The Ideal Diabetes Therapy: What Will It Look Like? How Close Are We?

The number of approved diabetes drug therapies (referred to hereafter simply as “therapies”) has grown significantly in the past two decades. In 1994, lifestyle-directed interventions, metformin, sulfonylureas, and insulin were the only glucose-lowering therapies in widespread use. Today, 11 classes of diabetes drugs are available, and many more are in development.

Despite this increase, diabetes was the leading cause of blindness, kidney failure, and nontraumatic lower-limb amputations in the United States in 2011,¹  and treating diabetes complications cost the United States $22.9 billion in 2006.²  These figures can be at least partially attributed to the use of highly imperfect diabetes therapy options that have limited efficacy, inconvenient side effects, and unfavorable delivery methods.

Yet, how we define an “ideal” diabetes treatment is not clear cut because new diabetes medications are evaluated for qualities ranging from their effects on glucose homeostasis to their effects on cardiovascular mortality. More closely defining the characteristics of an optimal therapy is of interest for evaluating the performance of current drugs and understanding how much extra benefit therapies in development will provide.

In this review, we explore what an ideal diabetes therapy would look like from the perspective of several key stakeholders in diabetes drug development. We then discuss the most commonly used current diabetes therapies and a number of therapies in development and assess how well each of these meets the criteria for an ideal therapy. Diabetes therapies are described by their brand names to differentiate new therapies with generic names similar to those of existing therapies and to draw distinctions between separate therapies that share the same generic names.

Hundreds of millions of dollars are spent each year in the pursuit of improved diabetes therapies. This pursuit is informed by a number of players, most notably patients, physicians, payors, and financial analysts. Although these parties would all broadly agree that an ideal diabetes therapy would eliminate suffering, death, and inconvenience resulting from diabetes, each has a variety of preferences regarding the specific characteristics of progressively more favorable diabetes therapies. These preferences drive diabetes drug design and serve as useful metrics for assessing how close current and upcoming diabetes therapies are to being “ideal.”

Fundamentally, patients want therapies that are effective, easy to use, safe, tolerable, and affordable. The drug that achieves the most favorable balance of these characteristics will have the highest appeal for patients.

Efficacy can mean a number of things in the context of diabetes. A maximally effective drug would address the underlying cause of diabetes. In type 1 diabetes, such a drug would stop the autoimmune process and regenerate β-cells or perfectly mimic the insulin-secreting functions of the pancreas with minimal patient input and hassle. In type 2 diabetes, it would tackle insulin resistance and loss of β-cell function. More commonly, a drug's efficacy is measured by how robustly it provides glucose control, how durable its effect is, and what benefits beyond glucose lowering it provides (e.g., lipid lowering, cardioprotection, preservation of β-cell function, and weight reduction).

Ease of use can be broadly defined by the amount of inconvenience associated with taking a drug, how much pain its dosing entails, the flexibility of its dosing, and its duration of action. From this perspective, an ideal therapy would have few steps involved in its administration, would be oral rather than injectable, could be administered any time of day (thus not demanding that patients adjust their schedule around their medication), and could be given once daily (or preferably much less often).

The optimally safe and tolerable antidiabetic drug would not cause nausea, vomiting, hypoglycemia, weight gain, or any other side effect that makes it difficult or undesirable to take, nor would it be associated with any increased risk for cancer, cardiovascular events, renal impairment, or other complications.

Finally, as with most other therapies, patients are sensitive to cost, and thus the ideal therapy would be inexpensive and reimbursable.

Many physicians' preferences for an ideal therapy are closely tied to those of patients, although physicians may weigh the importance of particular preferences differently. From the physicians' perspective, an ideal therapy would first and foremost improve patients' health and outcomes. It would possess strong efficacy and speak to patients' desires that a drug retain its efficacy over time, provide robust glucose control as well as nonglycemic benefits, and address the underlying cause of disease. It would also be safe, meaning that it would not increase the risk for cardiovascular events, cancer, or hypoglycemia.

Uniquely from the physicians' perspective, it would be easy to prescribe (e.g., dosing would not need to be based on kidney or liver function) and would not require insurance preapproval. Finally, per this group, the optimal drug would increase adherence, which requires that it be considered easy to use, effective, tolerable, and affordable by patients. It should be noted that physicians might be more willing than patients to accept a drug with side effects and poor tolerability that markedly improves outcomes, as long as the drug's unfavorable characteristics do not significantly hinder adherence.

