FDA Wants Artificial Pancreas Approval

An artificial pancreas, as commonly reported, is a medical device that is worn and attached to a person.  It autonomously changes the insulin administered to a person with diabetes in response to glucose measurements, which are measured automatically.

The system has a few basic components that include a continuous glucose monitor (CGM), a continuous subcutaneous insulin pump and a control system that determines the rate of insulin infusion given the measured glucose through time.  An artificial pancreas will also include a display, communication system, overrides, alarms, fail safes, and controls for user input (i.e., for indicating meals, exercise, etc).  However, the fundamental components are the three I mentioned.

Despite decades of research, the three basic components as a whole may not be ready for complete autonomous application.  Insulin pumps are quite reliable and control algorithms have reportedly been developed that effectively achieve a target range while avoiding hypoglycemia.  However, continuous measurement of blood glucose through current subcutaneous sensing technology has proven to be quite challenging .

That said, researchers are now entering what appears to be the final stretch to bring an artificial pancreas (or a system that at least reduces insulin in response to low glucose) to the FDA…. And, we want it:

Teenaged delegates from the Juvenile Diabetes Research Foundation testified on Capitol Hill, urging Congress to accelerate research and review of artificial pancreas systems for managing insulin for patients with type 1 diabetes.

“After participating in clinical research since I was three years old, I can honestly say the closed loop artificial pancreas trial was the most amazing experience of my entire life and holds so much promise for people living with this disease,” said Kerry Morgan, a 17-year-old JDRF Children’s Congress delegate from Glen Allen, Va., who testified before the Senate Committee on Homeland Security and Government Affairs this week.

Would you be surprised to learn that the FDA wants it too:


Charles “Chip” Zimliki, chairman of the U.S. Food and Drug Administration’s Artificial Pancreas Critical Path Initiative, which was created in 2006 to accelerate the availability of an artificial pancreas system, says he is eager to have a system approved.

“The FDA wants the artificial pancreas on the market as much as anyone else does. We just have to operate within U.S. laws to make sure it is safe and effective,” Zimliki said.

Last week, the agency released guidance for how to develop a low glucose suspend system, an automatic shut-off mechanism used with an insulin pump. Medtronic already sells pumps with this the feature in Europe. It safeguards against a dangerous drop in glucose levels by temporarily halting glucose delivery.

By year-end, FDA plans to release detailed guidance on more complicated closed-loop systems, Zimliki said.

“We think of this system, the artificial pancreas, as one unit. There is going to have to be agreement among various companies to determine who is the reporting party for submitting it,” he said.

“That is a relatively new idea with respect to these systems.”

Zimliki, who is a type 1 diabetic, thinks the first approved devices will be ones that deliver insulin only, but he is very encouraged by the system being developed by the team at Boston University and Massachusetts General.

“They have what I call the Cadillac of closed-loop systems,” he said. In addition to delivering insulin, the device also delivers an infusion of glucagon, a hormone released by the pancreas to raise blood sugar levels.

“They are showing some very promising results,” he said.

See the JDRF more more information on this project and the latest FDA guidance.

Diabetes Mine has an interesting post and interview with the principle investigators from Boston University and Massachusetts General Hospital (see the youtube video).

Low Glucose Suspend – FDA Guidance

While complications of diabetes are understood and have been tied to the cost of providing healthcare, trends indicate the incidence Diabetes Mellitus in the United States and other developing countries is growing at an alarming rate.

There is so much about diabetes that is understood, predicted and recommended and yet management continues to be a challenge.  Could it be that better tools are needed?

To meet the challenge, researchers have been developing technology based tools that will help manage the disease through automation. That is, automatically administering an appropriate amount of insulin in response to a person’s glucose level and carbohydrate intake.

Often called the “artificial pancreas”, such systems combine continuous glucose measurement systems (CGM), insulin pumps (giving continuous subcutaneous insulin infusion) and advanced algorithms to give insulin dosing recommendations and stop infusion when a hypoglycemic event is predicted. The latter is called a “low glucose suspend” (LGS) device and provides benefit as a result of its autonomous action aimed at avoiding low blood sugar (predictively) or reducing the impact of hypoglycemia in a reactive manner.

