The Artificial Pancreas Aces New Tests
“Bionic” volunteers venture into the real world of ice cream and red wine
For Edward Damiano, PhD, a biomedical engineer, developing the artificial pancreas is “a race against time.” Damiano has a son with type 1 diabetes and wants to send him off to college in 2017 equipped with an artificial pancreas. Though other researchers may not have as specific a target date as Damiano, scientists around the globe are focused on bringing the artificial pancreas to people whose lives will be forever changed by this technology.
Easing the Burden
To stay alive, people with type 1 diabetes need to take insulin multiple times a day, 365 days a year. In a person without diabetes, the beta cells in the pancreas make enough insulin, a hormone that helps convert food into energy. They detect blood glucose levels and deliver insulin based on those levels with little risk of the glucose levels ever going dangerously high or low. That’s what an artificial pancreas aims to do for people with type 1 diabetes, whose beta cells make little, if any, insulin.
Insulin users have to calculate appropriate doses based on what they eat, their physical activity, blood glucose, and other factors—intensive insulin therapy requires a great deal of physical, mental, and emotional attention. Even the most meticulous person can’t match insulin doses to the body’s needs perfectly all the time. The result: periods of high blood glucose (hyperglycemia), which raise the risk of long-term diabetes complications, and low blood glucose (hypoglycemia), which can trigger unpleasant side effects and, in severe cases, can result in seizures, coma, and even death. “For me, everything that I do, I have to think about it,” says Scott Scolnick, 53, of Nashua, N.H., who lives with type 1 diabetes. He participated in a recent study of an artificial pancreas, though he prefers the term “bionic pancreas.”
Scolnick, a foodie, loves going out for a decadent plate of pasta. When he does, though, he worries about what’s in the dish, how much insulin to take with the meal, how much insulin to take later, and whether he’s going to take a walk after dinner. “While I can enjoy the experience, I’m distracted by all of these things,” he says. “Being bionic means no more distractions.”
Measuring blood glucose levels is an essential part of staying healthy and making diabetes management decisions; it provides both immediate feedback and a way to figure out what works in the long term. But self-monitoring of blood glucose requires multiple finger sticks. Plus, factoring blood glucose levels, food, exercise, and other variables into choosing an insulin dose can involve tricky math.
The goal of the artificial pancreas is to relieve the person with diabetes from some of this burden, providing excellent blood glucose control with minimum effort. It’s not a cure, but it may be the next best thing.
Artificial Pancreas 101
The artificial pancreas bridges the gap between two pieces of diabetes technology that already exist: the insulin pump and the continuous glucose monitor (CGM). “The cool thing is it’s not brand-new technology,” says study participant Anna Floreen, 31, of Boston, who has type 1. “That part instilled a lot of hope in people.”
With an artificial pancreas, a computer program, instead of the person with diabetes, calculates how much insulin the pump delivers based on readings from the CGM. Such a “closed-loop system” requires little, or possibly no, input from the user. A person using an artificial pancreas would have the option to make dosing adjustments in certain situations, but the device would mostly act automatically to keep blood glucose levels within a target range.
The first artificial pancreases—clunky hospital-bound devices—were born in the 1970s. In the past decade, as technology has matured, researchers have gotten serious about developing a portable artificial pancreas for home use. The effort has accelerated in just the past few years, thanks in part to the efforts of the Food and Drug Administration (FDA) to expedite the process. The FDA is interacting closely with scientists and industry to ensure they know what types of studies are necessary to advance the artificial pancreas through each stage of approval.
Since 2010, about 40 high-quality clinical trials of some version of an artificial pancreas have been published, according to Francis J. Doyle III, PhD, department chair of chemical engineering at the University of California–Santa Barbara. He stresses that “there is no one single artificial pancreas,” but rather many variations in development by scientists around the world. This research is supported by a variety of public and private sources, including the National Institutes of Health and JDRF.
Within a few years, scientists are hoping to start the large clinical trials that will provide the data the FDA will use to decide whether or not to approve some version of an artificial pancreas. The people in those studies will use the experimental product in their daily lives, so scientists are now focusing on making the artificial pancreas robust, portable, and user friendly.
