Current and Future Strategies in Insulin Development and Treatment

The first therapeutic use of insulin by Best and Banting represents a milestone in diabetes therapy. More than 100 years have passed since then and there have been numerous advances in insulin development. From the first bovine insulin to the development of human insulin, neutral protamine Hagedorn (NPH), recombinant insulins, rapid-acting, ultra-rapid-acting insulins, and long-acting basal analogs of the first, second, and probably soon third generation (Fig. 1). By adding, removing, or exchanging amino acids and fatty acids, second-generation basal insulin analogs only need to be administered once a day to cover the basal insulin requirement during the night and between meals. They have a very long and consistent pharmacokinetic and pharmacodynamic profile. Amino acids are exchanged and additives such as nicotinamide and treprostinil are added for the fastest possible insulin action at mealtimes or when corrections are required (Fig. 2). In addition to insulins, the routes of administration have also evolved. Drawing up insulin from bottles and injecting it with syringes is used less and less for insulin delivery, while pens, “smart” pens, insulin pumps with and without systems for automated insulin delivery (AID) are becoming the standard of care in resourceful environments.

The advances in insulin therapy have shown to improve glycemic control, ease of administration, quality of life and reductions in the risk of diabetes-related complications. Nevertheless, diabetes therapy is still a major burden for those affected. Personalized insulin therapy with the dynamic adaptation to changing insulin requirements ensuring near-normoglycemic glucose levels without provoking hypoglycemia remains a challenge. This article focuses on the insulin products becoming available in recent years and those currently being investigated in studies.

Prandial insulin attempts to imitate physiological insulin secretion after a meal. Replacing the physiological two-phase insulin secretion into the portal circulation remains challenging. When glucose rises in people without diabetes, a lot of insulin is initially secreted quickly to intercept the initial rise in glucose. In the second phase, insulin is released in smaller quantities over a longer period. Over the last few years, mealtime insulins have been further developed from regular short-acting insulins to rapid-acting insulins and, in recent years, to ultra-rapid-acting insulins. These aim to mimic the endogenous insulin response as close as possible thereby reducing postprandial glucose spikes as well as the risk of hypoglycemia. In addition, ultra-rapid-acting insulins, but also the rapid-acting insulins, are used as corrective insulins to quickly reduce episodes of hyperglycemia.

Short-acting insulins (regular human insulin) have been an integral part of intensified insulin therapy with multiple daily injections (MDIs) reflecting a basal-bolus regime for many years. Currently, with the availability of rapid- and ultra-rapid-acting insulins they are increasingly losing ground in developed countries. Due to the slower action profile, an interval of at least 20, more likely 30 min must be observed between bolus insulin dosing and start of the meal. This requires meals to be planned and presents particular challenges for families with children. On the other hand, an insulin with a maximum effect after 2–4 h and a duration of action of more than 5–8 h makes it possible to cover a meal and a subsequent snack at the same time, for which no additional injections are required [1]. This can be useful for families with primary school children where the children do not yet calculate and inject mealtime insulin outside the family environment. Regular human insulin remains the standard for intravenous insulin therapy, as rapid-acting insulins offer no advantage here.

Rapid-acting insulin analogs are absorbed faster, enter the blood circulation faster and can therefore develop their glucose-lowering effect more quickly than regular human insulins. The duration of action is shorter, and the risk of hypoglycemia is lower. This rapid absorption from the subcutaneous tissue is achieved by modifying the amino acid sequences of the insulin. It accelerates the hexamere decomposition and prevents the association to new hexameres. Due to the predominant presence of insulin monomers and dimeres, the insulin action starts two to three times faster compared to regular human insulin. This makes it possible to reduce the time-lag of premeal bolus and start of the meal to 10–15 min which also makes it more suitable for corrections. There are three approved rapid-acting insulins for children and adults. First, insulin lispro approved without age limit, where the amino acids proline and lysine are interchanged. Second, insulin aspart approved from the age of 1 year, where the amino acid proline at B28 has been replaced by aspartic acid. The third is insulin glulisine, in which asparagine is replaced by lysine and lysine by glutamic acid in the B chain [1‒4]. Insulin glulisine is only approved for use from the age of 6, and higher rates of occlusion and symptomatic hypoglycemia with insulin glulisine than with either of the other rapid-acting analogs have been reported [5], so that it hardly plays a role in the use of insulin pump therapy. Insulin aspart and lispro are standard insulin approved for most insulin pumps. One advantage of insulin glulisine over the others is its positive effect on lipoatrophic injection sites [6].

