Twelve‐month results of the ADAPT randomized controlled trial: Reproducibility and sustainability of advanced hybrid closed‐loop therapy outcomes versus conventional therapy in adults with type 1 diabetes

To reassess the 6‐month efficacy and to assess the 12‐month sustained efficacy of the MiniMed™ 780G advanced hybrid closed‐loop automated insulin delivery (AID) system compared to multiple daily injections plus intermittently scanned glucose monitoring (MDI+isCGM) in people with type 1 diabetes not meeting glucose targets.


| INTRODUCTION
2][3] However, access to these technologies is often limited, as a stepwise approach to diabetes technologies is recommended.Hence, people with diabetes requiring intensive insulin therapy must start with MDI therapy and self-monitoring of blood glucose (SMBG) or intermittently scanned continuous glucose monitoring (isCGM) before progressing to MDI therapy with real-time continuous glucose monitoring (CGM), eventually moving on to automated insulin delivery (AID) system therapy if necessary. 4This can be a lengthy process, exposing people with diabetes to prolonged periods of suboptimal blood glucose levels.Following this stepwise approach, there is evidence demonstrating an improvement in glucose control with the MiniMed™ 780G Advanced Hybrid Closed-Loop AID system compared to its predecessors (MiniMed™ 670G system) 5 and to sensor-augmented pumps with predictive low glucose management. 6wever, to support a more direct approach to advanced diabetes management technology, it was of interest to compare the MiniMed™ 780G system to the most prevalent therapy, namely, MDI with isCGM. 7e Advanced Hybrid Closed Loop Study in Adult Population with Type 1 Diabetes (ADAPT) was the first randomized control trial assessing the efficacy and safety of the MiniMed™ 780G AID algorithm in comparison to MDI therapy with isCGM, in people with type 1 diabetes and suboptimal glucose control (HbA1c ≥ 8.0%). 8The primary results of the ADAPT trial showed that the AID system safely provided a 1.4% reduction in HbA1c and a 26.7% increase in time in range (TIR; 70-180 mg/dL) from baseline to the end of the 6-month study phase in participants randomized to AID treatment relative to those continuing with MDI and isCGM (MDI+isCGM) therapy. 9Participants randomized to AID treatment reached a mean HbA1c of 7.3% at the end of the 6-month study period, whereas those randomized to continue MDI+isCGM therapy had a mean HbA1c 8.9% at this timepoint. 9As such, these data suggest that the use of the AID system earlier in the type 1 diabetes treatment pathway would be beneficial, thus supporting wider access to AID therapy in people with type 1 diabetes who do not meet glucose targets.
In addition to the main study phase, the ADAPT study design included a continuation phase of an additional 6 months, during which participants randomized to receive treatment with the AID system continued with AID therapy (SUSTAIN group) and participants randomized to MDI+isCGM switched to AID therapy for 6 months (SWITCH group).The aims of the continuation phase of the ADAPT study were twofold: (1) to test if improvements in glycaemic control observed in the main study phase were reproduced in the SWITCH group, and (2) to confirm long-term sustained glycaemic control benefits of AID therapy up to 12 months in the SUSTAIN group.

| Study design and participants
Details of the ADAPT parallel randomized controlled study design are available elsewhere. 8In summary, the ADAPT study recruited 84 participants with type 1 diabetes to a prospective, multinational, open-label randomized controlled trial.Eligible participants had an HbA1c at baseline of at least 8.0% (64 mmol/mol) and were diagnosed with type 1 diabetes for at least 2 years, had been on MDI therapy for at least 2 years, and were using isCGM for at least 3 months prior to screening with an average of 5 or more isCGM scans per day and sensor readings at least 70% of the time during the month prior to screening.Exclusion criteria included the use of pramlintide, dipeptidyl peptidase-4 inhibitors, glucagon-like peptide-1 agonists/mimetics, metformin, SGLT2 inhibitors at screening, women who were pregnant at screening or planning to become pregnant during the study period, hearing or visual impairment that hindered the perception of glucose displays or alarms, or unresolved skin conditions near sensor placement areas.An exhaustive list of inclusion and exclusion criteria is available elsewhere. 8All participants gave their informed consent prior to inclusion in the study.
