CLIFFORD J BAILEY
Address for correspondence: Professor Clifford J Bailey
Health and Life Sciences, Aston University, Birmingham, B4 7ET, UK. E-mail: c.j.bailey@aston.ac.uk
https://doi.org/10.15277/bjd.2022.388
The origins of diabetes medications provide an intriguing catalogue of clinical serendipity and scientific design. Use of insulin (beyond 1922) gave recognition to insulin resistance and the categorisation of type 2 diabetes (T2DM). The first sulphonylurea (carbutamide, 1956) emerged from its use as an antibacterial sulphonamide prone to cause hypoglycaemia, and biguanides were first used to treat diabetes in 1957 despite their glucose-lowering properties having been known since the 1920s. Alpha-glucosidase inhibitors arose from a screening programme for amylase inhibitors by Bayer in the 1970s and acarbose was introduced in 1990. The first thiazolidinedione (ciglitazone; not developed) was identified in a screening programme for triglyceride-lowering compounds by Takeda in the late 1970s and gave rise to pioglitazone (approved 1999), although first to market was troglitazone (from Warner Lambert 1997, withdrawn 2000). Exendin, an analogue of the incretin hormone glucagon-like peptide-1 (GLP-1), was identified in 1992 in the saliva of a lizard (Heloderma suspectum), and took until 2005 to be marketed as exenatide. To promote the efficacy of endogenous GLP-1, its rapid inactivation by the enzyme dipeptidylpeptidase-4 (DPP4) was blocked by clever molecular design of the first DPP4 inhibitors (vildagliptin and sitagliptin, approved in 2006). SGLT2 inhibitors are based on phlorizin, identified in apple tree bark (1835) and modified (2000) to avoid intestinal degradation: further modifications to increase selectivity against SGLT2 gave dapagliflozin and canagliflozin - approved 2012 and 2013, respectively, in Europe.
Br J Diabetes 2022;22:112-120
Key words: type 2 diabetes, glucose-lowering agents, history, glycaemic control
This brief account of the origins of diabetes medications is based on the John Wales Lecture, Drugs by discovery and design, delivered at the Association of British Clinical Diabetologists conference in September 2022. Nine classes of glucose-lowering agents are routinely used in the UK (table 1), and several further agents are indicated for glucose lowering in other regions.1,2 Each drug class has an interesting history and has contributed to a succession of ‘fashions’ in the therapeutic management of hyperglycaemia in T2DM.
Although there are many excellent accounts of the discovery and subsequent development of insulin, the pre-insulin era of research that focused attention on the pancreas is less well known.3 Dating from the first description of the pancreatic islets by Paul Langerhans (Berlin) in 1869, a succession of innovative studies and astute observations (and some questionable findings) linked the pancreas with diabetes.4-6 Indeed, while pre-insulin treatment of T1DM relied on starvation diets and herbal preparations, research studies in pancreatectomised animals investigated the use of pancreatic tissue and pancreatic extracts to reduce glucosuria and blood sugar, and prolong survival (table 2). The animal studies were acknowledged by Frederick Banting and Charles Best when describing their own work in Toronto, 1921-22: these studies undoubtedly fostered the enthusiasm of Banting, informed the choice of animal model and emphasized the need to co-opt the expertise of James Collip to refine the extraction process.7
Of particular note, a preliminary trial in the early 1900s by Georg Zuelzer (Berlin) achieved temporary respite with a pancreas extract injected into young diabetes patients.8 The work of Nicolae Paulescu (Bucharest) also deserves its own footnote in pre-insulin research.9 Paulescu had observed in 1916 that blood sugar was reduced in diabetic dogs after injection of pancreas extracts, but he was unable to proceed or publish his work until 1921 due to World War I. Paulescu and Zuelzer each applied for patents for their pancreas extraction processes, and voiced their discontent with the award of the Nobel prize to Banting (he shared his prize money with Best) and laboratory head John Macleod (he shared with Collip). Notwithstanding the wrangles over prizes, patents and accounts of who did what, we are reminded that even such a momentous discovery as that of insulin was built upon a pre-history of (often unrecognised) research by many individuals and groups, several just short of the definitive last few steps to successful clinical use.4-6
After the momentous achievements of Banting and colleagues in 1921-22, the extraction of bovine and porcine insulin was quickly refined and commercialised, and treatment became increasingly available.3 However, the existence of insulin- insensitive presentations of diabetes (Harold Himsworth, 1936) then became recognised, and type 2 (maturity-onset) diabetes became distinguished from type 1 (juvenile-onset) diabetes (John Lister, 1951).10,11 This distinction indicated that insulin was not the ideal treatment for all presentations of diabetes and signalled the need for other therapeutic agents (figure 1), particularly to treat people with maturity-onset diabetes.
