Review of Automated Insulin Delivery Systems for Type 1 Diabetes and Associated Time in Range Outcomes

Automated insulin delivery (AID) systems play an important role in the management of type 1 diabetes mellitus (T1DM). These systems include three components: a continuous glucose monitor (CGM), an insulin pump and an algorithm that adjusts the pump based on the CGM sensor glucose readings. They are not fully automated and still require the user to administer bolus insulin doses for food. Some AID systems have automatic correction boluses, while others only have automatic basal or background insulin adjustments. As CGM has become more accurate and the technology has evolved, AID systems have demonstrated improved glycaemic outcomes. The clinical evaluation of AID systems in randomized controlled trials and real-world studies have shown their utility in helping glycaemic management. In this review, we compare AID systems that are commercially available in the US and summarize the literature, with a special focus on time in range in T1DM. The review also discusses new AID systems on the horizon and explores considerations for personalized care.

Automated insulin delivery (AID) systems play an important role in the management of type 1 diabetes mellitus (T1DM) by helping users achieve the recommended glucose targets while reducing the episodes of hypoglycemia. 1 First-generation AID systems, including MiniMed ™ 630G (Medtronic, Dublin, Ireland) and Basal-IQ (Tandem Diabetes Care, Inc., San Diego, CA, USA) featured prespecified glucose boundaries at which to suspend insulin pump delivery to prevent hypoglycaemia; in contrast, second-generation systems focused on "hybrid closed-loop" technology, which automatically delivers basal insulin based on sensor data while incorporating user-initiated mealtime boluses. Current systems in development aim to eliminate the need for meal announcements and, eventually, to fully "close the loop" by adding dual hormone capabilities that regulate blood glucose without user input. 2 The clinical evaluation of these systems through randomized controlled trials and real-world studies has demonstrated their utility in helping glycemic management. [3][4][5][6][7][8][9][10][11][12][13][14][15][16][17] Traditional AID systems have three components: • an insulin pump • a continuous glucose monitor (CGM) • an algorithm that titrates insulin by parsing sensor glucose levels.
While most commercially available and pipeline AID systems have similar components, they also have important feature differences. For example, some have automatic correction boluses, such as Control-IQ Technology ™ (Tandem Diabetes Care, Inc., San Diego, CA, USA) and MiniMed™ 780G (Medtronic, Dublin, Ireland), while others offer varying degrees of smartphone control (e.g., Control-IQ; Omnipod ® 5, Insulet, Billerica, MA, USA). 18 In this review, we highlight relevant outcomes from key research studies for five prominent AID systems while assessing the strengths and limitations of each system. We emphasize time in range (TIR), defined as the percentage of time spent between 70 and 180 mg/dl, other CGM-derived outcomes, and glycated haemoglobin A1c (HbA1c) as key metrics to assess the efficacy of the AID systems.
touchREVIEWS in Endocrinology uses the same MiniMed 670G algorithm with the updated MiniMed ™ 780G pump hardware (series 700 pump). It continues to work with the Guardian 3 sensor. 19 A non-randomized prospective pivotal trial, which led to the approval of the MiniMed 670G system, involved 94 adults and 30 adolescents and compared baseline data to 3 months on MiniMed 670G. 3 There was a 5.0% improvement in TIR for adults and a 6.8% improvement in TIR for adolescents, which correspond to 1.2 and 1.7 additional hours in range per day on average, respectively. Adult and adolescent HbA1c decreased by 0.5% and 0.6%, respectively, compared with baseline. Moreover, adult and adolescent time spent <70 mg/dL was reduced by 3.0% (43 minutes) and 1.5% ( Tandem's Control IQ Tandem's non-adjunctive t:slim X2 ™ pump with Control-IQ was the next AID system to launch in the USA in 2020 after the MiniMed 670G, and proved to be a very appealing option in many ways. 13 Besides being able to receive software updates and being calibration free, the Control-IQ also allows users to program their basal rates and bolus doses for meals and corrections; therefore, this system represents a far more customizable and personalized approach to AID. In particular, the Control-IQ's exercise and sleep modes, as well as the FDA-cleared smartphone bolus functionality, offer users a unique advantage compared with Medtronic's AID systems. 14 Tandem's Control-IQ system is composed of a Dexcom G6 ® (DexCom, Inc., San Diego, CA, USA) CGM and TypeZero "treat to range" predictive control algorithm integrated into the interoperable t:slim X2 pump. 13 During an adult pivotal trial, participants between 14 and 71 years old with T1DM who used Control-IQ witnessed a 10% improvement in TIR (from 61% to 71%) compared with those on SAP, who experienced no change from baseline. 13 The mean HbA1c reduction in participants using Control-IQ was 0.3%, whereas HbA1c did not improve in the SAP arm. Furthermore, time spent <70 mg/dL improved by 2.0% (from 3.6% to 1.6%) in the Control-IQ arm versus 0.6% (from 2.9% to 2.3%) in the SAP arm -a difference of 29 minutes per day versus 9 minutes per day spent <70 mg/dL, respectively.
A paediatric pivotal trial (age range: 6-13 years) also showed impressive improvements across all primary outcomes, with Control-IQ users experiencing a 14% improvement in TIR (from 53% to 67%) compared with the 4% improvement (from 51% to 55%) in SAP users. 15 There was an HbA1c reduction of 0.6% in the Control-IQ group (from 7.6% to 7.0%), compared with a 0.3% reduction in the SAP group (from 7.9% to 7.6%).
In a year-long real-world retrospective study, the 9,451 participants (83%     Manufacturers do not directly offer support for these systems.
There are no randomized controlled studies on DIY system use at the time of writing. In a systematic review, 6 observational studies, 2 case reports and 1 anecdotal study were found. 33  with T1DM via community-developed resources. 36 In this prospective real-world observational study, the mean TIR of both adult and paediatric participants improved by 1.6 hours/day to 73%.

