Advances in the Artificial Pancreas

by nick wilkie

Diabetes technology lovers! Have you ever wondered: why can’t my continuous glucose monitor (CGM) just tell my insulin pump when my blood glucose is high so that it gives me more insulin? CGM has been around for over a decade (in its earliest form), and pumps have been around 25 years, but automating the connection between them has been a work in progress. Since such a system would effectively represent a healthy pancreas with respect to glucose control, the goal has been termed “the artificial pancreas.” Unfortunately, creating the artificial pancreas has taken longer than most experts expected. Many difficulties are cited: delays in highly accurate CGM sensor readings (especially in hypogylcemia), rapid-acting insulin that just isn’t fast enough, delays in building workable algorithms, and individual biological variation, to just name a few.

In a recent paper published in The Lancet, Dr. Roman Hovorka (University of Cambridge) and colleagues tested the ability of a prototype artificial pancreas system that delivers insulin overnight to children with type 1 diabetes. The paper combines data from three studies, totaling 33 nights using the artificial pancreas and 21 nights using standard pump therapy. Despite the small size of the study, it has received significant attention in the media and in the medical community. The day after it was released, the paper was featured in the BBC and New York Times.

The major finding of the study was the reduction in hypoglycemia: the amount of time spent below the target range (hypoglycemic) was reduced by 50% in the artificial pancreas group compared to those in the standard pump therapy group. Furthermore, there were no incidents of severe hypoglycemia (defined as a blood glucose below 54 mg/dl) in the artificial pancreas group, while there were nine such incidents in the standard pump therapy group. Of course, nighttime hypoglycemia is a major fear for many patients and families, and any system that is able to reduce that risk would be welcomed. In addition to those findings, the study also found that patients using the system spent 60% of the time in the target range (normoglycemic) compared to 40% of the time for those using a standard insulin pump.

Dr. Hovorka’s group also tested the adaptability of the system, feeding the subjects rapidly absorbed (high glycemic-index) or slowly absorbed (low glycemic-index) meals before bedtime. The type of meal did not have a significant impact on the time in target range. Likewise, exercising patients before bedtime (a common cause of morning hypoglycemia) did not affect glycemic control.

These are impressive statistics, and after hearing about the artificial pancreas for years, it is wonderful to hear about the successes. That said, significant obstacles remain. Although nighttime use of the system is an excellent first step, glycemic variation during the day poses a far more complex problem. The eventual artificial pancreas must compensate for meals, exercise, and any other activities that could affect blood sugar in day-to-day life - difficult to achieve without some assistance from the user. However, we find the prospect of a “hybrid,” patient-assisted artificial pancreas very exciting as well, and this will come sooner. Effectively, this would require input from patients regarding food and exercise - but may well result in “smarter” equations that help patients reach lower A1cs more safety. We spoke to Dr. Hovorka after the publication of the article, and he said the next step is to test his system in patients outside the hospital setting - at home! We will keep you updated on when and where these trials take place - Dr. Hovorka said he expects that the first in-home studies will begin in the middle of this year. While we understand many patients have been frustrated by artificial pancreas timing and the seeming impossibility of reaching a perfect system, we do certainly feel that significantly more help for patients is on the way and that the vision for so many patients for so many years will, over time, ultimately become a reality.