Medtronic gets closer to an artificial pancreas


The artificial pancreas – a device combining a glucose monitor and an insulin pump that can perfectly control a diabetes patient’s blood sugar with no input from patients themselves – is still many years from market. But development of such a system proceeds stepwise, and Medtronic has just taken a stride.

Results of the pivotal US trial of the company’s – and arguably anyone’s – best attempt yet at a closed-loop insulin delivery system set the device up for an FDA submission this month. These systems are expensive, however, particularly compared with the dirt-cheap low-tech finger stick and self-injection used by the vast majority of diabetes patients. Approval might come, but reimbursement is not so certain.

Getting there

Medtronic’s Hybrid Closed Loop system consists of its MiniMed 670G insulin pump, blood sugar sensors – believed to be Medtronic’s Enlite 3 continuous glucose monitor – and an algorithm to link the two, delivering insulin when blood glucose reaches a certain threshold. The device measured glucose every five minutes, but it was not entirely effortless: patients still had to calibrate the sensors and take mealtime boluses or other corrective insulin doses.

The Hybrid Closed Loop was tested in 124 patients with type 1 diabetes who used the system for three months. Data presented at the annual meeting of the American Diabetes Association in New Orleans showed that safety was good, with no cases of diabetic ketoacidosis or severe hypoglycaemia, and no serious device-related adverse events.

As for the system’s performance, there was a significant decrease in patients’ mean haemoglobin A1c, from 7.4% to 6.9%. The normal range for A1c is between 4% and 5.6%, but the goal for people with diabetes is haemoglobin A1c of less than 7%.

There were significant improvements from baseline in the percentage of patients who had glucose levels over 180mg/dl – hyperglycaemia – on average over the three-month treatment period. Counting only those cases that occurred overnight, when it is harder to manage glucose levels, there was still a significant improvement.

The same was true for the percentage of patients whose glucose levels were under 70mg/dl – hypoglycaemia – and under 50mg/dl. Concomitantly, the percentage of patients in the safer range of 71-180mg/dl increased significantly.

Trial results
  24 hours 10pm to 7am
Parameter (mean values) Baseline Measurement period p Baseline Measurement period p
Total glucose 150.2mg/dl 150.8mg/dl 0.64 148.8mg/dl 147.5mg/dl 0.38
Glucose >180 mg/dl 27.4% 24.5% <0.001 26.9% 21.6% <0.001
Glucose >70 and ≤180 mg/dl 66.7% 72.2% <0.001 66.8% 75.3% <0.001
Glucose ≤70 mg/dl 5.9% 3.3% <0.001 6.4% 3.1% <0.001
Glucose ≤50 mg/dl 1.0% 0.6% <0.001 1.1% 0.6% <0.001
Total daily insulin dose 47.5U 50.9U <0.001      

Medtronic will be heartened by the fact that at the end of the study 99 of the 124 patients opted to enter the continued-access programme to keep using the device. Insulin pumps and glucose sensors are often popular with patients as they lessen the need for the cheap but onerous and painful daily routine of finger puncture and hormone injections. It is no surprise that a system incorporating both technologies would be readily accepted.

Still some way to go

But there are problems. Type 1 patients are the core market for an artificial pancreas, but much of the appeal of such a device rests on its potential utility in children. This population was not reflected in the study: the average age of the patients enrolled was 38, only 30 were under 18 and none was younger than 14. Insurers might wish to see further proof of the system’s use in children before agreeing to foot the bill.

The second problem is that bill itself. To get payers on board Medtronic will have to demonstrate that the system is cheaper than the low-tech alternatives in the long run thanks to limiting hyper and hypoglycaemic crises.

And the third is that the Hybrid Closed Loop system is still nowhere near an artificial pancreas. Such a device would need to be much more reliable, needing less calibration and fewer readjustments at mealtimes. The ideal system would also be able to deliver glucagon as well as insulin to head off hypoglycaemia – current systems can do little more than sound an alarm so the patient can eat something.

A true artificial pancreas would be able to predict peaks and troughs in blood sugar, delivering the appropriate hormone precisely as it is needed. That hypothetical device would almost certainly be cost-effective even at a pretty high price.

Artificial intelligence meets artificial pancreas

Still, perfection being a long way off is no reason to stop trying. It will be interesting to see what labelling the Hybrid Closed Loop device gets if it earns FDA approval next year, and how this compares to Medtronic’s first approval for a similar system nearly three years ago (Medtronic claims artificial pancreas win but tech is limited, September 30, 2013).

And development, of course, continues. Medtronic’s collaboration with IBM’s Watson artificial intelligence and its deals with companies such as DreaMed and Glooko are specifically aimed at improving these technologies (IBM collaboration is Medtronic's fifth diabetes deal in a month, April 14, 2015). If any commercial group cracks a true artificial pancreas it will almost certainly be Medtronic. But it is not there yet.

Share This Article