IN BRIEF The advantage of the insulin-and-glucagon artificial pancreas is based on the rapid effect of subcutaneous glucagon delivery in preventing hypoglycemia compared to suspension of insulin delivery. In short-term studies, the dual-hormone artificial pancreas reduced daytime hypoglycemia, especially during exercise, compared to the insulin-alone artificial pancreas, but the insulin-alone system seemed sufficient in eliminating nocturnal hypoglycemia. The comparative benefits of the single- and dual-hormone systems for improving A1C and preventing severe hypoglycemia remain unknown.
The benefit of tight glycemic control for reducing microvascular complications and all-cause mortality in type 1 diabetes is hindered by the increased risk of hypoglycemia. Fear of hypoglycemia often leads to permissive hyperglycemia, which leads to suboptimal A1C levels among individuals with type 1 diabetes (1).
The artificial pancreas is an automated insulin delivery system that combines a glucose sensor, an insulin pump, and an insulin dosing algorithm. The artificial pancreas improves time in the glycemic target range, reduces hypoglycemia, and improves quality of life in people with type 1 diabetes compared to conventional insulin pump therapy (2–4). However, clinically significant hypoglycemia and hyperglycemia still exist with the artificial pancreas (5,6). The dual-hormone artificial pancreas, delivering insulin and glucagon, has been proposed as an alternative to the insulin-alone artificial pancreas (7–10). In this article, we present the basics of the dual-hormone artificial pancreas and discuss its benefits and drawbacks compared to the insulin-alone artificial pancreas.
Rationale for the Dual-Hormone System: A Pharmacokinetics View
In healthy individuals, glucose levels are tightly controlled by the complex glucose-responsive activities of the pancreatic hormones insulin (glucose lowering) and glucagon (glucose raising) released into the portal vein, as well as cephalic insulin release, paracrine signalling between β-cells and α-cells, and exercise-induced glucagon secretion. In progressive type 1 diabetes, impaired glucagon response and gradual dysregulation of pancreatic α-cells and β-cells lead to elevated risk of hypoglycemia in the presence of exogenous insulin (11–14). Because subcutaneous glucagon is rapidly absorbed, prevention of hypoglycemia is more effective with administration of subcutaneous glucagon than with suspension of subcutaneous insulin (15). Rapid-acting subcutaneous insulin formulations have an onset of action of 10–15 minutes, time to maximal glucose excursion of 40–60 minutes, and duration of action of 4–6 hours (16). Thus, for some period after insulin suspension, previously delivered insulin via the subcutaneous route will still appear in the plasma. The relatively slow pharmacokinetic profile of subcutaneous insulin remains a limitation for avoidance of hypoglycemia during rapidly decreasing glucose levels. However, subcutaneous glucagon acts rapidly, with onset of 5 minutes and a time to peak plasma glucagon level of approximately 15–20 minutes (17–20). A dose-dependent, rapid, two- to threefold rise in glucose levels can be achieved with small doses of subcutaneous glucagon (18–20).
In the dual-hormone artificial pancreas, glucagon can be used in different approaches (21). In one approach, insulin optimizes glycemic control and mini-boluses of glucagon are added to prevent residual hypoglycemia, without a concomitant increase in insulin delivery. An alternative method provides more aggressive insulin delivery to target a lower blood glucose concentration and balances the increased risk of hypoglycemia with intermittent glucagon doses. A third option combines the two previous options, with use of glucagon to prevent hypoglycemia and to allow increased insulin delivery to reduce mean glucose levels.
The amount of daily glucagon dose depends on the specific approach used; algorithms aiming to reduce hypoglycemia use the lowest total daily glucagon dose, whereas those targeting a lower mean glucose use the most glucagon, and a combination of the two approaches leads to glucagon dosing in between these two extremes. The total daily dose of glucagon in tested dual-hormone systems has been observed to be between 0.145 and 0.82 mg/day (21). These amounts of delivery resulted in plasma glucagon levels mostly in the physiological range of 50–150 pg/mL, although occasionally induced transient hyperglucagonemia (9,17).
It is important to note that circulating plasma insulin levels affect glucagon action. In the presence of high insulin levels, the efficacy of glucagon is impaired, whereas at low or moderate insulin levels, glucagon effectively raises glucose levels in a dose-responsive manner (22–24). Thus, the addition of glucagon to a dual-hormone artificial pancreas does not allow liberal insulin dosing, nor does it completely eliminate the risk of hypoglycemia. For example, in a study of 11 individuals with type 1 diabetes using a dual-hormone system with an aggressive insulin dosing algorithm, five participants experienced hypoglycemia due to accumulation of excess insulin, despite delivery of large amounts of glucagon (17). When the experiments were repeated with a less aggressive insulin dosing algorithm, hypoglycemia was eliminated.
