Glycemic Outcomes in Baseline Hemoglobin A1C Subgroups in the International Diabetes Closed-Loop Trial

Abstract

Using a closed-loop system significantly improves time in range (TIR) 70–180 mg/dL in patients with type 1 diabetes (T1D). In a 6-month RCT, 112 subjects were randomly assigned to closed-loop control (Tandem Control-IQ) after obtaining 2 weeks of baseline Continuous glucose monitoring (CGM) data from sensor-augmented pump therapy. We compared glycemic outcomes from baseline to end of study among subgroups classified by baseline HbA1c levels. All HbA1c subgroups showed an improvement in TIR due to reduction of both hyperglycemia and hypoglycemia. Those with HbA1c <6.5% improved mostly by reducing nocturnal hypoglycemia due to the automated basal insulin adjustments. Those with HbA1c ≥8.5% improved mostly by reducing daytime and nocturnal hyperglycemia due to both automated basal insulin adjustments and correction boluses during the day. There does not appear to be any reason to exclude individuals with T1D from automated insulin delivery based on their HbA1c. Clinical Trial Identifier: NCT03563313.

Introduction

For many individuals with type 1 diabetes (T1D), reaching the glycemic target is difficult despite advances in technology. Multiple studies have shown significant improvements in the percent time in range (TIR) 70–180 mg/dL by using an automated delivery system.¹

The International Diabetes Closed-Loop (iDCL) trial led to FDA approval of Control-IQ technology on the t:slim X2 Insulin pump (Tandem Diabetes Care). Unlike most closed-loop clinical trials, there was no limit on HbA1c levels for subjects enrolling in the study and the goal was to enroll 50 subjects with HbA1c levels <7.5% and 50 with HbA1c levels ≥7.5% with T1D. The overall results of this trial have been published,² showing that the use of this closed-loop control (CLC) was associated with a greater TIR compared with sensor-augmented pump (SAP) therapy. We sought to understand whether these TIR improvements are consistent across different ranges of baseline glycemic control from tightly controlled to less well-controlled.

Methods

Participants

For this secondary analysis, the aim was to assess the glycemic outcomes among different subgroups of baseline HbA1c during the iDCL trial. The trial was conducted after approval by a central Institutional Review Board. One hundred sixty-eight T1D patients between the ages of 14–71 years participated in this study. All patients had T1D for at least 1 year. To collect baseline data, the trial began with a 2- to 8-week run-in phase. The participants were then randomized 2:1 to CLC using Tandem Control-IQ versus SAP. A baseline HbA1c laboratory was drawn during their randomization visit.²

Closed-loop system

The CLC system consisted of an insulin pump (t:slim X2 insulin pump with Control-IQ Technology; Tandem Diabetes Care) and a continuous glucose monitor (Dexcom G6; Dexcom). The CLC system has two main distinguishing features: (1) hypoglycemia prevention by the basal attenuation module and (2) hyperglycemia mitigation by the basal rate increase module, as well as an hourly automated insulin correction bolus.

Statistical analysis

Given that the study was not powered for this subanalysis, descriptive statistics are provided with no statistical tests to compare groups. For the purpose of this analysis, the participants in the CLC group (N = 112) were stratified into five subgroups based on their randomization HbA1c: <6.5%, 6.5%–7.0%, 7.0%–8.0%, 8.0%–8.5%, and ≥8.5%. Outcomes were summarized as mean ± standard deviation or median (interquartile range [IQR]) depending on the distribution of data. Analyses were performed using SAS 9.4.

Results

Participants' characteristics

A total of 168 participants were randomly assigned to either the closed-loop group (112 participants) or the control group (56 participants).² In this study, the baseline HbA1c levels for the participants in the CLC group ranged from 5.4% to 10.6% and they were grouped by their randomization HbA1c as follows: 14 were <6.5%, 24 between 6.5% and 7%, 45 between 7% and 8%, 15 between 8% and 8.5%, and 14 were ≥8.5%. The HbA1c ≥8.5% group had the lowest mean age (22 ± 10 years), while the HbA1c <6.5% group had the highest mean age (41 ± 13 years). The number of female participants was highest in the HbA1c <6.5% group. Among all the subgroups, most of the participants were white and non-Hispanic. Most patients in every subgroup had private insurance and an annual household income above $100,000 (Table 1).