Automated insulin delivery and personalized care: How to support patients
Supporting patients who choose to use AID systems will become more necessary as systems gain a wider audience of users and other AID systems come to market. In a study of MiniMed 670G users, 30% of youths who discontinued AID use stated difficulty with alarms, the number of calibrations to keep the system functioning and the time needed to keep the system functioning as key reasons for discontinuation, with higher HbA1c noted as a predictor of discontinuation. 35 Newer AID models require fewer to no calibrations and less intensive interaction. Discussions should be started before selecting a given AID system, and health care professionals must ensure that patients know that all current manufacturers require manual bolus doses for food and that there is currently no system that is completely free of patient interaction. 37

The future of automated insulin delivery
As noted above, AID systems cannot meet the needs of all people with T1DM, and multiple daily injections will remain an attractive option for many in the near future. Future iterations of AID systems will need to "close the loop" and remove the need for exercise announcements or meal boluses. One such example is the bihormonal iLet ® Bionic Pancreas (Beta Bionics ® , Irvine, CA, USA), a dual hormone pump that uses glucagon and insulin to regulate an individual's blood glucose level. During an open-label random-order crossover trial, 10 participants used the system and experienced statistically and clinically significant improvements while using this system, compared with the insulin-only configuration of the Bionic Pancreas. 42 While this technology is a promising step towards the "set it and forget it" goal of AID, it is far from receiving FDA approval. One of the other challenges to this goal is the speed at which insulin works. There have been strides toward faster insulin pumps, such as insulin lispro-aabc and the faster-acting insulin aspart. However, these insulins still require user-initiated boluses in advance of eating.
In the meantime, it is important to find material ways to improve AID for patients through increased education at the provider and patient levels.
Besides improvements in insurance coverage and device interoperability, there must be an increased awareness of how to customize AID system settings to meet the unique needs of each patient, for example, via tip sheets or customized glucose targets during different points of the day; furthermore, systematic approaches are needed for providers to understand which system may be most appropriate for which patient. On-going and future studies must also show the glycaemic benefits of AID use in specific groups, for example, in geriatric or pregnant populations with T1DM or those with new-onset T1DM for beta cell preservation.

Conclusions
AID has come a long way, starting with systems that could only suspend insulin delivery when glucose was low and evolving into systems that can suspend insulin delivery if glucose is predicted to go low. Now, we have systems that can increase or decrease insulin based on predicted glucose values and give automated correction bolus doses as needed.
There is plenty of evidence that supports the use of these systems as a standard of care for people with T1DM, as evidenced by the improvements in TIR and other glycaemic and patient-reported outcomes. Understanding the differences between each system and ensuring proper patient and provider education is important to ensure optimal outcomes. While many have used AID and the term "artificial pancreas" interchangeably for some time, we argue that there is no true artificial pancreas as of now, given that users must still initiate meal boluses and engage with the system to ensure consistent glycaemic management. The future does look very bright for AID systems, which will likely become even more automated by adding glucagon and incorporating faster insulins. These improvements may ultimately reduce the need for mealtime bolus insulin doses or additional carbohydrate intake for exercise. ❑