Amongst oral glucose-lowering agents, metformin probably has the oldest lineage. It stems from the use of Galega officinalis (Goat’s rue, French lilac) to treat thirst and frequent urination (reference to diabetes?) since the 1700s.12 Galega was found to be rich in guanidine, shown by Watanabe in 1918 to lower blood glucose in animals. Several derivatives were synthesized in the 1920s, and some were used as treatments for diabetes, but they were gradually discarded as insulin became more widely available.13 Metformin (dimethyl biguanide) was first synthesised in 1922 in Dublin by Emil Werner and James Bell, and in 1929 two laboratories in Breslau reported that it lowered blood glucose in non-diabetic animals (Hesse and Taubmann; and Slotta and Tschesche).14-16 Although side effects were minimal, its potency was deemed insufficient for clinical consideration.
Meanwhile, guanidine-based antimalarial agents such as proguanil were developed in the mid-1940s and reported to lower blood glucose in animals, and metformin was tested for antimalarial activity by Eusebio Garcia in the Philippines in 1949. Garcia noted that metformin was helpful in treating a local influenza outbreak, and metformin became used for a time as an anti-influenza agent (flumamine).17 Lowering of glucose was noted in some patients, but again this property was not taken further.
The trail now jumps to Paris where, in 1956, pharmaceutical laboratory owner Jan Aron recruited local physician Jean Sterne to re-assess the glucose-lowering properties of biguanides.18 Sterne must have been familiar with the field as he had assisted in a study of a guanidine derivative (galegine) as an intern. At Aron Laboratories Sterne worked in collaboration with pharmacist Denise Duval to examine the effects of several guanidine-based compounds (including metformin and phenformin) in animal models. Unknowingly they repeated studies from the 1920s, and were attracted by the effectiveness and tolerability of metformin. Reassured by accounts of flumamine use in humans, Sterne ventured to test metformin in the diabetes clinic and published a first account of this work in a Moroccan medical journal in 1957.19 To expand the clinical studies, Sterne co-opted colleagues in local hospitals and noted that metformin could reduce or replace the need for insulin in some individuals with maturity-onset diabetes, but could not eliminate the need for insulin in young individuals with diabetes.18
Metformin was introduced in Europe as a treatment for maturity-onset diabetes in 1958, and other biguanides were introduced at about the same time (phenformin widely: buformin in parts of Europe but not the UK). These other biguanides initially received preference over metformin due to their greater glucose-lowering efficacy, but were withdrawn in the late 1970s due to an unacceptably high occurrence of lactic acidosis.20 The therapeutic advantages of metformin were confirmed by extensive studies in Edinburgh in the 1960s and by the United Kingdom Prospective Diabetes Study which reported in 1998. However, it was not until 1995 that metformin was introduced into the USA, and several years later metformin replaced sulphonylureas as the primary oral glucose- lowering therapy for T2DM.20,21
Sulphonylureas were discovered as a side effect of sulphonamide antibacterial drugs, which were known since the 1930s to sometimes lower blood glucose. In 1942 in Montpellier (France) Marcel Janbon observed particularly severe hypoglycaemia causing convulsions and coma in some patients with typhoid and pneumonia whom he treated with the sulphonamide 2254RP (1-butyl-3- sulfonylurea), later termed carbutamide.22 Local physiologist, August Loubatières confirmed the hypoglycaemic effect in animal studies and by 1946 he provided evidence for a direct action on the pancreas to stimulate insulin secretion. Loubatières considered the therapeutic potential of the agent to treat diabetes, but this was not taken further.23 Indeed, the hypoglycaemia was regarded as a side effect that limited continued use of carbutamide for antibacterial purposes, and use of carbutamide was mostly confined to East Germany and eventually discontinued.24
Under the name BZ55, carbutamide was re-investigated by Boehringer Mannheim, with studies at a Berlin hospital by Hans Franke and Karl Fuchs in 1954.25 Franke and Fuchs were not aware of the ‘hypo’ history of the drug but quickly noted this side effect. Indeed, Fuchs tested the drug on himself, noted the potency of the hypoglycaemic effect, and conducted tests in people with diabetes, observing most effect in those with adult-onset diabetes. A race to market a treatment for adult-onset diabetes began with carbutamide in 1955, followed by several closely related (now called first-generation) compounds (tolbutamide, chlorpropamide, tolazamide and acetohexamide). However, controversy over long-term safety resulted in all but tolbutamide being discontinued over the next five decades, and equivocal findings of the University Group Diabetes Program (UGDP) trial brought into question the cardiovascular (CV) safety of tolbutamide.
