Multi-Hormone AID Systems (Insulin + Glucagon)

Introduction to Multi-Hormone AID Systems

A multi-hormone automated insulin delivery (AID) system uses one controller to dose insulin plus at least one other hormone. Most of the research adds glucagon, the hormone that raises blood glucose, because glucagon delivered under the skin heads off a low faster than pausing insulin does [1]. A smaller line of work adds pramlintide, a synthetic version of the hormone amylin, to soften the rise after meals [1]. Inreda Diabetic's insulin-and-glucagon system was approved for the European market in February 2020, yet as of October 2026 the company offers it only through projects in the Netherlands [2].

For the earliest glucagon trials and Beta Bionics' bihormonal program, see our companion article on dual-hormone AID systems.

What each extra hormone is for

An insulin-only system has one answer to falling glucose: deliver less insulin. That answer is slow, because insulin already under the skin keeps acting after the pump stops, and during aerobic exercise glucose can fall faster than an insulin pause takes effect [1]. Glucagon acts quickly, though it works poorly when insulin levels are already high [1]. In short trials it reduced daytime lows, especially during exercise, while insulin-only systems seemed enough overnight [1].

Pramlintide does a different job. Amylin is released together with insulin in people without diabetes and is missing in type 1 diabetes; given at meals, pramlintide slows stomach emptying and suppresses the body's own glucagon [1]. The FDA approved it as Symlin, an injection added to mealtime insulin. Its label carries a boxed warning that using it with insulin raises the risk of severe hypoglycemia, particularly in type 1 diabetes [3].

Keeping glucagon liquid

Here's what surprised us: the hardest part of an insulin-and-glucagon pump has been the glucagon itself. In water, native glucagon forms amyloid-like fibrils that turn into firm gels, and those gels will eventually clog a continuous infusion pump [4]. One fix is dasiglucagon, an analog that stays stable in water-based solution; another dissolves native glucagon in dimethyl sulfoxide (DMSO), a solvent already used in FDA-approved injectables [4]. The FDA approved dasiglucagon as Zegalogue, but for rescue treatment of severe hypoglycemia, not for pump delivery [5].

The difference shows up in daily handling. Inreda's system uses GlucaGen, a standard glucagon that is filled into its cartridge by hand and replaced every day [2]. In a 2021 home-use pilot of the bihormonal iLet, each participant's bihormonal week ran on a single prefilled dasiglucagon cartridge, with no infusion site reactions or occlusions reported [6].

Inreda AP: glucagon without meal announcements

Inreda's system is fully closed-loop, meaning the user doesn't tell it about meals or exercise. In a 2021 randomized crossover trial, 23 adults with type 1 diabetes spent two weeks on the system and two weeks on their usual pump therapy [7]. Time in range, the share of the day spent between 70 and 180 mg/dL, was a median 86.6% on the Inreda system against 53.9% on usual care [7].

A one-year trial then tested it inside routine care at eight Dutch clinics. Of 82 adults enrolled, 71 reached the 12-month analysis, where mean time in range had risen from 55.5% before the trial to 80.3% [8]. Median scores on a diabetes-distress questionnaire fell from 30 to 10 [8]. Five serious adverse events occurred, including two severe hypoglycemic events and one stroke [8]. The trial had no comparison group, and Inreda funded it [8].

The hardware shows what a second hormone adds. According to Inreda's FAQ, the system reads two Guardian CGM sensors, one for dosing and one to check the first, and connects through two Accu-Chek infusion sets, one per hormone [2]. As of October 2026, the FAQ says the device can't be bought and is used only by adults with type 1 diabetes in those projects [2].

The iLet: cleared as insulin-only

The Beta Bionics iLet has been tested with and without glucagon. In the 2021 pilot, 10 adults averaged 72% time in range in the insulin-only week and 79% in the bihormonal week, and the authors called for much larger, longer trials [6]. The version the FDA cleared in May 2023 runs in an insulin-only configuration for people with type 1 diabetes aged 6 and older [9]. It isn't fully closed-loop either: it needs no carbohydrate counting, but the user announces each meal as "Usual," "More" or "Less" than usual [9].

Insulin plus pramlintide

Pramlintide can't simply share a reservoir with insulin. The Symlin label says mixing the two changes how the body handles both drugs, so they must be given as separate injections [3]. Haidar's group at McGill University worked around this with a system that delivers pramlintide in a fixed ratio to insulin, imitating a co-formulation [10].

In their 2020 crossover trial, each arm was a 24-hour hospital stay. Rapid insulin plus pramlintide raised time in range from 74% to 84% compared with rapid insulin alone, mostly through better daytime control, and gastrointestinal symptoms followed 6% of meals [10].

The group then removed meal input entirely. In a 2021 trial of 24 adults, insulin plus pramlintide with no meal announcements reached 74.3% time in range, against 78.1% for insulin alone with carbohydrates counted precisely by research staff [11]. That wasn't enough to show non-inferiority, though fewer participants had a low (8 against 14) [11].

A 2024 trial of 30 adults and adolescents tried a middle road: a "simple meal announcement" that triggers a fixed-size bolus instead of a carbohydrate count [12]. Over 14 days per arm, insulin plus pramlintide with simple announcements was non-inferior on time in range to carbohydrate counting. Emotional burden didn't improve, and 47% of participants reported mild gastrointestinal symptoms [12].

Three hormones, then two

The most literal multi-hormone system we found came from the same team. In 2021 they tested a fully automated controller that gave insulin and pramlintide together and added glucagon when glucose ran low, refining it over 27-hour sessions in nine adults [13]. Two rounds of changes brought first unwanted glucose swings, then worse control than the insulin-alone comparison [13]. The final version added meal detection and dropped glucagon, roughly matching insulin alone with carbohydrate counting (81% against 83% time in range) in the last three participants [13].

Nine people can't settle anything. Still, every added hormone is one more thing the algorithm has to balance.

What still stands in the way

Hardware comes first. Haidar's review lists a dual-chamber pump, two infusion sites, extra drug handling and more training, and notes that two sites could be especially hard for small children [1].

Glucagon also depends on the liver's stored sugar, glycogen. The Zegalogue label names starvation, adrenal insufficiency and chronic hypoglycemia as states where those stores may fall short [5]. For the small repeated doses an AID system gives, limited human data suggest the stores hold up [1].

The biggest gaps are long-term. As of the 2019 review, no study of adequate size had shown whether glucagon lowers A1C or prevents severe lows better than insulin-only AID, and the safety of years of low-dose glucagon had not been established [1]. For pramlintide, the open question is tolerance, since stomach symptoms appeared in all three closed-loop trials above [10][11][12]. Whether joining a multi-hormone trial makes sense for you is a decision for your care team.