Good News First
This was a big (450+ people) study, and phase-III, so late in the processes of commercialization. The treatment was DiaPep277, which is a peptide (a short protein) which is related to a naturally occurring protein called "heat shock protein 60" or hsp60. The treatment is one under-skin injection every 3 months. People were followed for 2 years. The study was blind, random assignment; half got the treatment, half the placebo.
DiaPep277 works by increasing the activity of regulatory T cells (especially Th2 cells), which serve to control the autoimmune attack on beta cells in the pancreas. Earlier research described the mechanism this way: hsp60 is a protein found in the pancreas and very similar to hsp65, which is found in microorganisms. Your autoimmune system gets mixed up between hsp65 (sign of foreign invasion) and hsp60 (naturally there), and attacks the beta cells in your pancreas. DiaPep277 is part of the larger hsp60 molecule and teaches your immune system not to attack. The idea is similar to giving people tiny amounts of peanut protein to wean them off of peanut allergies. [r3,r4]
The important pieces of news from this study are:
- The primary outcome for this was C-peptide production. Measuring C-peptide is the same as measuring the body's production of insulin[d1]. The people who got the treatment generated (on average) 0.949 nmol/L/20 minutes more, which is 23.4% more than the untreated group. That was statistically significant. But remember that type-1 diabetics generate very little insulin, so even a tiny bit more will cause a big percentage change.
- Untreated honeymoon diabetics loose their ability to generate insulin as the disease progresses. People treated with DiaPep277 also lost this ability, but it happened more slowly, so that at each point in time, the treated group generated more of their own insulin as compared to the untreated group. Some of the news coverage refers to "preserving insulin production" but it is important to remember that insulin production was not preserved at the same level as when treatment started. Instead, production was higher in treated people then in untreated people. It's a big difference.
- A secondary outcome was lower A1c numbers in the treated group. The company did not report average A1c numbers [d2], but they did say that 45.5% of the treated group was below 7, but only 35.7% of the untreated group was that low. This suggests that the extra insulin produced was useful, and was lowering A1c, and was having a positive effect on the body.
- The "initial safety data also indicate that DiaPep277 was well tolerated", which means no serious side effects, which is always a good thing, and consistent with earlier testing. This drug never had a hint of safety issues, that I know of.
- Shlomo Dagan, CEO of Andromeda (the developer of the drug) said "I estimate that that the drug will be on the shelf by the end of 2014" (but this was quoted in Hebrew, which I don't understand, so I'm relying on a translation into English). But I think he's wrong about that. To get approved, a drug needs two trials. The second phase-III trial is supposed to finish enrollment in 2012, and it's a 2 year treatment program, so that study finishes in 2014. Then there is a year or two for marketing approval, so I'd estimate approval in 2015 or 2016, assuming further analysis of this trial is successful, and the second phase-III trial is also successful.
Now the Bad News
- This is not a cure, in it's current form. This drug has only been tested on honeymoon diabetics and only been found to extend the honeymoon duration.
- I'm not sure how big the effect really is, in terms of how much it would improve the life of a type-1 diabetic [d3].
Comparison with Other Results
Dr. Faustman's BCG results showed spikes of .004 to .005 nmols of C-peptide [d4], so the DiaPep277 results are about 200 times bigger an effect than her results [d5], although her results were in established type-1 diabetics, not honeymooners.
Dr. Orban's Abatacept (Orencia) results showed improvements of 60% in C-peptide production, as compared to 23.4% seen in DiaPep277 [d5]. However, Abatacept was in a phase-II trial, so it was smaller, and farther away from approval for type-1 diabetes. But it is already approved for use for another disease.
Dr. Pescovitz's Rituximab results showed improvements of about 20% in C-peptide production as compare to 23.4% seen in DiaPep277 [d5]. However, Rituximab was in a phase-II trial, so it was smaller, and farther away from approval for type-1 diabetes. But it is already approved for use for another disease.
More Data; Future Results
There were a couple of data points that I wish were in the press release, but aren't. I'll certainly be looking for them in the published paper. First, a comparison of A1c numbers in treated vs. untreated. See [d2] for details of what I'd like to see. Second, a time based comparison: if an untreated type-1 reaches the end of their honeymoon N months after diagnosis (on average), how many months will it take someone with this treatment to get to the end of their honeymoon? Third, some details on safety. It's great to say "well tolerated", but I'd like to see the details.