In the 1970s and 1980s the first-generation agents were superseded by agents designed to give higher potency (glibenclamide, glipizide, gliclazide) and there was a further addition (glimepiride) in the 1990s.26,27 Although much was known about the insulin secretory dynamics afforded by sulphonylureas, it was not until the late 1980s that the cellular mechanism of action was determined, namely binding to the so-called sulphonylurea receptor (SUR1) and closure of the inwardly rectifying K+-ATP channel (Kir6.2).28 Sulphonylureas remained the main oral glucose-lowering therapy for T2DM until around the turn of the 21st century when they were superseded by metformin, reflecting the weight gain with sulphonylurea therapy and the risk of hypoglycaemia consequent to continued stimulation of insulin secretion at low glucose levels.27
Although there had been reports of benzoic acid derivatives lowering blood glucose, it was particularly the work of Jean-Claude Henquin and his group in Louvain in the late 1970s that gave rise to the meglitinide class.29,30 They noted that the non-sulphonylurea benzamido moiety of glibenclamide (meglitinide; HB 699) could stimulate insulin secretion similarly to sulphonylureas, indicating what is now recognized as a separate binding site on SUR1. Based on the meglitinide molecule, repaglinide was designed by the Karl Thomae company to increase binding affinity, and (though not so closely related) nateglinide was designed by Ajinomoto. These agents were introduced in the late 1990s and the turn of the millennium, respectively, to serve as ‘prandial’ insulin releasers with a faster onset and shorter duration of action than sulphonylureas.31 Their use, however, was limited and declined in parallel with sulphonylureas.
As part of a screening programme for amylase inhibitors at Bayer in the 1970s, acarbose was isolated from cultures of a strain of Actinoplanes bacteria and found to be a potent competitive inhibitor of intestinal glucosidases.32,33 Acarbose was introduced in 1990, and two other alpha-glucosidase inhibitors (miglitol and voglibose) were introduced in some regions in the mid-1990s: these were synthesised, based on the discovery of sugar derivatives isolated from strains of Bacillus and Streptomyces.33 Though helpful in reducing the prandial glucose excursions from carbohydrate-rich meals, and gaining considerable use in some Asian countries, alpha-glucosidase inhibitors received little use in western countries due to their modest overall efficacy and gastrointestinal side effects.
