Thursday, December 8, 2011

Artficial Pancreas Updates

Here are a couple of recent updates on Artificial Pancreas testing and development. Remember that not everyone considers an AP a cure, but here is an update no matter if you consider it a better treatment or a cure.  Also there are a couple of "odds and ends" on other topics at the end of the posting:

Medtronic Starts Testing Veo technology in the US

Veo is a "stage 1" artificial pancreas, meaning it is an integrated pump, CGM device which has one small piece of intelligence in it: it will automatically turn off basil insulin if the BG numbers have been too low for too long.  It is the very first baby step to a commercial artificial pancreas.  They've been selling it commercially in Europe for years, and just got permission to start testing it in the US.

I could not find the clinical trial record for this study, so I'm basing my information off the press release.
    First, the study will be "pivotal" which usually means phase-III.
    Second, it will be an at home study, which is another sign that they are close to FDA approval.
But I do not know how large this test will be, how long it will run, or when results will be expected. Although device tests are often much quicker than drug tests, so these tests might only run for a few months, but they will still need to recruit a lot of people, which takes time in itself.

Press release:
News coverage: 

Artificial Pancreas Operating on Cellphone Tests Successfully
Type-1?  These guys got an app for that!  :-)

This is a university research group at the University of Virgina, which is running an aggressive artificial pancreas project.  They have completed four clinical trials, and  are recruiting for six more:

This specific trial involves 15 people (5 each Virgina USA, Padua Italy, Montpellier France), who will spend two nights in a hotel and the day between at the hospital.  This is a pilot study to see if it feasible to run a larger study.  I would consider this a phase-I trial.  The software uses standard CGM and pumps, but the thinking part of the artificial pancreas runs on a Android phone.

I like this approach for a number of reasons.  First, I think it will make it easy to make incremental improvements to the AP software.  It is much easier to download a new app, than to get a new pump.  Second, the easier it is to develop AP software, the more people will do it, and the faster development will move forward.  Third, my gut feeling is that anything that runs on a computer now (such as the Sansum software being tested in the next two trials) will be able to run on a smart phone in a few years.   Fourth, pumps tend to have crummy screens, buttons, and user interfaces in general, because the companies focus on the "functional" parts, such as the pump.  However, smart phones have great screens, buttons and user interfaces because those are very important for their success.  So any AP software running on a smart phone will get a better user interface "for free", as compared to anything running on a pump like device.  Fifth, smart phones are naturally networked which I think can lead to improved quality of care.   I look forward to a time when your smart phone will power your AP, and maybe once a week it will upload a week's data to a central computer ("in the cloud") which will run lots of data analysis on it, and then download some improvements to your AP.

Note: In real life (when not writing this blog) I'm a software engineer (actually a "technical lead") and the software I'm working on right now is an app for a smart phone, so I do know something about app development.  The software I develop is not part of the medical industry.

They expect this study to be completed by September 2012.  (Remember: device studies are often quicker than drug trials.)

Clinical trial records (one per site, I don't know why):

News coverage:

Phillip Artificial Pancreas Trial

This artificial pancreas is called MD-logic and this it it's second test on people (that I know of).  This trial is 18 teenagers, and is being done in a camp like setting in Isreal, for at least 24 hours.   The MD-Logic device comes in two types, but the one being tested here is the "SC" system which tests sugar levels just under the skin, and doses insulin just under the skin.  So it is like a current CGM system and a current pump system, connected  via a laptop computer.  (For this trial the laptop is being carried around by the patient, all the time.)  This "SC" system has been previously tested on 4 adults, in 8 hour sessions in a hospital.  The results of the previous trial was BG levels between 92 and 150, which in my opinion is very good.

MD-Logic also comes in an "IV" system, which measures sugar levels directly in the blood stream, and also doses insulin directly in the blood.  In theory this should lead to more accurate BG readings and faster insulin effectiveness.  This version has only been tested on pigs, for 1 hour at a time.  However, during those experiments it kept BG levels between 80 and 130, which I believe is the same levels as found naturally in a non-type-1 diabetic.

I think this is the clinical trial record for this study (if not it is for a closely related trial):

Note that some articles have claimed that this is the first trial outside of a hospital, but I don't think that is correct.  I think there have been two or more previous trials outside of hospitals, including the one right above.

Abstract of earlier research:
News coverage:

I'm not 100% sure, but I think that these guys are using the same Sansum Diabetes Research Institute software as the Beck group below.  I believe this team is also participating in the Beck trial below.