In recent years, ultra-rapid-acting insulins have become widespread available. Due to faster monomer formation after injection, a faster onset of action can be achieved and thus better mimic the physiological insulin secretion during and after a meal. This means that with premeal application, postprandial glucose excursions can be better intercepted, hypoglycemia can even be reduced and, with better glycemic control, glycated hemoglobin and more time in range can be achieved than with rapid-acting insulins [1, 4, 7]. In ultra-rapid-acting insulin aspart, nicotinamide was added to accelerate absorption and arginine was added to stabilize the insulin molecule. The insulin exposure is two times faster and the glucose-lowering effect stronger in the first 30 min after injection compared to rapid-acting insulin aspart [8]. Faster aspart shows a significant superiority in controlling postprandial glucose [9, 10]. The use in insulin pumps does not show any inferiority compared to rapid-acting insulins, but also no clear superiority. The use of faster insulin aspart versus standard insulin aspart could only achieve a comparable glycemic outcome when used in an AID system. But with the use in an insulin pump without AID, a better postprandial glycemic control could be observed [11‒13]. The ultra-rapid-acting insulin lispro contains the additives citrate and treprostinil. These increase the vascular permeability and vasodilation of the local blood vessels allowing the insulin to act more quickly. With ultra-rapid-acting insulins, comparable or better postprandial glycemic control, more time in range, less time below and above range can be achieved compared to rapid-acting insulins [12, 14].

Premixed insulins consist of a fixed ratio of basal and prandial insulin. Normally, NPH insulins and regular human insulins are used, but products using rapid-acting insulins are also available. As a rule, premixed insulins are administered twice daily. The combination of both insulins significantly reduces the frequency of injections but is also associated with a loss of flexibility in everyday life. Meals and sporting activities must be planned in advance and cannot be organized spontaneously according to individual needs. The risk of hypoglycemia and diabetic ketoacidosis is increased [15, 16]. They are of little value in the treatment of type 1 diabetes, especially in pediatrics, where the flexibility of the basal-bolus regimen is gold standard. Premixed insulins are mainly used in geriatric patients with type 2 diabetes who have repetitive, predictable daily routines. In the GINGER study, it was shown that the basal-bolus regimen is also superior in terms of glucose control in people with type 2 diabetes [17].

NPH contains zinc and protamine additives. This delays the dissociation of insulin into monomers and thus its absorption. It has been used as a basal insulin for many years but requires at least a twice-daily injection. It shows a peak action 1–1.5 h after application, so that the risk of hypoglycemic events is increased, especially in the first half of the night when insulin sensitivity is high. In the early hours of the morning, when insulin sensitivity is low, it has a diminishing effect which often leads to high morning blood glucose levels, known as “dawn phenomenon,” which is typically known for adolescents during pre-pubertal and pubertal stage. Thus, the insulin action profile of overnight NPH is the opposite compared to the secretion of endogenous insulin in people without diabetes [18]. Therefore, the NPH profile seems to be more useful for younger children and toddlers. A further disadvantage is that the suspension must be carefully rolled back and forth before each application, which means that the inter- and intraindividual insulin concentration can vary greatly and may cause major glucose fluctuations and hypoglycemia [4]. One advantage is certainly that it can be mixed with prandial insulins in a syringe. Also, the peak action after 4–6 h can be used to intercept meals that are difficult to regulate. Nevertheless, basal insulin analogs have become widely accepted replacing NPH insulin [1].