The study design consisted of a 2-week run-in phase followed by a 6-month study phase and a further 6-month continuation phase (Figure 1).During the 2-week run-in phase, baseline sensor glucose (SG) data were recorded using the Guardian Link 3 transmitter (data recorder only) with the Guardian Sensor 3 (Medtronic), while participants continued with their MDI therapy.Participants who demonstrated tolerance to continuous wear of the CGM device and compliance with study procedures were 1:1 randomized (investigatorblinded, block randomization procedure, conducted electronically) either to continue with MDI+isCGM therapy (control arm) or to use the Medtronic MiniMed™ Advanced Hybrid Closed-Loop system (treatment arm), henceforth referred to as AID, for the 6-month study phase.The AID system consisted of the MiniMed™ 670G (version 4.0) insulin pump, which is equivalent to the commercialized MiniMed™ 780G pump except for the absence of Bluetooth connectivity and the 110 mg/dL (6.1 mmol/L) glucose target, along with the Guardian Link 3 transmitter and Guardian Sensor 3 (Medtronic).Participants used the Guardian 3 CGM system according to the instructions for use, which require a fingerstick blood glucose measurement at least every 12 hours after initial sensor warm-up.For the continuation phase, participants in the control arm switched from MDI+isCGM to the AID system (herein referred to as the SWITCH group), whereas the treatment arm continued with AID treatment (herein referred to as the SUSTAIN group) for an additional 6 months.The findings presented here pertain exclusively to the continuation phase.
Unless there was a hypoglycaemia concern, participants were encouraged to use the optimal AID pump settings, consisting of a 100 mg/dL (5.6 mmol/L) glucose target and an active insulin time of 2 hours.The study was conducted at multiple clinical sites with experience in treating people with type 1 diabetes on insulin pump therapy in France (n = 6 sites), Germany (n = 4 sites) and the United Kingdom (n = 4 sites).The study was conducted in accordance with the Declaration of Helsinki, Good Clinical Practice guidelines, and local legislation.Competent authority and ethics committee approvals were obtained for all study sites.

| Outcomes and statistical analyses
The primary endpoint for the continuation phase was the withingroup change in mean HbA1c, measured in a central laboratory, between the end of the study phase (at 6 months) to the end of the continuation phase (at 12 months).The statistical analysis of the primary endpoint differed per treatment group: in the SWITCH group, change in HbA1c was tested for superiority, whereas in the SUSTAIN group, change in HbA1c was tested using a noninferiority test where the upper limit of the confidence interval (CI) for the mean change was compared to the noninferiority margin of 0.3%, as a 0.3% change in HbA1c is considered to be non-clinically significant. 10scriptive analysis of the within-group changes in HbA1c between baseline and end of the continuation phase (12 months) was conducted in both the SWITCH and SUSTAIN groups separately.
The secondary endpoints included CGM-based metrics of glycaemic control, system characteristics, metrics of insulin use, and patient-reported outcomes.A full overview of endpoints is listed in Table 2.In the SWITCH group, CGM metrics were compared between 2-week windows of CGM SG data at 3 and 6 months (study phase) and 2-week windows at 9 and 12 months (continuation phase), and in the SUSTAIN group, all available SG data over the entire 6 months of the study phase and 6 months of the continuation phase were compared.Patient-reported outcomes were assessed at baseline, 6 months and 12 months using the Diabetes Quality of Life Questionnaire (DQoL) 11,12 the Diabetes Treatment Satisfaction Questionnaire (DTSQ), 13 and the Hypoglycaemia Fear Survey. 14The statistical analyses of within-group changes in continuous endpoints used paired t-tests and 95% CI in cases where the normality assumption was met, otherwise endpoints were reported as median and 95% CI (estimated using the Hodges-Lehmann method) and compared using a Wilcoxon signed rank test.Noninferiority test margins were set at 6% for TIR (70-180 mg/dL [3.9-10.0mmol/L]) and time above range (TAR; >180 mg/dL [10.0 mmol/L] and >250 mg/dL [13.9 mmol/L]), and 5% and 2% for time below range (TBR) with SG <70 mg/dL (3.9 mmol/L) and SG <54 mg/dL (3.0 mmol/L), respectively.
To control for multiplicity, the following endpoints were tested in

| Baseline demographics
Of the 105 individuals assessed for eligibility, 82 were randomized at the start of the study phase (Figure 2).Of these, seven participants discontinued during the study phase, with 39 participants (15 female) in the SWITCH group and 36 participants (18 female) in the SUSTAIN group completing the 6-month study phase and entering the continuation phase.
