Related Documents
Hovorka et al., Manual closed-loop insulin delivery in children and adolescents with tpe 1 diabetes: a phase 2 randomized crossover trial. The Lancet. February 5, 2010.
The Artificial Pancreas is in the news, I think mostly because it has become a high profile cause of the JDRF. The concept is old. I remember when writing “Prospects in Diabetes Therapy” in 1980 I interviewed Dr. Robert Fischell at the Johns Hopkins Applied Physics Lab. He was working on an artificial pancreas (then called the ‘closed loop pump’) and predicted it would be ready in five years. Dr. Fischell is credited with inventing the implantable insulin pump, but he could never get the artificial pancreas to work.
This artificial pancreas concept is to take the readings from a continuous glucose monitor, and use an insulin pump to mimic the insulin secretions of islets of Langerhans.To me, this is like correcting nearsighted vision using a camera and TV monitor rather than eye-glasses. Compared with islet replacement, the artificial pancreas is complicated and distorting.
The distortion is mainly a matter of kinetics: the artificial pancreas is simply not fast enough to perform effectively. A healthy islet senses blood composition (glucose and other metabolites and hormones), integrates the information, and makes a determination of how much insulin to secrete. The reaction time is about one minute. In contrast, a glucose sensor is a few minutes behind, because it senses sugar in the fluid between cells and not in blood. That in itself is manageable. The big problem is that the insulin pump secretes insulin into the skin,and even the most rapid insulin takes about an hour to peak in the blood. So, for an artificial pancreas, the time from glucose to insulin action is an hour. And that is the problem.
The only technical fix would be to deliver insulin to circulation. But an indwelling vascular catheter is too dangerous to be used to treat diabetes.
Therefore, if the artificial pancreas is ever to work properly, the software algorithm connecting the sensor and the pump must compensate for the deficiencies of the sensor and the under-the-skin insulin delivery. If it simply follows the “program” of the islet and injects a lot of insulin when the glucose starts to rise the blood sugar is likely to plunge dangerously in two hours.
A study recently published in The Lancet carefully explored the use of closed loop technology in a hospital setting overnight. It was essentially a simulation of an artificial pancreas by a nurse. Every fifteen minutes the nurse entered the blood glucose reading into a computer. The algorithm computed the new insulin infusion rate, and the nurse would then adjust the infusion rate of the insulin pump.
The role of the nurse could be instantly automated. So why have the nurse involved rather than a true artificial pancreas? To protect the patient from insulin error. We’ll return to this safety concern.
[FIGURE 2D]
Figure 2D shows the results of the algorithm. The red line shows measured glucose at 15-minute intervals, and the blue line,the computed insulin infusion rate. Recall that the effect of insulin peaks an hour after injection. In the figure an increase in insulin infusion near 2100 results in a decline in blood glucose near 2200; a decline in insulin infusion near 2300 results in an increase in blood glucose near 2400. It is clear that the algorithm considers both glucose level and rate of change.
The primary end point of the trial was overnight glucose. How much of the time is it in the target range (dashed lines)?
Figure 4 shows the answer. This is a complex figure – start with the table at the bottom. The target range is 3.9 to 8 millimolar (equal to 70-144 mg/dL). The table is:
| Time in glucose range | 0-3.9 mM | 3.9-8 mM | > 8 mM |
| 0-70 mg/dl | 70-144 mg/dl | > 144 mg/dl | |
| Closed loop insulin infusion | 7% | 66% | 27% |
| Open loop insulin infusion | 19% | 39% | 43% |
This is impressive: the changes in insulin infusion rates – that is, changing from open to closed loop – increases the time in the target range of 70-144 mg/dL from 39% of the time during the night to 66% of the time. Perhaps more important is that the time of hypoglycemia drops from 19% of the time during the night to 7% of the time.
The curves in the top part of the figure pack in a lot more information. The blue curve shows the distribution of blood sugars overnight for a open-loop system and the red for a closed-loop system. Clearly any diabetic would prefer the red curve. Note in particular the blow up on the right, expanding the hypoglycemia domain. The closed loop pump nearly eliminates blood sugars between 2 and 3 mM (54-36 mg/dL), the more severe hypoglycemia.
So, from a believer in islet replacement, one cheer for the artificial pancreas.
Here are some of my reservations. Because the number of patients is small, the results are only borderline significant. But figure 4 is pretty persuasive to me. The setting, a research hospital, is artificial and more structured than real life. A big drawback is the study period is overnight and does not include meals, which would overpower this algorithm. Clearly the system would require an estimate of calories consumed; information from blood sugars alone would not be sufficient. Thus success can only be claimed for nocturnal hypoglycemia prevention.
Perhaps my biggest reservation is the fact that the artificial pancreas is being promoted by the nonprofit JDRF after being rejected by the for-profit medical device industry.
Al Mann is a successful medical device entrepreneur. His company Minimed spent hundreds of millions of dollars to develop a commercially competitive artificial pancreas. He sold Minimed to Medtronic for $3.7 billion in 2001. Based on Minimed’s insulin pump sales,financial analysts considered the price to be much too high; the only explanation for the price was that Medtronic considered Minimed to have a commercially viable artificial pancreas and was paying a high price for that. As of 2010 it has not been commercialized. Therefore by any standard Medtronic paid too much for Minimed.