Long-acting analog insulins have a very long, flat pharmacokinetic, and pharmacodynamic action profile and usually only need to be administered once a day. In combination with rapid- or ultra-rapid-acting insulins, they are now standard in MDI therapy in type 1 or even type 2 diabetes. The principle of prolonged action is to delay absorption from the subcutaneous tissue into the blood circulation. The first generation of long-acting basal analogs became available in the 2000s with insulin glargine 100 U followed by insulin detemir. In insulin glargine, asparagine at position A21 was exchanged for glycine and two arginine residues are added to position B30. These changes shift the isoelectric point so that insulin glargine is soluble at acidotic pH and precipitates at neutral pH. This results in slower absorption through a depot effect, i.e., a uniform, slower release is increased with a minimal peak action [7, 19]. In insulin detemir, available since 2005, a fatty acid was coupled to position B28. Although the insulin is absorbed relatively quickly and released into the bloodstream, it is bound to albumin by the fatty acid. This results in delayed clearance [4]. Compared to insulin glargine, it causes less pain and burning when administered, but the maximum effect is reached after 12 h. The duration of action depends on the dose and is only 20–24 h, so that two daily injections are usually necessary. Higher insulin doses are necessary [20], but unwanted weight gain as a side effect of insulin therapy appears to be less [21].

The second generation of basal insulin analogs was launched with insulin glargine U300 and insulin degludec 100 U. In the more concentrated formulation of insulin glargine U300, the reduced injection volume is leading to a smaller surface area of the subcutaneous insulin depot. This leads to a slower and more even release of insulin, which means that flatter, more stable, and longer action profiles can be achieved [19, 22]. It has a duration of action of 24 h but is only authorized for use from the age of 6. The EDITION clinical trial program showed a non-inferiority of insulin glargine U300 compared to U100 [23]. Glycated hemoglobin and glucose control were comparable. There are indications that there are fewer hypoglycemic events with U300, particularly severe nocturnal hypoglycemic events [24]. It has been shown to be beneficial in people with diabetes and a very high insulin resistance and insulin requirement [25].

Insulin degludec is a modified human insulin analog. The amino acid threonine was removed at position B30 and a fatty acid and glutamic acid were introduced. As a result, soluble multihexamers are formed after subcutaneous injection and zinc is only distributed slowly. This leads to stable, slow dissociation and release into the tissue. Due to the fatty acid, albumin binding is very high, which leads to delayed insulin clearance. Insulin degludec is authorized from the age of 1. The half-life is over 25 h and the duration of action is over 42 h. A steady state is reached after approximately three days. It has the advantage that it only needs to be administered once a day in a very variable time interval. An interval of 8 h must be observed between two injections. It is therefore very suitable in combination with ultra-rapid-acting insulins for adolescents with type 1 diabetes and adherence problems. Compared to glargine U100, it has a significantly lower rate of hypoglycemia, with fewer nocturnal hypoglycemia [26]. It leads to better glucose control, higher time in range and even time in tight rage [27]. Compared to insulin detemir, insulin degludec shows less frequent hyperglycemia with ketone body formation and thus has a potential preventive effect with regard to diabetic ketoacidosis [28].

Insulin icodec is the first representative of the third generation of basal analogs. A half-life of over 196 h is achieved by exchanging 3 amino acids in the insulin molecule and attaching a C20 fatty acid as a spacer that binds to the albumin. The very long pharmacokinetic effect is due to the stronger albumin binding and therefore delayed clearance as consequence to the low insulin receptor affinity. After 3–4 weeks of use, a steady state is reached in which the same amount of insulin is always released from the albumin depot. A maximum concentration is reached after approximately 16 h. Due to the long duration of action of 1 week, the injection frequency is reduced from 365 times to 52 times per year. It can be applied subcutaneously into the abdomen, upper arm or thigh. Although the maximum concentration after a single dose was higher in the upper arm and abdomen than in the thigh, there were ultimately no significant differences in exposure or results in glucose-lowering effects [29]. The effect of once-weekly insulin icodec appears to be superior to that of once-daily insulin glargine. It achieves better glycemic control, glycated hemoglobin, significantly higher time in range and time in tight range, with lower insulin doses. Hypoglycemic events seem to occur slightly more frequently, but not to a clinically relevant extent [30, 31]. The ONWARDS 4 trial demonstrated efficacy and safety, similar glycemic control with less bolus insulin. The rate of hypoglycemia was comparable to glargine U100 [32]. Compared to insulin degludec, insulin icodec also shows non-inferiority and superiority in glycated hemoglobin levels. A slightly increased rate of hypoglycemic events was also observed [33, 34]. In type 1 diabetes, a higher rate of hypoglycemia was seen in the ONWARDS 6 trial, most notably with self-measured blood glucose [35]. The comparator was daily administration of the second-generation basal analog degludec, which has a lower rate of hypoglycemia than insulin glargine [26], glargine being the basal analog with daily administration most frequently used worldwide. It is important to note that this risk in ONWARDS 6 is observed to decrease meaningfully when assessed with continuous glucose monitoring data. The median CGM-based hypoglycemia duration was also found to be comparable between treatment arms [36]. These results underline the need of CGM for all people using insulin products as this gives a better picture of glucose levels than blood glucose and provides better information for individuals to manage the risk of hypoglycemia.

