Artificial Pancreas is a technology wherein there is a system to administer insulin that can be automatically controlled by concomitant blood glucose measurements. Such a system is the closest that we can get to an endocrine pancreas and has a huge potential in reducing the burden of self-care as well as chronic complications associated with diabetes.
Diabetes is one of the oldest known diseases with earliest descriptions in both Indian and Egyptian texts dating back to the BC era. For a disease known since antiquity, there were no treatment options till the discovery of insulin by Banting and Best in 1921. Insulin discovery was hailed as a ‘miracle cure’ for diabetes but once the lifespan improved for diabetic patients with insulin treatment, physicians started to notice microvascular (retinopathy, nephropathy, neuropathy) and macrovascular complications (stroke, coronary artery disease) related to chronic hyperglycaemia. For pessimists, “Insulin is not a miracle cure folks; it lets you to
be a diabetic a little longer!” For optimists though, there is good quality evidence for reduction in both microvascular and to some extent macrovascular complications (legacy effect) with better glycaemic control both in Type 1 and 2 diabetes. However, intensifying insulin treatment to achieve better control of diabetes has an inherent risk of severe hypoglycaemia. In fact, the fear of severe hypoglycaemia is a principal deterrent for patients to adopt an intensified insulin therapy. Insulin therapy by subcutaneous injections, though lifesaving, is not a direct substitute to the body’s inherent physiological responses that involve highly evolved coordination between the pancreas, liver and peripheral tissues with moment-to-moment changes in insulin secretion resulting in very stable blood glucose levels. The endocrine pancreas secretes insulin continuously with rapid peaks during meals, and insulin
levels are automatically regulated by blood glucose levels. Thus, the Holy Grail for patients with Type 1 diabetes would be the development of an automated and accurate closed-loop system for insulin delivery based on accurate blood glucose monitoring, the so called ‘artificial pancreas’. Such a system would be the closest that we can get to an endocrine pancreas.
‘Artificial pancreas’: Goals, components and difficulties
The goal of ‘artificial pancreas’ is to control blood glucose within a narrow range, similar to what a pancreatic islet cell can achieve. To replicate that level of exquisite control, an artificial pancreas requires three principal components; an insulin pump, continuous glucose monitoring system and a mathematical control algorithm that adjusts the insulin infusion rate based on the glucose levels. It is interesting to note that the initial attempt at artificial pancreas was made in 1964. This further led to the development of ‘biostator’ in 1977, a device that combined continuous glucose monitoring with computer based mathematical algorithms to direct intravenous insulin delivery. This device was mainly used in a research setting as there were many limitations that made it impractical for home use.
Over the last decade, there have been significant advances in insulin delivery systems, especially insulin pumps and continuous glucose monitoring sensors (CGMs). The main limiting factor is the accuracy of CGMs, as precise glucose values are mandatory for safe, and efficacious closed loop control. As the sensors become more accurate, an optimal mathematical control algorithm will be needed to deliver the right amount of insulin. To add to the above challenges, the routes of glucose monitoring and insulin infusions need to be refined. There are two routes: the first model is based on using an intravenous glucose monitor and an intraperitoneal pump, and the second model uses a subcutaneous glucose monitor and a subcutaneous insulin pump. Each method has its advantages and disadvantages.
In terms of appeal, the latter approach is minimally invasive and the main components, an external insulin pump and the subcutaneous glucose monitoring system are in use anyway. But the main disadvantage is the inordinate time delay between insulin delivery, action and feedback through interstitial glucose measurements. To some extent, the above factors can be taken into account by the mathematical algorithm models. But there are a lot of other confounding factors such as exercise, subcutaneous blood flow and injection site issues like lipohypertrophy. On the other hand, the intravenous glucose monitor and intraperitoneal pump model has the advantage of reducing the time lag for plasma insulin metabolism in the liver. The main disadvantages are
intraperitoneal pump site infections, and clots over intravenous glucose monitors that can lead to embolic events and pump