Medical Device Daily Washington Editor

The artificial pancreas may never arrive, especially if stem cells and gene therapy ever make serious inroads on modern healthcare, but a number of parties are pressing forward with effort to put a device into play that will do a reasonable job all the same.

Earlier this week, FDA, the National Institutes of Health and the Juvenile Diabetes Research Foundation (JDRF; New York) hosted a symposium titled "Toward an Artificial Pancreas" that highlighted the progress to date and described the considerable terrain yet to be covered.

One of the recurrent ideas heard during the two-day gathering was that of a fast-track approach at FDA toward passing a premarket approval application for an artificial pancreas. But fast-track authority has boomeranged on the agency several times over the past decade, including a fast-track approach to anti-carcinogenic drugs.

Whether the agency is willing to risk more blistering attacks from Capitol Hill and the nation's major media outlets and bloggers remains to be seen, but any reticence on FDA's part to fast-track an artificial pancreas would surprise few.

In kicking off the symposium's first day, William Tamborlane, MD, deputy director of the Yale Center for Clinical Investigations (New Haven, Connecticut), reminded the audience: "The whole idea of a closed-loop system was proposed back in the 1970s." He said the Biostator, an oven-sized glucose monitoring system designed to perform as a "glucose clamping system," was a serious systematic effort, but that it was ungainly and of very limited utility.

Tamborlane made reference to a paper he co-authored in 1979 that addressed the use of a portable infusion pump in juvenile diabetic patients, which showed that small, pre-meal doses of insulin could bring down the post-meal (post-priandal) gyrations in blood glucose by substantial margins.

The abstract from that article, which appeared in the March 15, 1979, issue of the New England Journal of Medicine, noted – perhaps more presciently than Tamborlane and his co-authors might have appreciated at the time – that "the feasibility and value of the long-term application of this technique need exploration."

Aaron Kowalski, PhD, research director at JDRF, discussed the current state of affairs in the push toward an artificial pancreas, opening by stating that "the first generation of the artificial pancreas is pushing in some of the insulin some of the time." He said that while this is a sign of progress, "what's not debatable is that there are some important steps needed to realize" the potential of such technologies.

Even though the "big goal is to walk away from an artificial pancreases and cure diabetes," Kowalski said, researchers have to make progress quickly. He said the automated closed loop "needs to happen soon," listing algorithms in need of refinement as one of the big jobs. However, he also made the case that FDA needs to put more effort into clearing a device, and sponsors need to put more effort into making these devices.

He said Type 1 diabetes "is beyond just insulin, but there are opportunities to get beyond just insulin regulation. "If we could incorporate closed-loop control" into other cellular approaches to a cure, "closed loop control could give us a better chance of succeeding at that."

Stuart Weinzimer, MD, associate professor of endocrinology at Yale University School of Medicine (New Haven), discussed studies of closed-loop systems in children.

"One of the mantras of pediatrics is that children are not small adults," he said, noting that children and adolescents "are at greater risk of complications of hyperglycemia and hypoglycemia." There is "also a higher psychosocial burden of care" in connection with pediatric diabetes.

Weinzimer said the algorithm he chose for several of the studies he presented was a proportional integral derivative (PID) controller, which "has been around for a long time."

A Medtronic (Minneapolis) PID study at the UCLA Medical Center, the first fully closed-loop study in humans, suggested that glucose control was fairly good with the PID, he said, but a rebound after breakfast was sharp enough to require a dietary adjustment. "Some of this rebound is due to persistent elevations in insulin levels," he said, which in turn is attributed to slower absorption of serum insulin.

Weinzimer presented the results of a Yale pediatric study that tracked 17 subjects in two different study arms for 36 hours to assess a closed-loop system versus one that allowed patient intervention, termed a hybrid glucose loop system. They were all on sensors that transmitted to the hospital every 30 to 60 minutes.

"Half [eight] of our subjects received a priming dose of insulin five to 15 minutes before the meal," he said, but "not enough for the meal, just enough to get some insulin on board."

The study used the MiniMed insulin pump made by Medtronic.

While nighttime control of glucose was almost identical in both groups, glucose levels in the hybrid patients were an average of 149 milligrams per deciliter (mg/dl), a sharp drop from the 159 seen in the controls. A greater absolute difference was seen in post-priandal glucose levels, 194 mg/dl for the study subjects and 226 for the controls. The admittedly small test gives researchers confidence that minor tweaks in the delivery of insulin might make significant differences in the day-to-day life of the Type 1 diabetic.

"We believe the better performance of the hybrid was due to getting the insulin in faster" in the pre-meal doses, Weinzimer said.