Overnight Glucose Control with Dual- and Single-Hormone Artificial Pancreas in Type 1 Diabetes with Hypoglycemia Unawareness: A Randomized Controlled Trial

Abstract

Background:

The dual-hormone (insulin and glucagon) artificial pancreas may be justifiable in some, but not all, patients. We sought to compare dual- and single-hormone artificial pancreas systems in patients with hypoglycemia unawareness and documented nocturnal hypoglycemia.

Methods:

We conducted a randomized crossover trial comparing the efficacy of dual- and single-hormone artificial pancreas systems in controlling plasma glucose levels over the course of one night's sleep. We recruited 18 adult participants with hypoglycemia unawareness and 17 participants with hypoglycemia awareness, all of whom had documented nocturnal hypoglycemia during 2 weeks of screening. Outcomes were calculated using plasma glucose.

Results:

In participants with hypoglycemia unawareness, the median (interquartile range [IQR]) percentage of time that plasma glucose was below 4.0 mmol/L was 0% (0–0) on dual-hormone artificial pancreas nights and 0% (0–10) on single-hormone artificial pancreas nights (P = 0.20). Additionally, participants with hypoglycemia unawareness experienced two hypoglycemic events (<3.0 mmol/L) on dual-hormone artificial pancreas nights and three hypoglycemic events on single-hormone artificial pancreas nights. In participants with hypoglycemia awareness, the median (IQR) percentage of time that plasma glucose was below 4.0 mmol/L was 0% (0–0) on both dual- and single-hormone artificial pancreas nights. Hypoglycemia awareness participants experienced zero hypoglycemic events on dual-hormone artificial pancreas nights and one event on single-hormone artificial pancreas nights.

Discussion:

In this study, dual-hormone and single-hormone systems performed equally well in preventing nocturnal hypoglycemia in participants with hypoglycemia unawareness. Longer studies over the course of multiple days and nights may be needed to explore possible specific benefits in this population. ClinicalTrials.gov No. NCT02282254.

Introduction

The fundamental goal for type 1 diabetes (T1D) treatment is to develop a strategy of life-long insulin replacement therapy that reduces rates of hyperglycemia, and consequently the long-term risk of microvascular and macrovascular complications, while also minimizing iatrogenic hypoglycemia.¹ Unfortunately, most patients remain at high risk for life-threatening iatrogenic hypoglycemia and continue to experience chronic hyperglycemia.² This two-sided dilemma remains the main barrier for both patients and providers to achieving optimal glycemic control.³

Recurrent iatrogenic hypoglycemia is one of the main factors leading to hypoglycemia unawareness.⁴ In patients with T1D and suffering from hypoglycemia unawareness, autonomic warning symptoms are either not recognized or fail to occur before the development of neuroglycopenia. Hypoglycemia unawareness affects ∼20% of patients with T1D,⁵ and these patients are at a sixfold increased risk for severe hypoglycemia.⁴ Overnight control is an especially challenging task for many patients, with unperceived nocturnal hypoglycemia occurring approximately once every three nights.⁶ These nocturnal episodes contribute to the maintenance of hypoglycemia unawareness and are associated with devastating complications such as a fatal cardiac arrhythmia, the so-called dead in bed syndrome.^7,8

One strategy used to reduce nocturnal hypoglycemia is the sensor-augmented pump therapy with a threshold-suspend feature. In this system, continuous glucose monitoring detects hypoglycemia and triggers an automatic temporary insulin pump shutoff. In the ASPIRE (Automation to Simulate Pancreatic Insulin REsponse) trial, participants with T1D wearing insulin pumps and continuous glucose monitoring systems with the activated threshold-suspend feature had a 31.8% reduction in nocturnal hypoglycemia over a 3-month period.⁹ Despite this significant improvement, 91.7% of these participants had at least one nocturnal (22:00–8:00) hypoglycemic event lasting for 2 h during the 3-month study period.⁹

In closed-loop delivery systems, otherwise known as artificial pancreas systems, the insulin pump infusion rate is regularly adjusted by a dosing algorithm that relies on glucose sensor readings. Studies evaluating the use of single-hormone (insulin only) artificial pancreas systems have shown benefits compared with sensor-augmented pump therapy.^10

–15 Other studies have shown that adding glucagon to the artificial pancreas (producing a dual-hormone artificial pancreas¹⁶) provides additional reduction in hypoglycemia.^{17

–24} However, results have been mixed regarding the additional benefits of dual-hormone systems overnight. While some studies²³ have found a considerable benefit—with nocturnal hypoglycemia nearly eliminated, others^18,19 have demonstrated little to no benefit of dual-hormone over single-hormone artificial pancreas systems, mostly due to a single-hormone system that already sufficiently reduced nocturnal hypoglycemia to a very low frequency. It was recently reported that hypoglycemia awareness could be restored after continuous use of a closed-loop system for 5 weeks probably by significantly decreasing the frequency and duration of hypoglycemia.²⁵

Use of the dual-hormone artificial pancreas may be warranted in some, but not all, patients. In this study, we sought to compare the efficacy of single- and dual-hormone artificial pancreas systems in reducing nocturnal hypoglycemia in patients with and without hypoglycemia awareness. Patients in both groups had documented nocturnal hypoglycemia during a 2-week run-in period. Our hypothesis was that glucagon would provide additional nocturnal hypoglycemia prevention in patients with hypoglycemia unawareness.