The discovery of thiazolidinedione (TZD) peroxisome proliferator-activated receptor (PPAR) gamma agonists preceded the identification of PPAR transcription factors (circa 1990). While in search of clofibrate analogues with additional triglyceride-lowering activity, Takeda Chemical Industries in Japan observed in the late 1970s that some compounds (particularly thiazolidine derivatives) had glucose-lowering activity in insulin-resistant and diabetic mice. One such compound, ciglitazone, was evaluated in detail but was not considered potent enough to develop as a glucose- and lipid-lowering drug.34 However, the molecular template gave rise to pioglitazone, which was marketed in 1999.35 Other thiazolidine derivatives were developed by design, notably troglitazone by Sankyo in 1988 (with Warner Lambert), which became the first TZD to be approved for clinical use in the United States in 1997. It was withdrawn in 2000, however, due to idiosyncratic liver toxicity. Also using a thiazolidine structure, SmithKline Beecham (later GlaxoSmithKline) synthesized rosiglitazone in 1988 and brought it to market in 1999.35
Following a meta-analysis (2007) that raised cardiovascular (CV) safety concerns, rosiglitazone was withdrawn in Europe in 2010 and restricted in use in the USA.36 However, a large CV outcome study (RECORD) did not confirm the concerns; restrictions in the USA were lifted in 2013, although rosiglitazone has since received little use. Studies with pioglitazone have suggested possible reductions in myocardial infarction and stroke, but the increased risk of heart failure has limited prescriber uptake. Questions regarding bladder safety, although unconfirmed, have excluded use of pioglitazone in some countries.37
Since the discovery of the incretin hormones glucose-dependent insulinotropic polypeptide (GIP, 1970s) and glucagon-like peptide-1 (GLP-1, 1980s), the abilities of these hormones to enhance nutrient-induced insulin secretion have been viewed with interest as potential therapies for T2DM. Attention became focused on GLP-1 because (unlike GIP) it also reduced glucagon secretion and maintained greater insulin-releasing efficacy in people with T2DM as well as exerting a satiety effect and reducing body weight.38,39 However, rapid degradation by the enzyme dipeptidyl peptidase-4 (DPP4) precluded therapeutic use of GLP-1 itself.40
In 1992, when John Eng (New York) was investigating pancreatitis caused by reptile venoms he discovered the peptide exendin-4 in the venom of the Gila monster (Heloderma suspectum), and recognized this as having considerable (53%) sequence homology with GLP-1.41,42 Despite its ability to mimic the effects of native GLP-1 and resist rapid degradation by DPP4 (due to a glycine residue at N2), it was several years before a company (Amylin Pharmaceuticals) acquired the rights and formulated the molecule into an injection exenatide which was introduced in 2005 as a treatment for T2DM. Later, encapsulation of exenatide within polylactide-co-glycolic acid microspheres enabled once- weekly injection. Further members of the GLP-1 receptor agonist class have been based on either the exendin molecule (lixisenatide) or aligned more closely with native GLP-1 (liraglutide, dulaglutide, semaglutide) with modifications at residue N2 to avoid rapid breakdown by DPP4.43 Also, linkage to an immunoglobulin (dulaglutide) or inclusion of a fatty acid chain to enable attachment to albumin (liraglutide, semaglutide) have been used to extend time within the circulation. During their use in the treatment of T2DM and obesity, members of the GLP-1 receptor agonist class have shown reductions of blood pressure and in some CV complications as well as decreased albuminuria. Additionally, members of the class are under investigation for the treatment of fatty liver, dementia and low bone density.
With advances in custom peptide production, attention has been given to the design of single peptides that can interact with multiple receptors to achieve a greater lowering of blood glucose and body weight than can be achieved with existing GLP-1 analogues. The first so designed peptide to be approved (tirzepatide, 2022) to treat T2DM is a dual incretin agonist activating receptors for GLP-1 and GIP.44 The highest dose of tirzepatide tested (up to 15 mg once weekly over 40 weeks) in people with T2DM achieved reductions in HbA1c by >2% (>22 mmol/mol) and body weight by >9 kg. In obese people without diabetes the 15 mg once- weekly dose for 72 weeks achieved weight loss of 20%.
The DPP4 inhibitors provide us with an example of glucose- lowering drugs obtained entirely by design. Given the susceptibility of incretin hormones to rapid inactivation by DPP4, it was appreciated that inhibitors of the peptidase activity offered a therapeutic strategy to enhance the endogenous incretin effect via both GLP-1 and GIP concentrations.40,45 Encouraged by reports in the early-mid 1990s that DPP4 could be inhibited using various pyrrolidines and thiazolidines, Edwin Villhauer at Novartis in New Jersey screened a wide range of inhibitors to construct a topographical profile of the peptidase catalytic site. From this he designed vildagliptin (1998; the ‘vilda’ recognises his work) to block the site by reversible covalent bonding.46 Regulatory approval of vildagliptin (2007) was delayed to acquire additional phase 3 safety data, allowing a fast and effective development programme to bring sitagliptin to market (2006) to become the first in class.
