Saturday, January 17, 2015

Artificial Pancreas Update (Jan 2015)

Sound track for this blog is "Living After Midnight" by Judas Priest, but performed by The Donnas:!/s/Living+After+Midnight+Judas+Priest/3ASqUY

Research into the Artificial Pancreas (AP) continues to move forward.  The term "artificial pancreas" refers to using a continuous glucose monitor (CGM) to feed data to a computer, which controls an insulin pump, and in some models, a glucagon pump as well.  Artificial pancreas refers to using existing technology in all these areas, but connecting them together so that a person does not need to worry about counting carbs or blood glucose levels.  It is all done automatically.

Metronic Starts Selling a "Step 2" Artificial Pancreas in Australia

The JDRF uses a 6 step model to get to the fully featured artificial pancreas that we wall want.  You can read about that model here:

Right now, Medtronic makes the only commercially available "step 1" artificial pancreas.  However, just recently they started publicising a "step 2" artificial pancreas in Australia.  The difference is that "step 1" APs cut off insulin if you are already too low, while "step 2" cuts you off before you get too low.    This is an important improvement in functionality, but it is also an important regulatory hurdle. The model number is 640G.  I can't find any press releases, or any notice on their web site, but karri on CWD posted this link to a promotional video:
There are also several "Ambassador Reviews" on the Medtronic Diabetes ANZ youtube site.
It looks like they are taking orders for the device now.  However, I can't see anything on the web site related to the 640G.  Maybe they know I'm from the USA, and can't get it?

Of course, the next issue for us Americans is FDA approval.  Medtronic's "step 1" device took 31 months after European approval, before the FDA approved it!  It will be interesting to see if the FDA repeats that fiasco or not.  There has been a lot of progress, several meetings between the FDA and patient advocates (such as DOC, diaTribe, JDRF, etc.) and now we will see if any of that matters. But the clock is ticking now, and we will know a lot more when we see how long the FDA delay of approval is.

A Direct Single Hormone vs. Dual Hormone Comparison

In the past, I've tried to compare Single Hormone AP results to Dual Hormone AP results, by comparing similar results from different studies.  However, it's much better to compare the same result in the same study, rathern than combining data from different studies.  A group of researchers at Institut de Recherches Cliniques de Montreal, Montreal, QC, Canada (and elsewhere) recently did exactly this comparison.  They directly compared a regular pump plus CGM, to an insulin only AP, to a insulin and glucagon AP.  Patients (12 years or older) were treated for 3 24 hour periods.  The trial was not blinded. Funding was from the Canadian Diabetes Association, JDRF, and Medtronic (see above for involvement). They reported the following data:

Measure Pump + CGM Insulin AP Dual Hormone AP
Time spent in target range
Hypoglycaemic events
Symptomatic hypoglycaemic events
Nocturnal hypoglycaemic events


I think the clear result of this trial is that the dual hormone AP is very slightly better than the Insulin AP, and they are both noticeably better than the current standard pump and CGM.  This result is similar to previous studies.

Clinical Trial Registry:

Many Articles in the January Journal of Diabetes Science and Technology

The Journal of Diabetes Science and Technology had a special issue on glucagon therapy.  There were a couple of articles on stable glucagon, which is required for a bihormonal artificial pancreas (such as Dr. Damiano is working on):

From this abstract
Data are presented that demonstrate long-term physical and chemical stability (~2 years) at 5°C, short-term stability (up to 1 month) under accelerated 37°C testing conditions, pump compatibility for up to 9 days, and adequate glucose responses in dogs and diabetic swine. These stable glucagon formulations show utility and promise for further development in artificial pancreas systems.
From this abstract
Data are presented that demonstrate physical and chemical stability under presumed storage conditions ([over] 2 years at room temperature) as well as “in use” stability and compatibility in an Insulet’s OmniPod® infusion pump. Also presented are results of a skin irritation study in a rabbit model and pharmacokinetics/pharmacodynamics data following pump administration of glucagon in a diabetic swine model.
Here is a diagram of their results.  Note that they are comparing their glucagon to standard glucagon, so their trial is successful if the same colored lines are close to each other, and they are:

Although both of these are animal trials, for this kind of test, I think results in animals are likely to mirror results in people.  They are really measuring the stability of the glucagon, and the test subject species doesn't have a big impact on that.  So my normal worries about "works in mice, fails in people" are pretty small for this kind of research.

