Patients with diabetes mellitus type 1 (DM1) have lost their ability to produce and secrete insulin due to the selective destruction of beta cells in the pancreatic islets of Langerhans. Insulin is necessary for a wide range of physiological processes and is mandatory for keeping blood glucose concentrations within a narrow range. Thus, patients with DM1 depend entirely on exogenous insulin, and intensive insulin treatment and tight glucose control are recommended to reduce the risk of microvascular complications (Reichard, 1992; Nathan et al., 1993; Reichard et al., 1996). Most patients with DM1 use multiple daily subcutaneous (SC) insulin injections or continuous SC insulin infusions administered via a pump. Recently, double SC artificial pancreas devices have also been introduced as a treatment option, i.e., devices that measure SC glucose concentrations continuously and calculate and deliver suitable insulin doses automatically (Peyser et al., 2014). Currently, only hybrid artificial pancreas devices that require meal annunciations from the user are commercially available (Knebel and Neumiller, 2019). As they require frequent user intervention, patients using these hybrid devices are not fully relieved of the daily focus and stress of managing their disease, which is known to be a factor that causes distress and reduced mental well-being among patients with DM1 (Pallayova and Taheri, 2014).

Administering appropriate meal-time insulin doses is a major daily challenge for patients with DM1, and delayed or miscalculated insulin dosages often results in postprandial glucose excursions and poor glucose control (Robinson et al., 2021). Timing and adjusting insulin dosages is difficult because all insulins, even the most rapid-acting insulin analogues, are absorbed relatively slowly from the SC tissue, with an onset of effect after 10–30 min and a maximum glucose-lowering effect one to 3 hours after administration (AHFS Drug Information, Insulins General Statement, 2022a). Whereas in healthy individuals, postprandial blood glucose concentrations have returned to baseline within the same period (Hiyoshi et al., 2017).

Patients with DM1 are advised to inject meal-time insulin 10–20 min before they start eating to accommodate the insulin absorption delay. However, despite this precaution, glucose concentrations are often strongly elevated after meals (Birkeland, 2006). If fully automated artificial pancreas devices that require no user intervention are to become a reality, this issue must be resolved, as such devices evidently must detect ingestion of a meal before increasing insulin infusion.

A fully automated SC artificial pancreas device may only upscale the insulin infusion rate in relation to a meal once a meal is detected by a rise in the SC glucose concentration or through another sensing modality. As such, two major barriers exist that prevent the development of such a device: The delay in sensing changes in blood glucose concentrations from a sensor in SC tissue and, more importantly, the delayed absorption, and hence the glucose-lowering effect of SC delivered insulin (Christiansen et al., 2017; Gingras et al., 2018).

Although intensive insulin treatment is recommended to reduce long-term microvascular complications in patients with DM1, it increases the risk of severe hypoglycaemia, an acute and dramatic condition that can be life-threatening if left untreated (Nathan et al., 1993). The fear of hypoglycaemia significantly reduces the quality of life for patients with DM1 and prevents tight glucose control (Chatwin et al., 2021).

In healthy individuals, the alpha cells in the pancreatic islets of Langerhans secrete glucagon as a response to declining blood glucose concentrations to stimulate hepatic glucose output and prevent hypoglycaemia. In patients with DM1, this pancreatic response is reduced or even absent, which makes them susceptible to hypoglycaemic episodes (Gerich et al., 1973).

SC bihormonal artificial pancreas devices that utilise both insulin and glucagon can potentially allow more aggressive insulin treatment than unihormonal devices using only insulin, as glucagon could counteract insulin-induced hypoglycaemia.

Studies comparing bihormonal and unihormonal SC artificial pancreas devices have reported that the bihormonal approach reduces the incidence and duration of daytime hypoglycaemic episodes. However, evidence of long-term improvements in the time spent in the normoglycaemic range and effects on HbA1c is lacking (Bakhtiani et al., 2013; Weisman et al., 2017; Haidar, 2019).

For decades, it has been known that large doses of glucagon have physiologic effects beyond increasing glucose output. For instance, it may reduce gastrointestinal motility, increase heart contractility and cause vasodilation (Farah, 1983).

Glucagon has previously been shown to cause up to a 500 per cent increase in SC blood flow after SC injection of a large dose (1 mg) (Simmons and Williams, 1992). Our research group recently discovered that SC injections of much smaller doses of glucagon (0.1 and 0.01 mg) could also cause a substantial increase in local SC blood flow (Åm et al., 2022). In this study, glucagon was injected SC on the abdomen in healthy individuals, and blood flow was evaluated by laser doppler technology. The median increase in SC blood flow was 250 per cent after using a dosage of 0.1 mg glucagon. The blood flow peaked at 2–4 min after injection before declining slowly, and the effect was still present 30 min after injection. Although injection of glucagon caused increased SC blood flow in most of the study subjects, the effect was not observed in seven of the 22 included participants, which were classified as non-responders (Åm et al., 2022).

The effect of glucagon on SC blood flow appears to be local and is not present 2 cm lateral to the glucagon injection site as evaluated by both laser doppler technology and a thermal camera (unpublished data). This observation is perhaps unsurprising, as SC blood supply is organised vertical to the skin’s surface, with arterioles situated 1.5–7 mm apart depending on the area of the body (Braverman et al., 1992; Raju et al., 2012).

