Bringing an Artificial Pancreas to Market
By Adam Brown
by adam brown
As noted in the diaTribe #38 letter from the editor, the FDA released its much awaited draft guidance on the artificial pancreas (AP) by the promised December 1 deadline. This document provides the FDA’s recommendations and suggestions for carrying out artificial pancreas studies, which will culminate in an FDA review and hopefully approval of the system for use by patients. In the months leading up to the guidance release, JDRF put substantial pressure on the FDA (see New Now Next in diaTribe #38 for more details) – we were happy to see that the agency incorporated many of JDRF’s suggestions and hope that the FDA’s emphasis on a reasonable, flexible development and study process ultimately translates into faster regulatory approval of an AP.
The FDA has proposed a three-stage study progression to bring the AP to market: early feasibility studies, transitional studies, and pivotal studies. According to the guidance document, feasibility and transitional studies will require small numbers of participants, be of short duration, and take place in clinical research settings with careful medical supervision. The final pivotal study to approve the device will be at least six months in length and take place in a real-world setting with participants going about their daily lives. Whenever the FDA approves a medical device, it needs evidence to ensure that it is both safe and effective. Studies to approve the AP will thus be designed and evaluated with this perspective in mind.
The pivotal study to approve an AP will compare two groups: the “control group” will use a standard insulin pump and CGM as they are used
today, i.e., patients will manually bolus and set basal rates. They will be compared to a group wearing an artificial pancreas system that automatically changes insulin doses based on the CGM value. The first generations of the AP will be “control-to-range,” meaning the system will aim to keep patients in a range of glucose values (e.g., 70-180 mg/dl) rather than targeting a single number. These are also called “hybrid closed-loop systems” because patients will still need to bolus and set basal rates, but the AP will automatically increase or decrease basal insulin if the blood sugar is trending too high or too low. In the future, more advanced systems will require less patient intervention and will target a set glucose value instead of a range (e.g., 100 mg/dl); however, better CGM, even faster-acting insulin, and even glucagon (a hormone that quickly increases blood glucose) will likely be needed.
.JPG)
To gain FDA approval, studies must show that those wearing the artificial pancreas do not increase their A1c compared to those wearing a
traditional pump and CGM (this is known as ‘non-inferiority’). But the FDA could allow increases in A1c as long as AP-wearing patients have other improvements (e.g., a reduction in hypoglycemia). In addition to non-inferiority, superiority studies can demonstrate that the artificial pancreas is better than wearing just a standard insulin pump and CGM. For this sort of claim, the artificial pancreas must improve A1c by at least 0.4% and/or improve measures of hypo- or hyperglycemia (as measured by CGM) by at least 30%.
This draft guidance is an important step forward in artificial pancreas development. We are most concerned about study recruitment and cost given the safety requirements outlined in the guidance - an AP must not increase the rates of severe hypoglycemia, hyperglycemia, or diabetic ketoacidosis. While this is certainly a reasonable safety bar, the low rates of these events mean that long and large studies will be needed. Alternatively, studies must enroll individuals with recurrent severe hypo- or hyperglycemia (a fairly small population). The Agency also prefers enrollment of experienced CGM/pump users (>3-6 months), or at minimum, pumpers who undergo a four- to six-week CGM training period. With all of this mind, we anticipate that recruitment for these studies will be somewhat challenging, time-consuming, and relatively expensive. Nevertheless, at this point, we believe it’s less a question of “how” and more a question of “when” such a system will come to market.
The current version of the guidance document is a “draft” because it is open for public comment from patients, researchers, and industry representatives. If you’re interested in reading the fairly technical document, please click here. Public comments will be accepted until March 5, 2012, and can be submitted here. We hope that as the FDA finalizes the document and begins evaluating artificial pancreas trials, it keeps in mind the daily risks of living with type 1 diabetes and the potential for even first-generation AP systems to lower these risks. Despite the best technology today, insulin is still a hard medication to dose, and unfortunately mistakes are made all the time. We welcome anything that could reduce those errors and improve A1c.
For those unfamiliar with an artificial pancreas, please see Chapter 5 of our free book, Targeting a Cure for Type 1 Diabetes: How Long Will We Have to Wait?, available at www.diatribe.org/cure. For the most recent updates on the artificial pancreas, see our coverage of the recent Diabetes Technology Meeting in Conference Pearls in diaTribe #38. Finally, for those interested in artificial pancreas studies, we suggest visiting JDRF’s Artificial Pancreas Project Consortium and JDRF Clinical Trials Connection. –AB