For payors, who are focused on obtaining the best outcomes at the lowest cost, a drug's ability to provide novel or added benefit and decrease both short- and long-term treatment costs is paramount. This group will not see a “me-too” drug or a drug that provides only marginal improvements over existing options as ideal unless such a drug is much less costly than its counterparts.

To decrease treatment costs, a drug must be inexpensive or at least cost-effective, improve outcomes, decrease complications both in the short and long term, and be safe. Notably, such a drug would speak to physicians' and patients' preferences as well. A drug that provides significant novel benefits but does not meet the aforementioned cost-related requirements will only increase payors' outlays and thus is less likely to be reimbursed.

The views of financial analysts are important because this group garners monetary support for a drug's development and commercialization. Broadly, this group wants to support therapy options that will be approvable, widely adopted, profitable, and reliably manufactured.

How widespread a drug's adoption will be depends on how well it meets patients', physicians', and payors' demands and thus whether it fits the criteria mentioned above.

Approvability is determined by a mixture of safety and efficacy considerations. In the present U.S. regulatory environment, the risk tolerance for diabetes drugs is low. For example, the upper boundary of the 95% confidence interval of a drug's risk assessment for major adverse cardiac events must be < 1.8 in preclinical trials for the drug to be approved, regardless of its efficacy.³ 

A drug's profitability depends on the margins it can procure, which are determined by its potential audience, the price manufacturers can charge, and the costs associated with its production, and whether it can consistently bring in revenue, even if it addresses the underlying causes of disease. Profitability is also determined by how strong a drug's intellectual property protection is and thus how much competition it faces.

Financial analysts are finally concerned with a drug's manufacturing because production errors can potentially ruin its prospects, whereas ease of manufacturing reduces production costs and thus increases margins.

Although the development of new classes of diabetes medications has markedly increased the number of treatment options available for people with diabetes in recent years, the glucose-lowering interventions used today are still imperfect, meeting in entirety neither patients', physicians', payors', nor financial analysts' demands. Notable limitations of these therapies have limited efficacy, inconvenient side effects, and unfavorable delivery methods. Additionally, no diabetes therapy other than insulin can be titrated indefinitely with increasing biological effects. Rather, once glucose targets are no longer being met, noninsulin therapies must be combined or switched with other medications. In this section, we describe the benefits and limitations of the six most commonly used diabetes drug classes.

More than 90 years after insulin's discovery by Frederick G. Banting and Charles H. Best, it remains a staple of diabetes treatment used by both those with type 1 and those with type 2 diabetes. Insulin therapy has progressed considerably from the first bovine or porcine formulations that were initially sold as commercial insulins, and, uniquely, among all antidiabetic therapies, it can consistently lower A1C with upward titration. Yet, significant room for improvement in insulin efficacy, safety, and ease of use remains.

Today, five categories of insulin are in widespread use: regular insulin, neutral protamine Hagedorn (NPH) insulin, rapid-acting analogs, basal analogs, and premixed insulin products. All of these categories of insulin need to be injected, making insulin therapy more of a hassle and more painful to patients than several other forms of diabetes therapy. Additionally, all of these insulins are complex to use and prescribe, with the potential for dosing errors that could result in hypoglycemia or diabetic ketoacidosis.

Regular insulin and NPH are the two main types of recombinant human insulin. Regular insulin is short-acting human insulin meant to cover mealtime glycemic peaks.^4,5  It is lower in cost than alternative short-acting insulins and has been demonstrated to provide similar glycemic control to insulin analogs.⁶ 

NPH is an intermediate-acting human insulin that is used for basal insulin therapy^4,5  and is typically dosed once or twice daily. The glycemic control provided by NPH is similar to that provided by newer basal analogs, but NPH is lower in cost. However, the action profile of NPH features a slightly greater peak than those of long-acting analogs, and this increases the risk for nocturnal hypoglycemia (and possibly overall hypoglycemia, as well, although studies have not demonstrated this as consistently).^5,7  NPH must also be dosed at fixed times every day, which can make therapy burdensome.

