Diabetes treatment has improved over the last decades and diabetes technology in particular is developing rapidly. Type 1 diabetes therapy seems to be entering a new era as more and more automated insulin delivery systems (AID) are now available on the market and are thus increasingly becoming part of daily clinical practice. In addition to these commercially available systems, so-called do-it-yourself (DIY) or open source systems are used especially by technology-savvy people with diabetes.
Fig. 1
These DIY systems have become known mainly through the #WeAreNotWaiting movement, which was started by people with type 1 diabetes who no longer wanted to wait for commercially available closed-loop systems. Despite the systems that are now commercially available, the number of DIY system users continues to increase, as many T1D patients do not switch to the already available AID systems for various reasons. These reasons include limited access and related therapy costs as well as the limited customisability of commercially available AID systems. However, one issue with DIY systems compared to approved AID systems is the question of liability in the event of malfunctions and resulting damage to health [1]. Nevertheless, the satisfaction with either commercially available or DIY systems is very high as they reduce some burden of diabetes therapy of T1D patients.
Regardless of the type of system used, all systems calculate individual insulin doses by use of special algorithms in response to continuous glucose monitoring (CGM) values, which are transmitted to and administered via an insulin pump. Many clinical trials already demonstrated the beneficial effects of such AID systems such as significant improvement of glycemic control resulting in reduced HbA1c levels, increased time in range (TIR), and reduced time in hypoglycemia. Particularly TIR could improve already within short time frames as recently reported by Breton et al. This report based on real-world data showed that in patients who switched from an open to a closed-loop system, TIR could be increased by approximately 10% after only 14 days, reaching the recommended TIR ≥ 70% [2].This improvement, in contrast to other therapies, was sustained even after one year, hinting to the efficiency of AID systems. Moreover, the data showed that TIR increased throughout age groups including school children and adolescents although the treatment goal of TIR ≥ 70% could not be achieved in these young age groups. In general, diabetes management is very difficult in children and adolescents, who have the highest HbA1c levels of all age groups, so AID systems are very helpful in supporting diabetes management here, but further research and development is needed to achieve treatment goals in this age group.
Also in the adult age group, AID systems are not a guarantee that all patients reach the important treatment goal of HbA1c less than 7%. For example, in the iDCL trial almost half of the participants missed this goal after 26 weeks of treatment with an AID system [3].While current systems are successful in optimising nocturnal glucose control and reduces time in hypoglycemia, postprandial hyperglycemia is still a challenge for these systems. Currently, all commercially available systems require the announcement of meals and exercise. Although some algorithms already offer the possibility of simplified meal announcements without exact carbohydrate counting, the reduction of postprandial hyperglycemia is still a challenge particularly if a meal bolus is omitted. The development of AID systems that automatically identify meal intake would be a big step forward to a fully closed loop system.
The working group around Garcia-Tirado tried to step into this future as they recently published a trial testing an adapted algorithm that aimed to identify unannounced meals. The algorithm was designed to predict potential meals based on rapidly rising glucose values so that microbolus dosing could be increased more aggressively in order to respond more quickly to postprandial glucose excursions compared to a standard algorithm. The report showed that the adapted algorithm was able to increase insulin infusion more rapidly after an unannounced standardized meal so that peak glucose values were lower compared with the standard algorithm and TIR significantly higher within the observation period of 6 hours after the meal [4].
While engineers are working on improvements of AID algorithms to close the loop some day in the future, also alternative approaches are tested to optimize glucose control. As the limitations of current systems rely partly on pharmacokinetics and – dynamics of subcutaneous insulin administration, bi-hormonal closed-loop systems with glucagon as second hormone are as well under investigation. The addition of glucagon should further reduce insulin related hypoglycemia risk and may also allow for more aggressive insulin dosing by closed-loop algorithms [5]. However, due to a currently und available stable liquid glucagon formulation, it remains to be seen when such a bi-hormonal closed-loop system will enter the market.
Another option might be the addition of adjunctive pharmacotherapy to HCL treatment. In recent years, various studies have investigated different adjunct therapies, either with or without the use of an AID system. The approach of improving postprandial glucose excursions by means of glucagon suppression, for example by administering a GLP-1 receptor agonist or pramlintide, has shown particularly for GLP-1 receptor agonists conflicting results so that a clear benefit cannot be derived for this substance class. Pramlintide seem to be more benefitial as it blunts postprandial hyperglycemia but the burden of additional injections due to the lack of an approved co-formulation restricts this drug in its clinical use. The combination of SGLT inhibitors (SGLTis) with ADI systems has been tested only in small trials such as the DAPADream trial and a trial of Haidar et al with empagliflozin. Both trials demonstrated the already previously described benefits of SGLTis such as increased TIR and reduced postprandial glucose excursions without an increased hypoglycemia risk. However, the work of Haidar revealed that carbohydrate counting is still superior compared to additional pharmacotherapy also when an AID system is used [6, 7, 8].The major concern with the use of SGLTi is the risk of diabetic ketoacidosis (DKA), the risk of which is by 5 to 10 fold when using SGLT2i with concurrent insulin therapy [9]. As it is well known that insulin pump therapy increases the risk of DKA due to possible device malfunction and catheter occlusions, special attention needs to be paid to euglycemic ketoacidosis in combined therapy of AID systems and SGLTis.
In summary, technical development has advanced considerably with the market maturity of AID systems that have the potential to simplify type 1 diabetes therapy substantially. However, further research and advancements of current systems are needed to eventually move towards closing the loop in order to ease reaching treatment goals and thereby reduce the incidence of late complications.
Read more about our expertise in testing medical technology: Insulin pump development and design, artificial pancreas study and the AP@home project, glucose monitoring.