In terms of future work: obviously, these guys need to complete their analysis of this trial, complete their second phase-III trial, and get marketing approval from the FDA and EMEA. That's just to get it widely available in the market.
In addition there are several expansion paths which I hope they will take now, but might need to wait until after the treatment get approval. (Once a drug is approved for any use, it is a lot easier to run a trial for other uses.) Interesting lines of research would include:
- Trying it on patients before they were diagnosed with type-1 diabetes. Many believe that anything that works in honeymooners will work better in people who have not yet been diagnosed at all. Even a few years ago, however, we could not find those "not yet diagnosed" people, so we could not design a trial around them. Now however, thanks to the "natural history" trial run by TrialNet [r2], we can find people who have not yet been diagnosed, and recruit them for pre-diagnosis trials. So running a trial on people more likely to be diagnosed in the future, but not yet diagnosed, is an obvious thing to do.
- Combining it with other drugs. Testing multiple drugs is becoming a common technique when there is no one drug that is completely effective. For this drug, there are two ways to attack the problem. One is try a combination of drugs all aimed at extending the honeymoon. See if all of them together can actually stop the beta cell destruction and permanently preserve the beta cell function which is present at diagnosis. Another track is to combine this drug with another drug which grows beta cells.
- Trying it on established type-1 diabetes. Majority consensus is that stopping type-1 diabetes when in starts (during the honeymoon) is easier than stopping it later (once established). However, I'm still in favor of trying a drug that works on honeymooners, on established type-1 diabetics as well, especially when the side effects are very small or nonexistent. Especially researchers who believe that beta cells continue to grow back slowly throughout a person's life, they might be particularly interested in a drug that lowers the autoimmune attack, even if given late in life. Also, to put it bluntly, at this point established type-1 diabetics have little to loose, and few good options to try.
Extra Discussion and References
[d1] Researchers can not measure insulin directly, because that would just measure how much insulin was being injected. Measuring C-peptide tells them how much insulin the body is producing itself. C-peptide is the official (by FDA) marker for approval of drugs to help type-1 diabetes, so it is the right thing to look at for trials like this.
[d2] My personal requirement, is that changing A1c numbers by 0.5 is interesting and worth looking at, but that changing them by 1 is important, and obviously good without any further discussion. My understanding is that most diabetes researchers are happy with 0.5, and consider even smaller changes interesting.
[d3] Primarily, I'm not sure how much improvement there would be in terms of fewer side effects, less dangerous lows, etc. I think the most important point here is duration of effect. If the effect is long lasting, then it is reasonable to assume fewer long term side effects, but I have not seen any duration information. Separately, there is also some debate about the benefit of having a longer, stronger honeymoon. Some people hold that this just makes it tougher to treat type-1 diabetes: easier to have hypoglycemic events and harder to control BG and A1c numbers. From a cure point of view, however, generating more of your own insulin is clearly the path to a cure, so I consider a longer, stronger honeymoon to be a good thing.
[d4] Different research report C-peptide results in different units. Dr. Faustman reports [r1] her results in terms of pmols / liter, and a pmol is 1/1000th of an nmol, so 5 pmols is the same as .005 nmols / liter. On the other hand, Dr. Pescovitz reports in pmol / milliliter (which happens to be the same as nmol / liter).
[d5] Because there are several different ways to measure C-peptide production, I don't think this is a direct "apples to apples" comparison. However, it is the closest to a direct comparison that I can make. C-peptides can be measured in response to a meal, or while fasting. The exact mechanism of measurement can vary as well.
[r3] How it works in NOD mice: http://www.ncbi.nlm.nih.gov/pubmed/9133541
[r4] How it works in BB rats: http://www.diabetes-mellitus.org/p277_idf.htm
Thanks to Abush at CWD for posting the first news of this, including the English translation.
All the views expressed here are those of Joshua Levy, and nothing here is official JDRF or JDCA news, views, policies or opinions. My blog contains a more complete non-conflict of interest statement.