Currently available rapid-acting and basal human insulin analogs complement these human insulin choices. Rapid-acting analogs such as Humalog (insulin lispro) and Novolog (insulin aspart) act more quickly than regular human insulin, with an onset of action of 5–15 minutes compared to 30–60 minutes for regular insulin.⁸  Yet these products still have some time lag between administration and effect. This can result in both hyperglycemia during meals and hypoglycemia after meals, although less frequently than with regular insulin.

Basal analogs such as Lantus (insulin glargine) and Levemir (insulin detemir) have a more stable action profile than NPH and cause less weight gain and nocturnal hypoglycemia^7,8  than NPH. The principal drawbacks of current basal analogs are their increased cost compared to NPH and the fact that, like NPH, they should be dosed at the same time or times each day.

Finally, premixed insulins are pre-made combinations of long- and short-acting insulins. These preparations vary considerably, ranging from 50% long-acting/50% short-acting insulin to 75% long-acting/25% short-acting. Because they address both mealtime and fasting insulin requirements in a single injection, premixed insulins enable fewer daily injections than standard basal-bolus therapy. In patients with type 2 diabetes, premixed insulin analogs may help achieve lower A1C levels than long-acting insulin analogs used alone, but rates of hypoglycemia are typically higher.⁹  However, unlike with basal-bolus therapy, patients are unable to separately titrate the long- and short-acting components of their premixed insulin, resulting in less robust glycemic control.¹⁰  Premixed combinations of both insulin analogs and regular and NPH insulins are available.

Overall, because of their potential for continual up-titration, insulins are capable of lowering glucose to an extent that befits the ideal diabetes drug. However, they are associated with notable limitations in terms of safety, tolerability, and ease of use.

Metformin, an oral drug, is commonly accepted as the first-line therapy for type 2 diabetes in conjunction with diet and exercise. Metformin lowers glucose levels by decreasing gluconeogenesis, although the precise mechanism by which it does this is not well understood.¹¹  Metformin monotherapy causes A1C reductions of ~ 1.5%. Speaking to the preferences of all four diabetes drug stakeholder groups, it is weight neutral, delivered orally,¹²  and has been well studied (having been widely used in Europe since the 1950s and available in the United States since 1994). Additionally, metformin is generic and thus generally more affordable than newer diabetes therapies that remain on patent.

Metformin does have a few less-than-optimal attributes in addition to its positive ones. It causes gastrointestinal side effects such as nausea and diarrhea in nearly 30% of patients.¹³  More rarely (prevalence estimates range from 1 to 9 per 100,000 treated patients), metformin causes lactic acidosis, although mostly in people with severe kidney disease in whom treatment is contraindicated.^12–14 

This class of insulin secretagogues is well studied and, like metformin, lowers A1C by ~ 1.5%. It is generic and inexpensive. The drugs in this class, including glyburide, tolbutamide, glimepiride, and glipizide, are commonly used as second-line therapy after metformin.¹²  Mechanistically, they augment insulin secretion from β-cells by depolarizing β-cells via binding to adenosine triphosphate (ATP)-dependent potassium-ATP channels.¹⁵ 

Sulfonylureas are far from ideal, however, and of the regularly used antidiabetic therapies, these drugs in particular have many drawbacks. They cause weight gain (~ 2–3 kg in clinical trials) and hypoglycemia (incidence estimates vary depending on accompanying antidiabetic agents and patient characteristics).^12,15  Additionally, studies have suggested that they are associated with higher rates of monotherapy failure than alternative treatments, leading to the hypothesis that this class directly causes β-cell functional decline, or β-cell “burnout,” through the induction of insulin secretion.^12,16,17 

Glucagon-like peptide-1 (GLP-1) is an endogenous hormone that is secreted from L-cells and has a variety of glucose-lowering effects. These include increasing glucose-dependent insulin secretion, decreasing glucagon secretion, inhibiting gastric emptying, and increasing insulin sensitivity in some peripheral tissues.^18,19  In people with normal glucose tolerance, incretin gut hormones such as GLP-1 account for approximately two-thirds of mealtime insulin secretion. However, this “incretin effect” is severely diminished in people with type 2 diabetes, partially because of reductions in GLP-1 secretion.^20–22 