Methods

Study design

We conducted a randomized crossover trial comparing the efficacy of dual- and single-hormone artificial pancreas systems overnight in 18 adult participants with hypoglycemia unawareness and 17 adult participants with hypoglycemia awareness. All 35 participants had documented nocturnal hypoglycemia on a blinded continuous glucose monitor (iPro^™2 professional CGM; Medtronic) over a 2-week run-in period. Both 10-h interventions (single-hormone and dual-hormone artificial pancreas) occurred overnight (21:00–7:00) at Montreal Clinical Research Institute (IRCM, Quebec, Canada).

Study participants

Adults with T1D residing in the greater Montreal area were enrolled in this study. Eligible patients were referred by their local endocrinologist or approached by the research team at IRCM. Participants were required to be 18 years of age or older, on an insulin pump for at least 3 months, diagnosed with T1D for at least 1 year, and have two or more documented episodes of nocturnal hypoglycemia during the run-in period. Hypoglycemia unawareness was determined by the Clarke questionnaire²⁶ assessing patient symptom perception. Patients with poor glucose control (HbA1c > 12%) were excluded. The study was approved by the IRCM ethics committee. All participants provided written informed consent.

Randomization and masking

We used block-balanced randomization to determine the order of the interventions, as generated by www.sealedenvelope.com Participants and investigators were not masked to the allocation assignments. Participants were masked to sensor glucose readings and to hormonal infusion rates during both nights. For safety reasons, investigators had access to plasma glucose readings. Participants had access to their finger-stick glucose measurements.

Study procedures

The trial included up to six visits: the run-in period, admission, one or two sensor insertion visits, and two overnight intervention visits.

During the run-in period, participants' hypoglycemia symptom perception was measured using the Clarke questionnaire.²⁶ The questionnaire rates symptom perception on a scale between 0 and 7. Participants receiving a score between 0 and 2 were classified as hypoglycemia aware, while those who scored between 4 and 7 were classified as hypoglycemia unaware. Participants with a score of 3 were classified as neither hypoglycemia aware nor unaware and excluded from the study. To document nocturnal hypoglycemia, participants also completed up to 12 days of continuous glucose monitoring using either an iPro^™2 professional continuous glucose monitor (Medtronic) or the participant's own real-time glucose monitor (Enlite^™ sensor; Medtronic). To be included in the study, participants needed to have at least two nocturnal hypoglycemic events (≤3.5 mmol/L) between 22:00 and 7:00 for more than 20 consecutive min each.

During the admission visit, participants with documented nocturnal hypoglycemia underwent a medical evaluation, including a physical examination; measurements of weight, height, and waist circumference; and a blood draw for HbA1c (if a previous measurement taken within 1 month of the visit was not available). Participants also recorded their previous 3 days of insulin therapy (total daily dose, carbohydrate-to-insulin ratios, and basal rates). Study participants were then randomized to determine the order of interventions.

One or 2 days before the study's overnight visits, participants inserted a real-time glucose sensor (Enlite sensor; Medtronic) either at home or at the clinical research facility. They were provided with the study insulin pump (MiniMed^™ Veo^™ system) and a new cartridge containing their usual fast-acting insulin analog (Aspart, Lispro, or Glulisine). Participants already using an eligible Enlite sensor and Medtronic pump were permitted to use their own devices during overnight visits. To further minimize confounding factors that might influence overnight glucose fluctuations, participants were asked to refrain from exercise, alcohol, and caffeine consumption (each ≤1 beverage per day) on the days of the intervention.

For the intervention visits, identical procedures were followed on the single- and dual-hormone nights. Participants were admitted to the clinical research facility at 19:00. A cannula was inserted into an arm or hand vein for blood sampling purposes. During dual-hormone artificial pancreas visits, glucagon (Eli Lilly^®) was reconstituted according to the manufacturer's instructions and a MiniMed Veo insulin pump containing the glucagon solution was installed. At 20:30, a standardized snack containing 20 g of carbohydrate was given to all participants in an attempt to circumvent potential hunger overnight for participants. At that time, participants had the option to take a bolus dose of insulin, which was then repeated at the second overnight visit if taken. Venous blood samples were obtained from the preinserted cannula every 20 min starting at 21:00. If plasma glucose went below 3.5 mmol/L, samples were taken every 15 min until plasma glucose rose again above 3.5 mmol/L.