Until very recently, artificial-pancreas studies were conducted exclusively within the confines of hospitals. The computer programs that ran the experimental devices were housed on laptop computers, with wires snaking between participants’ pumps and CGMs. In the first studies, an artificial pancreas wasn’t even enabled to directly dose insulin. Instead, for extra safety, a nurse or doctor would check the laptop to see what insulin dose the artificial pancreas calculated and manually enter dosing instructions for the pump.
Study by study, as the artificial-pancreas prototypes demonstrated that they could be run safely, researchers have upped the ante, allowing some exercise here, a glass of wine or a doughnut there, all while trying to improve the computer programs.
Now, the artificial pancreas is venturing outside the hospital. At least three artificial-pancreas prototypes are being studied in semi-real-world settings. The first such study, led by Moshe Phillips, MD, and published in February 2013 in The New England Journal of Medicine, took place at diabetes camps in Israel, Germany, and Slovenia. For one night only, the researchers let the artificial pancreas take over the maintenance of campers’ blood glucose levels between dinner and breakfast the next morning. That night, the campers had better average blood glucose levels and fewer episodes of hypoglycemia (defined for the study as blood glucose levels below 63 mg/dl) than on a night they used a standard CGM-pump combination.
In more recent studies, the artificial pancreas has ventured into the Clara Barton diabetes camp in Massachusetts and sites in Italy, France, Charlottesville, Va., and Boston. Participants in these fledgling studies are closely watched, physically and digitally. During the studies, their blood glucose levels are beamed to websites that researchers can access. Much of this research is ongoing, with few published results. However, preliminary findings are raising hopes that an artificial pancreas is indeed coming, and soon.
The most important step toward an artificial pancreas, according to Boris Kovatchev, PhD, director of the University of Virginia Center for Diabetes Technology, was “replacing the laptop and all the wires.” Some prototypes have replaced the laptops with smartphones and the wires with wireless Bluetooth connections, while others have employed a tablet computer.
In Kovatchev’s experimental version of an artificial pancreas, the laptop was replaced by a smartphone running Android software, modified not to run games or allow the installation of new applications. “For the first studies, we even disabled phone calls,” he says, except for emergencies. “It’s more secure.” Also, the researchers took steps to make sure the phone batteries wouldn’t run out of juice halfway through the study. These modifications aren’t permanent, Kovatchev says. Once safety is assured, they’ll restore smartphone features.
The heart of Kovatchev’s artificial pancreas is the Diabetes Assistant, a smartphone app that contains the computer program that controls blood glucose. “The phone was easy to read and easy to use,” says Kate Jenks, 56, of Lexington, Va., who tested the device in a five-night study at a guesthouse on the University of Virginia campus. “I could check where I was at any time. It would show me my blood glucose, the direction I was heading, and the mini boluses I was receiving.” Diabetes Assistant wirelessly receives data from the CGM and gives commands via Bluetooth to an insulin pump. One advantage of the Diabetes Assistant is that it can be compatible with any CGM or insulin pump. “The phone is kind of impartial,” says Kovatchev. “It will work with any [CGM] sensor that wants to talk to the phone and any pump with Bluetooth.”
Results from the first two Diabetes Assistant artificial-pancreas studies were released in the summer of 2013. Twenty adults with type 1 diabetes from the United States or Europe stayed at a hotel or guesthouse for two 38-hour sessions during which they were free to eat out or exercise. For the first session, the participant controlled insulin pump settings through the Diabetes Assistant. For the second session, the artificial pancreas took over during the day (at night, it ran in safety mode, allowing the rate of insulin delivery to be decreased but not increased). During both sessions, study personnel stayed nearby to offer assistance as needed and monitored participants’ blood glucose levels remotely. The results showed that the artificial pancreas reduced hypoglycemia significantly, says Kovatchev. Additional studies of the device in increasingly real-world situations are under way.
Meanwhile, last spring 20 adults with type 1 diabetes, including Scolnick, participated in the ongoing Beacon Hill study, led by Damiano, of Boston University. Beacon Hill is a historic neighborhood in downtown Boston where participants stayed during a five-day test of Damiano’s artificial pancreas (Damiano prefers the term bionic pancreas, too). Damiano is running a similar study of children attending diabetes camps.