Table 1.

Baseline Characteristics and Glycemic Outcomes for All Participants in the Closed-Loop Control Group, by Baseline HbA1c Subgroups

	All, N = 112		<6.5, N = 14		6.5 to <7, N = 24		7 to <8, N = 45		8 to <8.5, N = 15		≥8.5, N = 14
Baseline characteristics
HbA1c, mean ± SD % (mmol/mol)	7.4 ± 1.0		5.9 ± 0.3		6.7 ± 0.2		7.4 ± 0.3		8.2 ± 0.1		9.1 ± 0.6
Age, mean ± SD (years)	33 ± 16		41 ± 13		34 ± 14		35 ± 17		26 ± 16		22 ± 10
Female, n (%)	54 (48%)		12 (86%)		9 (38%)		17 (38%)		8 (53%)		8 (57%)
Non-Hispanic White, n (%)a	94 (86%)		12 (86%)		21 (88%)		38 (86%)		11 (79%)		12 (92%)
Private insurance, n (%)b	102 (94%)		14 (100%)		22 (96%)		40 (93%)		14 (93%)		12 (86%)
Income ≥100K, n (%)c	55 (62%)		5 (56%)		12 (52%)		23 (64%)		9 (69%)		6 (75%)

	Base	Δ	Base	Δ	Base	Δ	Base	Δ	Base	Δ	Base
Glycemic outcomes, mean
% time 70–180 mg/dL	60.7	+10.7	78.9	+7.1	69.1	+8.7	61.9	+8.3	51.9	+13.4	33.6	+22.5
% time below 70 mg/dL	3.6	−2.0	6. 4	−4.4	4.5	−2.8	3.3	−1.6	2.2	−0.8	1.5	−0.6
% time above 180 mg/dL	35.7	−8.7	14.7	−2.7	26.4	−5.9	34.8	−6.6	45.9	−12.6	65.0	−21.9

^a Missing, unknown, or no answer n = 3.

^b Missing, unknown, or no answer n = 3.

^c Missing, unknown, or no answer n = 23.

Glycemic outcomes

All subgroups demonstrated an improvement in TIR. The highest reduction in percent time <70 mg/dL occurred in the group entering the study with an HbA1C <6.5%, with a 4.4% reduction (3.6% during the day and 6.8% overnight). The highest reduction in percent time >180 mg/dL occurred for those entering the study with an HbA1C ≥8.5%, with a 21.9% improvement (19.2% during the day and 30.4% overnight) (Table 1 and Fig. 1A, B).

FIG. 1.

Change in daytime (A) versus nighttime (B) glycemic control and median number of boluses per 24 h (C), by baseline HbA1c subgroups. (A) Solid black shows <70 mg/dL, solid white shows 70–180 mg/dL, and striped bar shows >180 mg/dL. (B) Solid black shows <70 mg/dL, solid white shows 70–180 mg/dL, and striped bar shows >180 mg/dL. (C) Solid blue bar shows manual boluses with Announced CHO (meal boluses), solid white bar shows manual boluses with No CHO (correction boluses), and solid red bar shows Automated Correction Boluses.

Bolus patterns

The number of user-requested boluses per day (including both meal related and corrections) was highest (median 6.8 [IQR: 5.2–9.2]) in the HbA1c <6.5% subgroup and lowest in the HbA1c ≥8.5% subgroup. The number of automated correction boluses was highest in the HbA1c ≥8.5% subgroup (median 5.7 [IQR 4.9–6.5]) and lowest in the HbA1c subgroup <6.5% (median 1.9 [IQR 1.2–2.7]) (Fig. 1C).

