The Bionic Pancreas: Getting Closer to Reality

Evidence-Based Diabetes Management, December 2014, Volume 20, Issue SP18

Many months typically elapse between fresh announcements about the “bionic pancreas” developed at Boston University (BU) and Massachusetts General Hospital (MGH), but the process of testing and refining the technology proceeds steadily. Indeed, one of the researchers who leads the project recently outlined the long series of successes that have brought the experimental system to the threshold of final-stage clinical trials, and suggested that it may even reach the market a little earlier than initially expected.

“We hope to be moving to pivotal studies in 2015, where we use a fully integrated device that has both an insulin pump and a glucagon pump, the continuous glucose monitor receiver, and the bionic pancreas algorithm, in a single device,” said Steven Russell, MD, PhD, assistant professor at Harvard Medical School and attending physician at MGH, in a webinar.1 “Once we finalize that design, we’re going to start the pivotal studies, and we hope to submit a premarket approval application to the FDA by late 2016 or early 2017, which means that we could commercially launch the device as early as 2017.”

Russell began his webinar by emphasizing that people with type 1 diabetes mellitus (T1DM) today have already benefitted from a true revolution in care. Until 1922, when researchers discovered how to purify insulin from animals and use it to combat diabetes in humans, T1DM was a relatively speedy death sentence. The best available treatment was virtual starvation, but the technique only extended lifespan modestly, and it reduced patients to living skeletons.2

Insulin represented such a monumental improvement over the status quo, Russell said, that medical authorities wrongly considered diabetes to be all but cured. The bionic pancreas may not be a breakthrough on par with insulin, but if it works as well in real life as it has in tests, it could produce the biggest revolution in T1DM care since Calvin Coolidge became president.

Russell predicted that the system will virtually eliminate hospitalizations and other short-term complications from hypoglycemia, while simultaneously reducing the long-term complications of excessive blood sugar by more than 80%. This alone would be great news for people with T1DM, a disease that continues to be the nation’s leading cause of blindness, kidney failure, and nonemergency amputations, despite incremental advances in care since the advent of insulin therapy.3

For many patients, however, the best news about the bionic pancreas may be the instructions for using the machine: calibrate the glucose monitor twice a day and ensure that the system is stocked with insulin, glucagon, and batteries. That’s it. Patients can inform the device that they’re about to eat a meal or snack, but they don’t have to. The only vital patient information that the system needs is their weight.

A number of teams around the world, including academic groups in Europe and at the University of Virginia, are working on similar technology. All seek to automate the

task of blood sugar control as much as possible by combining continuous glucose monitors (CGMs) with insulin pumps, computer analytics, and various other elements.4 Despite the basic similarities, all of these experimental devices differ somewhat. The bionic pancreas from the team at MGH and BU automates the process of increasing blood sugar, as well as the process of decreasing it, by administering both glucagon and insulin.

The size and shape of the team’s bionic pancreas has changed dramatically since human trials began in 2008. Back then, the system paired a full-sized laptop with the wireless transmitters that BU engineers attached to standard hospital glucose monitors and pumps for the insulin and glucagon. Today’s version combines a modified iPhone with a small glucose monitor and 2 pumps, each of which is just a bit larger than a pack of playing cards. The commercial model will likely pair each user’s existing smartphone with a single box that combines the monitor and the 2 pumps.

The algorithm that drives the system has also advanced throughout the years, but the technology has maintained the same fundamental advantages over self-medicating

humans. Machines effortlessly avoid the sort of mathematical errors that people make when they’re calculating insulin dosages and timing. They also work more diligently than any human can.

Indeed, Russell said, most people with T1DM struggle with recommendations such as, that they stop 10 times a day to note their blood sugar and consider making adjustments. The artificial pancreas, on the other hand, notes blood sugar levels and considers adjustments 288 times per day, every day, forever. It catches problems

far faster than people can, and uses very small doses of insulin and glucagon to fine-tune blood sugar very frequently.

These advantages have allowed the bionic pancreas to produce excellent results in trials to date, even as the development team has increased trial durations and reduced restrictions on patient behavior.

The first trial5 recruited 11 subjects with no endogenous insulin secretion for a series of 27 hour-long experiments undertaken at a hospital. Each subject had 3 carbohydrate-rich meals during the course of the experiment, but for 6 of the subjects, the system immediately achieved a mean blood glucose concentration of 140 mg/dL, which meets the ≤154 mg/dL target from the American Diabetes Association. When the study team adjusted the algorithm’s pharmacokinetic parameters to account for the varying speed of glucose absorption, the bionic pancreas got all 11 into the target range.

The second trial6 tested the system twice for 51-hour periods in 6 subjects, who, again, ate high-carbohydrate meals and exercised. Overall, the mean plasma glucose was 158 mg/dL, with 68% of glucose values in the range of 70 mg/ dL to 180 mg/dL. Hypoglycemia, defined as blood glucose less than 70 mg/dL, occurred only 8 times during the 576 hours of closed-loop blood sugar control, and accounted for just 0.8% of all the time that patients were using the system. Patients who medicate themselves often spend more than 10% of their lives in a state of hypoglycemia.

The so-called Beacon Hill Study of 2013 expanded the study period to 5 days, let patients leave the hospital and roam central Boston (with nurses shadowing them), and observed how patients fared with the bionic pancreas compared with a similar 5 days under their own care.7 After giving the system 1 day to adjust to each patient, mean glucose levels fell from 159 mg/dL when patients cared for themselves to 133 mg/dL when they used the bionic pancreas. The percentage of time patients had glucose levels below 70 mg/dL fell from 7.3% to 4.1%.