Although the native hormone has a half-life of a few minutes, GLP-1 receptor agonists with prolonged half-lives and thus significant therapeutic value for diabetes treatment have been developed.²³  In the United States, two GLP-1 receptor agonists are approved: Byetta (exenatide), which is injected twice daily, and Victoza (liraglutide), which is injected once daily. GLP-1 receptor agonists are included as less validated second-line therapies in the current American Diabetes Association (ADA)/European Association for the Study of Diabetes (EASD) consensus algorithm.¹² 

The GLP-1 receptor agonist class is appealing both in terms of efficacy and safety. Foremost, it provides robust reductions in A1C. Liraglutide has been shown to lower A1C by 0.5–1.6%, whereas exenatide is associated with A1C reductions of 0.5–1%.²⁴  Other attractive characteristics of GLP-1 receptor agonists include their induction of weight loss (2–3 kg with exenatide¹²  and 1–3 kg with liraglutide,²⁵  depending on whether the agonist is used alone or with one of several other antidiabetic therapies), induction of satiety, and minimal associated hypoglycemia.¹⁸ 

However, the class does face limitations with regard to tolerability and ease of use. GLP-1 receptor agonists must be injected subcutaneously and can cause significant but transient gastrointestinal side effects such as nausea and vomiting. In phase 3 clinical trials, 29.3% of patients taking liraglutide, 1.8 mg, as monotherapy initially experienced nausea,²⁶  whereas nausea was experienced by 36–51% of patients in the phase 3 exenatide trials.²⁷  With both currently available GLP-1 receptor agonists, nausea rates decline as treatment progresses.²⁸ 

Manufactured by Amylin, Alkermes, and formerly Eli Lilly, Bydureon is the first once-weekly GLP-1 receptor agonist to be developed. Bydureon has the same active ingredient as Byetta³⁷  but requires reconstitution before delivery. It was approved in the European Union in June 2011 and is now available in Germany and the United Kingdom.

The drug was first submitted to the U.S. Food and Drug Administration (FDA) in May 2009. The FDA has since requested (and Amylin provided) clinical evidence that supratherapeutic levels of exenatide do not pose a risk for QT interval prolongation and associated cardiac risk. Bydureon was approved in the United States on 28 January 2012; it became available in U.S. pharmacies on 13 February 2012.

Bydureon has numerous favorable characteristics. Speaking to patients' desires for efficacious, easy-to-use, and tolerable drugs, as well as physicians' and payors' preferences regarding drugs with potential for good adherence, it is dosed once weekly, provides A1C reduction in the range of 1.5% (significantly better control than Byetta,³⁸  but slightly worse than Victoza³⁹ ), causes less nausea than Byetta,³⁸  and results in weight loss.^40,41  Notably Bydureon's A1C and weight loss benefits are durable for up to 3 years.

Because the drug builds up slowly in the blood, no titration is needed, which makes dosing easier for doctors. Like the current GLP-1 receptor agonists, Bydureon has been demonstrated to provide improvements in lipid levels, triglyceride levels, and blood pressure,^41–43  and it may facilitate β-cell preservation,⁴⁴  which in addition to slowing the progression of diabetes could be useful in minimizing hypoglycemia and improving overall glycemic control.

The drug has a few drawbacks, however. First, Bydureon is delivered via a 23-gauge needle (compared to delivery via a 30- or 32-gauge needle for Victoza and 29- to 31-gauge needles for Byetta). Additionally, it is complex to administer with the current delivery device, involving six steps from reconstitution to injection; a pen is in development but is not projected to become available until late 2012. It is unclear to what extent these administration characteristics, which could potentially pose barriers to uptake and adherence, will overshadow the drug's many positive attributes.

Overall, if approved in the United States, Bydureon's once-weekly administration, decreased nausea profile, and easy titration will likely be appealing to patients, physicians, and payors. However, one can speculate that development of a Bydureon pen will be important for widespread adoption of this drug.

This class of drugs blocks dipeptidyl peptidase 4 (DPP-4), the enzyme that degrades circulating GLP-1 in the body, and thereby increases the bioavailability of endogenous GLP-1.^18,29  DPP-4 inhibitors available in the United States include Trajenta (linagliptin), Januvia (sitagliptin), and Onglyza (saxagliptin). Although not yet officially endorsed by ADA/EASD as a second-line therapy after metformin, DPP-4 inhibitors are commonly used as such in clinical practice.