Artificial pancreas interventions began at 20:00 and ended at 7:00 the next day. Every 10 min, a sensor reading was entered manually into a computer, which ran a dosing algorithm that generated recommendations for the basal rates of insulin and glucagon miniboluses (during dual-hormone nights). The participant's pumps' infusion rates were changed manually based on computer-generated recommendations using a predictive algorithm.

The dosing algorithm was based on model predictive control and adopted the compartmental approach to describe insulin–glucagon–glucose dynamics. The algorithm was initialized using daily insulin requirements and insulin-to-carbohydrate ratios. Glucagon delivery was based on heuristic logical rules that employed estimates of plasma glucose concentrations and their trends. Importantly, the aggressiveness of insulin delivery was identical in both the single-hormone and dual-hormone artificial pancreas algorithms, but glucagon-on-board was taken into account in the predictions during dual-hormone visits. This approach uses glucagon to further reduce the residual hypoglycemia that remains despite suspension of insulin delivery. This approach does not attempt to allow more aggressive insulin delivery.

Hypoglycemia was treated if plasma glucose levels fell below 3.0 mmol/L, irrespective of symptoms, or below 3.3 mmol/L with symptoms of hypoglycemia. Participants received 15 g of carbohydrates as treatment, and plasma glucose levels were measured every 15 min until they rose above 3.0 mmol/L; additional 15 g of carbohydrates was given every 15 min if plasma glucose remained below 3.0 mmol/L. Participants were discharged at 7:00 the next morning.

Outcomes

The primary outcome was the percentage of time that plasma glucose was below 4.0 mmol/L comparing single- and dual-hormone interventions in the hypoglycemia unaware group. Secondary outcomes included the percentage of time that plasma glucose was below 4.0 mmol/L comparing single- and dual-hormone interventions in the hypoglycemia aware group; percentage of time that plasma glucose was below 4.0 mmol/L compared between the hypoglycemia aware and unaware groups; percentage of time that plasma glucose was between 4.0 and 8.0 mmol/L, between 4.0 and 10.0 mmol/L, below 3.5 mmol/L, below 3.3 mmol/L, above 8.0 mmol/L, and above 10.0 mmol/L; mean plasma glucose levels; total insulin delivery; total glucagon delivery; and the number of hypoglycemic events. Outcomes were calculated from 23:00 to 7:00 using plasma glucose levels.

Statistical analyses

The sample size was calculated to achieve enough power for the comparison between single- and dual-hormone artificial pancreas interventions in the hypoglycemia unaware group. Based on our previous studies, we anticipated that compared with a single-hormone artificial pancreas, a dual-hormone artificial pancreas would reduce absolute time spent at glucose levels below 4.0 mmol/L by 3.0%.^{17
–19,22,23} The standard deviation of paired differences was estimated to be 4.0%. Therefore, 17 subjects in each group were needed to achieve 80% statistical power to detect this level of improvement at a 5% significance threshold. For the secondary analysis, assuming that the benefit of added glucagon is higher in the hypoglycemia unaware group, an equal number of hypoglycemia aware participants were expected to provide enough statistical power.

Analyses were performed on an intention-to-treat basis. A multivariate linear mixed effect model was used to estimate the differences between the two treatments (dual-hormone artificial pancreas vs. single-hormone artificial pancreas) on the main end points (percentage of time plasma glucose was below 4.0 mmol/L), adjusting for study periods as fixed effects. A separate linear mixed model was also built with the inclusion of a period by treatment interaction term to test for the presence of carryover effects. The model is suited for repeated observations (i.e., adjusts for patient-level intracorrelation). A linear model adjusting for study periods was also used to estimate the difference between hypo aware and unaware participants in the percentage of time below 4.0 mmol/L in each artificial pancreas system, the paired difference between systems, and total glucagon delivery in the dual-hormone system. Statistical significance was defined as a two-sided level of 0.05. A similar statistical strategy employed for analysis of the primary outcome was used to compare treatment effects for all continuous secondary outcomes.

Results

We invited 116 patients to participate in this study and 58 declined or did not answer. Five participants were excluded for scoring 3 on the Clark questionnaire²⁶ (neither hypo unaware nor aware). Thirteen participants were excluded for having fewer than two hypoglycemic events during the run-in period. One hypoglycemia unaware participant was excluded due to the inability to cannulate a vein. Two hypoglycemia aware participants were excluded for completing only one of two study intervention nights. Table 1 shows the baseline characteristics of the 35 participants who participated in the study from November 28, 2014, to August 2, 2016. All 35 analyzed participants completed both study intervention nights.

Table 1.

