (as measured by C-peptide production [d1]).  Higher points on the left hand scale show more insulin production, so are good.  The bottom scale is the number of months from treatment, and treatment started about a month after diagnosis.
  • In this study, about 55% of the people who were treated, had almost no effect.  This is the green line above, which is right next to the blue line (untreated people).
  • However, about 45% of the treated people had a good response to the treatment, and this data is the red line.
  • If you look at the responders (the red line), you can see that their insulin generation ability actually went up in the 6 months after their first dose.  And even about 20 months later, they were still generating as much of their own insulin as they were when first treated.   This is in stark contrast to the placebo group, and the non-responsive group, which steadily dropped for that whole time.
Comparing results from different studies is never straight forward, but here is my attempt to compare the recent TOL-3021 results to Teplizumab.  For this comparison, TOL-3021 results are for established type-1 diabetics and include everyone who got the best dosing regimen, while the Teplizumab results are for honeymoon type-1 diabetics, and only include "responders" (about 45% of total):
  • Both treatments resulted in a maximum improvement of +20% in the body's ability to produce it's own insulin, as compared to their production at the start of the study.
  • The duration of effect (until insulin production dropped back down to baseline) was about 6 months for TOL-3021 and about 20 months for Teplizumab.
  • The absolute change was higher for Teplizumab than for TOL-3021.  The Teplizumab patients started out generating higher levels of insulin since they were in their honeymoon phase, so their maximum was 20% of a larger number.

Background

How Teplizumab Works

Teplizumab is a humanized monoclonal antibody [d2] which targets CD3 cells in the immune system in order to lower (or stop) the body's autoimmune response. This drug tries to prevent type-1, or lessen it's severity, by "turning down" [d3] the immune system's attack on the body's own pancreas cells. This basic approach has resulted in treatments (but not cures) for other autoimmune diseases. It does carry the possible risk that the body's immune system will not properly attack a real threat in the future.

The History of Teplizumab Clinical Trials

Teplizumab has a long history of clinical trials, and not all of it successful.  There are 9 clinical trials of Teplizumab either completed or ongoing at this time.  The quick summary is this:  A company called MacroGenics (partnering with Eli Lilly) put Teplizumab through a full round of testing, culminating in two large phase-III clinical trials.  It looked good in the phase-II study, but failed in the first phase-III study.  The second phase-III study was canceled.  In addition to the MacroGenics studies, there were other studies done by universities' researchers that were not directly related to FDA approval.  This study is the most recent of those.

Discussion

There are a lot of interesting issues here, but I'm gong to touch on some of the bigger ones:

Why did previous studies fail, while this study succeeded? This is a major question.  Remember that the study that failed had twice as many people as this one.  On one hand, it is reasonable to expect more accurate results from the larger study.  On the other hand, later trials can learn from the earlier trials, and the successful trial was later.  So should we side with the larger trial that was unsuccessful, or the later trial that was successful?  I think the answer is: we do more research, run more clinical trials.

The researchers in the study suggested that their better results were caused by treating generally younger people, and starting treatment closer to diagnosis [d4].  If these two factors did cause the difference, it is a mixed bag, because on the one hand it suggests that this really is a honeymoon-only treatment, but on the other hand, if approved as a regular treatment, then it is reasonable to assume that it will be given within a day or two of diagnosis, and even better results might be seen.

Could it continue forever? / Did the second dose help?  Another question which is critical to this research is this: Did the second dose help?  Because answering that question is critical to answering the question: can this treatment be extended for longer periods of time?  If the best this treatment can do is extend the honeymoon by 12 extra months, that has some advantages, but it's not an (obvious) path to a cure.  On the other hand if repeated dosing results in longer and longer honeymoons, that might someday lead to a cure.  And even more quickly: a preventative [d5].

Unfortunately, the news here doesn't look too good.  The second dose was given at about 12 months, so pretty close to the second row of dots in the graph above.  You'll notice that there is no improvement at that point.  The C-peptide measures don't change much (i.e. the slope of the line doesn't change at that point).  If the second dose had an obvious good effect, I would expect to see a bump up from about 12 to 18 months, much as there is a bump up between 0 and 6 months. Unfortunately, I don't see that.  An interesting trial would be to randomly give some people one dose, others two doses (as done here), and others three, four, or more doses [d6].  That would give very strong data telling us if extra doses resulted in longer effectiveness.

