Story of Discovery—Artificial Pancreas for Managing Type 1 Diabetes: Cutting-edge Technology 50 Years in the Making
Aristotle once said: “the whole is greater than the sum of its parts.” This could be said for a revolutionary new management tool for type 1 diabetes: artificial pancreas technology. This technology builds on decades of advances in type 1 diabetes management—such as the advent of insulin pumps and continuous glucose monitors (CGMs). Now, with long-standing support from NIDDK, private funding entities, and industry, researchers have combined these individual “parts” to develop a “whole” device that can benefit people more than each component alone.
A major milestone occurred in 2016 with the first U.S. Food and Drug Administration (FDA) approval of a commercial “hybrid” artificial pancreas device developed by industry; NIDDK-supported research contributed to the early development of the approved device. This milestone promises to be just the beginning of a new era in the development of cutting-edge artificial pancreas technologies that could benefit people with type 1 diabetes.
What Is an Artificial Pancreas?
Artificial pancreas technology strives to mimic the exquisite control of blood glucose (sugar) maintained by a healthy pancreas. The pancreas produces a number of hormones, including insulin and glucagon, to enable the body to use glucose for fuel and store glucose when it is in excess, or release it from stores when glucose levels get too low. In people with type 1 diabetes, the immune system destroys insulin-producing pancreatic β cells. Thus, people with the disease need to administer insulin for survival and measure blood glucose levels to determine how much insulin to take.
An artificial pancreas, or “closed-loop system,” is technology that would replace the function of the pancreas that does not work in people with type 1 diabetes: delivering insulin in response to blood glucose levels. It links three technologies: (1) a sensor, such as a CGM, that measures blood glucose levels and sends data to a computer; (2) an insulin delivery device, such as an insulin pump; and (3) a computer that calculates the amount of insulin needed and instructs the pump to deliver it. The goal is that the technology would require minimal human input, thus acting much like a healthy pancreas.
The need for improved type 1 diabetes management tools is underscored by the 1993 results of the NIDDK-supported landmark Diabetes Control and Complications Trial (DCCT). This trial and the ongoing follow-up study showed that intensive blood glucose control greatly reduced the development of eye, kidney, and nerve complications in people with type 1 diabetes, and even suggested that the increased mortality historically seen in those with type 1 diabetes can be reduced or eliminated through careful management of blood glucose. The ongoing follow-up study has continued to demonstrate long-term, improved health benefits of early and intensive glucose control. The intensive control regimen is difficult, however, because it requires that people frequently and carefully monitor blood glucose levels throughout the day with finger sticks or a CGM and determine how much insulin to administer. The calculation is a “best guess” that requires monitoring and modification, particularly when everyday experiences like eating, exercising, and illness can affect blood glucose levels in unexpected ways. Additionally, the DCCT showed that intensive control increased people’s immediate risk for hypoglycemia, or dangerously low blood glucose levels. With the difficulties of intensive glucose control and the threat of hypoglycemia, people with type 1 diabetes still rarely achieve recommended glucose levels to reduce the risk of complications. Thus, there is a critical need for better management tools, such as artificial pancreas technology that could help people achieve recommended blood glucose control while preventing hypoglycemia and reducing management burden.
Early Artificial Pancreas Development
The idea of replacing the pancreas with an artificial one is not new. In fact, the first experimental artificial pancreas was developed in 1964. The device—about the size of a large backpack—measured blood glucose levels and delivered insulin intravenously and thus was not intended for everyday use in free-living conditions. Despite this early effort, the technologies it used to measure blood glucose levels and deliver insulin would take decades to develop for daily, home use. Additional progress was made in the mid- to late-1970s with several research groups publishing results of their closed-loop studies, as well as with the introduction of the first commercial artificial pancreas device intended to be used for inpatient hospital use—the Biostator. These big and bulky devices also used intravenous glucose sensing and inulin administration. While these advances were important for showing that this technology was feasible, much more research was needed to develop artificial pancreas technology that could be used in everyday life.