Basal insulin Fc (BIF) is a single-chain insulin variant fused to a domain of human IgG2. This results in a molecule that is absorbed very slowly due to its size. The delayed effect occurs in the subcutaneous tissue and through reduced insulin receptor affinity. The half-life is approximately 17 days therefore a higher initial loading-dose is required [37]. Similar to icodec a better or even comparable glucose control with only once-weekly injection should be achieved. BIF has comparable efficacy and safety to insulin degludec in patients with type 1 and type 2 diabetes. It is not inferior in glycemic control. In most trials, the glycated hemoglobin level remains stable or even improves. The same applies to the time in range. Data on fasting glucose differ. Both higher and improved fasting glucose values were found in comparison with insulin degludec. The rate of hypoglycemic events is lower with BIF. All in all, there is a non-inferiority of BIF versus insulin degludec with significantly lower injection frequency [38‒40].

Third-generation basal insulin is currently only used in clinical trials but shows promising results and non-inferiority to other basal insulins as described above. One must keep in mind that nearly 40% of people with T1D miss at least 1 basal insulin dose per month. Reducing the number of injections required means more convenience, better adherence, less burden, less risk for injection site reactions, and less potential for dosing errors [41]. International real-world data from 3,945 adults with CGM coverage and smart pen data have shown that missing two basal insulin doses over a 14-day period would be associated with a more than 5% decrease in percentage of time in range, which is considered a clinically relevant change. In addition to better blood glucose control, once-weekly basal insulin injections mean a higher quality of life and flexibility for people who receive metered-dose inhaler therapy or even only basal insulin for type 2 diabetes.

Once-weekly insulins have also the potential to be a breakthrough for unmet needs in low and middle-income countries where access to insulin is just one of the barriers to achieving acceptable diabetes management, particularly for type 1 diabetes (Table 1) [43]. Possibly the complex interplay of issues with access, quality, control, and stigma necessitates scientifically evaluation of new approaches tailored to the resources available in low- and middle-income countries (LMIC’s) instead of copying the approaches of MDIs that have become the standard of care since the DCCT trial. Indeed, decentralized once-weekly drug delivery in tuberculosis and other health threats has been shown to be more effective in LMIC’s than more frequent dosing or self-dosing at home [44]. It remains to be seen if a medium-low dosed once weekly insulin (to avoid hypoglycemia) even without meal-boluses (acknowledging that detailed diabetes education is not feasible and stigma is a barrier to regular self-injection) would be able to avoid DKA and premature death. Now that the first once-weekly insulin has received approval in EMA, it remains to be seen if a coalition of stakeholders will enable to study such an approach in a pilot study. This pilot needs to assess the efficacy and safety of an ultralong once weekly insulin without mealtime injections an alternative to T1D standard of care in low-resourced settings (premixed insulin, MDI w/glargine or NPH, limited test strips and education).

According to the recently published T1D index unrestricted access to insulin and blood glucose test strips, as well as support for self-management of the disease, could avoid premature death of around 2 million people with type 1 diabetes by 2040. In the USA, there is a 13-year difference in life expectancy for those with type 1 diabetes compared to those without. Globally, this gap is approximately 24 years and as high as 46 years in low-income countries [45]. As an example, how to fight shortages in insulin availability in the US the Juvenile Diabetes Research Foundation (JDRF) support the manufacture and distribution of low-cost insulin through the Civica Insulin Project. Three biosimilar insulins, for USD 30 per vial and USD 55 for a box of five pens will become available with the CA’s CalRx Biosimilar Insulin Initiative [46]. It remains to be seen, how biosimilar insulins will address insulin shortages even in the more affluent countries.