Efficacy of Dual-Hormone Systems
Eight published studies have directly compared single- and dual-hormone artificial pancreas systems (8,9,25–30) (Table 1). In each of these studies, the dual-hormone system used a glucagon-dosing algorithm aimed at reducing hypoglycemia, as opposed to one that allows more aggressive insulin delivery.
Summary of Results from Eight Randomized Crossover Controlled Studies With Direct Comparison of Dual- and Single-Hormone Artificial Pancreas Systems
Year (Ref.) . | Setting and Duration of Intervention . | n . | Participant Characteristics, mean (SD) . | Study Arms . | Insulin Delivery . | Glucagon, mg . | Time in Hypoglycemia (<70 or <72 mg/dL), % . | Hypoglycemia Events¥, No. per participant . | Mean Glucose, mg/dL . | |||
---|---|---|---|---|---|---|---|---|---|---|---|---|
A1C, mmol/mol;% . | Age, years . | Median (IQR) or mean (SD) . | P . | Mean (SD) . | P . | |||||||
2010 (8) | Inpatient, 9 hours or 28 hours | 13 adults | 60 (3); 7.6 (0.3) | 37 (4) | Dual-hormone | 54 U/day | 0.62/day | 1.0 (0.4)* | 0.04 | 1.1/day* | 145 (14) | NS |
Single-hormone | 48 U/day | — | 2.8 (0.7)* | 2.3/day* | 135 (16) | |||||||
2015 (9) | Inpatient, 24 hours | 30 (20 adults, 10 adolescents) | 60 (11); 7.7 (1.0) | 33 (18) | Dual-hormone | 43 U/day 9 U/night | 0.18/day 0.052/night | 1.5 (0–3.5) | 0.018 | 0.3/day 0/night | 144 (25) | 0.14 |
Single-hormone | 44 U/day 9 U/night | — | 3.1 (0.6–8.7) | 0.43/day 0/night | 139 (25) | |||||||
2015 (27) | Outpatient (camp), overnight (three nights) | 33 children | 67 (8); 8.3 (0.8) | 13 (3) | Dual-hormone | 0.9 U/hour | 0.04/night | 0.0 (0–2.4) | 0.032 | 0.0/night | 139 (31) | 0.066 |
Single-hormone | 0.9 U/hour | — | 3.1 (0–6.9) | 0.04/night | 146 (31) | |||||||
2016 (29) | Outpatient, overnight (two nights) | 28 (21 adults, 7 adolescents) | 58 (11); 7.5 (1.0) | 33 (17) | Dual-hormone | 1 U/hour | 0.07/night | 1.0 (0–8.0) | 0.34 | 0.05/night | 111 (29) | 0.32 |
Single-hormone | 1 U/hour | — | 5.0 (0–13.0) | 0.11/night | 111 (31) | |||||||
2016 (26) | Inpatient, 90 minutes† (exercise study) | 17 adults | 64 (10); 8.0 (0.9) | 37 (14) | Dual-hormone | 0.4 U/hour | 0.11/90 minutes | 0.0 (0–0) | <0.001 | 0.09/90 minutes | — | — |
Single-hormone | 0.3U/hour | — | 11.0 (0–46.7) | 0.46/90 minutes | — | |||||||
2017 (28) | Outpatient, day and night 60 hours) | 23 adults | 58 (9); 7.5 (0.8) | 41 (15) | Dual-hormone | 0.8 U/hour | 0.58/60 hours | 3.6 (1.4–4.9) | 0.072 | 0.26/60 hours 0.04/night | 142 (29) | 0.98 |
Single-hormone | 0.8 U/hour | — | 3.9 (2.2–7.0) | 0.61/60 hours 0.11/night | 142 (22) | |||||||
2018 (30) | Inpatient, overnight | 18 adults with hypoglycemia unawareness and 17 with hypoglycemia awareness | 60 (8); 7.7 (0.7) | 46 (18) | Dual-hormone | 1.0 U/hour | 0.023/night | 0.0 (0–0.1) | 0.349 | 0.06/night | 137 (33) | 0.92 |
Single-hormone | 1.0 U/hour | — | 0.0 (0–1.6) | 0.11/night | 136 (27) | |||||||