Unlike insulin, glucagon is absorbed rapidly after SC administration, reaching maximum plasma concentrations in humans at approximately 10–20 min (Simmons and Williams, 1992; Graf et al., 1999; El-Khatib et al., 2010; El Youssef et al., 2014; Blauw et al., 2016; Castle et al., 2016; Ranjan et al., 2016; Hövelmann et al., 2018; Hövelmann et al., 2019). Previous animal trials conducted by our research group demonstrated no significant difference in the absorption speed or glucose elevating effect of glucagon after SC compared to intraperitoneal administration (Dirnena-Fusini et al., 2018; Åm et al., 2020; Teigen et al., 2022). This observation contrasts with the major delay in insulin absorption and effect after SC administration compared to intraperitoneal administration (Christiansen et al., 2017). Therefore, we hypothesise that the local vasodilatory effect of glucagon promotes its absorption from the SC space, as this would explain why there is no significant difference in absorption or the time to onset and maximum effect of SC versus intraperitoneally delivered glucagon (Teigen et al., 2022).

As previously addressed, it is probably impossible to achieve excellent glucose control by a fully automated SC artificial pancreas device without any feature that can enhance insulin absorption.

If it is confirmed in future trials that micro-doses of glucagon increase local SC blood flow in patients with DM1, this effect can be utilised in SC bihormonal artificial pancreas devices where tubes that deliver insulin and glucagon could lie in immediate proximity (i.e., the hormones are delivered via a dual-lumen line). Micro-doses of glucagon could be given at the same time as meal boluses of insulin to promote postprandial insulin absorption and larger doses of glucagon may be administered when indicated to treat and prevent hypoglycaemia.

The glucagon dosages that have been used previously in SC bihormonal artificial pancreas devices to counteract hypoglycaemia resemble those we have used to demonstrate increased SC blood flow in healthy subjects (Haidar, 2019). An obvious objection to using glucagon to enhance insulin absorption is that glucagon exerts antagonistic pharmacodynamic effects on glucose metabolism to insulin and may increase blood glucose concentrations, which is highly undesirable in the postprandial state. However, in studies investigating the potential benefit of adding glucagon to artificial pancreas devices, glucagon has been administered to prevent hypoglycaemia and not with insulin infusions or in relation to meals. Glucagon’s ability to increase blood glucose concentrations seems to be influenced by the concentration of circulating insulin, and it is most effective in situations with impending hypoglycaemia and low systemic insulin concentrations (Castle et al., 2010; Bakhtiani et al., 2015). This is the opposite of the situation in which we suggest using micro-doses of glucagon to increase blood flow, where insulin concentrations are increasing.

Further, there is evidence supporting a meal-induced glucagon release from the intestine in subjects with DM1 (Cryer, 2012; Knop, 2018; Yosten, 2018; Bengtsen and Moller, 2021; Ito et al., 2021). This glucagon release appears within minutes of oral glucose or food ingestion. Glucagon originating from the intestines is drained directly via the portal system to the liver, where it might increase glucose output. Results from a previous study by our research group support a saturable first-pass metabolism of glucagon in the liver (Teigen et al., 2021). Therefore, we hypothesise that micro-doses of glucagon administered with meal-time insulin would probably have no additional effects on postprandial glucose excursions as the liver will already be saturated with glucagon from the intestine.

In the study by Åm et al., there was a substantial proportion of non-responders to glucagon (Åm et al., 2022). The extent of interindividual variation and the mechanisms causing non-response are not yet known and should be elucidated in future studies. Individual response to glucagon will need to be confirmed before using it as part of a patient’s artificial pancreas algorithm.

We hypothesise that bihormonal SC artificial pancreas devices may use micro-doses of glucagon to induce local vasodilation and accelerate SC insulin absorption after meals. We suggest investigating the effects of micro-dosages of glucagon that are administered together with insulin boluses via a dual-lumen delivery line by the artificial pancreas when it detects a meal. The glucagon dose should be the minimum dose required to achieve adequate vasodilation while the insulin dose should be adjusted according to the effect on glucose levels. The same glucagon delivery line may also be used to deliver conventionally sized doses of glucagon during hypoglycaemic events when the insulin infusion rate is low or discontinued.

Human glucagon is relatively unstable and may aggregate and form amyloid fibrils quite rapidly, depending on the solution’s temperature and pH-value. Therefore, human glucagon is recommended to be used within 24 h after reconstitution in artificial pancreas devices, although there is some evidence indicating that reconstituted human glucagon could be chemically stable for up to 7 days (El-Khatib et al., 2007; Taleb et al., 2017). The stable glucagon analogue dasiglucagon was approved for clinical use by the FDA in 2021, but it has not yet been granted marketing authorisation in Europe (AHFS Drug Information, Dasiglucagon Hydrochloride, 2022d). Questions regarding the vasodilative properties of different glucagon formulations, dosages of glucagon or glucagon analogues to be used, chemical compatibility with different insulin preparations and technical details of the glucagon delivery in an artificial pancreas device are obvious subjects for future animal trials and clinical research in patients with DM1.

The recent discovery of a pronounced local vasodilative effect of SC administered micro-doses of glucagon should be studied further in patients with DM1. If confirmed, a disruptive change in the use of glucagon in patients with DM1 may soon emerge as micro-doses of glucagon could be used to accelerate the absorption of meal boluses of insulin from SC tissue. This could potentially improve the performance of SC bihormonal artificial pancreas devices. Thus, a fully automated bihormonal artificial pancreas that can provide excellent glucose control without daily input from the patient may be achievable.