And glycemic control "went from good control to excellent control," with the average hemoglobin A1c declining from 7.4% at baseline to 6.9% at study end, they reported here at the European Association for the Study of Diabetes meeting and simultaneously online in a research letter in the Journal of the American Medical Association.
More than 75% of nighttime levels were in the target range and over 70% during the day, which Bergenstal said represents some work to be done, but "in the real world with 124 patients using this system, this is the data we achieved."
"For over 30 years, I've been promising my patients we're going to have an artificial pancreas someday," he told MedPage Today. Now these data from "the largest, longest study using a hybrid closed loop system in type 1 diabetes" make it finally "potentially ready to be approved."
The multicenter study used the Medtronic Hybrid Closed Loop system, which automates basal insulin delivery, with patients entering mealtime carbohydrate intake and exercise and periodically calibrating the sensor. Medtronic has applied for clearance of the device by the FDA.
The single-arm study included patients ages 14 to 75 who had type 1 diabetes for at least 2 years and more than 6 months' experience using an insulin pump. Bergenstal noted that patients with HbA1c levels of 10% or greater were excluded as were patients with hypoglycemia unawareness -- two groups that might particularly benefit from automated treatment and should be studied.
Patients had a 2-week run-in period to learn the devices without the automated features turned on and then the device collected each patient's insulin and sensor glucose data for the hybrid closed-loop algorithm for the first 6 days of the study.
Patients uploaded their device data through the CareLink system, which was reviewed each week. The study also included a day of frequent venous blood glucose sampling once in the 3-month study period to verify the accuracy of the system.
While the system features a decision-support mode, "most patients run with the system and let it decide," Bergenstal said. The closed loop was active 87% of the time, he noted, saying that was important as patients can run into trouble when they flip the system on and off.
He cautioned that the before-and-after design without randomized comparators was a limitation.
The study was funded by Medtronic.
Several co-authors were employees of Medtronic.