Baseline characteristics for all randomized participants (n = 82, previously presented 9 ) and for those participants who entered the continuation phase (n = 75) are presented in Table 1.
During the continuation phase, seven participants dropped out from the SWITCH group, four withdrew themselves, one was withdrawn due to physician decision, and one was withdrawn due to an adverse event.From the SUSTAIN group, one subject withdrew themselves from the study during the continuation phase.
From 6 to 12 months, mean SG and standard deviation of SG both significantly decreased in the SWITCH group (P < 0.001).In the SUS-TAIN group, all glycaemic control endpoints met noninferiority criteria for changes from 6 to 12 months of the continued use of the AID system.
In the SWITCH group, during the continuation phase, mean ± SD percentage sensor use was 79.0% ± 24.7%.The SUSTAIN group had a percentage sensor use of 88.8% ± 9.8% during the continuation phase as compared to 92.2% ± 4.2% during the study phase.During the continuation phase, participants in the SWITCH group spent 77.1% ± 24.1% time in automation, whereas SUSTAIN group participants spent 95.8% ± 3.4% time in automation during the study phase and 93.7% ± 8.1% time in automation during the continuation phase.
Of the participants in the SWITCH group, 15 (44.1%) selected the glucose target of 100 mg/dL (5.6 mmol/L), and no participants selected the glucose target of 120 mg/dL (6.7 mmol/L) for at least 95% of the continuation phase; the remaining 19 participants (55.9%) used a mixture of glucose targets.Ten SWITCH group participants (29.4%) selected an active insulin time of 2 hours, and five (14.7%) selected 2 to 3 hours, for at least 95% of the continuation phase; the remaining 19 participants (55.9%) used a mixture of active insulin times.In the SUSTAIN group, 21 (58.3%)selected the glucose target of 100 mg/dL (5.6 mmol/L), and eight participants (22.2%) selected the glucose target of 120 mg/dL (6.7 mmol/L), for at least 95% of the continuation phase; the remaining seven participants (19.4%) used a mixture of glucose targets.Nineteen (52.8%)SUSTAIN group participants selected an active insulin time of 2 hours, 11 (30.6%)selected 2 to 3 hours, and two (5.6%) selected 3 to 4 hours, for at least 95% of the continuation phase.The remaining four participants (11.1%) used a mixture of active insulin times.
The mean ± SD number of SMBG readings per day was 3.5 ± 0.9 for the SWITCH group during the continuation phase; in the SUSTAIN group, the number of SMBG readings per day was 3.   No other significant changes between 6 and 12 months in patient-reported outcomes in the SUS-TAIN group were observed.All patient-reported outcome data can be in Table 2.

| Safety
Three severe hypoglycaemic episodes occurred in two participants in the SWITCH group during the continuation phase while receiving treatment with the AID system, two of which were classified as serious adverse events.In one patient, both severe hypoglycaemic episodes occurred due to incorrect use of the cannula fill function on the pump.
In these cases, the participant used the pump feature that is intended for filling the cannula with insulin when initiating a new infusion set in order to administer a bolus that is not factored into the AID algorithm.
No cause was determined for the third severe hypoglycaemic episode.
No severe hypoglycaemic episodes occurred in the SUSTAIN group during the continuation phase.No DKA, serious adverse events, serious adverse device effects, nor unanticipated serious adverse device effects were reported during the continuation phase.

| DISCUSSION
The continuation phase of the ADAPT trial reproduced the results observed during the study phase, 9 demonstrating that, in people with type 1 diabetes using MDI+isCGM therapy and experiencing suboptimal glycaemic control, switching to the AID system significantly improved HbA1c, with a À1.4% reduction from 8.9% (74 mmol/mol) to 7.5% (59 mmol/mol), and increased TIR by 28.1% from 43.6% to 70.7%.Switching to AID from MDI+isCGM increased the proportion of participants achieving the recommended glycaemic targets of HbA1c <7% and >70% TIR 15 from 6.5% to 60.6% and from 0% to 15.6%, respectively.Additionally, the continuation phase data confirmed the sustained efficacy of the AID system up to 12 months, with no change in HbA1c (from 7.3% to 7.4% [56 to 57 mmol/mol]) nor proportion of time spent in range (from 70.4% to 69.7%) between the study phase and the continuation phase in those participants randomized at baseline to receive AID treatment.Importantly, the observed sustained glycaemic outcomes were achieved in a setting consistent with clinical practice, in which participants had only one in-clinic follow-up visit during the 6-month continuation phase at 9 months (in addition to an end-of-study visit at 12 months to return study materials).The results of the ADAPT continuation phase corroborate previously observed glycaemic control benefits of the MiniMed™ 780G AID system in clinical trials 6,[16][17][18] and real-world settings, [19][20][21][22][23] and up to 1 year after initiation, 17 demonstrating that the AID system can provide sustained benefits to people living with type 1 diabetes.