In contrast to GLP-1 receptor agonists, DPP-4 inhibitors are orally administered and have much improved tolerability profiles. They are also weight neutral.²⁹ 

However, DPP-4 inhibitors have more limited efficacy than GLP-1 receptor agonists (showing A1C reductions of 0.5–0.8% in clinical trials,²⁴  although this has interestingly not stifled payors' willingness to reimburse for them) and have been associated with an increased incidence of nasopharyngitis and upper respiratory tract and related infections.^30,31 

Thiazolidinediones (TZDs), including Avandia (rosiglitazone) and Actos (pioglitazone), increase insulin sensitivity via activation of peroxisome proliferator–activated receptor-γ, a nuclear receptor that regulates production of proteins related to glucose and lipid homeostasis. These drugs are potent insulin sensitizers and enhance the effectiveness of both endogenous and exogenously injected insulin.³²  As demonstrated by the Avandia arm's cumulative 5-year incidence of monotherapy failure being significantly lower than that of the metformin or glyburide arms in ADOPT (A Diabetes Outcome Progression Trial),¹⁶  they provide effective long-term glycemic control in type 2 diabetes.

In the few clinical trials examining their use as monotherapy, TZDs have been associated with a 0.5–1.4% decrease in A1C. TZDs are also considered less validated add-on therapies to metformin and are often used in triple combination therapy with sulfonylureas.¹² 

Despite the durable efficacy of TZDs, their use has decreased significantly in the United States because of increasing concerns about their numerous non-ideal side effects. The drugs are associated with significant weight gain (~ 3 kg), as well as fluid retention³³  and about a twofold increased risk of fractures.³⁴  Furthermore, recent studies have prompted specific concerns about a possible cardiovascular and bladder cancer risk associated with Avandia and Actos use, respectively.

Tight restrictions have been placed on rosiglitazone use in the United States, and, accordingly, its U.S. sales have fallen 57% in the past year. As of November 2011, patients wishing to take Avandia in the United States must document that they understand the drug's risks; doctors have to attest that no other therapy can provide a particular patient with adequate glucose control; and rosiglitazone-containing products are available only through mail order.³⁵  Meanwhile, an interim analysis of an ongoing 10-year, 200,000-patient epidemiological study of pioglitazone conducted by Takeda showed long-term pioglitazone use to be associated with an increased risk of bladder cancer. As a result, the FDA has strongly discouraged Actos for patients with active bladder cancer or a history of bladder cancer.³⁶ 

A number of noteworthy diabetes drug candidates are in development. Knowledge of current drugs' drawbacks and the preferences of various drug development stakeholders helps put the strengths and weaknesses of these possible future drugs in perspective. Each of these drugs has the potential to markedly improve the treatment of diabetes for at least some individuals. However, none yet integrates all of the characteristics of an ideal diabetes therapy.

Amylin and Alkermes are also developing exenatide once monthly, potentially the first-ever once-monthly diabetes drug. This therapy is a GLP-1 receptor agonist that uses the same extended drug-release technology as Bydureon. The companies are currently meeting with regulatory agencies to plan phase 3 trials.

Notably, exenatide once monthly is soluble and, unlike Bydureon, will not require reconstitution. In phase 2 trials, it has shown similar A1C-lowering efficacy to Bydureon. However, the drug has been associated with slightly higher nausea and a worse weight loss profile than its once-weekly counterpart.⁴⁵  Dosing-related concerns, including patients forgetting to take the drug and excess drug remaining in the body if too much is taken, have also been raised.

At this point, drawing a verdict on this drug would be premature. If, however, exenatide once monthly is demonstrated to have an efficacy, weight loss, and tolerability profile similar to that of Bydureon in phase 3 trials, its once-monthly administration would likely make it highly popular among patients, physicians, and payors and bring it significantly closer to the ideal than most other therapies have come thus far.

