Artificial Pancreas Reduces Overnight Hypoglycemia at Camp

Miriam E. Tucker

February 28, 2013

(Updated) An investigational artificial-pancreas system reduced nighttime hypoglycemia and improved glucose control among kids at a diabetes camp compared with treatment with a sensor-augmented insulin pump, a new study has shown.

The findings were published in the February 28 issue of the New England Journal of Medicine by Moshe Phillip, MD, from the National Center for Childhood Diabetes, Schneider Children's Medical Center of Israel, Tel Aviv, and colleagues.

Nocturnal hypoglycemia episodes account for 75% of total hypoglycemic seizures in children and have been linked to deaths in type 1 diabetes patients younger than 40 years of age, the authors write.

Fully automated artificial-pancreas systems currently in development combine an insulin pump with a glucose sensor via a computerized control algorithm, which directs insulin delivery in response to sensor data. Previous studies have shown that these systems can normalize glycemic control and reduce or eliminate overnight hypoglycemia among children and adults with type 1 diabetes, but these studies have taken place in hospital settings.

The current study compares the MD-Logic artificial pancreas — a closed-loop system with real-time alarms — with a sensor-augmented pump system in 56 children with type 1 diabetes. The setting of a camp was chosen because it combines real-life elements with the availability of a healthcare team.

"This study evaluates for the first time an artificial-pancreas system outside the hospital in [the] challenging setting of a diabetes camp in a prospective randomized crossover study," Dr. Phillip told Medscape Medical News.

Asked to comment on the findings, Stuart A. Weinzimer, MD, from Yale University School of Medicine, New Haven, Connecticut, who was not involved in the study, said: "This is a very important paper…in that it evaluated a large number of children in an ambulatory setting without relying on laboratory-quality glucose levels overnight. It is by no means a pivotal or effectiveness study, but it's a wonderful first study [and] works as a major proof-of-concept of [an] outpatient [closed-loop system] in kids, no less." Nevertheless, he pointed out, "We're still a long way from rolling this out to [general use]."

Meanwhile Steven J. Russell, MD, from Massachusetts General Hospital, Boston, who is also developing an artificial pancreas, questioned the practicality of the system used in this study and the fact that the work assessed only overnight control, something that has been accomplished previously, he said. To gain Food and Drug Administration approval, such an artificial pancreas would likely have to be able "to handle daytime control as well," he observed. He also queried the relevance of the camp setting given the intensity of the interventions delivered and the number of times the children were woken up and given carbohydrates.

Data From 2 Consecutive Overnight Sessions

Dr. Philip explained that the MD-Logic artificial-pancreas system was developed at Schneider Children's and tested in the multinational study, called Diabetes Wireless Artificial Pancreas Consortium (DREAM), at 3 camps, 1 each in Israel, Slovenia, and Germany.

The 56 children were aged 10 to 18 years, with type 1 diabetes of at least 1-year duration and insulin-pump treatment for at least 3 months. Each child was randomized to either the artificial pancreas or a sensor-augmented insulin pump (Paradigm Veo, Medtronic) on the first night and then switched to the other treatment the next night.

The artificial-pancreas results were significantly better than the sensor-augmented-pump results for 2 of 3 primary end points, the number of episodes of blood glucose levels less than 63 mg/dL (7 vs 22, P = .003) and the time periods during which the blood glucose level was below 60 mg/dL, which were shorter for the artificial pancreas (P = .02).

However, there was no significant difference in the remaining primary end point, overnight glucose levels, at 126.4 mg/dL for the artificial pancreas compared with 140.4 mg/dL for the sensor-augmented pump.

There were also significant improvements with the artificial pancreas in several other measures of glycemic control and variability, with the investigational system providing overall tighter glucose control, say the researchers.

Of 39 hypoglycemia alarms in the artificial-pancreas group, 13 were false, compared with 27 false alarms out of 41 with the sensor-augmented pump. There were a total of 26 carbohydrate interventions overall during the artificial-pancreas nights and 25 during the control nights.

More adverse events, including day and nighttime hypoglycemia, headaches, dizziness, and "feeling ill" were reported during sensor-augmented-pump nights than during artificial-pancreas nights. No serious adverse events were reported during the study.

"In our study, the significant improvement in patients' overnight glucose control and the reduction in the number of events and duration of hypoglycemia appeared to be related to the combined effect of better control of the amount of insulin provided and better control of the timing of insulin delivery, together with the presence of an alarm module, in the artificial pancreas," they observe.

Dr. Weinzimer said that before this approach can progress to the real world, it will be necessary to examine how robust the system is in accounting for changes in insulin sensitivity from day to day, given that this study assessed the system for only a night.

And Dr. Russell observed, "Because it's in a camp setting and it's only overnight, it's a little unclear to me how much of a reduction in supervision this entails. They're actually doing more supervision in this outpatient setting in the nighttime than we do in inpatient studies."

And he pointed out that the majority of the interventions were given when the sensor predicted a low glucose value, rather than for actual lows. "The way I look at this paper, the reduction of hypoglycemia they saw has way more to do with their alarm criteria that are built into their artificial-pancreas system than the way the system gave insulin. It's not really a closed-loop system in the way I think of a closed-loop system. They're giving carbohydrates many times, and they're recalibrating the system frequently. It's far beyond anything that would be practical in a real-life setting."

Progress Being Made for Use in Home Setting

Dr. Phillip noted that his team has already moved to testing the artificial pancreas in the home setting. "We believe that research in this field is very progressive. Our team has managed in a 6-year research period to develop a system safe and efficient enough to be tested at home."

They are developing a "Glucositter" product for taking overnight control for patients with type 1 diabetes, which will be available in "the foreseeable future," he said.

"There is a hope [for patients] that better metabolic control without the fear of hypoglycemia and therefore an improvement in their quality of life is coming soon," Dr. Philip told Medscape Medical News.

This study was supported in part by Sanofi. Medtronic Diabetes supplied software to interface with the Paradigm Veo system, and Intel Israel, Dell Israel, and Yael Software and Systems donated laptops for the study. Dr. Phillip discloses consulting for Bristol-Myers Squibb, Ferring Pharmaceuticals, D-medical Industries, and Andromeda Biotech and receiving payment for lectures including service on speaker's bureaus from Sanofi, Novo Nordisk, Roche, and Pfizer; his institution has received grants from Medtronic Diabetes, Novo Nordisk, Abbott Diabetes Care, Eli Lilly, Roche, Dexcom, Sanofi, Insulet, Animas, Andromeda Biotech, and Macrogenics. Disclosures for the coauthors are listed at . Dr. Weinzimer has received research support from Medtronic and has consulted for Animas in the past. Dr. Russell is developing an artificial-pancreas system and obtains funding from the National Institutes of Health, the Helmsley Foundation, and private donors.