Objective
Progression towards fully closed loop systems capable of controlling blood glucose without the need for patient input is one of the key objectives of Type 1 diabetes research. These systems include an insulin pump, continuous glucose monitoring and an algorithm which receives continuous glucose levels and translates these into insulin infusion rate changes. The objective of this study is to assess the safety and efficacy of an advanced algorithm, android APS (AAPS), including use with standard and ultrafast insulin, in a representative, at home, patient population with Type 1 diabetes. This will include adults with established Type 1 diabetes on insulin pump therapy, or multiple daily injections of insulin. We will assess of the feasibility of a fully closed loop system with this algorithm, progressing participants from the standard AAPS algorithm with meal announcements to management with minimal patient input and automated post meal insulin delivery. Efficacy will be determined by measures of glucose control and safety analysed by adverse outcomes associated with the use of this system. It is anticpated that there will be improvement in blood glucose control with this system, however also significant change to quality of life given the reduced requirement for patient input with a fully closed loop system. Participants will be directed to be'hands free', leaving insulin adjustment, as far as possible, to the AAPS system. Measurement of changes to quality of life metrics will therefore be a key component of the study, and the proportion of participants able to safely achieve fully closed loop without any meal announcements will be assessed.
Background Rationale
Type 1 diabetes is the most complex chronic condition to manage both for patients and clinicians, and is associated with financial and quality of life burden. The impact of Type 1 diabetes on patients and clinical services is largely driven by the development of acute and chronic complications, and it is recognised that advances in technology have reduced, and will continue to improve outcomes and costs. The use of insulin pumps and continuous glucose monitoring is increasing, and have led to significant improvements in glycaemic control, however to date these improvements remain modest. The combination of these technologies leading to auto regulation of insulin pump infusions rates is progressing with a number of commercially available systems now incorporating partial or hybrid closed loop systems with artificial intelligence technology. The ultimate goal is to produce a fully closed loop system which will safely manage blood glucose levels with minimal requirements for patient involvement or adjustment.
Commercial develoment of closed loop systems has progressed slowly mainly due to the requirements of regulatory bodies and lengthy approval processes. Frustration with this has led to action from members of the diabetes community to develop an open source algorithm, giving access to advanced algorithms and do it yourself (DIY) systems to assist with management of their diabetes. DIY technology has advanced rapidly and open source algorithms are available for use albeit at the patients' risk given the unapproved nature of these systems. Use of these algorithms has been supported by real world experience, however clinical trials with these systems has been lacking, and therefore the true safety and efficacy is not clear. Investigators Cohen and De Bock have both independently completed at home trials with android APS (AAPS) which is the android phone version of the algorithm. Results are soon to be published and expected to show safety and significant benefit compared with standard insulin pump use.
One of the key areas of difficulty in Type 1 diabetes is managing meal time insulin, and reducing post meal glucose without causing low blood glucose ot hypoglycaemia. This traditionally has required attention to carbohydrate counting and the implementation of complex calculations involving insulin to carbohydrate ratios and insulin sensitivity factors. New ultra fast insulins were developed to improve onset and offset of insulin action in order to safely reduce post meal spikes in glucose levels, however these have not shown great clinical benefit to date. Recent development of the AAPS algorithm has led to advanced features enabling super microboluses, or the addition of rapid, automatic, frequent correction dosing according to glucose levels. These are designed to correct glucose levels quickly after meals and, anecdotally, are showing promise with reduction in the need for accurate carbohydrate counting and improvement in post meal glucose levels. There are a growing number of patients using this system, many of whom have attempted to fully close the loop without meal announcement.
There is now sufficient expertise and real world experience with AAPS and advanced features in a clinical trial setting, to test the hypothesis that it can safely provide a fully closed loop system as an option for patient management in Type 1 diabetes.
Description of Project
Type 1 diabetes is a complex chronic condition which involves skilled management of lifestyle, diet and insulin dosing, with self-management estimated to require around 180 decisions per day. This is known to lead to diabetes distress and burnout in many patients, which may contribute to poor quality of life and suboptimal glycaemic outcomes. New technologies involving continuous glucose monitoring (CGM) and insulin pump therapy have resulted in improved glycaemic control. The addition of safety features to this system including insulin delivery suspension at low glucose levels (low glucose suspend), predictive algorithms resulting in suspension with predicted low glucose levels (predicitive low glucose suspend), and more recently hybrid closed loop systems, with auto adjustment of insulin infusion rates with high and low glucose levels. These have improved outcomes, however there is still a requirement for frequent patient input and decision making, including announcement and dosing for meals, counting carbohydrate intake and correction dosing for high glucose levels. A fully closed loop insulin delivery, or artificial pancreas system (APS), in which automatic adjustment of insulin infusion rates with no requirement for manual bolus dosing is the ultimate aim and focus of technological advances in this area, however to date, there is no current commercially available system with this capability.
AndroidAPS (AAPS) is an open-source “DIY” (Do It Yourself) closed loop system, consisting of an application on an Android device that works as a hybrid closed loop system, using an algorithm that adjusts insulin infusion rates according to glucose levels and trends. A growing movement of DIY closed loop technology has emerged with beginnings in 2013, and this is now associated with a community of users worldwide. This was initially developed to improve availability and choice for people with Type 1 diabetes without reliance on slowly developing approved systems, and without waiting for regulatory approval. There is limited clinical trial safety and efficacy data using this open source algorithm, however published real life data appears very positive. Our group has completed a safety study using AAPS in conjunction with an Ypsomed pump and Dexcom G5 continuous glucose monitoring system in patients with type 1 diabetes. To date, most participants have had a very positive experience with this hybrid closed loop system, and our results are to be published in the near future.
