Closed-loop insulin delivery in adults with type 1 diabetes

Abstract

Achieving tight glucose control safely in type 1 diabetes with currently available methods of insulin delivery is challenging. Aggressive regimens carry an increased risk of hypoglycaemia, particularly overnight. Both alcohol consumption and exercise predispose further to low glucose levels. The demands are even greater in pregnancy where, in addition to limiting hypoglycaemia, avoidance of postprandial hyperglycaemia is critical to minimising adverse obstetric outcomes. The aim of my studies was to evaluate feasibility and safety of a closed-loop or ’artificial pancreas’ system linking insulin delivery with continuous glucose monitoring (CGM), in adults with type 1 diabetes in a controlled setting. Three randomised crossover studies compared closed-loop insulin delivery with conventional insulin pump therapy on two separate occasions, matched in meals and activities. During closed-loop visits, CGM values were entered into a computer containing a model predictive control algorithm which advised on basal insulin infusion for subcutaneous delivery, every 15 minutes. During control visits, usual insulin pump regimen was continued. The feasibility study evaluated overnight closed-loop in 12 adults (seven females, mean age 37.7 years, HbA1c 7.8%) following 60gcarbohydrate evening meal. A follow-up study assessed overnight closedloop in 12 further adults (seven females, mean age 37.2 years, HbA1c 7.8%) following 100g-carbohydrate meal and (mean 564 ml) white wine. The third study evaluated 24 hours of closed-loop in 12 pregnant women (mean age 32.9 years, 19 to 23 weeks gestation, HbA1c 6.4%) during normal daily activities, including low and moderate intensity exercise. Activity and glucose levels were also measured during free-living. CGM performance during exercise was evaluated. Overnight closed-loop insulin delivery in adults, compared with conventional pump therapy, increased time spent with plasma glucose in target range (3.9− 8.0 mmol/l) following both standard meal (81% versus 57%; p = 0.012) and large meal accompanied by alcohol (70% versus 46%; p = 0.012). Glycaemic variability, and time spent in hypoand hyperglycaemia were lowered. In pregnant women, day and night closed-loop insulin delivery was as effective as usual pump regimen (81% versus 81% time spent with plasma glucose 3.5 − 7.8 mmol/l; p = 0.754). Hypoglycaemia occurred following exercise, although closed-loop prevented nocturnal episodes. Glycaemic control during free-living was suboptimal, compared with controlled diet and exercise conditions. Accuracy of CGM was lower during exercise. In conclusion, these studies confirm the feasibility and efficacy of overnight closed-loop insulin delivery in adults with type 1 diabetes. Closed-loop is safe during pregnancy and may be beneficial in women with suboptimal glycaemic control. Meals and physical activity currently limit optimal daytime use of closed-loop.