Baseline Characteristics of Study Participants (Mean ± Standard Deviation)

	Total study group (n = 35)	Hypo unaware group (n = 18)	Hypo aware group (n = 17)
Age (years)	45.6 ± 17.7	51.4 ± 16.2	39.5 ± 17.5
Gender (females)	17	9	8
Clarke score	—	5.0 ± 1.0	1.4 ± 0.9
BMI (kg/m²)	26.0 ± 3.3	26.7 ± 3.8	25.2 ± 2.6
HbA1c (%)	7.7 ± 0.7	7.8 ± 0.7	7.6 ± 0.5
Total daily dose (U)	43.6 ± 17.1	42.0 ± 19.2	45.3 ± 15.0
Total daily dose (U/kg)	0.58 ± 0.19	0.54 ± 0.18	0.63 ± 0.18
Diabetes duration (years)	26.9 ± 16.1	31.0 ± 16.0	22.5 ± 15.5

BMI, body–mass index.

During the run-in period, participants in the hypoglycemia unaware group had a mean of 0.36 hypoglycemic events per night (sensor glucose <3.5 mmol/L for at least 20 consecutive min), which is comparable with previously reported rates of nocturnal hypoglycemia.⁶ The hypoglycemia aware participant group had a mean of 0.25 hypoglycemic events per night, and overall, the average across the two groups was 0.31 hypoglycemic events per night.

During the intervention nights, 22 of the 35 participants chose to deliver an insulin bolus for the bedtime snack (20 g of carbohydrate). The boluses were determined based on the participants' insulin-to-carbohydrate ratio and pre-snack glucose levels. Median (interquartile range [IQR]) insulin delivery from the snack bolus was 1.8 U (0.3–4.5 U) during single-hormone artificial pancreas nights and 1.8 U (0.4–4.5 U) during dual-hormone artificial pancreas nights. Figure 1 shows plasma glucose profiles on study interventions.

FIG. 1.

Plasma glucose levels of participants with hypoglycemia unawareness (left) and hypoglycemia awareness (right). Bold line: median plasma glucose; dotted lines: interquartile range of plasma glucose; dashed lines: median basal insulin infusion; vertical lines: total glucagon infused. The gray zone represents the target range. Units are shown in the figure. The figure gives the impression that insulin and glucagon were administered simultaneously to participants. The lines represent medians for all participants, and insulin infusion is stopped by the AP algorithm before glucagon is administered. AP, artificial pancreas.

In participants with hypoglycemia unawareness, none of the hypoglycemic outcomes differed between the single- and dual-hormone artificial pancreas systems (P = NS; Table 2). The median (IQR) percentage of time that plasma glucose was below 4.0 mmol/L was 0% (0–0) during dual-hormone interventions and 0% (0–10) during single-hormone interventions (P = 0.20). In the dual-hormone nights, 15 (83%) participants had no glucose measurement below 4.0 mmol/L, 1 (6%) participant had less than 15% of glucose measurements below 4.0 mmol/L, and 2 (11%) participants had more than 15% of their glucose measurements below 4.0 mmol/L (Table 2). In the single-hormone nights, the number of participants with no, <15%, and >15% of glucose measurements below 4.0 mmol/L were 11 (61%), 4 (22%), and 3 (17%), respectively (Table 2). There were three hypoglycemic events (<3 mmol/L) on single-hormone nights and two events on dual-hormone nights. We treated each event with 16 g of oral carbohydrate. A second treatment was needed for three hypoglycemic events, one of which needed a third treatment. Glucose variability, hyperglycemic outcomes, mean glucose levels, and the times spent in the target range also did not differ between the interventions (P = NS; Table 2).

Table 2.

Comparison of Primary and Secondary Outcomes Between Single-Hormone and Dual-Hormone Artificial Pancreas in Hypoglycemia Unaware Participants (N = 18) During Both Nocturnal 10-H Intervention Periods

