The latest FDA guidance is out on the Artificial Pancreas, and I’ve been taking a look to get a better idea of what it’s going to take to bring something like this to the market here in the USA.
I haven’t had a look at previous statements from the Food and Drug Administration on this issue, but after looking at the 63 page document recently released, I can tell you that it’s going to take a lot.
There are a number of reasons for that assessment, so let me see if I can break it down for you a bit. This isn’t a complete, detailed look, and I’m still seeking some other input on what some of the finer points really mean. For now, I’ll just try to give you some of the highlights.
When I think Artificial Pancreas, I think (like many others do) that we just need a device that will use data from our CGM to tell our insulin pumps what to do to help manage our glucose. Simple, right? Now, let’s think of everything having anything to do with that.
From the FDA’s point of view, that means anything and everything to do with your BG meter, your CGM, your insulin pump, and then your Artificial Pancreas device, which would probably be a modified phone of some sort.
If your meter, CGM, and pump have all been tested and approved before, great. Just provide all of the documentation that supports that, including the models, the way they each interact with each other, their adaptability to heat, cold, being dropped on the floor, your line getting kinked, getting wet, and on and on, ad infinitum.
Ready to move on to the AP itself? Okay, let’s talk about that. You’ll still need to detail how each piece of your AP system is going to work in a real-world setting. Detail how your device will interact with the CGM and pump, including how you plan to calibrate the CGM with each change, and how you plan to communicate that to the AP itself. Detail your algorithms, who they’re designed for.
Will the device be unidirectional (information moving in one direction only)? Or multi-directional (information moving between each piece of the system)?
A big question: How will the device combat hypoglycemia and hyperglycemia? And in that vein, will the device help in the proper delivery of just insulin? Or possibly insulin and glucagon? Or, as written here before, the delivery of insulin and amylin?
What about security? How can you ensure that the device won’t be hacked in some way? How can you be sure that the radio frequencies being used for the system won’t impede other systems? And while you’re at it, please detail how your system won’t be impeded by other systems like magnetic resonance imaging (MRIs), or even cell phones, computers, iPads, etc.
Once you’ve documented all of that, you can get down to the business of detailing the human factor. How will humans interact with your system? Will someone be allowed to override parts of your system? What about incorrect entry of key elements? What happens when someone leaves their CGM in “beyond the sensor wear period, when CGM results may be compromised” (actual language in the document)?
Now let’s talk about Clinical Study progression… including Early Stage testing, which includes proof-of-concept testing and studies regarding the effects of modifications to algorithms. Then later-stage feasibility studies, including some basic outpatient testing. After that, there’s “Studies to Demonstrate Performance of Risk Mitigation Strategies” (I think we all know what that means).
Finally, the Pivotal Clinical Study, designed to “gather data to support the safety and effectiveness of the device”. Real-world testing including individuals that the device is intended for. The patient population information detailed in the guidance on page 31, 32, and 33 is very important and worth the read.
Also worth the read are the very detailed Study Endpoints noted from page 33 through 35 that should include “objective characteristics or variables that reflect how a patient feels, functions, or survives. Surrogate endpoints should predict meaningful clinical outcomes and be based on valid scientific evidence”. This includes information on the use of CGM data, how the device handles hypoglycemia and hyperglycemia, how the device helps A1c results, how much the device helps the patient keep BG within range and what that range is, and how much time is spent within the range. Add in detail about safety mechanisms like low-glucose suspend and how high glucose and elevated ketones are dealt with. A final statistical analysis is expected.
But wait… you’re not finished yet. You have to include all of the documentation you intend to give the end user in:
- A User manual, written at an 8th grade level
- Training materials
- Professional documentation for prescribing physicians
- Package inserts for the AP device and any components that are packaged separately from the device itself
- Box and container labels for the AP device and any components that are packaged separately from the device itself
Don’t forget to describe how your device will be manufactured, including anything having to do with sterile parts of your system, if applicable.
And finally, you’ll want to include detail of the protocol of any Post-Approval studies you’ll want to conduct, assuming you’ll get approval in the first place.
That’s the very basic, 50,000-foot level view of what needs to be included in anyone’s application for an Investigational Device Exemption (IDE) or premarket approval (PMA) for their artificial pancreas device. That’s a lot, isn’t it? And to tell you the truth, I can’t find a whole lot in there that isn’t useful in defining how something will work, how it will be studied, and how it will keep us safe. But despite everything that’s necessary to bring this closer to the everyday patient, I’m still looking to the future with a hopeful eye. I’ll let you know if I find out more.