Advances in the “Parts” of Artificial Pancreas Technology
Before the artificial pancreas as a whole could be improved, it was necessary to improve the individual parts. One key component of artificial pancreas technology is the insulin pump, which was developed with NIH and industry support. Insulin pumps are small devices that deliver insulin continuously in a small basal amount and provide larger boluses when needed, for example at mealtime. They provide an alternative to multiple injections of insulin and deliver insulin subcutaneously (i.e., under the skin). The first commercially available insulin pump was introduced in 1978, and was bulky and not very user-friendly, relative to more recent devices. However, clinical studies of insulin pump therapy beginning in the late 1970s, and supported by the NIDDK and others, demonstrated that pumps were an effective alternative to multiple daily insulin injections, which helped promote their adoption into clinical practice. The development of insulin pumps made the DCCT possible, and the results of the DCCT demonstrating the importance of intensive blood glucose control stimulated future development of improved pumps.
Insulin pumps are widely used today and have improved over the years. For example, in 2013, the FDA approved a device, consisting of a pump linked to a sensor, that suspends the delivery of insulin when glucose levels reach a preset low threshold. This approach helps reduce the risk of hypoglycemia from excess insulin. Further research has led to newer devices that predict, based on trends in the sensor data, when this level will be reached and preemptively suspend insulin delivery. Such devices have been tested in NIDDK-supported studies and are being used in today’s artificial pancreas systems. Advances in insulin engineering, which has produced faster acting and longer duration forms to use in insulin pumps, have also been important to artificial pancreas development.
Another critical component of artificial pancreas technology is a CGM, which provides a way to measure glucose levels every few minutes. For years, people with diabetes monitored their glucose levels with urine tests, which recognized high but not dangerously low glucose levels and reflected past, not current, glucose levels. By the 1980s, blood glucose meters had been developed and were widely used, enabling people to self-monitor blood glucose levels at home with finger sticks.
By the late 1990s, measurement of glucose in the blood had proven useful for several checks per day, but it was not readily amenable to continuous monitoring by a device that people could use in their daily lives. Researchers supported by the NIDDK and others found another route that would be safe and practical for continuous monitoring: glucose levels in “interstitial” fluid in tissues under the skin reflected glucose levels in the blood. One aspect of this advance was the development of miniaturized electrodes to measure the reaction of glucose with an enzyme called glucose oxidase to assess glucose levels in the small fluid volumes present under the skin. A continuous monitor was first approved by the FDA in 1999. However, the glucose values obtained from this device could only be assessed retrospectively, not in real time. Further research led to FDA approval, in 2006, of next-generation continuous monitors that allowed real-time glucose sensing. Since then, CGMs have continued to improve, are used by some people with diabetes, and are part of today’s artificial pancreas systems.
With the advances in the two major parts of artificial pancreas systems, also needed was a way to link them together. Thus, at the same time, researchers were working on the mathematical part of artificial pancreas technology by developing algorithms so that a computer can calculate how much insulin is needed based on CGM readings and tell the insulin pump to deliver it. Many algorithms being tested in some of today’s artificial pancreas devices, called “model predictive control algorithms,” have their roots in an NIDDK-supported 1979 study that evaluated the feasibility of using a mathematical model of glucose disappearance to estimate insulin sensitivity. Additionally, computers themselves have significantly advanced over the last few decades, becoming smaller, more portable, faster, and capable of advanced processing. These technological advancements laid the foundation for research to link together these individual technologies to develop next-generation artificial pancreas devices.
The NIDDK supported development of these components—and continues to support the next generation of these devices—through grant awards to small businesses, as well to academic institutions. Many of the businesses were created by academic investigators to work toward commercializing the new technologies. Often these NIDDK-supported small businesses became larger companies, or the technologies were sold to larger businesses toward bringing the technologies to market. These important investments in small businesses played a significant role in ensuring that these advances moved from the laboratory to the people who could benefit from them.
A New Era of Artificial Pancreas Research—Making the Whole Greater Than the Sum of Its Parts
A new era of research began in the 2000s as these technologies were becoming available. A 2006 NIDDK-supported study conducted by researchers in industry and academia was the first to show the feasibility of simultaneously using a continuous glucose monitor for subcutaneous glucose sensing and a subcutaneous pump for insulin delivery in an artificial pancreas system in people with type 1 diabetes. This study propelled the field toward the advanced trials being conducted today.