I am not aware that Medtronic has ever said so publicly, but my sources tell me that the reason the artificial pancreas was not launched is that the liability is too great. There is a significant probability of the artificial pancreas permitting or causing serious hypoglycemia. If an artificial pancreas user died or suffered from hypoglycemia-induced damage, the manufacturer would likely be subject to a lawsuit and might well pay a large settlement or judgment. I am told that the risk was judged too great and the Minimed artificial pancreas did not go to market. Medtronic now seems interested in the “semi-closed loop” device which would function by stopping insulin in response to low blood sugar only. I suppose that preventing hypoglycemia is not burdened with litigation risk.
Here is a summary of the situation in 2010. There are commercial insulin pumps with increasingly sophisticated programming capabilities. And there are continuous glucose monitoring devices with increasing accuracy and ease of use. And the gap between them to complete the artificial pancreas is a computer algorithm – and it is JDRF, a nonprofit – that is attempting to fill the gap. Could it be that one day might you purchase a pump from one manufacturer, a meter from another, and a device from JDRF that communicates with both? Is the JDRF protected from litigation by its nonprofit status? It is a curious anomaly that the conjunction of FDA policy, American liability law and the peculiarities of type 1 diabetes means that a therapeutic device cannot be sold by a for-profit company. It will instead by sponsored by a nonprofit entity. Insulin pump companies must be very pleased.
When I was first diagnosed with type 1 diabetes in 1966, during my patient education classes at the Joslin’s Clinic in Boston we were urged to tolerate the admittedly severe demands of strict glucose control because “in the next 5 years an artificial pancreas will be available to mimic the actions of a real pancreas, so you don’t want any damage before then.” Of course, we are still waiting.
While I agree with all your concerns about the closed-loop pump, I would add one further consideration. Of all the devices which have ever come into therapeutic practice in the history of medicine, no artificial mechanism has ever performed well in imitating the natural, organic processes of the human body. The iron lung provided polio patients with a life which amounted to a living death, as does the modern hemodialysis machine, which gives its users in the United States a 23% death rate per year, perpetual exhaustion, and a host of medical complications, all in addition to the 20% of patients who eventually opt for death rather than continuing with the dialysis regimen. The results with the very different approach to replacing natural renal clearance of toxins with peritoneal dialysis are similarly dismal. The artificial heart of the early 1980s essentially tortured poor Barney Clark to death, and when the details of the kind of ‘life’ it offered him were revealed, there was a public scandal over the unethical nature of the experiment. All of this experience should teach us one thing — that the solution to an organic problem must itself be organic if it is to succeed.
I agree that there are steps that still need to be overcome and brought to market before an ideal true closed loop system can be achieved. Without glucose sensors that can detect current glucose in the blood system within a few minutes and tranfer that to the insulin pump, and then probably a faster acting insulin that reaches the blood quick, peaks within minutes and then dies out soon after, and of course a few more safety protocols added to the pump, I think a true closed loop insulin pump system would be possible.
DZ-
You are exactly right. But I am told by insulin people that the absorption is limited by how fast it gets into the blood from the fat depot, and significant improvement is not possible. Let’s hope they find a way to make it faster.
When I was first diagnosed with type 1 diabetes in 1966, during my patient education classes at the Joslin’s Clinic in Boston we were urged to tolerate the admittedly severe demands of strict glucose control because “in the next 5 years an artificial pancreas will be available to mimic the actions of a real pancreas, so you don’t want any damage before then.” Of course, we are still waiting.
While I agree with all your concerns about the closed-loop pump, I would add one further consideration. Of all the devices which have ever come into therapeutic practice in the history of medicine, no artificial mechanism has ever performed well in imitating the natural, organic processes of the human body. The iron lung provided polio patients with a life which amounted to a living death, as does the modern hemodialysis machine, which gives its users in the United States a 23% death rate per year, perpetual exhaustion, and a host of medical complications, all in addition to the 20% of patients who eventually opt for death rather than continuing with the dialysis regimen. The results with the very different approach to replacing natural renal clearance of toxins with peritoneal dialysis are similarly dismal. The artificial heart of the early 1980s essentially tortured poor Barney Clark to death, and when the details of the kind of ‘life’ it offered him were revealed, there was a public scandal over the unethical nature of the experiment. All of this experience should teach us one thing — that the solution to an organic problem must itself be organic if it is to succeed.
Hi Steve–
I have never thought about it is such global terms, but I agree. I once read that if the NIH had set out to cure polio it would have come up with a portable iron lung.
I remember going on the “Biostator” a sort of primitive artificial pancreas, probably in 1979. My blood was pulled out through a cannula to get continuous blood sugar and a computer ran an insulin pump. After a couple of hours I was pulled off the machine and my blood sugar was uncontrollable for a day. Dr. Drexler (that’s when I first met him) said that was typical of the Biostator.
I’ll take a cell therapy over any mechanical device.