The design of sitagliptin involved molecular modelling of the peptidase site using X-ray crystallography imaging. This enabled Nancy Thornberry and Ann Weber at Merck (MSD) in the USA to evaluate the ability of a series of piperazine derivatives to block the site non-covalently.47 Other DPP4 inhibitors are either covalent inhibitors (e.g. saxagliptin) or non-covalent inhibitors (e.g. linagliptin, alogliptin). Most of these agents are given as once daily (vildagliptin is given twice daily) tablets, but very long- acting (once-weekly, e.g. omarigliptin) DPP4 inhibitors are available in some regions. DPP4 inhibitors gained a reputation for their safety profile (which has been generally neutral in CV outcome studies) and have replaced TZDs as add-on therapy to metformin.
The isolation of salicylic acid from willow tree bark in the 1820s stimulated chemists to investigate other trees for compounds of potential medicinal interest. In 1835 the professor of chemistry at Louvain (Jean-Baptiste Van Mons) was moving his apple tree nursery. Two of his assistants, Laurent-Guillaume de Koninck and Jean Stas, identified phlorizin in root bark from the trees.48,49 Although phlorizin didn’t show any obvious medicinal value, in 1886 Josef von Mering (Strasbourg) described its blood glucose- lowering and renally-induced glucosuric effects. However, the glucosuria became viewed as indicative of a form of diabetes and although many groups investigated the effects of phlorizin it was not assigned a medicinal use.50
In the late 1950s Robert Crane in Saint Louis added the intestine to the sites of action of phlorizin when he used it as an inhibitor of intestinal sodium-dependent glucose transport.51 However, it was not until the mid-1980s that studies at Yale by Ralph DeFronzo, Gerry Shulman, Luciano Rossetti and colleagues showed that phlorizin could reduce the hyperglycaemia of partially pancreatectomised diabetic rats.52 This attracted a rethink of the potential medical value of phlorizin but its low solubility and low potency hampered development. Several O-glycoside derivatives of phlorizin (e.g. Tanabe T-1095 and Kissei’s remogliflozin) showed improved activity, but it was William Washburn and colleagues at Bristol Myers Squibb in 2000 who determined that C-glycoside derivatives avoided hydrolysis by intestinal glucosidases and thereby increased bioavailability.53 With further molecular manipulation the selective SGLT2 inhibitor dapagliflozin was produced (approved in Europe 2012) and other C-glycoside derivatives were produced with varying degrees of selectivity for SGLT2 inhibition (e.g. canagliflozin, approved in the USA in 2013, followed by empagliflozin and ertugliflozin).54 The cardio-renal effects of these agents, notably to reduce blood pressure and to reduce risk and progression of heart failure and chronic kidney disease independent of glycaemic control, have generated indications beyond the management of diabetes.55
The emergence of each class of glucose-lowering agents has created a succession of ‘fashions’ in the therapeutic management of hyperglycaemia, especially for T2DM. For example, sulphonylureas and meglitinides generated focus on beta cell dysfunction, which was the main target for glycaemic management of T2DM up to the early 1990s. Acarbose generated focus on prandial glucose excursions and their link to cardiovascular risk, before attention shifted to insulin resistance with the rise of metformin and thiazolidinediones - each of the latter three classes emphasizing the avoidance of hypoglycaemia. The DPP4 inhibitors facilitated straightforward combination therapy with metformin, while the weight-lowering properties and cardiorenal benefits of GLP-1RAs and SGLT2 inhibitors have moved CV and renal considerations to the fore in the most recent treatment guidelines.56
Dietary fashions, too numerous to elaborate here, have featured alongside drug-related fashions in the management of diabetes and cardiovascular risk. Interestingly, before insulin was discovered, type 1 diabetes and unspecified type 2 diabetes were treated with starvation diets: a hundred years later the low- calorie approach is again favoured for type 2 diabetes.57
Most non-insulin glucose-lowering agents (DPP4 inhibitors excepted) have arisen from chance clinical or scientific observations that were followed up with extensive experimental refinements (Table 3). Nearly all have incurred a long ‘adoption time’ from initial circumstantial evidence to therapeutic purposing, but once a lead agent is established there is invariably a race to design ‘look-alike’ compounds. All drug classes have experienced early safety scares and most have shown that effective clinical advantage can be gained before there is a full understanding of the mechanisms, provided the implementation of safe prescribing practice is respected.
Conflict of interest None.
Funding None.