The summer camp study of 2013 tested the bionic pancreas against normal care for 5-day periods in 16 boys and 16 girls, 12 to 20 years of age.7 Mean glucose levels for the children, who were all attending special summer camps, were 158 mg/dL under normal standards of care and 142 mg/dL with the bionic pancreas (after the 1-day adjustment period). The time spent with low glucose levels fell from 7.6% to 6.1%, and the frequency of interventions for hypoglycemia fell from 1 per 0.8 days to 1 per 1.6 days.

The development team is currently analyzing results from 2 more studies: a 2014 summer camp study that tested the bionic pancreas on children as young as 6 years of age, and a 2-week long home trial in 48 adults at 4 sites around the country: MGH, Stanford University, the University of North Carolina at Chapel Hill, and the University of Massachusetts.

The home-use trial places a few restrictions on participants—like asking them not to leave the metropolitan area around each participating hospital—but otherwise asks them to live normally. Researchers do not receive any information from their devices during the course of a study unless a machine breaks or the patient’s blood sugar gets dangerously low.

Until very recently, the bionic pancreas from Boston had a major weakness that struck some observers as the sort of critical flaw that might prevent it from ever achieving widespread usage: every commercially available glucagon formulation deteriorated so rapidly that users would have to mix it up and restock their machines every day. In recent months, however, a company called Xeris has begun providing the project with an experimental nonaqueous form of glucagon8 that lasts for months and has performed, in tests to date, almost identically to fresh glucagon from Eli Lilly. Now, Russell said during his presentation, several other companies are working on similar formulations, so users of the artificial pancreas should benefit from price competition. Russell believes that the advent of a more stable form of glucagon makes the use of the hormone an unquestionable advantage for his team’s bionic pancreas.

Systems that primarily use insulin have no way of raising blood sugar levels, which often fall below the desired level when people exercise, accidentally use too much insulin, or go too long without eating. Those single-hormone systems, therefore, must sound an alarm when glucose levels begin to fall and urge users to eat some form of sugar, even if it means waking them up from a sound sleep. The bionic pancreas simply releases a bit of glucagon into the bloodstream, and the user never knows. Indeed, Russell said, the 2-hormone design of his team’s device is a necessary precondition for making it operate like a real pancreas, which also secretes both insulin to lower blood glucose levels and glucagon to raise them.

Another potential concern that came up during the question-and-answer session following Russell’s presentation was the risk that users would develop tolerance to glucagon if they received dozens of small infusions of the hormone every day.

Russell said that patients who participated in trials so far had shown no signs of developing any such tolerance. He then added that while future trials would certainly look for indications that people might develop glucagon tolerance over longer time frames, the chance of such a problem occurring appears remote, both because the body typically uses that hormone to boost blood sugar over the course of a lifetime and because of the very small doses involved.

Russell went on to note that the bionic pancreas typically uses far less insulin per day than patients use on their own, and when raising glucose levels, it typically uses far less glucagon than patients use sugar. Better still, he said, patients who use the bionic pancreas may well spend so little time with low blood sugar levels that they regain the hypoglycemic awareness that people with T1DM typically lose, thus developing an extra safeguard against the condition.

To date, bionic pancreas systems have worked reliably, and the team behind the technology believes not only that their creation will prove durable in real life, but also that it employs enough safeguards to give patients plenty of warning should any problems arise. The only real danger, Russell said, lies in the calibration of the glucose monitor: if patients calibrate it badly enough, they risk problems.

Russell concluded his presentation by addressing 1 other serious concern that haunts every potential medical breakthrough: price. Russell predicted that the bionic pancreas will likely cost more than a top-of-the line insulin pump—but not that much more. He also predicted that every insurance program would happily cover the device.

“It will incorporate 2 pumps, the CGM, the bionic pancreas algorithm, and a dual infusion set. We don’t know what it will be priced at, but I would guess not double what an insulin pump costs,” said Russell, who added that the cost to most patients would be low or nonexistent because insurers would have so much incentive to promote usage.


“The biggest cost of diabetes care is treating the complication. There also is a tremendous cost of hospitalizations due to severe hypoglycemia,” Russell said. “We believe the system would be able to eliminate almost all those hospitalizations and reduce other complications by 80% to 90%, or even more. We think this is going to be the future of diabetes care.” References

1. Russell S. A bionic endocrine pancreas for automated management of glycaemia in diabetes mellitus. Mercodia website. Accessed October 28, 2014.

2. The discovery of insulin. website. Published February 2009. Accessed October 28, 2014.

3. Type 1 diabetes. The New York Times. Accessed October 28, 2014.

4. Berg E. The artificial pancreas aces new tests. Diabetes Forecast website. Published March 2014. Accessed October 28, 2014.

5. El-Khatib FH, Russell SJ, Nathan DM, Sutherlin RG, Damiano ER. A bihormonal closed-loop artificial pancreas for type 1 diabetes. Sci Transl Med. 2010;2(27):27ra27.

6. Russell SJ, El-Khatib FH, Nathan DM, Magyar KL, Jiang J, Damiano ER. Blood glucose control in type 1 diabetes with a bihormonal bionic endocrine pancreas. Diabetes Care. 2012;35(11):2148-2155.

7. Russell SJ, El-Khatib FH, Sinha M, et al. Outpatient glycemic control with a bionic pancreas in type 1 diabetes. N Engl J Med. 2014; 371:313-325.

8. Xeris Pharmaceuticals, Inc. announces dosing of first patient in phase 2 clinical trial of its investigational soluble glucagon for the treatment of mild-to-moderate hypoglycemia [press release]. Austin, TX: Xeris Pharmaceuticals; June 5, 2014.