2018 (25) | Outpatient, four days | 20 adults | 58 (9.8); 7.5 (0.9) | 34 (5) | Dual-hormone | 44 U/day | 0.51/day | 1.3 (1.0) | <0.001 | — | 155 (16) | 0.062 |
Single-hormone | 43 U/day | — | 2.8 (1.7) | — | 148 (12) |
Year (Ref.) . | Setting and Duration of Intervention . | n . | Participant Characteristics, mean (SD) . | Study Arms . | Insulin Delivery . | Glucagon, mg . | Time in Hypoglycemia (<70 or <72 mg/dL), % . | Hypoglycemia Events¥, No. per participant . | Mean Glucose, mg/dL . | |||
---|---|---|---|---|---|---|---|---|---|---|---|---|
A1C, mmol/mol;% . | Age, years . | Median (IQR) or mean (SD) . | P . | Mean (SD) . | P . | |||||||
2010 (8) | Inpatient, 9 hours or 28 hours | 13 adults | 60 (3); 7.6 (0.3) | 37 (4) | Dual-hormone | 54 U/day | 0.62/day | 1.0 (0.4)* | 0.04 | 1.1/day* | 145 (14) | NS |
Single-hormone | 48 U/day | — | 2.8 (0.7)* | 2.3/day* | 135 (16) | |||||||
2015 (9) | Inpatient, 24 hours | 30 (20 adults, 10 adolescents) | 60 (11); 7.7 (1.0) | 33 (18) | Dual-hormone | 43 U/day 9 U/night | 0.18/day 0.052/night | 1.5 (0–3.5) | 0.018 | 0.3/day 0/night | 144 (25) | 0.14 |
Single-hormone | 44 U/day 9 U/night | — | 3.1 (0.6–8.7) | 0.43/day 0/night | 139 (25) | |||||||
2015 (27) | Outpatient (camp), overnight (three nights) | 33 children | 67 (8); 8.3 (0.8) | 13 (3) | Dual-hormone | 0.9 U/hour | 0.04/night | 0.0 (0–2.4) | 0.032 | 0.0/night | 139 (31) | 0.066 |
Single-hormone | 0.9 U/hour | — | 3.1 (0–6.9) | 0.04/night | 146 (31) | |||||||
2016 (29) | Outpatient, overnight (two nights) | 28 (21 adults, 7 adolescents) | 58 (11); 7.5 (1.0) | 33 (17) | Dual-hormone | 1 U/hour | 0.07/night | 1.0 (0–8.0) | 0.34 | 0.05/night | 111 (29) | 0.32 |
Single-hormone | 1 U/hour | — | 5.0 (0–13.0) | 0.11/night | 111 (31) | |||||||
2016 (26) | Inpatient, 90 minutes† (exercise study) | 17 adults | 64 (10); 8.0 (0.9) | 37 (14) | Dual-hormone | 0.4 U/hour | 0.11/90 minutes | 0.0 (0–0) | <0.001 | 0.09/90 minutes | — | — |
Single-hormone | 0.3U/hour | — | 11.0 (0–46.7) | 0.46/90 minutes | — | |||||||
2017 (28) | Outpatient, day and night 60 hours) | 23 adults | 58 (9); 7.5 (0.8) | 41 (15) | Dual-hormone | 0.8 U/hour | 0.58/60 hours | 3.6 (1.4–4.9) | 0.072 | 0.26/60 hours 0.04/night | 142 (29) | 0.98 |
Single-hormone | 0.8 U/hour | — | 3.9 (2.2–7.0) | 0.61/60 hours 0.11/night | 142 (22) | |||||||
2018 (30) | Inpatient, overnight | 18 adults with hypoglycemia unawareness and 17 with hypoglycemia awareness | 60 (8); 7.7 (0.7) | 46 (18) | Dual-hormone | 1.0 U/hour | 0.023/night | 0.0 (0–0.1) | 0.349 | 0.06/night | 137 (33) | 0.92 |
Single-hormone | 1.0 U/hour | — | 0.0 (0–1.6) | 0.11/night | 136 (27) | |||||||
2018 (25) | Outpatient, four days | 20 adults | 58 (9.8); 7.5 (0.9) | 34 (5) | Dual-hormone | 44 U/day | 0.51/day | 1.3 (1.0) | <0.001 | — | 155 (16) | 0.062 |
Single-hormone | 43 U/day | — | 2.8 (1.7) | — | 148 (12) |
Hypoglycemia events were defined as <54–59 mg/dL, except if indicated otherwise.
