Bionic pancreas better controls glucose levels than standard care in T1D

Insulin delivery via a bionic pancreas better controlled HbA_1c levels compared with standard of care (SoC) in patients with type 1 diabetes (T1D), according to a recent multicentre trial from the US.

The 326 participants were aged 6–79 years (mean age 28 years, 74 percent non-Hispanic White) with T1D and were using insulin for ≥1 year. They were randomized 2:1 to receive either a bionic pancreas (administering insulin aspart or insulin lispro) or SoC (any insulin-delivery method with unblinded, real-time continuous glucose monitoring [CGM]).

The bionic pancreas algorithms were initiated by entering the patient’s body weight with no run-in period prior to automated insulin delivery. Meal doses of insulin were delivered in response to meal announcements and glucose-correction doses based on entered blood-glucose values.

At 13 weeks, mean HbA_1c levels decreased from 7.9 to 7.3 percent in the bionic pancreas group and remained unchanged at 7.7 percent in the SoC group (mean adjusted difference, –0.5 percentage points, 95 percent confidence interval [CI], –0.6 to –0.3; p<0.001). [N Engl J Med 2022;387:1161-1172]

The outcomes did not differ by age (between-group difference, –0.5 percentage points in patients aged ≥18 and 6 to <18 years). The treatment effect was more evident in analysis of patients with baseline HbA_1c >7.0 percent (between-group difference, –0.7 percentage points).

At 13 weeks, CGM-assessed glucose levels were <54 mg/dL for a comparable amount of time in the bionic pancreas and SoC groups (adjusted difference, 0.0 percentage points, 95 percent CI, −0.1 to 0.04; p_{noninferiority}<0.001). There was a greater reduction in mean glucose levels at 13 weeks in the bionic pancreas vs SoC group (adjusted difference, −16; p<0.001).

Patients in the bionic pancreas group were in the glucose target range of 70–180 mg/dL for 2.6 hours/day longer than those in the SoC group (adjusted difference, 11 percentage points; p<0.001). Conversely, patients in the bionic pancreas group were in the glucose ranges of >180 and >250 mg/dL for a shorter time than those in the SoC group (adjusted differences, −10 and −5 percentage points, respectively; p<0.001 for both). There was no difference between groups in the amount of time patients had glucose levels <70 mg/dL (adjusted difference, −0.1; p=0.51).

More adverse events (AEs) occurred in the bionic pancreas than SoC group (244 vs 10 events). A total of 214 and two incidents of hyperglycaemia with or without ketosis were reported in the bionic pancreas and SoC groups, respectively. These were primarily due to infusion-set failure in the bionic pancreas group, though infusion-set failure was not reported as an AE in the SoC group.

Severe hypoglycaemia incidence did not significantly differ between patients in the bionic pancreas and SoC group (17.7 vs 10.8 events per 100 patient-years; p=0.39). There were no incidents of diabetic ketoacidosis.

The authors noted that the small number of patients with hypoglycaemia at baseline prevented any conclusion being made on the effect of the bionic pancreas in reducing CGM-detected hypoglycaemia risk.

“Many patients closely manage their T1D, and do extremely well,” said study author Professor Irl Hirsch from the University of Washington, Seattle, Washington, US. “But there’s another, very large group of patients who don’t pay as much attention, for a variety of reasons.” [https://newsroom.uw.edu/news/study-bionic-pancreas-improves-type-1-diabetes-control, accessed 7 Dec 2022]

“Our observation that this system can safely improve glucose control to the degree we found, and do so despite requiring much less input from users and their healthcare providers, has important implications for children and adults living with diabetes,” said study chair Associate Professor Steven Russell from the Harvard Medical School and Massachusetts General Hospital, Boston, Massachusetts, US. [https://www.nih.gov/news-events/news-releases/bionic-pancreas-improves-type-1-diabetes-management-compared-standard-insulin-delivery-methods, accessed 7 Dec 2022]

“This pump uses more artificial intelligence than the other pumps on the market. It identifies trends of the CGM to give insulin that keeps blood glucose in the normal range — without patients having to count carbohydrates,” highlighted Hirsch on the benefits of this particular device.