Thursday, November 11, 2010

My Fascination With The Artificial Pancreas: Episode 3.

Yesterday, the FDA and NIH held a public meeting on the Artificial Pancreas (AP) technologies; specifically, to discuss the next step - which is to take clinical studies from in-hospital to out-patient.  The day-long meeting was shared through an online webcast, and I listened to nearly all of it while at work yesterday, hoping to absorb as much as I could.

I felt like I learned so much - as I did the last time I tuned into a webcast on AP technology - but I'm not sure how to properly convey it all here.  I tried to explain everything I learned to A when I got home; and, as he is prone to do, kindly listened to my Excited Fast Talking for as long as needed. 

My mind stove.

I'll pre-emptively apologize for the likely scatterbrained organization of this post.  I honestly have more information than I know how to handle, so I'm just doing a short post of the main points that stuck with me.  I need to let the rest simmer a bit on the back burner of my mind stove.  Also, I'll be exploring more of my AP concerns in my post appearing tomorrow... somewhere else!

There are just so, so many intricacies to getting an AP in the hands of those who need it here in the US - just in the technology development itself, not even counting the FDA approval process.  I knew it was a complicated technology unlike any that's been developed before (the technology required for the AP was often compared to flight engineering - but even then, that doesn't cover it all), but I underestimated how many hundreds of small decisions have to go into getting it "finished".  And while I still find the wait irritating, I'm starting to understand the need for it a bit more.

For example, it sounds like the first version of the AP that might get approved would be a system with only the ability to shut off insulin delivery, for probably a two-hour period, to mitigate hypoglycemic events.  It sounds a lot like what we insulin pump users already do with temp basals, right?  Except, it's not that easy.  At what point will this feature kick in - and do you really trust the CGM's accuracy in acknowledging these events?  Do you wait for the reading to cross a certain BG threshold for this feature to engage (which is a reactive approach), or do you find a way for it to respond to a specified downward trend (which is proactive)?

This version of the AP is missing a key component though - it doesn't do anything to correct high BG levels.  And when you suddenly give the AP the ability to increase insulin dosing, instead of merely being able to reduce it, you're entering a whole new world of risk.

One issue that got a good deal of air time was the matter of who the AP trial clinicians should be looking for to participate in this round of study.  Who could "benefit most" from the technology?  (I'd argue "all of us", but I guess that's why I wasn't on the panel.)  You could argue for those with hypoglycemic unawareness.  Or for those who experience a significant number of "severe hypoglycemia excursions" each year (Their words, not mine - although I kind of love this phrase.  It makes me think of giraffes, and safari hats.)  Many suggested a group who was already "well controlled", with an A1C under a certain number, and who was already familiar with CGM and insulin pump technology.

Which leads me to my last point - one of the panelists expressed concern over the amount of education that would need to be provided for these out-patient clinical trial participants.  They were mostly concerned that the problem would lie with the amount of knowledge and skill in using the technology away from a hospital environment.

I think they're missing a big point.

The use of an artificial pancreas as a treatment option presents a huge paradigm shift for people with diabetes.  We're used to the ones being in charge of our own care; of making educated guesses based on our personal experiences with our disease.  It's hard to imagine a place in time where someone who is used to that amount of mental involvement could suddenly put complete trust in a mostly-autonomous diabetes management system.  I think the acceptance rate for the AP is going to be a big obstacle to overcome.  Which means that a question the researchers should be looking to answer is this:  How do we get people to trust this thing?

**The title of "Episode 3" might be a little bit misleading.  I wrote a post about the AP back in August (hence, episode 1), and I've written another (episode 2) which hasn't been published yet.  It will show up Friday.  :)


  1. It took me an embarrassingly long time to figure out what you guys were twitterin' about yesterday with the APP hashtag. Looking forward to reading your next posts on the subject!

  2. Because of the amount of testing needed, I'd probably look for educated and active participants for the first round of testing: people who already know what a pump and a CGM should do, and how to access and interpret the data they get from both. They will want to be able to download the insulin dosage information and compare the CG information with their own CGMs and blood-prick meters. They will understand that these prototype devices will be larger than the eventual market devices (or will come with separate, wireless read-out modules), and they will be willing to put forth the effort needed to say "this works, that doesn't, and the average joe may need these additional types of training".