A 2016 UK study investigated the safety and efficacy of artificial pancreases for managing blood sugar in type 2 diabetics in a non-intensive care unit setting. Compared with conventional insulin therapy, artificial pancreases were found to be more effective at maintaining the target range of blood sugar and were well-received by patients.
Persistent high blood sugar in patients with type 2 diabetes is associated with greater disease severity and an increased risk of death. For many type 2 diabetics, high blood sugar results from the pancreas’ inability to produce sufficient levels of insulin, the hormone which allows cells to absorb sugar from the bloodstream. Patients with advanced diabetes and persistent high blood sugar are typically monitored for significant changes in blood sugar and insulin is administered by health professionals alongside other blood sugar-lowering medications, according to various regimens typically used in intensive care units (ICU) .These regimens are labour-intensive and expensive to maintain, and as such, are not well-suited outside of the ICU. One proposed alternative to conventional insulin regimens is an automated system of insulin delivery – an artificial pancreas – which would monitor blood sugar and accordingly dispense insulin in real time. Early trials have suggested such systems are feasible in an ICU setting, though more research is necessary to determine their feasibility, safety, and efficacy outside of the ICU.
A 2016 UK trial published in the Lancet Diabetes & Endocrinology examined the safety and efficacy of a closed-loop, automated insulin delivery system in type 2 diabetics. Between February 20, 2015, and March 24, 2016, a total of 40 adults diagnosed with type 2 diabetes for at least 1 year receiving insulin treatment were recruited from Addenbrooke’s Hospital in Cambridge, UK. Those admitted to an ICU, who received medication or had a psychological or physical condition likely to affect results who were pregnant or breastfeeding, or who were scheduled to receive surgery within the study period were excluded from the trial. Participants were assigned in equal proportions (20 per group, matched with respect to blood glucose, body mass index, and the usual amount of insulin received) to either have insulin delivered by an artificial pancreas or through conventional insulin therapy (the control group).
The artificial pancreases consisted of a control device which, according to blood sugar data provided by an implanted monitor, maintained or adjusted insulin delivery every 12 minutes using an insulin pump. An alarm would sound in the event of malfunction or disconnection. The target range for blood sugar was between 5.8 and 7.2 mmol/L. Upon the start of the study period, participants in the artificial pancreas group discontinued insulin therapy or use of sulfonylureas – a class of drugs which promote insulin release from the pancreas. All participants were followed for up to 72 hours, during which standard hospital meals were provided at regular mealtimes and patients were free to consume snacks at their leisure. Efficacy was assessed based on the amount of time spent at 5.6 to 10.0 mmol/L blood sugar compared to other concentrations (i.e., <3.5 mmol/L, <5.6 mmol/L, >10 mmol/L, or >20 mmol/L), average blood sugar concentration, and total insulin dose. Safety was assessed based on low blood sugar episodes, and mild and serious adverse events.
One participant, in the artificial pancreas group, withdrew from the study due to unrelated medical complications. A higher proportion of time was spent at blood sugar concentrations between 5.6 and 10.0 mmol/L by the artificial pancreas group (59.8%) than by the control group (38.1%). The proportion was 20.1% higher at night and 21.0% higher during the day for the artificial pancreas group than the control group, respectively. The proportion of time spent at concentrations higher than 10.0 mmol/L was 19.0% lower for the artificial pancreas group than for the control group, without significant increases in total daily insulin delivery. Time spent under 5.6 and under 3.5 mmol/L were similar for both groups. A total of 5 mild low blood sugar events were experienced across 3 patients in the artificial pancreas group, treated with 20mg of carbohydrates taken orally. Blood sugar levels varied less for the artificial pancreas group.
Blood sugar monitors were replaced 7 times in the artificial pancreas group and 3 times in the control group due to malfunction or removal for hospital procedures such as MRI scanning. Two insulin pumps were removed for MRI scanning during the study.
The results of this study suggest the closed-loop, automated insulin delivery system tested is feasible, safe, and effective for managing high blood sugar in type 2 diabetics. Participant satisfaction was very high in the artificial pancreas group, with upwards of 85% pleased with their blood sugar levels, appreciative of the automated system, or willing to recommend the artificial pancreas to other diabetics. The high proportion of type 2 diabetics admitted into the trial with foot ulcers (75%) – who were more likely to have a longer hospital stay – enrolment and study at only one hospital, and short study period may limit the applicability of results. During periods when insulin pumps had to be disconnected, individuals in the artificial pancreas group were given low doses of a synthetic form of insulin to reduce the risk of high blood sugar. The presence of alarms for the artificial pancreas group may have reduced the number of adverse events by alerting staff who would then intervene. Overall, these findings provide a good foundation for future research on artificial pancreases for managing blood sugar in diabetics.
Written By: Raishard Haynes, MBS