Adding Physical Activity Measurements to Aritifical Pancreas Calculations

Also in The Journal of Diabetes Science and Technology, there is this paper:
(People in the San Francisco bay area will notice that the authors are mostly local: Drs. Stenerson, two Paynes, Ly, Wilson, and Buckingham.)

This paper attempted to use data on physical activity to improve an artificial pancreas's BG numbers. The idea is simple enough: we all know that physical activity lowers BG numbers.  If an artificial pancreas knew how active you were, could it do a better job of controlling BG numbers?   The kids in the study (average age 13), did a soccer workout ("football workout", in the rest of the world) on two separate occasions.  In one case, data from an accelerometer was used to help calculate real time insulin dosing, and in the other case, this data was not used.  BG and hypoglycemic events were measured both during the soccer, but also after it, until the next morning.  There were 18 kids in the study.

Overall BG numbers were similar in each group.  Hypoglycemia events were higher when not using the accelerometer, but the difference between groups was not statistically significant. (This is the scientific way of saying "close, but not quite".)


From my point of view, there are a couple of ways to interpret this result:
  1. One could say that the study was just too early and too small to interpret the results, and it's really more of a proof of concept of how accelerometers could be tested in the future.  The most that can be said is that more research should be done. 
  2. Or, one could say that accelerometer data is not needed for an artificial pancreas, because it had no statistically significant effect here.  And this is a good thing, because it means we will not have to burden type-1 diabetics with accelerometers (in addition to pump(s) and a CGM), because the extra information they provide is not needed.
  3. Or, one could say that existing algorithms and accelerometers are not good enough, and we need to develop better ones in order to take advantages of this information.
  4. Or, one could sort of split the difference, and say that most people don't need accelerometer data (as also supported by the good BG numbers reported by other AP tests which don't use them). However, for serious athletes who do need this extra data, we need to develop better algorithms (or accelerometers), if we are going to successfully use this data. 
  5. Or, we can view this research as answering the question "how much physical exertion is needed before an AP needs exertion data to work well?"  The idea is that, of course at some level of vigorous exercise, accelerometer data will be needed.  This trial just shows that the soccer exercises wasn't enough, and we need to do something even more vigorous until it does matter.
No matter which interpretation appeals to you, we are still very early in the testing of accelerator enhanced APs.  I'm sure there will be more clinical trials before there is any consensus on the proper way to integrate accelerator data into APs.  

Interview with Dexcom

Here is a two part interview with the CEO of Dexcom, a big CGM maker.  He talks about future development, both in terms of CGMs for APs and CGMs as a replacement for finger sticks.  I found part 2 more interesting than part 1, but here are links to both:

Another Summary

Here is another writer's summary of artificial pancreas progress for 2014.

Homebrew Artificial Pancreas

As Dave Berry used to say: I'm not making this up!

This project is being called "Hacking an Artificial Pancreas" or "DIY Artificial Pancreas", but in the tradition of Silicon Valley, I would called it "Homebrew Artificial Pancreas".  It looks like we have reached the point where people can cobble together a functional AP in their garage.

You can see pictures here (two quite different paths):

If you speak twitter: #DIYPS #WeAreNotWaiting #dexdrip

Less than $100 in parts. A little soldering (sometimes).  What could possibly go wrong?
When Steve Jobs and Steve Wozniak made the Apple I they were basically making something that they could not buy, and that's what these guys are doing.  Of course the Steves were not replacing  an organ of the body.

I want to thank Dominik for sending me this. I will never look at a Raspberry Pi the same way, again.

Joshua Levy 
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.


Malcolm Mackay said...

I have also tried to look at Medtronic Diabetes Australia and get redirected to I think they'd rather not have people outside Australia knowing there is a new pump on the way.

Descoindia said...

I read your post Which was really good