	Single-hormone	Dual-hormone	P
Hypoglycemic outcomes
Percent time below 4 mmol/L (primary outcome)			0.20
None, n (%)	11 (61)	15 (83)
≤15%, n (%)	4 (22)	1 (6)
>15%, n (%)	3 (17)	2 (11)
Percent time below 3.5 mmol/L			0.23
None, n (%)	13 (72)	16 (89)
≤10%, n (%)	3 (17)	0 (0)
>10%, n (%)	2 (11)	2 (11)
Percent time below 3.3 mmol/L			0.52
None, n (%)	15 (83)	16 (89)
≤10%, n (%)	1 (6)	1 (6)
>10%, n (%)	2 (11)	1 (6)
Area under the curve below 4.0 mmol/L			0.22
None, n (%)	11 (61)	15 (83)
≤3, n (%)	3 (17)	1 (6)
>3, n (%)	4 (22)	2 (11)
Area under the curve below 3.5 mmol/L			0.29
None, n (%)	13 (72)	16 (89)
≤2, n (%)	3 (17)	0 (0)
>2, n (%)	2 (11)	2 (11)
Area under the curve below 3.3 mmol/L			0.54
None, n (%)	15 (83)	16 (89)
≤1, n (%)	1 (6)	1 (6)
>1, n (%)	2 (11)	1 (6)
No. of subjects experiencing a hypoglycemic event requiring oral treatment			NA^a
N (%)	3 (17)	2 (11)
Target ranges, median (IQR)
Percent time between 4 and 8 mmol/L	78 (37–99)	61 (38–100)	0.45
Percent time between 4 and 10 mmol/L	97 (81–100)	82 (68–100)	0.22
Hyperglycemic outcomes, median (IQR)
Percent time above 8 mmol/L	15 (0–54)	29 (0–62)	0.72
Percent time above 10 mmol/L	0 (0–5)	5 (0–32)	0.17
Area under the curve above 8 mmol/L	11.5 (0.0–25.8)	32.0 (0.0–92.1)	0.46
Area under the curve above 10 mmol/L	0.0 (0.0–0.2)	1.9 (0.0–44.1)	0.12
Glucose variability, mean ± SD
SD of glucose level (mmol/L)	1.4 ± 0.9	1.5 ± 1.0	0.61
CV of glucose level (%)	21 ± 14	20 ± 11	0.82
Mean glucose level (mmol/L), mean ± SD	6.8 ± 1.4	7.7 ± 2.2	0.09
Total insulin delivery (U), mean ± SD	7.0 ± 2.4	8.0 ± 3.5	0.05

All reported P-values are from a linear mixed model adjusting for study periods as fixed effects. Area under the curve (mmol/L × min/h).

Formal statistical comparison was not conducted due to low incidence of hypoglycemia.

CV, coefficient of variation; IQR, interquartile range; SD, standard deviation.

Similarly, in participants with hypoglycemia awareness, none of the hypoglycemic outcomes differed between the single- and dual-hormone artificial pancreas systems (P = NS; Table 3). The median (IQR) percentage of time plasma glucose stayed below 4.0 mmol/L was 0% (0–0) during dual-hormone interventions and 0% (0–0) during single-hormone interventions (P = 0.95). There was one hypoglycemic event (<3 mmol/L) on single-hormone nights and zero events on dual-hormone nights. The median percentage of time spent in the target range—plasma glucose between 4.0 and 8.0 mmol/L—was greater during dual-hormone interventions compared with single-hormone interventions (84% vs. 58%, P = 0.007; Table 3). Hyperglycemic outcomes—plasma glucose above 8.0 mmol/L (P = 0.02) and plasma glucose above 10.0 mmol/L (P = 0.04)—were reduced during dual-hormone interventions compared with single-hormone interventions (Table 3). Mean plasma glucose was 6.8 ± 1.1 mmol/L during dual-hormone interventions and 7.9 ± 1.3 mmol/L during single-hormone interventions (P = 0.01; Table 3).

Table 3.

Comparison of Primary and Secondary Outcomes Between Single-Hormone and Dual-Hormone Artificial Pancreas in Hypoglycemia Aware Participants (N = 17) During Both Nocturnal 10-H Intervention Periods

	Single-hormone	Dual-hormone	P
Hypoglycemic outcomes
Percent time below 4 mmol/L			0.95
None, n (%)	14 (82)	14 (82)
≤15%, n (%)	2 (12)	2 (12)
>15%, n (%)	1 (6)	1 (6)
Percent time below 3.5 mmol/L			0.68
None, n (%)	16 (94)	15 (88)
≤10%, n (%)	1 (6)	2 (12)
>10%, n (%)	0 (0)	0 (0)
Percent time below 3.3 mmol/L			0.79
None, n (%)	16 (94)	16 (94)
≤10%, n (%)	1 (6)	1 (6)
>10%, n (%)	0 (0)	0 (0)
Area under the curve below 4.0 mmol/L			0.95
None, n (%)	14 (82)	14 (82)
≤3, n (%)	2 (12)	2 (12)
>3, n (%)	1 (6)	1 (6)
Area under the curve below 3.5 mmol/L			0.82
None, n (%)	16 (94)	15 (88)
≤2, n (%)	0 (0)	2 (12)
>2, n (%)	1 (6)	0 (0)
Area under the curve below 3.3 mmol/L			0.65
None, n (%)	16 (94)	16 (94)
≤1, n (%)	1 (6)	1 (6)
>1, n (%)	0 (0)	0 (0)
No. of subjects experiencing a hypoglycemic event requiring oral treatment			NA^a
N (%)	1 (6)	0 (0)
Target ranges, median (IQR)
Percent time between 4 and 8 mmol/L	58 (41–79)	84 (72–95)	0.007
Percent time between 4 and 10 mmol/L	77 (66–100)	100 (85–100)	0.04
Hyperglycemic outcomes, median (IQR)
Percent time above 8 mmol/L	40 (21–59)	16 (0–25)	0.02
Percent time above 10 mmol/L	17 (0–33)	0 (0–14)	0.04
Area under the curve above 8 mmol/L	49.6 (7.6–69.8)	4.3 (0.0–33.3)	0.02
Area under the curve above 10 mmol/L	11.3 (0.0–34.2)	0.0 (0.0–8.8)	0.04
Glucose variability, mean ± SD
SD of glucose level (mmol/L)	1.8 ± 0.9	1.4 ± 0.8	0.09
CV of glucose level (%)	23 ± 10	20 ± 12	0.33
Mean glucose level (mmol/L), mean ± SD	7.9 ± 1.3	6.8 ± 1.1	0.01
Total insulin delivery (U)^b	9.0 ± 3.5	7.8 ± 2.7	0.09

