The Artificial Pancreas Does Its Homework

By Michelle A. Cissell, Ph.D.
September 24, 2013

JDRF-funded researchers across the globe are currently testing artificial pancreas systems. Some clinical trials are inpatient, in which participants stay overnight or for a few days in a hospital or research clinic, while others take place in an outpatient setting. In both cases, a medical team carefully monitors the daily life of each trial volunteer wearing an artificial pancreas (AP) system—examining factors such as diet, activity level, and sleep periods—to evaluate how the system performs under various conditions and extreme blood-glucose levels. But everyday life is not always as neat as it is in a clinic, especially during the childhood and teenage years.

That’s why Roman Hovorka, Ph.D., director of research in the Department of Pediatrics at the University of Cambridge Metabolic Research Laboratories in the United Kingdom, recently tested at-home use of an AP system for overnight glucose control in adolescents ages 12 to 18 who have T1D. This JDRF-funded study was one of the first to test an AP system outside the controlled environment of a research clinic. It drew on Dr. Hovorka’s unique expertise in mathematical informatics and modeling to develop control algorithms—the crucial part of the computer program that uses data from the AP system’s continuous glucose monitor (CGM) to control its insulin pump.

Working from home

For three weeks, volunteers used a specific type of insulin pump and CGM that are compatible with the closed-loop system as they would normally during the day. But at bedtime, participants wirelessly connected the two devices to a laptop running a control algorithm, and the system began modulating insulin delivery from the pump according to the CGM reading. The study tracked how well the blood-glucose level was controlled during the three weeks of nighttime AP system use compared to that during a separate three-week period when the volunteers practiced conventional pump therapy around the clock.

The at-home study recently completed, and Dr. Hovorka relates that the study team observed wide variations in insulin requirements from night to night when the system was tested at home. Some nights, a volunteer needed half the amount of insulin he or she would have taken on conventional treatment. Other nights, the opposite could be true—a participant’s CGM data might cause the system to administer twice the amount of insulin that the person would have taken normally. This observation highlights the difficulties of overnight glucose control as well as the potential benefit of a closed-loop system that helps people with T1D significantly improve their diabetes management overnight, since better control during sleep reduces the risk of hypoglycemia. Putting the nighttime hours—which account for one-third of a day—into better control should pay long-term dividends in terms of reducing the risk of diabetes complications.

If for any reason the laptop lost the CGM signal or its connection to the pump, the system automatically returned to a preset insulin-infusion rate to restart conventional treatment. Dr. Hovorka says the system is simple and safe to use and seemed to be well accepted by the study participants. “The safety mitigation system allows us to be confident that the system is safe,” he states. “What is most challenging for the volunteers is using a new pump. Generally, they are quite relaxed, because they find it’s just a matter of pushing a button on the laptop to start closed loop.”

Since the launch of its Artificial Pancreas Project, JDRF has led the way in research on closed- loop systems, and this study is no exception. Aaron J. Kowalski, Ph.D., JDRF’s vice president of treat therapies, views Dr. Hovorka’s at-home trial of an AP system as “transformational.” Dr. Kowalski explains, “The trial is a huge advance because the studies are happening in outpatients who are unsupervised, meaning there is not a doctor or nurse in the next room—and that’s the real world. This study is blazing a trail.”

Glucose management: anything but child’s play

Dr. Hovorka is also applying his knowledge to better improve glucose control in young children with T1D. Many children, especially those too young to talk, are not able to recognize or communicate symptoms of dangerously high or low blood-glucose levels. Moreover, young children have small bodies, so they often need only minute amounts of insulin, which can be difficult to dispense accurately. These factors can make glucose control in young children extremely challenging for parents and caregivers.

Dr. Hovorka is conducting a clinical trial to test the use of standard-strength and diluted insulin in children with T1D who are two to six years old. The goal of the study is to determine whether diluted insulin can improve the safety and efficacy of closed-loop control in these young children.

In this study, children stay overnight in a research clinic on two separate occasions. On one night, the children are treated with a closed-loop system that uses insulin in the standard concentration; on the other night, the system uses an insulin formulation that has been diluted fivefold with a saline solution. Ideally, the diluted insulin should increase the accuracy of insulin delivery by the pump, reduce glucose fluctuations, and improve control in very young children compared to the standard insulin concentration.

Leading the way to better glucose control

JDRF’s goal of developing safe and effective artificial pancreas systems relies on partnerships of scientists working in many disciplines, including fields that are not traditionally associated with human disease research, like mathematics. Dr. Kowalski notes, “Dr. Hovorka is a brilliant mathematician. Working with an excellent clinical team at Cambridge, he is leading the way on control-algorithm development for the artificial pancreas.”

Dr. Hovorka himself admits to being somewhat skeptical about closed-loop systems when he first started research in the field in the mid-1990s. But over time, he says, “it became an evolving journey of realizing that we can use existing devices to much greater benefit for people with type 1 diabetes.” His persistence and creativity are now paying off in the form of innovative algorithms that forge a strong link between insulin pumps and CGMs.

With those algorithms, Dr. Hovorka and his team are tackling two of the most worrisome times for diabetes management in children—very young ages and nighttime. If his current trials successfully demonstrate improved control with a closed-loop system, then children and adolescents with T1D, along with their parents, may be able to sleep a little easier.