Background

The artificial pancreas is an emerging technology for the treatment of type 1 diabetes and two configurations have been proposed: single-hormone (insulin alone) and dual-hormone (insulin and glucagon). We aimed to delineate the usefulness of glucagon in the artificial pancreas system.

Methods

We did a randomised crossover trial of dual-hormone artificial pancreas, single-hormone artificial pancreas, and conventional insulin pump therapy (continuous subcutaneous insulin infusion) in participants aged 12 years or older with type 1 diabetes. Participants were assigned in a 1:1:1:1:1:1 ratio with blocked randomisation to the three interventions and attended a research facility for three 24-h study visits. During visits when the patient used the single-hormone artificial pancreas, insulin was delivered based on glucose sensor readings and a predictive dosing algorithm. During dual-hormone artificial pancreas visits, glucagon was also delivered during low or falling glucose. During conventional insulin pump therapy visits, patients received continuous subcutaneous insulin infusion. The study was not masked. The primary outcome was the time for which plasma glucose concentrations were in the target range (4·0-10·0 mmol/L for 2 h postprandially and 4·0-8·0 mmol/L otherwise). Hypoglycaemic events were defined as plasma glucose concentration of less than 3·3 mmol/L with symptoms or less than 3·0 mmol/L irrespective of symptoms. Analysis was by modified intention to treat, in which we included data for all patients who completed at least two visits. A p value of less than 0·0167 (0·05/3) was regarded as significant. This trial is registered with ClinicalTrials.gov, number NCT01754337.

Findings

The mean proportion of time spent in the plasma glucose target range over 24 h was 62% (SD 18), 63% (18), and 51% (19) with single-hormone artificial pancreas, dual-hormone artificial pancreas, and conventional insulin pump therapy, respectively. The mean difference in time spent in the target range between single-hormone artificial pancreas and conventional insulin pump therapy was 11% (17; p=0·002) and between dual-hormone artificial pancreas and conventional insulin pump therapy was 12% (21; p=0·00011). There was no difference (15; p=0·75) in the proportion of time spent in the target range between the single-hormone and dual-hormone artificial pancreas systems. There were 52 hypoglycaemic events with conventional insulin pump therapy (12 of which were symptomatic), 13 with the single-hormone artificial pancreas (five of which were symptomatic), and nine with the dual-hormone artificial pancreas (0 of which were symptomatic); the number of nocturnal hypoglycaemic events was 13 (0 symptomatic), 0, and 0, respectively.

Interpretation

Single-hormone and dual-hormone artificial pancreas systems both provided better glycaemic control than did conventional insulin pump therapy. The single-hormone artificial pancreas might be sufficient for hypoglycaemia-free overnight glycaemic control.

Funding

Canadian Diabetes Association; Fondation J A De Sève; Juvenile Diabetes Research Foundation; and Medtronic.