Discussion

The use of CLC led to the improvement in TIR in all the subgroups of baseline HbA1c. For those with a higher baseline HbA1c, the increased TIR originated mostly from hyperglycemia reduction; for the participants with lower baseline HbA1c, the use of the closed loop reduced their overnight hypoglycemia by the greatest percent.

As reported based on T1D Exchange data, average HbA1c levels vary with age, race/ethnicity, and socioeconomic status, and the average HbA1c reaches the highest point during adolescence.^3,4 Similarly, among our study participants, the adolescents and young adults had the highest HbA1c levels. Previous studies report poorer glycemic control in women with T1D compared with men.^5,6 However, in our study, there was no clear trend. Gender disparity reinforces the need for a deeper understanding of gender-specific attitudes and barriers and for new gender-based psychosocial approaches in the use of closed-loop systems.

In the iDCL trial, the participants were recruited from seven clinical centers in the United States without regard to gender, race, or ethnicity. This study was designed to be broadly inclusive to represent the larger population living with T1D.

The hyperglycemia reduction in the higher HbA1c subgroups and the reduction of hypoglycemia in the lower HbA1c subgroups are likely due to automated basal insulin adjustments both during the day and night reducing hypoglycemia and hyperglycemia and the hourly automatic correction boluses correcting hyperglycemia during the day, in apparent compensation for missed meal boluses. Missed meal insulin boluses represent a major challenge for suboptimal glycemic control in T1D patients.⁷ In a study of 48 children with T1D using an insulin pump, 65% missed more than one mealtime bolus per week.⁸ Two missed boluses resulted in a half-point rise in the HbA1c, while four missed boluses per week resulted in a full 1% rise in HbA1c levels.

The users of the commercially available hybrid closed-loop systems are recommended to administer meal boluses 10 min before eating, however, premeal bolusing does not always occur. The automated hourly correction doses provided by the Control-IQ system assists with managing postprandial hyperglycemia, compensating for missed, late, or underestimated meal boluses, which can reduce the burden for T1D patients.

In our study, participants with lower baseline HbA1c reduced their overnight hypoglycemia by the greatest percentage, likely due to the system's automated basal insulin adjustment. It is well known that the risk of hypoglycemia increases with tightening glycemic control.⁹ Hypoglycemia has been proven to be the main barrier to tight control.¹⁰ The use of the closed-loop system is an effective means in reducing the risk of nocturnal hypoglycemia while increasing the percentage of time spent in target range.

Our study's main limitation is the lack of diversity among all the subgroups, with most of the participants being white and non-Hispanic. In this study, the great majority of patients in every subgroup had private insurance with an annual household income above $100,000. Future clinical trials should include a wider range of socioeconomic levels and broader ethnic diversity.

The results of this analysis are a strong argument for not having restrictions on who should have access to this technology. Potential users should not be excluded because they have already achieved recommended HbA1c targets, or because they have high HbA1c levels. In fact, people having the most difficulty adhering to current diabetes treatment recommendations will have the greatest improvements in TIR by using this technology.

Authors' Contributions

L.E. wrote and edited the article. D.R. performed statistical analysis and wrote/edited the article. M.T., J.W.L., S.A.B., and B.A.B. contributed to discussions and edited the article.

Author Disclosure Statement

L.E. has received consulting fees from Tandem Diabetes and Ypsomed. M.T. has no disclosures to report. D.R. reports receiving grant support and supplies, paid to his institution from Tandem Diabetes. J.W.L. reports receiving consulting fees, paid to his institution from Animas Corporation, Bigfoot Biomedical, Tandem Diabetes Care, and Eli Lilly and Company. S.A.B. has received research support from Tandem Diabetes Care, Insulet, Dexcom, Roche, and Tolerion. B.A.B. has received research support from Medtronic, Insulet, Tandem, Dexcom, and ConvaTec.

Funding Information

This work was supported by the NIDDK grant UC4 DK 108483, as well as material support from Tandem.