An advanced AAPS algorithm has now emerged, open APS reference design 1 (oref1), which includes the availability of supermicroboluses, or frequent small corrective doses of insulin. This allows for more frequent, rapid doses of insulin post meal, minimising the need for meal dosing and accurate carbohydrate counting. This has the capability to administer micoboluses every 5 minutes based on glucose response. Clinical experience is limited with this advanced algorithm, however patients choosing to trial this have reported excellent results with almost no requirement for meal announcement. There is an urgent need for safety and efficacy data using this advanced algorithm and we propose a six month in home trial with 3 months on AAPS leading to 3 months on the advanced oref1 algorithm, and no meal announcements. This will be performed with standard rapid acting insulin , and an ultrafast rapid insulin FiAsp to assess relative efficacy and safety of this fully closed loop approach.
Anticipated Outcome
The successful completion of this clinical trial is likely to have implications for Type 1 diabetes treatment and provide evidence for a changed approach to management. In contrast with the currently available hybrid closed loop systems and studies, it will be the first long term at home trial to attempt management of Type 1 diabetes without calculations for food intake and meal announcement. Observed changes in glucose control will help inform as to whether the AAPS algorithm is safe and effective for unnanounced meals, or may require adjustment going forward. Quality of life changes will be assessed and central to published outcomes of the trial. It is anticipated that such a 'hands off' system will reduce stress and burnout associated with constant monitoring and decision making required on a day to day basis for the people with type 1 diabetes.
Education of patients with Type 1 diabetes currently includes carbohydrate counting, insulin dosage adjustment and correction bolus dosing for fluctuations in diet and glucose levels. It is known that these are effective and important measures to optimise glucose control given the day to day variations experienced by patients with Type 1 diabetes. This type of management approach is however time intensive for patients, and an ongoing expense for clinicians and services. Furthermore the calculations required for insulin dosing for meals are imperfect and do not adequately take into account types of carbohydrate, and the glucose effects of non carbohydrate foods containing fat and protein. Correction doses for high glucose levels are also dependent on many factors including fluctuating levels of insulin sensitivity from day to day and week to week. All of these factors contribute to frustration for patients who struggle to normalise glucose levels, and consume a large amount of the resources of clinical services.
The highly adaptable AAPS algorithm provides a much higher level of personalisation of management than is currently available in commercial hybrid closed loop systems. This system has the potential of providing much greater choice for patients who do not wish to use the commercial algorithms. The commercial algorithms are approved with fairly rigid settings that are not suited for all patients, and may not be fully adjustable for patients and changed circumstances. AAPS is designed by people with diabetes, for people with diabetes, and has therefore been adapted fo a wider range of needs and circumstances. The quality of life implications of a safe and effective fully closed loop system are likely to be significant given the marked reduction of patient management requirements and adjustments that are associated with such an automated system.
Publication of this study would be anticipated to have implications for recommendations by peak bodies around the world and the approach to this unapproved technology. Furthermore it may provide evidence and potentially accelerate opportunities for some of the commercial providers to progress closed loop technology.
Relevance to T1D
The management of Type 1 diabetes has evolved over the past 100 years from the discovery of insulin through to multiple daily injections, blood glucose monitoring, and the use of technology. Landmark clinical trials have now proven the ongoing benefits of intensive glucose control in preventing long term damage from diabetes. This is however limited by the risks associated with low glucose or hypoglycaemia, which is a commmon occurence, and can lead to severe consequences including loss of consciousness and death. We have progressed from management designed to maintain moderate blood glucose control to more intensive therapies that are able to achieve much tighter control with lower risks of precipitating severe episodes of hypoglycaemia.
One of the ultimate goals of Type 1 diabetes research is to provide a cure or preventative treatment for this condition, and these are likely to be available in the distant future. Technological advances including insulin pump therapy and continuous glucose monitoring have already provided significant improvements for patients, however these are still not used by many patients with type 1 diabetes. The majority of patients with Type 1 diabetes are currently using multiple daily injections and finger prick glucose testing. Advances in technology are needed to improve blood glucose control, but also to offer quality of life improvements for people with Type 1 diabetes.
Real world studies with AAPS have shown remarkably good long term results with very high rates of tight glycaemic control nearing normal glucose levels. This has provided hope that complications associated with long term Type 1 diabetes may be minimised in the future with systems that safely auto adjust insulin dosing. Anecdotes of improved results with this sytem using ultra fast insulins are also emerging and worthy of clinical studies.
The development of a new treatment algorithm by the DIY diabetes community is a demonstration of the frustration with current processes put in place by regulatory bodies that have slowed the progress of development. It is well recognised that these important processes and standards have been implemented in order to reduce risks to consumers, however many patients now who are not satisfied with results from commercially available systems have explored unapproved options such as AAPS. It is likely that this trend of social media based, community approach to chronic disease management will increase, and it is important that science and medicine keep pace. Clinical trial evidence is important in this setting and will hopefully lead to more informed choices for consumers, and evidence based guidelines for clinicians.
Moving forward it is likely that there will be a number of fully closed loop systems approved for use in Type 1 diabetes, and this should give rise to improved availability, flexibility and competetive pricing for these systems and associated hardware. This trial will lead to important knowledge regarding the feasibility, limitations and practicality of advanced AAPS use, and lead to ongoing research in this important area of clinical research. It will provide a key step towards full automation of insulin delivery which will ultimately impact outcomes and costs for patients with Type 1 diabetes.