Achieving a Closed-Loop Artificial Pancreas

Home Clinical Background Current Approaches Design Criteria References

by Kameel Abi-Samra (BME 240, Spring 2009, UC Irvine)

Current Approaches

Surgical Approaches      Biological Approaches     Mechanical Approaches



Surgical Approaches

  • Whole Pancreas Transplantation

  • 400 - 500 whole pancreas transplantations performed every year in the United States.

  • Only recommended by doctors for patients with  hypoglycemia unawareness and severe glucose instability.

  • Recipients must take immunosuppressants for all their life

  • Less than 50% graft survival rate after 10 years

  • Conclusion:  Not currently a good approach because of its poor survival rate, immunosuppressants required, and difficult availability of an organ donor.

  • Islet Cell Transplantation

  • A less complicated surgical procedure than Whole Pancreas Transplantation

  • 225 patients received this treatment between 1999 and 2005

  • 75% of recipients of this treatment no longer produce insulin after 2 years.

  • Immunosuppressants are required but they are considered to be not as strong as those required for a patient who had a whole pancreas transplant.

  • It is thought that the implantation failure has to do with the same autoimmune response that destroyed the Beta Cells of the patient as a child.

  • Conclusion: A good approach in theory, but maybe limited by the patient's autoimmune response and the rate of success is currently too low.

  • Islet Microencapsulation (Immuno-isolation)

  • Basic idea is to put islet cells into microcapsules that have pores too large for cellular or humoral immunity to pass through, but still large enough for insulin to pass outward and glucose inward. Thus, immunosuppressants would not be necessary.

  • In a study published in 2009, pig islets were immuno-isolated  and implanted in to spontaneously diabetic dogs and the researchers observed a 20-80% decrease in insulin needs 6 to 12 months after transplantation

  • In a study in 2001 performed in Mexico neonatal porcine islets were combined with Sertoli cells and they were placed in a subcutaneous autologous collagen-covered device which was implanted into 12 subjects. Four years after implantation, half of the subjects showed a significant decrease in their insulin needs.  Six years after implantation the patients still showed tolerance towards the implantations.

  • Conclusion: A very promising approach that gives acceptable results; however, this approach will still require much more research to optimize and implement commercially.

Biological and Tissue Engineering Approaches

  • Stem CellsStem Cells induced to release Insulin

  • Researchers have been able to induce stem cells in vitro to release insulin in response to ambient glucose concentrations.

  • This approach has not yet been tested in vivo.

  • Our understanding of stem cells is still very premature to be considered a good approach.

  • Conclusion: Approach could lead to promising results, but is still very poorly understood and the behavior of insulin induced stem-cells is still very premature.

  • Interruption of the Autoimmune destruction

  • Because Type 1 diabetes is simply due to the autoimmune destruction of the B-Cells researchers have tried to interrupt the autoimmune destruction as soon as it is visible.

  • Proof of principle report in 2002 started a series of researchers looking into this approach.

  • No successful trials to date.

  • Conclusion: A novel idea, but like the Stem Cell approach it is very premature and has yet to have any real results.

  • Insulin Delivery with a Glucose-Responsive Matrix

  • Goal is to develop a hydrogel that releases insulin by responding to ambient glucose concentrations.

  • Constructs being explored basically are made up of a lattice of glucose oxidase, that acts as glucose sensor, and a pH sensitive hydrogel. When the construct reacts with glucose the pH is lowered because glucose oxidase converts glucose to gluconic acid. Then the insulin is released by diffusion.

  • Recent  promising progress in terms of biocompatibility, selectivity, and pharmacokinetics

  • Current challenges include: glucose sensor specificity, biocompatibility, toxicity, and in vivo application.

  • Conclusion: Very interesting and unique approach that could lead to promising results with continued research. It will be many decades before this approach would be medically available because of the many challenges it faces.

Mechanical Approaches

  • Basic approach is to have a:

1) Glucose Sensor

2) Insulin pump

3) A set of algorithms that relate the glucose sensor to the insulin pump

         Having all three of these elements in a long-term implantable device is considered to be a "closed-loop" system, the holy grail of battling Type 1 diabetes. The insulin pumping mechanism has been well worked-out and is not really considered a "challenge". In fact, implanted insulin pumps currently can last up to five years. The following is an external link comparing the currently available external insulin pumps on the market in the U.S.

Comparison of Insulin Pumps

        The real problem lies in determining the set of algorithms that synchronizes the glucose sensor with the insulin pump as well as creating an implantable continuous glucose sensor. It is well agreed upon that there is not a linear relationship between the glucose concentrations and the amount of insulin that Beta cells release. There is a whole field of research dedicated to determining the algorithms and they will without a doubt improve with time. 

        Achieving a continuous implanted glucose sensor is another feat of its own. There are currently no available long-term implantable continuous glucose sensors. Currently the longest life span of an FDA approved subcutaneous implantable glucose sensor is at 2 weeks and one is produced by DexCom ( and another MiniMed Medtronic ( Because these subcutaneous implantable glucose sensors are based on the enzyme glucose-oxidase their lifespan is currently limited to around two weeks, after which the enzyme has been degraded too much to be considered accurate. In addition, these sensors are no where near ideal. They must be calibrated 4-6 times a day by the patient taking an "old fashioned" blood prick. Moreover, these are subcutaneous glucose sensors and, thus, they are not truly a fully implantable sensor.  Currently the MiniMed Paradigm REAL-Time System is the only commercially available system that integrates an insulin pump with real time continuous glucose monitor. The following is a diagram of the MiniMed's Paradigm REAL-Time system:

Conclusion: Currently the most common commercial approach to this problem. While the development of an insulin pump is considered a completed task, we are still far from having a long-term implantable glucose sensor, and having a set of algorithms that fully relate the insulin pump to the glucose sensor. Optimizing this approach might be the most commercially successful approach.


Next Section: Design Criteria