But the insulin development is still not over. For example, ideal pump insulins with rapid on- and offset of insulin action and low probability of infusion set-failure are under development. Ideally, future developments try to mimic physiological glucose control achieving dynamic insulin secretion that responds to glucose fluctuations without causing hypoglycemia. The endogenously released insulin is subject to the first-pass effect in the liver, where it is metabolized. Thus, endogenous insulin can take effect immediately. Exogenously subcutaneously applied insulin does not undergo a first-pass effect and must first be absorbed from the subcutaneous tissue. This results in a delayed effect even with increasingly fast-acting insulin formulations. In addition, the amount of insulin injected is calculated according to individually estimated parameters. The closest commercially available therapy options to endogenous insulin secretion are currently AID pumps. A glucose-responsive insulin that can adapt its biological activity to the fluctuating glucose levels in the blood would be desirable but must fulfil many requirements. It must contain a glucose-binding or reactive molecule that enters the blood quickly to mediate a rapid selective glucose response. Furthermore, a long half-life is required to cover the basal insulin requirement without leading to hypoglycemia. To achieve this, various strategies are pursued. Many rely on polymer structures as carriers containing insulin molecules and glucose-responsive proteins such as glucose-binding proteins. These include concanavalin A (ConA), glucose oxidase (GOx), and phenylboronic acid (PBA). ConA has a mitogenic effect and is therefore not used pharmacologically. GOx is often used instead. It causes a glucose-dependent pH change which alters the permeability of the polymer but is not stable in aqueous formulations. The time of action of insulin is related to the molecular weight. This is the principle behind the conjugation of PBA and polyol to a single insulin molecule. PBA and polyol can bind to each other and therefore last for a long time. When glucose binds, the formation dissolves, and an insulin molecule is released. Another approach is the use of a glucose transporter. For example, glucosamine conjugated insulin can bind to the glucose transporter of the erythrocyte membrane. At high glucose levels, there is a competitive binding of glucose, and the insulin is released. Despite promising approaches, there are still many challenges to be solved such as biocompatibility, adequate insulin release, toxicity of carrier molecules and particle stability. Clinical trial experience is therefore limited [7, 42, 47].

Diabetes therapy is a particular challenge for those affected and can significantly impair their quality of life. The aim must therefore be to make insulin therapy as simple as possible without sacrificing glucose control. This includes a lower hypoglycemia rate as well as lower postprandial glucose peaks. Due to the remarkable advances since the first insulin therapy, we have insulins at our disposal that can mimic endogenous insulin secretion more and more precisely with their pharmacological and dynamic properties. Together with advances in diabetes technology, it is possible to minimize the risk of acute and long-term diabetes-related complications and should be a goal for all people with diabetes. However, one point that must not be overlooked is that the insulins and technical devices that are standard therapy in industrialized countries cannot be used everywhere. The latest insulins and devices are often not affordable, accessible, or useable in developing countries because, for example, cold chains or digital infrastructure are lacking. It is therefore important that such needs are considered in the further development of insulins.

An ethics statement was not required for this study type since no human or animal subjects or materials were used.

Jantje Weiskorn has received speaker’s honoraria from Amryt Pharma. Banshi Saboo has received speaker’s honoraria from Sanofi and Novo Nordisk. Thomas Danne has received speaker’s honoraria and research support from or has consulted for Abbott, AstraZeneca, Boehringer, DexCom, Lilly, Medtronic, Novo Nordisk, Provention Bio, Roche, Sanofi, and Vertex and is a shareholder of DreaMed Ltd.

There are no funding sources for this article.

Jantje Weiskorn made significant contributions to the conception and design of the review and to the interpretation of the current literature. She contributed to the review and revision of the manuscript and approved the final version. She is the guarantor of this work and takes responsibility for the integrity and accuracy of the scientific information. Banshi Saboo made significant contributions to the conception and design of the review and to the interpretation of the current literature. He was involved in the review and revision of the manuscript and approved the final version. Thomas Danne provided substantial contributions to the conception and design of the review and the interpretation of the current literature. He contributed to review and revision of the manuscript and approved the final version.