All reported P-values are from a linear mixed model adjusting for study periods as fixed effects. Area under the curve (mmol/L × min/h).

Formal statistical comparison was not conducted due to low incidence of hypoglycemia.

A significant carryover effect was observed for total insulin delivery.

More glucagon was delivered to participants with hypoglycemia awareness compared with participants with hypoglycemia unawareness (P = 0.03; Table 4). The percentage of time that plasma glucose was below 4.0 mmol/L was not different between the two groups in either artificial pancreas system (P = NS; Table 4).

Table 4.

Comparisons of Outcomes Between Hypoglycemia Aware and Hypoglycemia Unaware Participants

	Hypoglycemia aware (N = 17)	Hypoglycemia unaware (N = 18)	P
Percent time below 4 mmol/L
Single-hormone			0.15
None, n (%)	14 (82)	11 (61)
≤15%, n (%)	2 (12)	4 (22)
>15%, n (%)	1 (6)	3 (17)
Dual-hormone			0.96
None, n (%)	14 (82)	15 (83)
≤15%, n (%)	2 (12)	1 (6)
>15%, n (%)	1 (6)	2 (11)
Paired difference in percent time below 4 mmol/L between AP systems			0.30
<0, n (%)	3 (18)	7 (39)
None, n (%)	11 (65)	9 (50)
>0, n (%)	3 (18)	2 (11)
Total glucagon delivery (U) in the dual-hormone AP system			0.03
None, n (%)	14 (82)	9 (50)
≤4, n (%)	3 (18)	3 (17)
>4, n (%)	0 (0)	6 (33)

All reported P-values are from a linear model adjusting for study periods.

AP, artificial pancreas.

The paired difference between the two artificial pancreas systems in the percentage of time that plasma glucose was below 4.0 mmol/L did not differ between participants with and without hypoglycemia unawareness (P = NS; Table 4). When combining data from participants with and without hypoglycemia unawareness, none of the glucose outcomes differed between the single-hormone and the dual-hormone artificial pancreas nights (data not shown).

Discussion

We conducted a randomized crossover trial comparing the efficacy of dual- and single-hormone artificial pancreas systems in controlling plasma glucose levels in adults with T1D, and for the first time, we specifically examined patients with hypoglycemia unawareness and documented nocturnal hypoglycemia.

The argument most supporting of dual-hormone artificial pancreas systems is that unopposed insulin action can lead to hypoglycemia. Patients with hypoglycemia unawareness have a defective counter-regulatory hormone response to hypoglycemia (including impaired glucagon secretion²⁷) and thus they could, in theory, derive greater benefit from a dual-hormone artificial pancreas. However, we were not able to demonstrate a significant benefit from the dual-hormone system in hypoglycemia unaware participants, and hypoglycemic outcomes were low and comparable between both artificial pancreas systems and between participant groups. Each intervention was conducted over the course of a single night. Hypoglycemia remains common during the day, particularly among patients with varying meal, exercise, and work routines. It is possible that adding glucagon may benefit hypoglycemia unaware patients during the day or that longer studies are required to show improvements in glycemic control and reduction in hypoglycemia. Adding glucagon may also benefit T1D patients in specific situations, for example, by reducing late postprandial hypoglycemia due to the prolonged action of current insulin or by targeting lower mean glucose more aggressively without fear of hypoglycemia.²⁸ It may also be helpful during periods of greater glucose variability, such as illness or intensive exercise.²⁹ This warrants further investigation.

Our study had several limitations. First of all, each study intervention lasted only one night and was conducted in a controlled inpatient setting, which can only partially simulate a real at-home overnight environment and thus limit the generalizability of our results to outpatient settings. Five hypoglycemia unaware and two hypoglycemia aware participants used their own real-time continuous glucose monitors during the run-in period. It is unlikely that the use of continuous glucose monitors by some participants affected our results because we still required participants to experience two or more hypoglycemic events to be included in the study (removing the sensor would likely have led to more hypoglycemia and thus these patients would have been included anyway). Exercise was not advised on the day of the study intervention, and we cannot exclude the possibility of variable endogenous glycogen storage among participants since nutritional intake was not controlled before the overnight visit. Finally, hypoglycemia unaware patients generally avoid hypoglycemia through intentional underinsulinization and/or preventative snacking. Providers often target higher HbA1C levels as well.^30,31 These factors may have inherently limited hypoglycemic events, which our study was trying to capture.