Is this a preventative, a honeymoon cure, or a cure for established type-1 diabetics?  In a sense, this study extended the honeymoon from one year to two years (approximately) in the 45% of the patients who responded to the treatment.  So right now the biggest question is: can this result be extended for longer periods of time?  If the answer is "yes", then it can be used as an important adjunct treatment.  It would lower insulin use, and presumably lower all sorts of long-term bad side effects of having type-1 diabetes.  Furthermore, if the results can be extended, then this might very well turn into a preventative.

Who is going to respond?  The difference between those who responded to the treatment as compared to those who did not respond, was large.  From a clinical point of view, it brings up a different question: can we tell ahead of time who is going to respond, so we don't even bother to treat people who will not benefit.  From a basic research point of view, this brings up a bunch of interesting questions which boil down to: why do some people respond and others don't?

The researchers did see differences in people who responded versus those who did not.  They do not have a single test that gives a simple yes/no answer, but by combining the results of several tests, they can say that some people are a lot more likely to be responders, and others much less likely. However, if the side effects are small enough, then it might be worthwhile to dose everyone.  We will need to know more about the rate adverse effects to be sure.

We know that many different genes are involved in type-1 diabetes, and several different autoantibodies, so studying Teplizumab may give insight into immunological differences between different populations of type-1 diabetics.

Discussion Footnotes

[d1] When your body makes insulin, it comes with a C-peptide attached, but if you inject insulin, there is no C-peptide.  So measuring C-peptide will measure just the insulin generated by the body.  Neat trick; very useful in all kinds of diabetes research.

[d2] Monoclonal antibodies is a method of creating therapies by finding one cell that attacks the cell you don't want, and cloning it.  You end up with a vast number of identical cells, all of which attack the cell you don't want.  By carefully chooing the starting cell, you can "target" the monoclonal antibody to attack a very specific type of immune cell.  Because of how they are produced, early monoclonal antibodies were targeted at mouse cells, rather than human cells, and this sometimes caused problems.  Later techniques were developed to make the cells more human like, but still not completely human.  These were called "humanised" or "chimeric" monoclonal antibodies.  Finally, techniques were developed to create fully human monoclonal antibodies, and these are called "human" or "fully humanized".

There are scores of monoclonal antibodies (of all three types) approved for use in the United States, for a wide variety of illnesses.  Different monoclonal antibodies have different safety profiles.  Most have names ending in "mab".

[d3] The medical term for this is "modulate".  Teplizumab is said to modulate the immune system.

[d4] The researchers also point out that this trial did not use a placebo and the previous trial did.

[d5] The treatment given here, was given just after diagnosis, and kept insulin production at or above current levels for 24 months or so.   So if we could predict that someone would come down with type-1 diabetes a few months before they actually did, then we could give them Teplizumab, and they would (theoretically, if all worked well) preserve their insulin production high enough to not be diagnosed, for 24 months.  Furthermore, if we could extend this treatment for longer periods of time, we could delay diagnosis for longer periods of time.  There is an ongoing clinical trial called "The Natural History of Type-1 Diabetes" which is specifically aimed at gathering data from one who gets type-1, when, and how to predict it.

[d6] Some patients in this study were given only one dose, while others got two, but it was not random.  The patients who had better initial results got the second dose.  This makes it hard to separate out the effect of one dose versus two doses, from the effect of "responder" versus "non-responder".

Joshua Levy  -- http://cureresearch4type1diabetes.blogspot.com
publicjoshualevy at gmail dot com
All the views expressed here are those of Joshua Levy, and nothing here is official JDRF or JDCA news, views, policies or opinions. My daughter has type-1 diabetes and participates in clinical trials, which might be discussed here. My blog contains a more complete non-conflict of interest statement. Thanks to everyone who helps with the blog. 

No comments: