Major Finding: A closed-loop insulin delivery system produced more in-target glucose levels and less hypoglycemia than did standard continuous insulin infusion.

Data Source: A three-part randomized crossover study of 17 children and adolescents with type 1 diabetes.

Disclosures: The study was funded by the Juvenile Diabetes Research Foundation and three European research foundations. Dr. Hovorka has received lecture fees from Minimed Medtronic, Abbot Diabetes Care, Lifescan, Novo-Nordisk, and BBraun. He reported two patent applications. Two coinvestigators reported disclosures, while 12 others reported no competing interests. Dr. Renard stated he had no conflicts of interest.

A closed-loop system linking continuous glucose measurements to insulin delivery reduced the risk of nocturnal hypoglycemia compared with standard continuous subcutaneous insulin infusion in a three-part randomized crossover study involving 17 children and adolescents with type 1 diabetes.

Previous studies have assessed the feasibility of closed-loop insulin delivery systems that used various types of control algorithms, but this is the first to compare closed-loop delivery with traditional continuous subcutaneous insulin infusion (CSII) and the first to assess the effect of evening meals and exercise, noted Dr. Roman Hovorka, who is with the University of Cambridge (England) and his associates (Lancet 2010 Feb. 5 [doi:10.1016/S0140-6736(09)61998-X]).

“Closed-loop systems could transform management of type 1 diabetes, but their introduction is likely to be gradual, starting from straightforward applications such as shutting off the pump at low glucose concentrations or overnight closed-loop delivery, proceeding to more complex applications providing 24-[hour] control,” the investigators said.

The study subjects were aged 5-18 years and had type 1 diabetes for a mean duration of 6.4 years.

Thirteen of the children were assigned to be treated with overnight (8:00 p.m. to 8:00 a.m.) closed-loop delivery (using Medtronic's Guardian Real-Time) or standard CSII on two separate occasions.

Seven of those 13 children were also evaluated overnight with the closed-loop device (Abbott's Freestyle Navigator) on two further occasions, this time after having consumed either rapidly or slowly absorbed large meals matched for total carbohydrates (129 g), but differing in glycemic load.

In a third overnight evaluation, 10 adolescents aged 12-18 (including 4 from the first evaluation) rode a treadmill from 6:00 to 6:45 p.m. after having eaten a light meal at 4:00 p.m., again comparing closed loop (Navigator) with CSII. (All patients used Smiths Medical's Deltec Cosmo insulin pumps for the study.)

During the closed-loop nights, glucose measurements were fed every 15 minutes into a control algorithm calculating rate of insulin infusion, and a nurse adjusted the insulin pump. Standard pump settings were used on control nights.

Primary outcomes—the time for which plasma glucose concentration was in the target range of 3.91-8.00 mmol/L (70.4-144 mg/dL) or a hypoglycemic level of 3.90 mmol/L (70.2 mg/dL) or lower—did not differ significantly between closed loop and CSII for the first evaluation, Dr. Hovorka and his associates said.

The time spent in target range was 52% with closed loop versus 39% for CSII, and the time spent with hypoglycemia was 1.0% vs. 2.0%, respectively. For the exercise evaluation, the proportions for closed loop vs. CSII were 78% vs. 43% of time in target range and 10.0% vs. 6.1% of time with hypoglycemia, respectively.

For the meal comparison, there was no significant difference in overnight control between the two meals: Time spent in target range after midnight was 86% for the rapidly absorbed meal and 83% for the slowly absorbed meal. From the start of the closed-loop delivery, those proportions were 53% and 55%, respectively, Dr. Hovorka and his associates reported.

In a secondary analysis pooling the data from the first and third evaluations, closed-loop delivery significantly increased the time for which plasma glucose was in the target range (60% vs. 40%) and significantly reduced the time in hypoglycemic range (2.1% vs. 4.1%). The difference was even more significant after midnight, when closed-loop became fully effective (79% vs. 35% in target range, 3.0% vs. 6.1% hypoglycemic), they said.

In an accompanying editorial, Dr. Eric Renard of Universitaire de Montpellier, France, commented that although overall mean blood glucose concentrations were not significantly different with closed-loop delivery, the fact that the closed loop stabilized blood glucose levels overnight while keeping hypoglycemia to a minimum is “an important step forward for young patients and their parents.”

The nurses' intervention could easily be replaced by a wireless direct connection, he noted, and “the absence of any sensor or pump failure during the trials allows realistic extrapolation towards a home environment.”

However, the model predictive control algorithm used in the closed-loop system in this study will make glucose control at mealtimes more difficult than overnight control, due to the more complex effects that will need to be integrated into the model.

Proportional-integral-derivative algorithms may offer more flexibility around meals, noted Dr. Renard, who uses proportional-derivative algorithms in his own closed-loop research.

Closed-Loop System Improves Glucose Control