Since that time, there has been tremendous progress in developing artificial pancreas technologies and testing them in people, which has been supported, for example, by the NIDDK, JDRF, the Leona M. and Harry B. Helmsley Charitable Trust, and industry. The first small clinical trials, in the early 2010s, were conducted in hospital or clinical settings using laptops to control the artificial pancreas device and with constant monitoring by researchers and study staff, restricting participants’ activity. The trials showed very promising results. For example, in 2010, results from the first randomized controlled trial showed that use of an artificial pancreas device reduced the risk of nighttime hypoglycemia in children and adolescents with type 1 diabetes compared to standard therapy. Two years later, NIDDK-supported studies showed that artificial pancreas technology could improve blood glucose levels without increasing hypoglycemia in adolescents and adults with type 1 diabetes.
The promising results from hospital-based trials, as well as the advent of portable smartphone technology to replace laptops, enabled the next step—moving to “transitional” outpatient settings, such as hotels, that more closely mimicked free-living conditions but still allowed the participants to be monitored closely by study staff. For example, in 2013, NIDDK-supported researchers working with adult participants who had type 1 diabetes showed that smartphone technology could be used to run a closed loop system. Another group of NIDDK-supported researchers tested a “bihormonal” artificial pancreas system that delivers two different pancreatic hormones—insulin to lower blood glucose levels and glucagon to increase them—with the hope of more closely mimicking the function of a healthy pancreas. They tested their wearable, automated device in adults staying at a hotel, and in adolescents at a diabetes camp. Results reported in 2014 showed that compared to usual care, participants had lower mean glucose levels and reduced episodes of hypoglycemia. In fact, the device allowed nearly all participants to achieve recommended levels of blood glucose control. Armed with positive results from transitional settings, scientists took the next step: moving to at-home studies to replicate real-life conditions. The first studies tested overnight artificial pancreas use, as it is easier for the artificial pancreas to control blood glucose levels when it is not being challenged with daytime activities, like meals and exercise, that cause unpredictable blood glucose swings. In 2014, NIDDK-supported researchers studied adolescents with type 1 diabetes who participated in school and other activities during the day using standard diabetes management tools, and wore an artificial pancreas at home at night. Results showed that unsupervised, i.e., not monitored continually by study staff, closed-loop control at night improved glucose control during the day and night, and reduced the number of episodes of nighttime hypoglycemia.
More recently, researchers have begun to test day and night artificial pancreas use under free-living, home-use conditions, for a short duration in small numbers of people who went about their daily lives (e.g., going to school or work) and had no restrictions on their diet and exercise. For example, in 2016, NIDDK-supported researchers showed that unsupervised day-and-night artificial pancreas use was feasible and safe in adolescents with type 1 diabetes. In 2017, NIDDK-supported researchers found that day-and-night use of a bihormonal artificial pancreas improved blood glucose control compared to conventional insulin pump therapy in adults. The positive results of these and other studies testing 24/7 closed-loop control at home under free-living conditions underscore the promise of this therapy for everyday use. Encouraged by this progress in moving from carefully controlled studies in hospitals to hotels to free-living conditions, researchers are now testing artificial pancreas technologies in even more challenging environments, like the extended vigorous outdoor exercise of snowboarding camp.
First FDA Approval of a Hybrid Artificial Pancreas Device
A key milestone in artificial pancreas technology development was achieved in 2016 with the publication of results of an industry-supported trial. It was the largest outpatient artificial pancreas study done to that date, including 124 people with type 1 diabetes ages 14 and older who used the device for 3 months. The studied tested a “hybrid” system: although the device automatically adjusts insulin levels, it requires users to count and enter mealtime carbohydrates. Thus, it is not completely automated and still requires human input. The study showed that the hybrid artificial pancreas system was safe for use.
Based on the results of the trial, in September 2016 the FDA announced that it approved the device for use in people 14 years of age and older with type 1 diabetes—the first commercial hybrid artificial pancreas to be FDA approved. The NIDDK supported early research that contributed to the development of the approved device, such as testing some initial versions of the different components and algorithms that preceded the currently approved device. To improve upon this approved device, the NIDDK is supporting research comparing the current computer algorithm used in the device to another algorithm that might provide better control.
