Technical Challenges in Developing an Artificial Pancreas Device System (APDS)
Recreating the precise and dynamic glucose-regulating function of a healthy pancreas is a challenging task. This task is further complicated by the numerous behavioral and biologic factors that impact diabetes management, and by the current limitations of the devices that are components of the artificial pancreas device systems being developed.
Current FDA approved or cleared diabetes management devices used as intended have dramatically changed the quality of life for people with type 1 diabetes.
Researchers have made significant progress towards combining these devices into a more dynamic system and the FDA will continue to prioritize the development of an APDS. Here are some examples of areas where more research will help advance the development of these important devices:
Continuous glucose monitoring systems (CGMs) incorporate sensors that measure glucose levels and change in glucose levels in the fluid around the cells (interstitial fluid). These levels may differ from the levels of glucose circulating in the blood. As of today, the FDA has not approved CGM values alone to determine insulin dosing.
CGM sensors often build up organic matter on their surfaces (biofouling) or in close proximity (fibrosis) once they are inserted under the skin. Biofouling or fibrosis could potentially impact the effectiveness and implantation duration of the sensors, which may lead to inaccurate readings or cause sensors to stop working.
More research and advances in sensor technology will help continuous glucose monitoring systems more accurately and quickly measure blood glucose levels and resist biofouling and fibrosis.
Improvements in blood glucose device technology can sharpen the accuracy of the overall APDS. For example, it is possible for blood glucose meter readings to be influenced when a patient is sick or taking other medication. Research in this area could improve these readings.
Blood glucose monitors are also used to calibrate Continuous Glucose Monitors. The more accurate the blood glucose monitor, the more accurate the CGMs.
Improvements to the test strips could make them less susceptible to storage conditions and to variations in how different patients take and test their blood samples.
The FDA has identified problems with the mechanical components and software of many insulin infusion pumps. These problems have led to improper insulin dosing and compromised patient safety.
These known problems with infusion pump software present challenges to creating the computer programs that connect insulin infusion pumps and CGMs in an artificial pancreas. The FDA has taken steps to bring safer infusion pumps to market, but more research and innovation would improve the overall safety and effectiveness of the pumps, and thus of any APDS.
The insulin currently used in APDSs takes hours to completely absorb, and absorption rates can vary among patients as well as for the same patient throughout the day. Creating a control algorithm that allows the CGM to take both these factors into account is difficult. Thus the development of faster-acting insulin would compress the variability of bolus insulin absorption and improve the effectiveness of APDSs by enabling them to calculate more effective insulin doses.
There is no FDA-approved glucagon formulation that can be used in an APDS outside of a research study. There is a need for a glucagon formulation that is stable over several days and can be delivered by an infusion pump.
Further development of control algorithms could lead to safer, more reliable and robust APDSs. For instance, adding redundancies in the control algorithm can help the system check itself and prevent dosing errors. In the absence of a fast-acting insulin, a control algorithm may need to address the variability of insulin absorption. In addition, algorithms that can allow for variations between patients by incorporating additional physiologic measures, such as physical activity, could lead to more precise control of blood glucose.
Control algorithms can reside on any number of devices that are able to receive input from the continuous glucose monitor and deliver commands to the insulin pump. These could include the pump itself, or other devices such as a computer or cellular phone. The communication between these devices should be evaluated based on a risk analysis that incorporates all of the potential hazards that could affect the communications between the various components.