The paradigms for oral pharmacological therapy in type 2 diabetes are shifting as we attain new insights into the optimal metabolic control in our patients. Each drug category has unique advantages and disadvantages, and their proper use necessitates a full understanding of their mechanisms of action, glycemic and nonglycemic effects, and prescribing indications. This article reviews published clinical trial data and places them into the context of contemporary, rational therapeutic strategies for this increasingly common condition.
Type 2 diabetes is a multifactorial metabolic disease characterized by abnormalities at multiple organ sites. These defects include insulin resistance and insulin deficiency.1,2 The former is primarily represented by decreased insulin-stimulated glucose uptake in skeletal muscle, augmented endogenous glucose production(predominately in the liver), and enhanced lipolytic activity in fat.3 The latter is an apparent progressive process with both functional defects in islet cell function and, eventually, apparent loss of β-cell mass.4,5 These defects are intimately linked, with derangements in one system exacerbating those in the others.6 Understanding the defects is important, because addressing them forms the cornerstone of current and future therapy for this disease.
Several studies, including the U.K. Prospective Diabetes Study (UKPDS),have now unequivocally shown the benefits of tight glucose control for patients with diabetes.7-9 In these studies, microvascular complications were significantly and consistently reduced in the more aggressively controlled groups of patients. As a result, various professional organizations have proposed increasingly stringent metabolic targets for their management.10,11 The extent to which glycemic control affects macrovascular end points, which is the major cause of death in patients with diabetes, remains incompletely understood. Most studies have yet to show definitive benefit. Potential explanations for this discrepancy include study methodology, the potential influence of the degree and timing (postprandial versus fasting) of glucose lowering, or the failure to adequately address insulin resistance in studies published to date. Several investigations are underway to address these issues.
At present, when lifestyle changes fail to reduce glucose levels to the desirable range, the conventional approach is to begin therapy with an oral antihyperglycemic agent. In 2002,12 we reported a systematic review of oral agents for type 2 diabetes, in which we assessed the efficacy of these drugs, both as monotherapy and in combination, and discussed evidence-based treatment strategies. Now, 3 years later, with a >50% increase in the number of published trials in this area, we reassess the literature. Have there been any new developments in this field that might alter the therapeutic approach to this increasingly common disease?
A MedLine search was performed to identify all English-language articles of randomized, controlled, clinical trials involving currently and previously available oral agents for type 2 diabetes published after our initial report. As in our earlier analysis, studies were included if they met the following criteria: study period of at least 3 months, minimum of 10 subjects in each group at the conclusion of the study, and hemoglobin A1c (A1C)reported as a major end point. Studies were excluded if they involved insulin,triple oral agent combinations, or investigational drugs, or if the study was limited to a specific subpopulation of type 2 diabetic patients.
There remain five classes of oral antihyperglycemic drugs approved by the U.S. Food and Drug Administration (FDA).
Sulfonylurea (SU) drugs (e.g., glyburide, glipizide, and glimepiride)improve glucose levels by stimulating insulin secretion by the pancreaticβ-cell,13 with elevated circulating insulin levels partially overcoming peripheral insulin resistance. It is well recognized, however, that with time, patients on SU monotherapy experience a progressive loss of glucose control. Because of this,the question of islet cell “burnout” has been raised. This same phenomenon, however, is noted in patients taking metformin, a drug that does not increase insulin secretion.14 Therefore, β-cell failure may simply be a fundamental feature of type 2 diabetes itself that is not substantially affected by the type of therapy used.
Treatment with SU agents generally yields a mean absolute A1C reduction of 1-2%.9,15,16 Several published SU studies completed since our original report was published confirm these points (Tables 1and 2).17-20 SU agents are effective both as monotherapy and in combination with agents that have different mechanisms of antihyperglycemic action(Table 3).
Side effects of SUs include weight gain9,15,21,22 and hypoglycemia.9,21,23 Weight gain is of particular concern, given that patients are often obese before therapy initiation. Hypoglycemia risk becomes a more important issue as patients' overall glucose control approaches the normal range. During the past several years, the cardiology community has become disquieted because of the potential effect of SUs on myocardial ischemic preconditioning.24 The actual importance of this issue in clinical practice remains unclear, but it has likely been exaggerated.
There are no new outcomes data on vascular end points from prospective clinical trials. In some retrospective analyses,25-27 but not in others,28-30 worse cardiovascular outcomes have been observed in groups of patients taking SU agents compared to groups taking metformin or thiazolidinediones (TZDs). Without prospective data, it is not possible to make firm conclusions,especially in light of data from previous randomized studies showing no significant increases or decreases in macrovascular risk in SU-treated patients, despite an apparent benefit on microvascular end points.9
Non-SU Secretagogues (Meglitinides)
Drugs in this class work similarly to SU agents but have a more rapid onset and shorter duration of action. As a result, insulin secretion is stimulated to a greater extent immediately after administration. When drug ingestion is timed with meals, the result is more physiologically appropriate control of postprandial glucose concentrations. The benefit of this effect remains unclear, although epidemiologically, postprandial hyperglycemia is more closely related to cardiovascular morbidity than is fasting glucose.