Yesterday, the FDA release a new guidance document that will help medical device manufacturers submit their artificial pancreas-like product for review. The move is encouraging for a number of reasons. First, the agency is agreeing that automation has a role in the marketplace and is encouraging a path forward. Second, progress has been made and interest expressed to the point that the FDA felt is necessary to invest in guidance.

On the other-hand, there are a number of challenges the agency suggests must be remedied involving CGMs:

  • Using the same CGM to measure success and to make decisions about if and when to turn the pump off will introduce bias. Although the size of the bias may or may not be large, determining the extent of the bias will be impossible without an independent measure.
  • Although CGMs have been successful in improving diabetes management through their tracking and trending functions, these devices have not been shown to be accurate enough to support use for insulin dosing.
  • The glucose meters used to calibrate the CGMs also have inaccuracies that can compound the errors in the glucose values reported by the CGM and are part of the device system.
  • Use of retrospective signal calibration using reference blood glucose values or introducing a reference method to be performed by the patients may be possible solutions if the approach is appropriately validated.
  • CGMs have periods of sensor irregularities and signal drop out. These sensor performance problems arise in addition to sensor accuracy challenges and would need to be resolved and/or mitigated.

But are they effective enough to “pause” an insulin infusion?  Perhaps, but what if users begin to rely on a “pause” as their safety net even though CGMs have the issues listed above?

From an altruistic standpoint, advances in this area will help those suffering from Diabetes and in particular individuals who are unable to properly treat their disease (esp. children). From a business stand-point, a revolutionary product that is intuitive, effective and safe will provide a huge advantage in a market that seems stuck on the stick meter.



Artificial Pancreas and Closed Loop Control – is it really that far off?

For many years academic researchers and device companies alike have investigated, developed and studied that artificial pancreas.   The system isn’t really a pancreas per se but rather a multi-component system that  continuously measures glucose (via a subcutaneous sensor), uses a computerized closed loop controller to determine the proper insulin infusion rate and and then automatically adjusts insulin delivery via a subcutaneous insulin pump.

Many of these automatic feedback control systems are in clinical trials and there has been enthusiasm generated by NIH and JDRF sponsored research in this field.

I don’t think the artificial pancreas is far off at all, although the first generation may not be the fully automated high performing system that we dream about.  Insulin pumps are reliable and sophisticated control systems have been developed that optimize insulin delivery decisions using a personalized predictive model (i.e., determines future glucose response via an adaptive nonlinear PK/PD equation).   The weak link, in my opinion, is the glucose sensing system.  Continuous glucose monitors have improved but may not be at the point were tight glycemic control is possible.  However, some level of autonomous action does seem feasible – today.


A Non-invasive Glucose Sensor?

non-invasive-glucose.jpgAnother report of the long sought after non-invasive glucose meter… this one is the size of a cell phone and will be available in a year:

Hong Kong scientists have invented a device to help diabetics measure their blood sugar painlessly for the first time – without pricking their fingers.
The size of a mobile phone, the instrument emits a weaker form of infrared, or near-infrared, which penetrates the skin on the finger and homes in on the bloodstream.
Out of the many components in the blood, the beam is able to identify bits of glucose through the frequency, or wavelengths, they transmit and the amount of blood sugar present would be displayed on the instrument in 10 seconds. 

Considering the near-infrared train wrecks that have preceeded this announcement, the device (if real) is quite an accomplishment. Unfortuantely, there is precious little information available regarding the system although their patent suggests the device does the following:

To predict the blood glucose level of the person using the NIR spectral scan an evaluation model was found based on a population of test subjects. The evaluation model relates NIR absorbance of blood vessels at a first and a second group of wavelengths to a glucose level of the blood vessels . The graph in figure 2 shows the relationship between the NIR absorbance of blood of a body part of the test subjects and the mean value of laboratory tested glucose levels of the test subjects at three sample wavelengths of 940nm, 1310nm and 1550nm. The reference glucose levels are obtained using proven standard laboratory tests on blood samples from the test population. The linear relationship is found using a least squares method. The standard deviation on the mean value of the tested glucose levels for the test population was 10% to 20%. 

Quite remarkably, the measurement system claims to use only six wavelenghts to measure glucose, a trace constituent, amid a background of interferring analytes.
I’m sorry, but I’m not going to hold my breath on this one but will certainly wish the scientific team “good luck, your technology will provide a substantial benefit to glucose testers.”