Participants in the Beacon Hill study had free range over 3 square miles, though they were required to be in their hotel rooms from 11 p.m. to 7 a.m. During the day, participants were constantly shadowed by a nurse. Apart from those restrictions, they could basically do and eat whatever they pleased. Scolnick embraced the opportunity. “I wanted to see what this thing could do,” he says. “There are things I do not eat and times I do not eat, because I know what it’s going to do to my blood sugar and it’s not worth it to me.” With the bionic pancreas in the driver’s seat, Scolnick ate pasta with abandon, french fries four days in a row, Chinese food late at night, and a hot fudge sundae. “I couldn’t break it,” he says. “After every meal, my blood sugars went up but came right back down.” The goal, of course, is not to enable people to eat to excess, but the device needs to be ready for any challenge, including large meals with lots of carbohydrate and fat.
The main difference between this system and many experimental artificial pancreases is the addition of a second hormone, glucagon (which raises blood glucose)—and thus a second pump. “I already wear a pump and a CGM, so going from two to three wasn’t a big jump,” says Floreen, who also participated in the Beacon Hill study. “I can see that someone using injections might think this was very robotic. The benefits outweigh the costs.” Children, of course, might have more trouble finding space on their bodies for a sensor and two pump cannulas.
Damiano believes a successful artificial pancreas may need glucagon to compensate for people’s active lifestyles and impulsive natures because “once insulin is given, it can’t be taken back.” He notes that, in the human body, the pancreas has evolved to rely on glucagon to prevent hypoglycemia.
The main obstacle for a dual-hormone pump is the form of glucagon that’s currently available. The powdered glucagon must be dissolved in water before being injected or infused and, once dissolved, has a short shelf life. In the Beacon Hill study, nurses have to whip up fresh batches of glucagon daily for each participant’s glucagon pump. “What we need is a glucagon that is stable for several days,” says Damiano. More stable forms of glucagon are being developed. For example, a company called Xeris has already tested its version of stable glucagon in pigs with diabetes for up to three days. The results have been promising, paving the way for studies in people.
Although artificial-pancreas prototypes are chugging along with existing pump and CGM components, enhancements in insulin delivery and speed of action are needed. Doyle says that getting tight control with a truly hands-free artificial pancreas may depend on using insulin that works faster. That’s because the artificial pancreas calculates insulin doses based on sensor readings, which lag behind rapid fluctuations of blood glucose. So, by the time blood glucose starts to increase after eating, existing insulin formulations that are delivered in response have too little time to prevent blood glucose from going too high.
Doyle is studying one faster-acting insulin, an inhaled version made by MannKind Corp. that is still in clinical trials. In a preliminary study, a dose of inhaled insulin before a meal combined with an artificial pancreas kept blood glucose from spiking afterward, Doyle says. Faster-acting injected insulins, which may provide the same benefit, are also under development.
Another way to make insulin faster is by getting it more quickly into the circulatory system, where it can work. That could be done, Doyle says, by delivering insulin into the abdominal cavity using an experimental Roche product called a DiaPort. It gets the hormone into the bloodstream faster than insulin that’s delivered under the skin by pumps, pens, or syringes. In a study, people using an artificial pancreas along with a surgically implanted DiaPort kept blood glucose levels within target range 47 percent of the time after a meal, compared with 23 percent among those using the artificial pancreas with a standard pump.
The artificial pancreas would also benefit from further development of CGM technology. CGMs now measure glucose levels in the interstitial fluid, the liquid that bathes the body’s cells. While that measurement reflects blood glucose levels, there is a time lag between when blood glucose levels change and when those changes show up in the interstitial fluid. What’s more, better CGM accuracy would improve artificial-pancreas performance.
Regardless of the form the first artificial pancreas takes or whether it’s called a bionic pancreas, the device is inspiring hope in people with type 1 diabetes. “I was told there would be a cure in 10 years. For the first time, I see that light,” says Scolnick. “People need to know that this really is on the horizon.” According to Damiano and Doyle, an artificial pancreas is likely to hit the market within three to five years—and that might be in time for Damiano’s son to take one to college.