This possibly first-to-market sodium-glucose transporter-2 (SGLT-2) inhibitor, which is being developed by Bristol-Myers Squibb and AstraZeneca, promotes urine glucose excretion by inhibiting glucose reabsorption via the kidney's SGLT-2 receptors.⁴⁶  In trials, dapagliflozin has shown relatively good glycemic efficacy; when used as monotherapy in a phase 3 trial with treatment-naive adults, dapagliflozin, 5 mg and 10 mg, reduced A1C by a placebo-adjusted 0.77 and 0.89%, respectively.⁴⁷ 

The major benefit of SGLT-2 inhibitors is that they act through a separate glucose-lowering mechanism than currently available antidiabetic drugs, which means they can provide additive benefit beyond other therapies. This makes them particularly appealing to patients, physicians, and financial analysts looking for extra A1C reduction, as well as to payors who are looking for therapies that will add true extra value.

Additionally, dapagliflozin is orally administered, which eases inconvenience associated with dosing. By increasing urinary glucose excretion, dapagliflozin enables some weight loss.⁴⁸  Finally, dapagliflozin decreases blood pressure, is associated with low rates of hypoglycemia,⁴⁶  and, according to retrospective analyses of phase 2 and phase 3 data, may also provide some cardioprotection, although whether that protection is associated with the observed weight loss is unclear.⁴⁹ 

Unfortunately, use of dapagliflozin is associated with an increased risk of urinary tract and genital infections (in the placebo-controlled short-term pool from phase 3 trials, 10% of females and 3.5% of males experienced one of many types of genital infection, and 10% of females and 2.7% of males experienced a urinary tract infection), an effect that seems directly related to the mechanism of SGLT-2 inhibition. Additionally, imbalances in breast and bladder cancer occurrence in dapagliflozin-treated patients and one incident of drug-related liver injury were seen in the drug's clinical development program.⁵⁰ 

Because of these imbalances, an FDA advisory panel committee voted 6–9 in July 2011 to recommend against the drug's approval; a final decision on the drug will be made on 28 January 2012. Although the committee did not put much emphasis on the urinary tract and genital infections associated with dapagliflozin use at this meeting, these types of infections will make the drug less than ideal for patients. These infections are also likely to make physicians and payors more wary of prescribing and reimbursing for dapagliflozin, respectively. Moreover, because dapagliflozin does not directly address the underlying cause of diabetes, it is unlikely to slow disease progression.

Because of its low risk of hypoglycemia and associated weight loss, dapagliflozin could be a useful addition to other regularly used antidiabetic drugs. Yet the drug's modest A1C-lowering efficacy and side effects may limit its appeal for patients, physicians, and payors and subsequently its adoption and reimbursement.

Degludec is the next-generation Novo Nordisk basal insulin, and degludecPlus is a combined formulation of degludec and aspart. Degludec's half-life is significantly longer than that of glargine (25.4 vs. 12.5 hours, respectively).⁵¹  This prolonged half-life allows degludec to be dosed less frequently and regularly than glargine. Specifically, flexible once-daily dosing (administration at alternating 8- and 40-hour intervals) of degludec has been shown in a type 2 diabetes trial to be noninferior compared to standard glargine treatment.⁵²  Based on these data, Novo Nordisk is pursuing a flexible dosing indication for the compound, which could help increase adherence by lowering the burden associated with insulin therapy.

Degludec is also peakless and has a much less variable action profile than glargine.⁵³  As a result, it causes statistically significantly less overall hypoglycemia and nocturnal hypoglycemia than glargine.^54,55  Novo Nordisk filed for approval of degludec and degludecPlus in the United States and the European Union in late September 2011.

The dosing flexibility of degludec and degludecPlus and lower associated hypoglycemia rates could make them attractive treatment options from the perspectives of all parties with a stake in diabetes drug development. However, because their price has not yet been set, glargine is already well established among physicians and patients, and “biosimilar” insulins (insulins that appear on the market after the patent for the reference insulin product has expired and are similar, but not identical, to the original product), which would be much lower in cost, could enter the insulin market as early as 2015. It is difficult to predict how widely these drugs will be used if approved.