Beck In-patient Evaluation of an Artificial Pancreas

This is a 50 person study, which should be completed by March 2012, so quite soon.  (It started back in March 2011, and I'm sorry that I did not blog on it back then.)  It uses a DexCom CGM, an Omnipod pump and a laptop.

For those in the bay area: Drs. Buckingham and Wilson at Stanford are involved.  There are also sites in Virgina, Denver, Padova Italy, Montpellier, France, and Israel.

I believe that the "Phillip" trial listed above, is using the same basic software, but in a more aggressive setting.  The "Phillip" trial is camp like, while "Beck" is in a hospital.  On the other hand, "Beck" is larger and multi-site, while "Phillip" is smaller and only one site.  "Phillip" is on children, "Beck" on adults.  Finally, "Beck" is over half way done, while "Phillip" is just starting.

The FDA's New Guidance for Artificial Pancreas Testing

Last Thursday the FDA issued new guidance for artificial pancreas testing, and several readers have asked my thoughts on it.  As you read my opinions, remember that I'm not an expert in understanding FDA technical documents, and it is a very specialized field with much specific knowledge needed to do a good job.   

My general opinion is that FDA guidelines don't matter.  What matters is how they are interpreted in actual use.  So reading a guideline is nice, but the important thing is what happens when the FDA actually uses that guideline to approve or delay a medical device.  So my basic reaction to this news (that the guidelines have been released) is to shrug and wait for them to be used, and see what happens then.  The guidelines are a necessary step forward, but can't be evaluated on their own merits.  Not issuing them delays the process, and now that delay has ended, and that's a very good thing, but it says nothing about the quality of the guidelines.

I did read parts of the guidelines, and skim other parts.  It's tough going, but I have the following comments based on my understanding of the guidelines.  These are all improvements over the current rules, and point 2 especially would be a huge improvement:
1. There need to be three phases of testing (much like new drugs), and the first is usually in a hospital, the second usually in a camp or similar controlled environment, and the third in the real world.
2. There seem to be two alternate paths to approval, one being testing that the device is better than current methods ("Superiority"), and the other is that the device is not worse than current methods ("Non-Inferiority").  Either path would lead to device approval, but with different marketing claims being allowed.  Proving superiority would allow marketing literature saying that the device was better, and so on.  If A1c is used as the primary end point for the phase-III study, then showing a 0.4 improvement would be proof of superiority. 
3. Computer simulations (referred to a "in silico" testing), may be used to replace some animal testing, but is not a replacement for human testing.
4. A1c data or BG data from a CGM may be used as primary end point data, although the FDA recommends A1c data. 
5. There is a lot (my opinion) of flexibility in the secondary data that an applicant may choose to collect in their study.
6. In some cases, trials from other countries can be considered in approving devices.

News coverage:

Unrelated News Items, Which I Found Interesting

Measuring Pre-Type-1 Diabetes

Quote from the press release:
[Lead researcher Kevan C. Herold and team] at Yale University have developed a method to detect and measure the destruction of beta cells that occurs in the pancreas by measuring DNA expression in the blood. The destruction of beta cells leads, over time, to type 1 diabetes. 
If this research pans out, it is likely to have two large and quick effects on research aimed at curing type-1 diabetes.  First, it will make it easier to test treatments aimed at preserving or regrowing beta cells.  Right now, it is hard to tell if these work, because we generally measure them indirectly (via C-peptide production).  This might allow us to measure it directly, and see if the treatments are working a little, a lot, or not at all.  Second, it might make it easier to prevent type-1 diabetes, by agressively treating type-1s right when the beta cell destruction starts.   Right now, we know when it is about 80% complete (that's when type-1 is diagnosed), and we can see when antibodies start to be generated (but that might be too early).  This gives us another way to intervene early for purposes of prevention, if not cure.

Press release:

Overview Article on Status of a Cure

The following article is worth a read.  It discusses the recent failures of CD3 based clinical trials, and the general state of research into a cure:

More Evidence that "Dead in Bed" is Slow

In my previous post on "Dead in Bed", one of the points I made, very briefly, was that type-1 diabetics who died in their sleep did not "spike low".  They did not have really low BG for a really short period of time before dying.  Quite the opposite, in the one case history presented there [r13], the person was low for many hours before dying.  The study below is a similar one, but it covers four people who had seizures.  These people did not die.

The take home point, is that three of them had low BG levels for four hours before their seizure, and the forth for over two hours.  That suggests to me that a low BG cut off feature would have plenty of time to work and prevent seizures (and eventual death).  

Full paper:

Joshua Levy
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.

1 comment:

genetech said...
This comment has been removed by a blog administrator.