Such sustained improvements in glycaemic control represent clinically meaningful changes 15 that could profoundly reduce the risks of longterm cardiovascular and microvascular damage associated with type 1 diabetes. 24,25e patient-reported outcome data also support the overall benefits of the AID system.Participants in the SWITCH group reported reduced fear and perceived frequency of hypo-and hyperglycaemia, were generally more satisfied with their diabetes treatment, and had improved quality of life after having used the AID system for 6 months.Furthermore, previously reported improvements seen at 6 months in the SUSTAIN group were maintained at 12 months. 9[28] Patient satisfaction with the system was also reflected in the attrition rates, which mimicked those seen in the study phase.Of the 41 participants randomized to receive AID, five dropped out during the study phase (12% attrition). 9In comparison, in the continuation phase, seven of the 39 participants who switched from MDI +isCGM to AID dropped out (18% attrition).Importantly, only one of the 36 SUSTAIN group participants dropped out during the continuation phase, signifying sustained, long-term satisfaction with system use.
It can be noted that, while there were significant improvements in glycaemic outcomes with AID therapy, mean HbA1c remained greater than the recommended target of <7%. 15However, previous studies show that people with type 1 diabetes consistently achieve HbA1c levels <7% with this AID system, 16 and these data indicate the need for enhanced education for the studied population, especially regarding pump settings and CGM usage.In fact, only six (17.6%) of the SWITCH participants and 11 (30.6%) of the SUSTAIN participants used the recommended glucose target of 100 mg/dL (5.6 mmol/L) and active insulin time of 2 hours for at least 95% of the time, and these settings have been shown to be significant predictors of better glucose control. 20Furthermore, SWITCH group participants spent only 77.1% of time in automation in the continuation phase, compared to the 95.8% in the SUSTAIN group during their first 6 months of system use (study phase), which may be attributable to the lower CGM sensor usage time in this group (79.0%).While study visits and training sessions were identical between treatment arms when starting the AID system, these results emphasize the need for adequate and recurrent training on pump settings and CGM use for optimal results. 29ile the mean TIR was unchanged in the SUSTAIN group between 6 and 12 months, the achievement rate of >70% TIR was lower in the SUSTAIN group at 12 months compared to 6 months (44.4% vs. 52.8%).This decline was not statistically significant (P = 0.169), with the mean TIR (69.7% at 12 months) being nearly equivalent to the 70% target threshold.Indeed, sensitivity analyses indicated that 50.0% of participants achieved >69% TIR during the continuation phase (Figure S3).It can thus be concluded that achievement of glycaemic control remained stable in the SUSTAIN group between 6 and 12 months of AID use.
In general, the reported safety events were relatively low during the continuation phase, with no reported DKA events nor serious adverse device effects.In the 12-month follow-up period, no severe hypoglycaemic events occurred in the SUSTAIN group.However, three severe hypoglycaemic events occurred during the continuation phase, all occurring in the SWITCH group, as compared to no such events occurring during the study phase in either treatment arm.With the cause of two of these three severe hypoglycaemic episodes being the incorrect use of the cannula fill function, these findings highlight the importance of patient education surrounding infusion set changes and setup.
Analysis and interpretation of the continuation phase data is inherently limited by the lack of an independent control group for direct comparison and because the study was not specifically powered for the continuation phase.However, participants switching from MDI+isCGM to AID therapy served as their own controls and confirmed the benefits of the AID system in this population with significant, clinically meaningful improvements to diabetes treatment outcomes.Additionally, data presented here were only from participants on MDI and isCGM starting with the AID system, rather than those on MDI and real-time CGM, which may be more representative of standard of care in some geographical regions.Results from this cohort (MDI+ real-time CGM users) of the ADAPT study are presented elsewhere. 30The ADAPT study is strengthened by its testing of a representative sample of people with diabetes who are engaged with MDI+isCGM therapy yet experiencing suboptimal glucose control.Furthermore, the study design (randomized control trial), duration, and primary outcome (HbA1c) ensure robust and relevant results that can be readily translated into clinical practice.