Treatment of hypoglycemia was administered at glucose levels <3.9 mmol/L.
60 minutes exercise and 30 minutes recovery. IQR, interquartile range; U, units.
Reduction of Hypoglycemia
Most, but not all, studies reported less overall hypoglycemia with the dual-hormone artificial pancreas than with the insulin-alone artificial pancreas (8,9,25,28). One outpatient study in 23 adults over 60 hours reported six hypoglycemic events (<54 mg/dL) with the dual-hormone system compared with 14 events with the single-hormone system (28). Another outpatient study in 20 adults over 4 days reported lower time spent in hypoglycemia (<70 mg/dL) with the dual-hormone artificial pancreas (1.3%) compared to the insulin-alone artificial pancreas (2.8%, P <0.0001) (25).
Prevention of Hypoglycemia During and After Exercise
Aerobic physical activity leads to rapid decline in glucose levels due to accelerated glucose uptake, increased insulin sensitivity, and faster insulin absorption (31). Attempting to prevent hypoglycemia by suspending insulin delivery is not always effective because the glucose-lowering effect of physical activity is usually faster than the glucose-raising effect of insulin suspension (32,33). In this context, the rapid onset of glucagon delivery may be beneficial for preventing hypoglycemia during exercise. One study compared single- and dual-hormone artificial pancreas systems during 60 minutes of moderate continuous and interval aerobic exercise sessions and showed less time in hypoglycemia (<72 mg/dL) with the dual-hormone system (time in hypoglycemia median 0 vs. 11%, P = 0.0001) (26). It is worth noting that glucagon administration in the dual-hormone system increased the plasma glucagon concentration by a factor of 2.5 in comparison to the single-hormone artificial pancreas, which is comparable to the 1.4- to 3-fold increase observed in healthy individuals performing exercise at 40–70% VO2peak (34,35). Another study over 4 days that included three moderate-intensity aerobic exercise sessions showed less time spent in hypoglycemia (<70 mg/dL) during the exercise period with the dual-hormone artificial pancreas (3.4%) compared to the single-hormone artificial pancreas (8.3%, P = 0.0009) (25).
Late-onset post-exercise hypoglycemia affects individuals with type 1 diabetes through the alterations in insulin sensitivity and glucose uptake for up to 11 hours after exercise (26–38). Two studies compared the single- and dual-hormone artificial pancreas on nights after daytime exercise and observed no difference between the two systems (9,29). Both systems nearly eliminated post-exercise nocturnal hypoglycemia (<54 mg/dL); the sufficiency of the insulin-alone system can be explained by the slow decline in glucose levels during the post-exercise period.
Prevention of Nocturnal Hypoglycemia
Several studies in adults and pediatric patients showed no difference in nocturnal hypoglycemia with the dual-hormone artificial pancreas compared to the single-hormone artificial pancreas (9,25,28,29), including in patients with hypoglycemia unawareness and documented previous nocturnal hypoglycemia (30). This observation is not related to the inefficiency of glucagon delivery overnight, but rather to the sufficiency of the single-hormone system in counteracting slow variations in nocturnal glucose levels. However, in diabetes camp settings with unusually high levels of physical activity, one study in children and adolescents showed that a dual-hormone system was more effective than a single-hormone system at reducing nocturnal hypoglycemia (27).
Severe Hypoglycemia
It is currently unknown whether the dual-hormone artificial pancreas reduces the incidence of severe hypoglycemia compared to the single-hormone artificial pancreas. It is important to note that the risk of severe hypoglycemia will likely be low with the single-hormone artificial pancreas (5,6). Thus, studies assessing the additional benefits of the dual-hormone artificial pancreas in reducing severe hypoglycemia might be more desirable in high-risk populations such as those experiencing recurrent severe hypoglycemia (39).
Reduction of Mean Glucose Levels
Adding glucagon to the artificial pancreas could reduce mean glucose levels and consequently A1C by allowing more aggressive insulin delivery (21). However, long-term studies comparing the effects of single- and dual-hormone artificial pancreas on A1C levels are lacking. Moreover, short-term studies assessing dual-hormone systems versus single-hormone systems did not use glucagon to allow aggressive insulin delivery, and thus, the comparative efficacy of each system to reduce mean glucose levels is still unknown.