Conceivably, dual-hormone systems could provide additional safety from potentially life-threatening hypoglycemia, but single-hormone systems benefit from reduced cost and device complexity. While certain patients and settings may justify the use of a dual-hormone artificial pancreas, such as in exercising adults²⁹ or pediatric patients at camp,²³ our study suggests that single-hormone artificial pancreas systems may be sufficient for overnight control, even in a high-risk group of hypoglycemia unaware participants with documented nocturnal hypoglycemia.

Footnotes

Acknowledgments

The authors thank their study participants for their time and involvement and the JDRF and Fondation JA DèSeve for providing funding for this project. The authors extend special thanks to the staff, study coordinators, and nurses at l'Institut de Recherches Cliniques de Montréal for their support and to Elena Weissmann for helping edit the manuscript. Dr. Abitbol would also like to mention that the undertaking of this study began during his endocrinology residency training and was made possible through the guidance of the Division of Endocrinology and Metabolism at the McGill University Health Centre. Funding source: Medtronic provided pumps and sensors in-kind. No sponsor had any role in the study design, data collection, data interpretation, or writing of the report. Study investigators held responsibility for the final decision to submit for publication. The corresponding and senior author had access to all data and takes responsibility for the integrity of the data and accuracy of data analysis.

Author Disclosure Statement

A.A.: Research Support: Astra-Zeneca, Boehringer Ingelheim, Eli Lilly, Gilead, Lexicon, Merck, Novo Nordisk, Pfizer, Sanofi, Senseonics. Speaker/Consulting/Honoraria: Astra-Zeneca, Eli Lilly, Janssen, Merck, Novo Nordisk, Sanofi, Valeant. N.C.: Grants: JDRF. A.H.: Consulting/Grants: AgaMatrix, Eli Lilly, Medtronic. Purchase fees from Eli Lilly. L.L.: Consulting/Grants: Eli Lilly, Medtronic, Merck, Novo-Nordisk, Sanofi. Purchase fees from Eli Lilly. R.R.-L.: Consulting/Grants: Astra-Zeneca, Becton Dickinson, Bohringer, Eli Lilly, Janssen, Insulet, Lifescan, Medtronic, Merck, Novartis, Neomed, Novo-Nordisk, Roche, Sanofi-Aventis, Takeda. Purchase fees from Eli Lilly. All other authors have no competing financial interests.

References

Diabetes Control and Complications Trial Research Group, Nathan

, Genuth

, et al.: The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. N Engl J Med, 1993; 329:977–986.

Yale

: Nocturnal hypoglycemia in patients with insulin-treated diabetes. Diabetes Res Clin Pract, 2004; 65 Suppl 1:S41–S46.

Weinstock

, Xing

, Maahs

, et al.: Severe hypoglycemia and diabetic ketoacidosis in adults with type 1 diabetes: results from the T1D exchange clinic registry. J Clin Endocrinol Metab, 2013; 98:3411–3419.

Cryer

: Hypoglycemia in Diabetes: Pathophysiology, Prevalence and Prevention. Alexandria, VA: American Diabetes Association, 2016.

Geddes

, Schoppman

, Zammitt

, et al.: Prevalence of impaired awareness of hypoglycaemia in adults with type 1 diabetes. Diabet Med, 2008; 25:501–504.

Desjardins

, Brazeau

, Strychar

, et al.: Association between post-dinner dietary intakes and nocturnal hypoglycemic risk in adult patients with type 1 diabetes. Diabetes Res Clin Pract, 2014; 106:420–427.

Sovik

, Thordarson

: Dead-in-bed syndrome in young diabetic patients. Diabetes Care, 1999; 22 Suppl 2:B40–B42.

Gill

, Woodward

, Casson

, et al.: Cardiac arrhythmia and nocturnal hypoglycaemia in type 1 diabetes—the “dead in bed” syndrome revisited. Diabetologia, 2009; 52:42–45.

Bergenstal

, Klonoff

, Garg

, et al.: Threshold-based insulin-pump interruption for reduction of hypoglycemia. N Engl J Med, 2013; 369:224–232.

10.

Dauber

, Corcia

, Safer

, et al. Closed-loop insulin therapy improves glycemic control in children aged <7 years: a randomized controlled trial. Diabetes Care, 2013; 36:222–227.

11.

Hovorka

, Allen

, Elleri

, et al.: Manual closed-loop insulin delivery in children and adolescents with type 1 diabetes: a phase 2 randomised crossover trial. Lancet, 2010; 375:743–751.

12.

