Medtronic (MDT) made a big leap toward the future of healthcare in diabetes treatment innovations when it launched what media have hailed as the first artificial pancreas, the MiniMed 530G, earlier this fall. But it’s actually just an “artificial pancreas device system.” It’s not a true artificial pancreas. The MiniMed 530G is only Step 1, DexCom CEO Terry Gregg said. It will still be about five years or more before such a device can truly function as an artificial pancreas in an optimal way for Type 1 diabetes patients, he said.
Gregg was the President and COO of MiniMed when Medtronic acquired the medical devices company for $3.4 billion in 2001. He’s been a Juvenile Diabetes Research Foundation Angel Award recipient, and serves on the Scripps Translational Science Institute advisory board. He’s a definite thought leader in diabetes treatment innovation.
So what will it take to move from where we are today to a true artificial pancreas? The first step was creating an insulin infusion pump that could suspend insulin delivery when a certain glucose threshold had been passed, which Medtronic achieved. It shouldn’t be discounted. It’s significant.
“This first step is important. It gives everybody that’s worked so hard at least a glimpse of the receptiveness of the Food and Drug Administration,” he said. But we still have further to push.
For instance, one of the major barriers, Gregg said, is that today’s insulins aren’t yet good enough to operate as an artificial pancreas.
Here are the next four steps required for creating a true artificial pancreas, according to Gregg.
Step 2: Predictive capability.
“Suspend insulin delivery by the insulin infusion pump based on a predictive glucose value generated by a continuous glucose sensor.”
Today’s product is reactive. “You’ve created a failure mode recovery, which is not an artificial pancreas,” Gregg said. The endpoint for an artificial pancreas should be able to anticipate highs (and lows–more on that later) of glucose levels, rather than just respond to an emergency. This is the next crucial step.
Step 3: Overnight control.
“Establish a range of glucose for the patient overnight and the system delivers or suspends the insulin based on that range.” Again, this acts more like a real pancreas and also eases the worry and sleeplessness of patients and caregivers.
Step 4: Same as step 3 but during the daytime.
Establishing that range and delivery is more challenging during the day because the patient is active, mobile and eating, Gregg said. “This is still an ‘open loop’ loop system, since the patient is very involved,” he said.
Step 5: Make the device bi-hormonal. Add glucagon.
“You need a drug that will actually cause the glucose to come up,” Gregg said.
And that will be tricky. Because insulin and glucagon can’t be mixed, it would require two pumps and infusion sets or a dual chamber pump with two infusion sets plus a continuous glucose sensor (or maybe even more than one) plus a new algorithm, he said.
The closer the device can move toward automatic restart (while, of course, maintaining accuracy), the better.
Seems like Step 5 is more easily done than step 2.
While a second infusion site would be bothersome, the technology is well-developed.
As for Step 2:
Blood glucose tests are 15 minutes behind consumption of glucose, and current CGM's are 30 minutes behind (they sense interstitial glucose).
Fast-acting insulin takes 15 minutes to start acting and 45 minutes to peak.
So to be more accurate, we need faster sensing and action, not prediction.
That is, what stimulates insulin production?
Rather than sense increases in blood sugar (at which point it's too late) we should look for markers or signaling hormones that correlate to the stimulation of insulin secretion.