Glycemic Control in the Prandial Period
Using glucagon to allow more aggressive prandial boluses to reduce postprandial glucose peaks is unlikely to be an effective strategy. One study in children investigated using 60% higher prandial insulin boluses along with glucagon mini-boluses to prevent late postprandial hypoglycemia and showed no reduction in early postprandial hyperglycemia compared to usual insulin dosing (40). Instead, the system delivered glucagon 3–4 hours after the meal to prevent late postprandial hypoglycemia due to the increased prandial insulin bolus.
Positive Behavior Change and Improved Quality of Life
More than 60% of adults with type 1 diabetes are sedentary (41). Fear of hypoglycemia has been reported as a major barrier to exercise (42). Adding an extra layer of hypoglycemia protection with glucagon may promote physical activity in patients with type 1 diabetes. In one qualitative study, participants reported that the insulin-alone artificial pancreas allowed them to exercise more freely (3). It is currently unknown whether the dual-hormone artificial pancreas would bring additional benefits in patient experience, behaviors, and psychosocial outcomes related to exercise.
Practical Considerations of the Dual-Hormone Artificial Pancreas
The dual-hormone artificial pancreas will require dual-chamber pumps, which are currently in development, but they will also require two infusion sites, additional drug manipulation, and additional patient education. The need for two infusion sites and frequent sites rotation might be a particular limitation for younger, smaller patients. Moreover, current glucagon formulations are not suitable for continuous pump use (i.e., not stable in liquid form), but several alternative, stable liquid glucagon formulations are under development (43).
Common side effects reported with high-dose rescue glucagon include nausea, vomiting, and headaches, but glucagon has the potential for pleiotropic effects as it affects gastrointestinal motility, cardiac contractility, renal function, the central nervous system, and lipolysis in adipose tissue (44). The safety of long-term low doses of novel glucagon formulations is yet to be established and will be required by the U.S. Food and Drug Administration before the approval of any dual-hormone artificial pancreas system.
In animal studies, long-term glucagon exposure resulted in increased hepatic glycogen stores (45–47). In one human study in individuals with well-controlled diabetes, glycogen stores and glucagon response were maintained after repeated glucagon administration under fasting and fed conditions (48). More data are needed, but it appears unlikely that repeated small doses of glucagon will deplete glycogen stores.
Other Dual-Hormone Systems
Pramlintide is an analog of amylin, a hormone that is co-secreted with insulin in healthy individuals but is deficient in type 1 diabetes. Injecting pramlintide at mealtimes delays gastric emptying, suppresses glucagon secretion, and increases satiety (49). Recent studies assessing the addition of adding prandial pramlintide injections to an insulin-alone artificial pancreas (50–52) reported a reduction in postprandial glucose excursions (50,51). Despite its benefits, injecting pramlintide at mealtimes may be a barrier to patient acceptance and long-term compliance (53), and thus, attempts to develop stable co-formulations of insulin and pramlintide are underway (54–56). A co-formulation would deliver pramlintide during fasting as well as at mealtimes, which aligns with the natural physiology of the pancreas. A dual-hormone artificial pancreas system that delivers basal-bolus pramlintide along with insulin with a fixed ratio, mimicking a co-formulation, has been recently tested with promising preliminary results (57).
Conclusion
The glucose-raising effect of subcutaneous glucagon delivery is faster than the effect of insulin suspension in the prevention of hypoglycemia, and thus the insulin-and-glucagon artificial pancreas is more potent in preventing hypoglycemia than the insulin-alone artificial pancreas. At times when glucose levels vary slowly, such as overnight, the insulin-alone artificial pancreas seem sufficient in preventing hypoglycemia, but the dual-hormone artificial pancreas has apparent benefits during times of rapidly changing glucose levels, such as during exercise. The dual-hormone artificial pancreas should, in theory, lead to lower A1C compared to insulin-alone artificial pancreas by allowing more aggressive insulin delivery, but studies of appropriate size and duration are needed to assess this approach.
Article Information
Duality of Interest
A.H. has received research support, intellectual property purchase fees, or consulting fees from AgaMatrix, Dexcom, Eli Lilly and Company, and Medtronic and has pending patents in the artificial pancreas area. No other potential conflicts of interest relevant to this article were reported.
Author Contributions
A.H. is the sole author and guarantor of this work and, as such, takes responsibility for the integrity and accuracy of the data presented.