Collaborations Stimulate Progress
Critically important to the success of this effort have been partnerships that the NIDDK has forged with FDA, and voluntaries such as JDRF and the Leona M. and Harry B. Helmsley Charitable Trust, all of whom are working toward the same goal of making the technology available to people with type 1 diabetes. For example, the NIDDK, FDA, and JDRF have sponsored a series of scientific workshops that have brought together experts in the field to stimulate discussion of the current state of the art, technical challenges and possible solutions, safety issues, and next steps. Since 2007, the NIDDK and FDA have also worked together on a federal Interagency Artificial Pancreas Working Group, which has been instrumental in promoting the field. NIDDK collaboration with JDRF and the Leona M. and Harry B. Helmsley Charitable Trust has also resulted in coordinated efforts with more efficient investment of resources and has greatly expedited progress. Many of the studies described above have also had support from these funding entities.
Future Research Directions
The FDA approval of the hybrid artificial pancreas device is a critical milestone, moving the technology from a research setting to the marketplace. However, there are still limitations with the approved device. For example, it requires users to prick their fingers to determine blood glucose levels to calibrate the device’s glucose sensors. Other approved sensors, however, do not require this type of calibration and may be incorporated into future devices. Also, the device requires users to count carbohydrates and put that information into the system at mealtimes. Thus, the device is not completely automated and still places management burden on patients. Still needed are improved components of artificial pancreas systems, as well as research testing different systems so that people could have choices about which system may best suit their needs and lifestyle.
Toward these goals, the NIDDK continues its vigorous support of artificial pancreas research through grants to small businesses and academia. For example, the Institute is supporting four new clinical trials to test technologies in larger groups, with wider age ranges, over longer periods of time, and in largely unrestricted conditions. These trials are generating data to determine whether the technologies meet the safety and efficacy requirements for FDA approval of these systems. Research is also ongoing toward improving the “parts” of artificial pancreas technology, such as developing implantable glucose sensors and novel hormone formulations, which could help overcome some of the limitations of current systems toward making fully automated devices. Additionally, developing implantable components is important toward achieving the goal of giving people with diabetes freedom from having devices attached to their bodies and having to manage the devices. Because new technology will only benefit people if they can use it, research optimizing incorporation of artificial pancreas technologies into clinical care and enhancing their “usability” is essential, so that people are not overwhelmed with excessive data volume or complexity. This includes research on special populations—such as young children, people with more difficult-to-control diabetes or vision complications, pregnant women, and older adults—to ensure that all people with type 1 diabetes could use and benefit from this new technology.
This technology could also be beneficial beyond type 1 diabetes. Many people with type 2 diabetes manage their diabetes with insulin. Hypoglycemia in this population is a major cause of morbidity, and artificial pancreas technology may help people using insulin to manage their type 2 diabetes to avoid this dangerous complication while improving their blood glucose control. A recent study found that adults who used continuous glucose monitoring on a daily or near-daily basis for 6 months showed improved blood glucose control. Another study tested artificial pancreas technology in people with type 2 diabetes who had not previously used insulin to treat their diabetes. This study found that participants spent more time in the target glucose range and less time in hyperglycemia. This small, short trial suggested that artificial pancreas use in people with type 2 diabetes could improve glucose levels, especially overnight, without increasing the risk of hypoglycemia. There is also potential for these devices to be used in hospital settings to control glycemia; lack of glycemic control is a major problem, even in people who do not have diabetes, but who have hyperglycemia due to trauma or a critical illness.
While the artificial pancreas represents near-term technology that could improve the health of people with type 1 diabetes, it is not a cure. Therefore, the NIDDK also supports research toward a biological cure for type 1 diabetes, such as developing therapies to replace pancreatic β cells that have been destroyed by the immune system.
With the significant progress to date and with additional research, it is expected that the FDA approval of the first hybrid artificial pancreas device is just the beginning of a new era of technologies that can improve the short- and long-term health of people with type 1 diabetes, while also reducing management burden. Artificial pancreases truly represent a life-changing advance for people with this chronic disease.