In summary, the continuation phase of the ADAPT study reproduced the glucose control benefits of the AID system versus MDI +isCGM previously seen in the study phase and revealed the sustained efficacy of the AID system after 12 months of use without meaningful attrition nor reduced treatment satisfaction.
Furthermore, these benefits were observed after 6 months of AID use, regardless of beginning AID immediately after randomization or 6 months later, suggesting that a lengthy observation period using a therapy providing suboptimal glucose control is not necessary before beginning AID therapy.Taken together, the ADAPT study data indicate that AID therapy should be considered early on for people with type 1 diabetes on MDI therapy with suboptimal glucose control.
an ordered sequence at level α = 0.05 until a first non-statistically significant result was observed: (1) Change in HbA1c; (2) TAR with SG >250 mg/dL (13.9 mmol/L); (3) TAR with SG >180 mg/dL (10.0 mmol/L); (4) TIR 70-180 mg/dL (3.9 to 10.0 mmol/L); (5) TBR with SG <54 mg/dL (3.0 mmol/L); and (6) TBR with SG <70 mg/dL (3.9 mmol/L).The full sequence of statistical testing per treatment group is available in Appendix S1.Adjustment for multiplicity was not applied to other endpoints.All continuation phase analyses were exploratory and P values less than 0.05 were considered statistically significant.Safety endpoints included the number of episodes of severe hypoglycaemia, diabetic ketoacidosis (DKA), serious adverse events, serious adverse device effects, and unanticipated serious adverse device effects.Study registration is available at ClinicalTrials.gov:NCT04235504.F I G U R E 1 Study design consisting of run-in, study and continuation phases.CGM, continuous glucose monitoring; HbA1c, glycated haemoglobin; isCGM, intermittently scanned continuous glucose monitoring; MDI, multiple daily injections.
8 ± 1.2 during the study phase and 3.3 ± 0.8 during the continuation phase.Change in weight from 6 to 12 months was 2.25 kg (95% CI 1.20 to 4.40 kg) in the SWITCH group and 0.55 kg (95% CI À0.25 to 1.50 kg) in the SUSTAIN group.The total daily insulin dose delivered by the pump was 47.7 ± 16.1 units in the SWITCH group during the continuation phase, while in the SUSTAIN group it was 54.6 ± 21.8 units in the study phase and 54.3 ± 22.5 units in the continuation phase, with no change in total daily insulin dose between the study phases (À0.3 units [95% CI À2.1 to 1.5]).Twenty (60.6%) of the SWITCH group participants achieved >70% TIR (70-180 mg/dL [3.9-10.0mmol/L]) during the continuation phase, compared to two participants (6.5%) during the study phase.Thirty of the SWITCH participants (90.9%) achieved <4% TBR (<70 mg/dL [<3.9 mmol/L]) during the continuation phase, compared to 24 participants (77.4%) in the study phase.In the SUSTAIN group during the continuation phase, 16 participants (44.4%) achieved >70% TIR and 33 (91.7%) achieved <4% TBR, compared to 19 participants (52.8%) achieving <70% TIR and 30 (83.3%) achieving <4% TBR during the study phase.In the SWITCH group, patient-reported outcomes significantly improved from 6 to 12 months based on the Hypoglycaemia Fear Survey (mean change in total score of À6.5 [95% F I G U R E 2 Flowchart of study inclusion.AID, automated insulin delivery; isCGM, intermittently scanned continuous glucose monitoring; MDI, multiple daily injections.T A B L E 1 Participant characteristics at time of randomization Baseline characteristics of all randomized participants, n = 82 Baseline characteristics of participants entering the continuation phase, n = 75 MDI+isCGM, n = 41 AID, n = 41 MDI+isCGM (SWITCH), n = 39 AID (SUSTAIN), n = 36 Abbreviations: CI, confidence interval; DQoL, Diabetes Quality of Life questionnaire; DTSQs/c, Diabetes Treatment Satisfaction Questionnaire status/ change; SG, sensor glucose; TAR, time above range; TBR, time below range; TIR, time in range.a Noninferiority tested with noninferiority met.
Continuation phase glycaemic control and patient-reported outcomes in the SWITCH and SUSTAIN groups Note: Values reported as mean ± SD unless otherwise noted.a n = 39.b n = 37. c n = 40.d n = 38.e n = 35.T A B L E 2