Home / Resources / Featured Writers / The Coming Debate on Closed-Loop Insulin Delivery Systems

The Coming Debate on Closed-Loop Insulin Delivery Systems

Feb 24, 2018
 

Guest Post by David Kliff, Editor, Diabetic Investor

As I left Vienna after attending the 11th International Conference on Advanced Technologies & Treatments for Diabetes, my thoughts turned to what does this all mean, what was the most significant takeaway? By far the most significant development coming out of ATTD is the long quest to have a REAL closed-loop insulin delivery system is no longer a pipe dream. This is going to happen and for once I believe it will happen in my lifetime. It is difficult to overstate the power of the data I witnessed. The reality here is that some very smart people have come a very long way.

That being said I do see a debate looming on the horizon, a debate which has business implications; a debate which I hope will be solved as more studies are done and more data is released. To fully understand the nature of this debate, let’s review for moment just what a closed-loop insulin delivery system is. Simply put, this system is a collection of devices combined with software all working together. There is the insulin pump, which delivers the insulin; there is the CGM; which collects glucose data and insulin dosing algorithm.

Like everything else in diabetes devices, there are several iterations of how this system is put together. The best data I have seen came from a Tandem insulin pump, a Dexcom G6 continuous glucose sensor combined with a TypeZero insulin dosing algorithm. The hub of this system is the Tandem pump; it collects the CGM data with the insulin dosing algorithm contained on the pump.

Other systems working with other CGMS and other insulin pumps add another link in the chain — the patient’s smartphone. There are also other insulin dosing algorithms. Many are speculating that the outstanding results seen with the Tandem/Dexcom/TypeZero system are partly due to the fact there are fewer connectivity issues; all the information is collected and/or stored on the pump. The information does not have to be collected on one device and then sent to another device. Simply put, there are fewer links in the chain.

Yet this is just the first part of the debate, as depending on who you listen to, the straw that stirs the drink is either having an accurate CGM or the “best” insulin dosing algorithm. Or put another way, which of these links in the chain is the most vital and therefore by extension the most valuable? To illustrate what we are talking about: some wondered would the results seen with the Tandem/Dexcom/TypeZero system have been different had it been a Senseonics CGM or Dreamed insulin dosing algorithm?

I have long maintained that the most important link in the chain is the CGM — that if the CGM does not deliver accurate data the whole system falls apart. Without accurate data the insulin dosing algorithm is making decisions based on faulty data. Or as my college professors used to say, garbage in garbage out. However, while Dexcom maintains their accuracy lead, all the CGM systems are becoming more accurate.

Given I do not have an advanced mathematics degree, I have spoken with people much smarter than I am to better understand insulin dosing algorithms. While I won’t say that these algorithms are the secret sauce, they are critically important to the system. With accurate CGM data these algorithms do some amazing calculations and, even better, they learn. And I was somewhat surprised to learn that different algorithms might work better for different types of patients; that one may be better for children, another better for adults.

What I did not hear is anything about the insulin pump that  acts on all this information that has been collected. About all I heard was, whichever pump is used, the company behind it better be financially sound. I have long maintained that all insulin pumps do basically the same thing the same way. Yes, there are minor differences between these systems, but nothing so great as to indicate that with a well-trained patient you would see different outcomes using different pumps. However — and this is where the devil is in the details — the minor differences in these pumps come into play with a closed-loop system. Here’s why.

Without becoming overly complex, the issue comes down to whether the pump is delivering the correct amount of insulin at the correct time; that when the system tells the pump to deliver 3 units of insulin, it better deliver those 3 units accurately and on time. Even if the CGM delivers accurate data and the algorithm works its magic, the whole system falls apart if the pump doesn’t do what it’s told, how it’s told to do it.

This is where running an insulin pump company becomes critically important, as part of this job requires the ability to manufacture reliable systems. Again, it does no good to have this whiz-bang way-cool toy if one of the components of the toy keeps breaking. And let’s be honest, the insulin pump, while not an overly sophisticated device, has the most moving parts and is the most susceptible to failure. Yes, sensors occasionally fail but not nearly at the rate pumps fail.

Anyone who disputes this fact only need to review the FDA’s MAUDE database.

Just to provide some perspective, in the early days the OmniPod from Insulet — they are also involved in this process — was susceptible to a higher than acceptable failure rate. Thankfully the company has become much better at manufacturing pods, and while pods still fail, they fail at much lower rates. And to be honest, every insulin pump company has had some sort of issue in the past, so this issue impacts more than just Insulet.

Just to complicate matters further, as if we needed to do that, there is a growing voice that the links in this chain should become interchangeable, that a Dexcom CGM should work with a Medtronic insulin pump, and that the patient or patient’s physician choose which insulin dosing algorithm is used. The theory is the patient and the patient’s physician could select the best combination of components and assemble a closed-loop system that fits the patient’s needs. While I can understand this mix and match approach, I can also see lots of problems too.

But the additional complexity does not end there, as another debate is whether these systems would receive premium reimbursement, two words that don’t exactly go together these days. One school of thought, touted understandably by the insulin pump companies, is that given the results these systems will receive premium reimbursement. They also add that once available, they will do what hasn’t been done in years and increase the number of patients using an insulin pump.

Let’s tackle premium reimbursement first as I just don’t see it. I have no doubt these systems will be reimbursed, but given their cost and payors’ zest for low cost devices I don’t see them getting anything more than what payors pay today. But there’s another reason I don’t see premium reimbursement: market dynamics, or what I call good old-fashioned competition. At the moment Tandem may have the best data, but Medtronic also has skin in this game and we all know how they operate.

As to increasing the number of patients using insulin pump therapy — to me that’s one big fat maybe. Although the insulin pump companies don’t like to admit it, there is a large population of patients who just don’t want to be hooked up to a machine, no matter how way-cool whiz-bang it is. As I keep saying, only 30% of type 1 patients use an insulin pump and there are only about 1 million or so pump patients worldwide, even though pumps have been around forever. Greater pump adoption isn’t just about technology; it’s also to large extent about reimbursement. Yes, I know this seems like a vicious circle, because it is.

Think of this scenario as it could happen. First, we get a system all the way through the FDA and onto the market. Second, payors do agree to reimburse such a system. Then the question becomes at what level, tier 1, tier 2, etc. The next question is, will payors put any restrictions on reimbursement, meaning they will pay for it only AFTER its shown the patient has used another system that didn’t work? Then, just to make matters even more complex, what about patients with high deductible plans, something that is common nowadays? The possibilities get even more complex when you consider that a payor may pay for it, but even with that the patients’ out of pocket may make the system unaffordable. What, you thought this was going to be easy?

The fact is we are at the beginning of what’s going to be a long and arduous process. The first step, which looks be closer than ever before, is getting the damn thing approved by the FDA — something I suspect will be easier given the approval of the 670G from Medtronic. Once that happens, step two is getting this system into the hands of patients, and that is a much more difficult step as it involves multiple factors. Step three will be what happens when there is more than one of these things, as you can take it to the bank there will be more than one.

Still I must admit I did not believe this day would arrive during my lifetime. There is finally light at the end of the tunnel, and that light is not a freight train, but possibly the dawn of a new day for insulin pump patients. The coming debate will be interesting, and the market dynamics will definitely play a role here, but even with the debate and market dynamics these dedicated people deserve a round of applause. Thank you for proving me wrong; this is one prediction I’m glad didn’t come true