Closed loop insulin delivery

Interesting tandem cyclists on their Grand Tour, July 2018

In the world of Type 1 diabetes, the holy grail (apart from a cure) is the 'artificial pancreas' - a way to control blood glucose levels so that they resemble as closely as possible those of a non-diabetic person. There are a few barriers, however. One is the subcutaneous delivery of insulin analogue into the peripheral circulation rather than endogenous secretion of insulin from the pancreas directly into the blood vessel serving the liver. Another is the role played by other pancreatic endocrine hormones, principally glucagon, which may or may not be affected by autoimmune beta cell destruction. The third is the minute by minute nature of physiological insulin and glucagon adjustment, which cannot reasonably be replicated by a human being.

The best case scenario which is licensed and available to a person with diabetes (PWD) at the moment consists of a Continuous Glucose Monitoring (CGM) system linked to an insulin pump, which alerts the PWD when glucose levels are rising or falling outside certain parameters. One system available in the UK will suspend insulin delivery when low glucose levels are predicted (low glucose suspend), but the PWD is always expected to manage the situation. They still have to pay close attention to carbohydrate intake, estimating insulin for meals, drinks and snacks, taking account of any number of other factors such as weather, health, activity, location of insulin delivery site, time of day, when you last ate, what you've just done, what you're going to be doing next, and more. It's still a full time job for someone with diabetes.

The situation described in the paragraph above is known as 'open loop' - the CGM and pump provide information to the PWD who is the third party in the loop, and who must make the vast majority of the decisions on insulin delivery. If it were possible to monitor glucose levels minute by minute and automatically deliver the 'right' amount of insulin without consulting the PWD then the huge burden of continuous glucose management and insulin delivery would be lifted. The pharmaceutical companies are working on this artificial pancreas idea, which is more accurately known as 'closed loop' (because it isn't really anything like an artificial pancreas at all). There is one system (not yet available in the UK) will adjust the background insulin up and down according to the CGM results, which is the first licensed partial closed loop option.

Insulin is what keeps people with Type 1 diabetes alive, but it can also do great harm if the wrong quantity is delivered. As you can imagine, ensuring that the closed loop algorithm used by the CGM and pump combination is 100% safe (or safe within whatever boundaries are required by the licensing authorities) is a huge regulatory burden, given that this equipment cannot be restricted to sensible or intelligent people.

So there is a movement which has adopted the slogan #WeAreNotWaiting. People with programming knowledge and open source software are getting on with the job, and have produced their own unlicensed closed loop solutions using the technology that's already available. The three necessary components are a CGM system, an insulin pump, and a device to communicate with both and run the algorithm. A fourth element is cloud storage, mostly for reporting and analysis, but also for third party monitoring (a parent seeing real-time results for a child, for example).

Anyone with a modicum of technical skill and some disposable income can create a closed loop system using these components, and the results I've seen are sometimes astonishing - not quite non-diabetic blood glucose levels, but so much closer. The reason for the disposable income is obviously that being unlicensed, the total solution is not available or supported within the NHS, although the pump and the CGM system might be, and the algorithm and cloud storage are free and open source and can run on a mobile phone.

I've been trying to understand the technology for a while, and I joined a Facebook group relating to looping in the UK. There I discovered that one of our patients has set himself up with a closed loop, and I invited him to come and show us what he's been doing. He didn't respond to the invitation, but another slightly less local person did, so we arranged for him to come to the Diabetes Centre at a time convenient for the two consultants, three nurses and myself. And he did come, and so did I, and the nurses were there too, but neither of the consultants turned up.

Our guest was extremely helpful, and brought along some examples of the different bits of kit that can be used, as well as a presentation that took us through it all.

The choice of other components depends on the type of pump, so that is where to start. Then there are options for how to run the algorithm, which can be on a bit of specialist kit (RileyLink, Linux) and/or a phone (iPhone, Android). The AndroidAPS is the one I understand best, but perhaps that's because I'm not familiar with the iPhone, Linux or the Apple watch and have no idea what RileyLink is.

There isn't usually a problem with the CGM end because almost all of them will do, it's just a question of budget and availability - all the Dexcom CGM systems work, as does Medtronic. The Libre needs to be adapted to turn it from 'Flash glucose monitoring' to true CGM, which can be done using one of three devices: MiaoMiao, Blucon or LimiTTer. Our guest brought the MiaoMiao option so I've seen that, but I don't know what the other two are like.

There are some variables to consider, including three levels of glucose - minimum, maximum and target - and at least two levels of temporary basal rate, so that the algorithm can adjust its behaviour according to these parameters. The AndroidAPS option is more structured than the other two, because it guides the user through 'gates', introducing more features gradually and providing access to the next feature only when the previous step is successfully implemented.

The potential benefits of closed loop are longer times in range, fewer hypos, improved HbA1c, and less effort day-to-day required to achieve these results. The downside is the cost, the effort required to set the system up in the first place, and the possible mental strain of maintaining your unlicensed technology. Your diabetes team may have little or no knowledge of what is involved, and this may be a problem, but I hope that awareness is rising. It is very likely that a looper would be asked to sign some sort of waiver which will absolve the professional team of liability should something go wrong - these documents are being considered and consulted on in my region as I write.

Joining the 'Looped UK' Facebook group has given me a lot of information and access to UK expert and non-expert volunteers. Clearly each PWD is expected to manage their own technology, there is no commitment from the group's volunteers to provide help or support, but they can be extremely helpful in answering questions, showing how it's done and helping with the trickier aspects of the setup. 'Information days' and 'Build days' take place occasionally to help new loopers, and I'm hoping to attend one soon to find out even more.