Made by MannKind, this rapid-acting inhalable insulin is delivered in a small device, does not require injections, and, because of its fast-on, fast-off action profile, causes less hypoglycemia than a usual insulin regimen. When used as a prandial insulin in combination with glargine in previous trials, Afrezza, which is also known as Technosphere insulin after the Technosphere pulmonary therapeutic administration technology it employs, has been shown to cause A1C reductions similar to those from premixed biaspart insulin (70% insulin aspart protamine suspension and 30% insulin aspart of recombinant DNA origin).⁵⁶ 

Several factors may limit Afrezza's use and uptake. Despite concerns regarding alterations in lung function with inhalable product use, no decrease in this metric has been associated with Afrezza after more than 2 years of treatment.⁵⁷  Nevertheless, people with decreased lung function will likely not be able to use the product. Because the insulin is inhaled rather than injected into the body, there is a greater risk for improper insulin administration with Afrezza; physicians and patients may therefore shy away from it. Moreover, because insulin is such an important medication and its therapeutic index is so narrow, unfamiliarity with the product may limit its initial uptake. Finally, significantly more people do experience a mild and transient cough with Afrezza than with injectable insulin. It is difficult to predict how much of a nuisance this will be in everyday treatment.

These concerns aside, Afrezza has the potential to improve care, both for insulin-naive patients who might be more likely to start on Afrezza than on traditional injectable insulin, and as a mealtime supplement for patients already using injections or insulin pumps. Furthermore, if it proves to have the same or better safety and efficacy as current insulin options, its inhalable dosing will mark it as an important step forward in the simplification of insulin therapy.

Afrezza's approval process is currently on hold. Although MannKind originally filed a new drug application (NDA) with the FDA for Afrezza in March 2009, the company received a request from the FDA in January 2011 for two new phase 3 studies examining delivery of Afrezza with the company's next-generation inhaler. Both trials are expected to be complete by the end of 2012, and MannKind expects to resubmit Afrezza's NDA in early 2013.

TAK-875 is a once-daily oral agonist of the G-protein–coupled receptor (GPR-40), which is found on islet cells. This receptor is activated by fatty acids and stimulates glucose-dependent insulin secretion. TAK-875 is being investigated in phase 3 trials by Takeda. In a 12-week phase 2 trial, 6.25-, 25-, 50-, and 100-mg TAK-875 doses given with or without metformin were shown to confer significant reductions in A1C beyond placebo; efficacy plateaued for doses > 50 mg at around the 1.0% A1C reduction seen with glimepiride. TAK-875 treatment was not associated with significant weight gain and carried a significantly lower risk of hypoglyemia than glimepiride.⁵⁸ 

Generally, TAK-875 appears to be a type of improved sulfonylurea. Sulfonylureas and TAK-875 provide similar A1C reductions and bind to receptors on β-cells to stimulate insulin secretion. However, GPR-40 agonists such as TAK-875 induce secretion in a glucose-dependent manner and therefore carry significantly less risk of hypoglycemia than sulfonylureas.

TAK-875 is also weight neutral, improves lipids, and seems to improve homeostasis model assessment-B (a measure of β-cell function).⁵⁹  It is not yet understood whether TAK-875 will lead to long-term β-cell burnout like sulfonylureas or how durable its effect is; phase 3 studies should be telling in this regard.

It will be interesting to glean from future trials exactly what benefits TAK-875 confers, to what extent it can improve β-cell function, and where it fits in with GLP-1 receptor agonists and DPP-4 inhibitors. The combination use of a DPP-4 inhibitor and TAK-875 would be intriguing because it would be completely orally administered and could potentially provide effective glycemic control, a low risk for hypoglycemia, weight neutrality, a low risk for gastrointestinal side effects, β-cell preservation, and cardioprotection. Such a combination of features would be quite appealing to patients, physicians, and payors and could bring the diabetes field significantly closer to the ideal therapy.

Steady progress has been made in the treatment of diabetes since the discovery of insulin in 1921. Although more treatment options are available for diabetes patients today than ever before, no therapy in widespread use can consistently halt, reverse, or cure type 1 or type 2 diabetes. Moreover, current diabetes treatments and those in late-stage development still have significant drawbacks such as limited efficacy, unwanted side effects, and inconvenient dosing.

With diabetes treatment costs estimated to comprise ~ 10% of total health care spending by 2020,⁶⁰  diabetes represents one of the highest health priorities in the United States and an area of considerable unmet need. Although a cure for both types of diabetes is the ultimate goal of diabetes innovation, the development of improved drugs that approach the ideal, whether in terms of efficacy, safety, tolerability, bettering adherence, or reduced long-term costs, is the intermediate important goal.

The authors thank Dr. Michael Dougan and James S. Hirsch for outstanding editorial and review assistance.