Posted by Al Gordon on September 03, 1999 at 10:02:23:
In Reply to: Re: Closed Loop Electromechanical Pancreas posted by jeff on September 02, 1999 at 17:16:54:
Jeff,
I appreciate your well-thought out responses to the issues that I raised. I hate to start a lengthy back-and-forth on this one, but I remain skeptical on the prospects for a closed-loop electromechnaical pancreas in the next decade. Looking at your ideas:
1. Check out the Minimed page. They have a continuous monitor that was approved by the FDA for three days of usage. I don't know what technology they are using, but three days of test results(every five minutes) sounds pretty usable to me. And this is the first version; it will only improve.
As I understand it, the user is not able to see the readings, and make insulin adjustments based thereupon. The readings are stored and interpreted by a technician. Perhpas this is a precaution only during the testing phase of the device. Also, since this device requires that the user take action and change the interstitial probe every 3 days, the possibility of human error giving rise to false highs is very real. I believe that for a closed-loop system, the sensor technology will have to be optical, or some other method that does not consume an enzyme and does not require user intervention.
2. Single point of failure: as long as the blood sugar monitor is designed to give "no-result" or a spurious low-blood sugar and never a spurious high-blood sugar, there is no danger. Your blood sugar will just go high. (This is similar to the way pumps work now, designed so they never give too much insulin even when they fail - the worst that can happen is not enough insulin). And the monitor can beep or shut down if the results
seem inconsistent. I don't think implementing these safe-guards is a seriously hard programming effort.
Since you have a glucose monitor that must be accurate over a range of 30 to 500 mg/dl, how do you design it such that errors are always in one direction (never a false high)? The sensor has to know it has an erroneous reading before it can "fail safe". The only way it can know there is an error is if there is a second sensor, and the 2 do not agree. Now we are starting to accumulate a lot of hardware. Any system that is this critical is always redundant (car brakes, aircraft controls, power steering, process control systems, etc). Injecting insulin into a living, breathing, driving human being is certainly critical enough to demand redundancy at least at the sensor level.
3. I don't think this is as nearly bad as you suggest, especially in light of the safety features you would need for your single-point-of-failure concern. Already people push a bolus button on their pumps without fear of low blood sugar. And the monitor can be programmed to look for patterns (i.e. three consequitive results with blood glucose rising, and compute the rate of increase). In fact this seems to me to be eminently programmable.
You're right that it is easy to program to detect suspicious patterns, and to reject, for example, a sudden rapid rise in glucose levels. However, many sensors do not fail this way. Their failure mode is to slowly drift over time, generating a rising signal that is indistinguishable from a normal glucose rise after a meal. Of course, the resulting overdose of insulin could be very dangerous for the user.
I agree that existing glucose pumps need to accomodate a wide range of secretion rates, and can problably be made sufficiently reliable that they do not overdose unless told to do so by a drifting sensor.
Don't forget that a non-diabetic human body gets a high blood sugar after eating too, until the unknown amount of food consumed has been covered.
Agreed
4. I agree that an external device is not great. That is why I still inject insulin rather than use a pump - for me the benefits of a pump only are not great enough to justify the downside of an external device. But if I had a virtual artificial pancreas - i.e. a system that freed me from all the mickey mouse testing and guessing due to glucose tests, then I would be the first to sign up for an external device.
Me too, if it were reliable and I didn't need a backpack to carry it around. I have no argument with the idea of a reliable artificial pancreas. I just question our ability to make something sufficiently safe and small to be practical in the next decade.
5. I agree that an implantable artificial pancreas is unrealistic for the near-term future. And I would not want to wait for one to be implantable - if I could get an external one I would go for it.
Agreed. See above.
6. But you have to agree that an external device would be no worse than whatever regime you are using now: i.e. you are already replenishing your insulin bottle and test strips when they run out. Would it be nicer if you didn't have to? Yes, but I don't find this a reason to avoid an artificial pancreas. The point is that an artificial pancreas would remove you from the decision making loop. It would give you insulin when you need it, and only when you need it, and it would give you the right amount regardless of the amount of food, and exercise, and stress and the innumerable other factors that make it so difficult for us diabetics to predict our insulin needs.
Agreed. The present regimen requires as much expensive consumables as an artificial pancreas, and is less effective because it is really open-loop control. But islets require none of that stuff!
I agree with you that public research money does not need to be put into this work. It is being adequately researched by the private sector.
Agreed.
But as far as where research money should be invested, my point was that it is important to put most of the public money into the areas that will do the most public good, and this is prevention. It bothered me that significant research that could lead to the prevention of diabetes was being dismissed as not important.
I think the 2 cornerstones of diabetes research should be prevention and cure. On that we agree absolutely. However, I believe we are closer to a cure than prevention. Why?
1. We know that diabetes is a condition wherein the insulin-producing islets have died. To fix it, we need to replace the islets. That is a much clearer target than prevention, and we will know immediately if we have managed to restore normoglycemia.
2. On the other hand, how close is prevention when we have no idea what causes autoimmune diabetes in the first place? Every few months there is a new "cause", be it coxsackie, stomach virus, cow's milk, or vaccines. I expect it will be a long time before we can avoid or immunize against such a host of "causes".
3. The issues associated with vaccinating everybody against diabetes are enormous (it is useless to vaccinate only relatives of diabetics, since over 80% of new diabetes cases have no relatives with the disease). Every vaccine has adverse effects, and will the people saved from diabetes be a greater benefit than the people hurt by the vaccine? A cure, on the other hand, targets ONLY those who will benefit, namely people with diabetes.
PREVENTION RESEARCH IS ESSENTIAL! But, the best prospect for a breakthrough, I believe, is in the restoration of normoglycemia -- in other words, a cure.
Al