Hovorka

, Kumareswaran

, Harris

, et al.: Overnight closed loop insulin delivery (artificial pancreas) in adults with type 1 diabetes: crossover randomised controlled studies. BMJ, 2011; 342:d1855.

13.

Kovatchev

, Cobelli

, Renard

, et al.: Multinational study of subcutaneous model-predictive closed-loop control in type 1 diabetes mellitus: summary of the results. J Diabetes Sci Technol, 2010; 4:1374–1381.

14.

Phillip

, Battelino

, Atlas

, et al.: Nocturnal glucose control with an artificial pancreas at a diabetes camp. N Engl J Med, 2013; 368:824–833.

15.

Steil

, Palerm

, Kurtz

, et al.: The effect of insulin feedback on closed loop glucose control. J Clin Endocrinol Metab, 2011; 96:1402–1408.

16.

Haidar

.: The artificial pancreas: how closed-loop control is revolutionizing diabetes. IEEE Control Systems, 2016; 36:28–47.

17.

Haidar

, Legault

, Dallaire

, et al.: Glucose-responsive insulin and glucagon delivery (dual-hormone artificial pancreas) in adults with type 1 diabetes: a randomized crossover controlled trial. CMAJ, 2013; 185:297–305.

18.

Haidar

, Legault

, Messier

, et al.: Comparison of dual-hormone artificial pancreas, single-hormone artificial pancreas, and conventional insulin pump therapy for glycaemic control in patients with type 1 diabetes: an open-label randomised controlled crossover trial. Lancet Diabetes Endocrinol, 2015; 3:17–26.

19.

Haidar

, Rabasa-Lhoret

, Legault

, et al.: Single-and dual-hormone artificial pancreas for overnight glucose control in type 1 diabetes. J Clin Endocrinol, 2016; 101:214–223.

20.

Haidar

, Farid

, St-Yves

, et al.: Post-breakfast closed-loop glucose control is improved when accompanied with carbohydrate-matching bolus compared to weight-dependent bolus. Diabetes Metab, 2014; 40:211–214.

21.

Gingras

, Rabasa-Lhoret

, Messier

, et al.: Efficacy of dual-hormone artificial pancreas to alleviate the carbohydrate-counting burden of type 1 diabetes: a randomized crossover trial. Diabetes Metab, 2016; 42:47–54.

22.

Haidar

, Messier

, Legault

, et al.: Outpatient 60‐hour day‐and‐night glucose control with dual‐hormone artificial pancreas, single‐hormone artificial pancreas, or sensor‐augmented pump therapy in adults with type 1 diabetes: an open‐label, randomised, crossover, controlled trial. Diabetes Obes Metab, 2017; 19:713–720.

23.

Haidar

, Legault

, Matteau-Pelletier

, et al.: Outpatient overnight glucose control with dual-hormone artificial pancreas, single-hormone artificial pancreas, or conventional insulin pump therapy in children and adolescents with type 1 diabetes: an open-label, randomised controlled trial. Lancet Diabetes Endocrinol, 2015; 3:595–604.

24.

Taleb

, Coriati

, Khazzaka

, et al.: Stability of commercially available glucagon formulation for dual-hormone artificial pancreas clinical use. Diabetes Technol Ther, 2017; 19:589–594.

25.

Kingman

, Robic

, Buckingham

, et al.: Restoration of Hypoglycemia Awareness with Closed-Loop Therapy. Alexandria, VA. American Diabetes Association, 2017.

26.

Clarke

, Cox

, Gonder-Frederick

, et al.: Reduced awareness of hypoglycemia in adults with IDDM. A prospective study of hypoglycemic frequency and associated symptoms. Diabetes Care, 1995; 18:517–522.

27.

Cryer

: Mechanisms of hypoglycemia-associated autonomic failure in diabetes. N Engl J Med, 2013; 369:362–372.

28.

Gingras

, Taleb

, Roy-Fleming

, et al.: The challenges of achieving postprandial glucose control using closed‐loop systems in patients with type 1 diabetes. Diabetes Obes Metab, 2018; 20:245–256.

29.

Taleb

, Emami

, Suppere

, et al.: Efficacy of single-hormone and dual-hormone artificial pancreas during continuous and interval exercise in adult patients with type 1 diabetes: randomised controlled crossover trial. Diabetologia, 2016; 59:2561–2571.

30.

Canadian Diabetes Association Clinical Practice Guidelines Expert Committee, Imran

, Rabasa-Lhoret

, Ross

: Targets for glycemic control. Can J Diabetes, 2013; 37 Suppl 1:S31–S34.

31.

Desjardins

, Brazeau

, Strychar

, Rabasa-Lhoret

: Are bedtime nutritional strategies effective in preventing nocturnal hypoglycaemia in patients with type 1 diabetes?. Diabetes Obes Metab, 2014; 16:577–587.