Currently, people with type I diabetes have to monitor their blood glucose frequently and inject themselves several times a day with the correct doses of insulin, as their pancreases have ceased to produce this hormone which is essential to life. Wouldn’t it be great if the whole process could be automated, reducing the frequency of these inconvenient tests and dose calculations, thus allowing patients to get on with their lives? Could a wearable artificial pancreas that monitors glucose levels continuously, calculates how much insulin is needed and then inject the correct amount at the right time, be the answer?
Of course, it is not just a matter of convenience, a so-called ‘closed loop’ system such as this, which does not require any human input, will allow a patient’s blood glucose to stay within the physiological range as long as possible; thus, avoiding side effects like hypoglycaemia and life-threatening complications like diabetic ketoacidosis.
A meta-analysis, published in the British Medical Journal in April 2018, showed that a ‘closed loop’ system versus standard treatment increased the time spent within physiological range (normoglycaemia) by almost 2.5 hours a day, reduced hyperglycaemia by about 2 hours and hypoglycaemia by 20 minutes. In principle, the artificial pancreas would also reduce long-term complications of the cardiovascular, renal, visual and nervous systems, amongst others.
Are we there yet?
Back in 1977, Diabetes UK purchased the country’s first artificial pancreas, which enabled stabilisation of blood glucose levels during surgery and childbirth for some patients with type I diabetes. While innovative at that time, it was the size of a filing cabinet and was hardly suitable for routine use by patients. Fast-forward 40 years and the ‘machine’ has shrunk in size, making it suitable for everyday use by the majority of patients with the condition.
So, what does an artificial pancreas look like? It is not so much a single machine per se, but a system of devices (the hardware) linked to each other and controlled by an algorithm (the software), without any daily input from the patient (except routine maintenance of the devices).
An artificial pancreas device system (APDS) is comprised of a continuous glucose monitor (CGM) – a sensor inserted subcutaneously that measures glucose levels in the interstitial fluids around cells (which is an indirect measurement of blood glucose levels). The measurements are transmitted to a receiver every few minutes, which displays the estimated blood glucose levels as well the predicted levels in the next few hours. The measurements are then analysed by a computer-controlled algorithm, which instructs an insulin pump to inject the correct doses of background basal insulin at the right time; the algorithm could be housed within a smartphone or the insulin pump.
At present, the patient still needs to measure the blood glucose levels using a ‘traditional’ blood glucose device (BGD) before meals (for the bolus insulins) and to calibrate the CGM periodically. Because a BGD is still needed, this type of APDS is described as a ‘hybrid closed loop system’. It is a matter of time before a truly automated ‘closed loop system’ is developed, which does not need any regular BGD checks by the patient. One such effort is the International Diabetes Closed Loop Trial, led by the University of Virginia in collaboration with other European institutions and companies like TypeZero Technologies, Tandem Diabetes Care, Dexcom and Roche Diagnostics.
Nearly there now
Many companies and academic institutions have been working together over the last few decades to make a wearable artificial pancreas become a reality. At the forefront of this endeavour is California-based Medtronic. It has been incrementally developing an automated artificial pancreas, using its long-standing experience of manufacturing glucose meters, insulin pumps and analytical software. Its latest system, the MiniMed 670G, was approved by the FDA in September 2016 as a ‘hybrid closed loop system’; ‘hybrid’ because it requires a degree of manual input by the patient before meals and for calibration. It is a relatively simple system, comprising a CGM sensor and insulin pump, with the latter also acting as a CGM receiver and hosting the algorithm. The system automates the delivery of background basal insulin but not the pre-meal bolus insulin. Besides Medtronic, companies like Insulet (Omnipod) and Bigfoot are also developing their own APDS.
Interestingly, researchers at the University of Cambridge, led by Professor Roman Hovorka, are taking a slightly different tack. Instead of focusing on the hardware devices, their sole focus is on developing the software, an algorithm that could be used with any insulin pump and any CGM. Their recent study, published in The Lancet in October 2018 (carried out in the UK & US following approval by regulatory authorities in both countries), showed a reduction in HbA1c versus standard treatment, amongst other positive glycaemic measures. In this hybrid closed loop study, their ‘model predictive control algorithm’ was held in a Samsung Galaxy G4. They are looking to commercialise their algorithm, thus making the artificial pancreas one step closer for the 400,000 and 1.2 million people with type I diabetes in the UK and US, respectively.
A fully automated ‘closed loop’ artificial pancreas promises to function as closely as possible to a real pancreas. The conveniences for patients are obvious: fewer finger pricks, no need to calculate one’s doses of insulin on a daily basis and removing the fear of
hypoglycaemia or diabetic ketoacidosis. However, it remains to be seen if such a system will perform as well during times when the demands for insulin fluctuates widely, eg unpredictable activities and meals during holidays.
On another note, some companies are developing a dual-hormone APDS, which delivers glucagon as well. Glucagon’s physiological action is to counter that of insulin by raising the levels of glucose when severe hypoglycaemia occurs, thus acting as a safeguard to possible overdose of insulin.
Who will pay for it?
As with all new technologies, payers would need to be convinced that costly APDS offers more than mere convenience and quality of life for patients. They will want to see hard data that better glucose control will translate into short-term budgetary savings like hospitalisation reductions for hypoglycaemia and diabetic ketoacidosis; this prerequisite cost-effectiveness data may be difficult to generate for a novel technology with trials involving limited numbers of subjects and short durations. The full retail price for the MiniMed 670G is said to be around $8,000.
Nevertheless, the expectation is that a fully automated ‘closed loop’ APDS will reach the market sometime in the next 5-10 years, allowing people with type I diabetes to live the life that others without the condition take for granted.
James Huang is a policy researcher and Stephen Huang a pharmaceutical medical consultant, both at SCP Medical.