There are efficacy differences between the two agents within this group,repaglinide and nateglinide. The former has an A1C-lowering effect similar to most other antihyperglycemic agents in both placebo-controlled and head-to-head trials,31-35 whereas the latter appears to be less efficacious36-39 (Tables 1 and 2). Both are approved for use as monotherapy and in combination with most other oral agent classes(Table 3).
Side effects are otherwise similar to other secretagogues, including weight gain and hypoglycemia. These likely occur to a lesser degree than with SUs. Meglitinides must also be taken shortly before each meal and therefore have a more frequent dosing schedule than most other agents. Their cost is generally higher than that of SUs.
Long-term outcomes data are still unavailable for this drug class. However,the effect on long-term complication rates is likely to be at least similar to that observed with SUs. It is unlikely that such long-term outcomes studies will ever be conducted.
Metformin, a biguanide, acts mainly by decreasing hepatic glucose production40,41 —primarily gluconeogenesis—probably through effects on AMP-kinase. Circulating glucose levels are thereby reduced. Improved peripheral insulin resistance may also occur, but study results are inconsistent.40,42-44 Metformin is commonly referred to as an “insulin sensitizer,”because glucose levels improve without stimulation of insulin secretion.
Notably, metformin as monotherapy remains the only agent associated with the potential for weight loss.40,42,45 Other nonglycemic benefits have been reported, including modest lowering of lipid levels,46,47 and improvements in fibrinolysis,47 inflammatory markers, and endothelial function.48
Numerous studies during the past several years have continued to demonstrate a benefit of metformin on these cardiovascular risk markers.49 In fact,to date, metformin is the only oral antihyperglycemic agent shown to reduce macrovascular events in patients with type 2 diabetes, in a relatively small substudy of the UKPDS involving overweight subjects.50 When the drug was added to the regimens of patients no longer responding adequately to SUs, there was a puzzling increase in mortality, an association that remains essentially unexplained.50 Such a finding appeared to be confirmed in a more recent retrospective analysis.51 Another group of investigators has suggested a cardiovascular benefit in patients undergoing percutaneous coronary intervention.52
In placebo-controlled trials, metformin consistently lowers A1C by 1-2%.19,46,50,53-58 Trials published since our original report confirm that this drug, while having a unique mechanism of action, reduces A1C to a similar degree as most secretagogues59,60 (Tables 1 and 2). Metformin is approved for use alone or in combination with all other antidiabetic agents(Table 3). It is also gaining in popularity as a treatment option for women with polycystic ovary syndrome and has been demonstrated by multiple investigators to improve ovulatory capacity and metabolic parameters in this group of insulin-resistant women.61-63
Since our original report, the results from the Diabetes Prevention Program64 have also been published. Metformin was used as one strategy to prevent or delay the development of type 2 diabetes in one arm of this study involving patients with impaired glucose tolerance. The relative risk of progressing to diabetes in metformin-treated patients was reduced by 31%.64 While less impressive than the 58% risk reduction with lifestyle change, such data have given encouragement to the notion of using pharmacological therapy in patients with prediabetes, at least in the subset who cannot or will not undertake a diet and exercise program.
Gastrointestinal side effects of metformin are common42 but can be minimized by slow dosage titration. Because of the rare risk of lactic acidosis, several contraindications limit this drug's use, including renal and liver dysfunction, heart failure, dehydration or hemodynamic compromise, and alcohol abuse. Several studies have described a surprising proportion of metformin-treated patients with active contraindications for its use.65-67 Despite this, complication rates are few, suggesting that current prescribing guidelines may be overly stringent. In fact, a recent retrospective analysis involving heart failure patients demonstrated actual improved outcomes in those treated with this drug.68
TZDs are activators of the nuclear transcription factor peroxisome proliferator-activated receptor-γ (PPAR-γ) and modulate the activity of a host of genes that regulate carbohydrate and lipid metabolism.69 Currently available TZDs are pioglitazone and rosiglitazone.
Most notably, TZDs improve insulin sensitivity and enhance glucose utilization by adipocytes and skeletal muscle.70-73 Some investigators have also demonstrated a reduction of hepatic glucose production,44,72 although not to as significant a degree as with metformin. PPAR-γ is most highly expressed in fat cells, and TZD therapy is associated with prominent effects on circulating fat-derived factors that influence insulin sensitivity, such as free fatty acids, adiponectin, and tumor necrosis factor-α.74 TZD action in muscle tissue may indeed derive indirectly through these effects.
Since our original report, many more studies indicate that TZDs have beneficial effects on a variety of cardiovascular risk determinants, including cytokines and inflammatory markers,75-77 lipids,78-81 blood pressure,78,82,83 endothelial function,78,84-87 and certain cellular and molecular events that control the atherosclerotic process.88-91 Recently, pioglitazone has been shown to have better effects than rosiglitazone on plasma lipids,92 although the ultimate role of the lipid changes induced by TZDs remains uncertain,given these agents' apparent potential widespread vascular benefit. Provocative data regarding a suppressive effect on carotid intimal media thickness,93 a surrogate for atherosclerosis, as well as coronary artery restenosis after angioplasty94,95 have also emerged.
Long-term outcomes studies with TZDs are not yet available. In some retrospective analyses thus far presented mainly in abstract form, benefit is suggested on cardiovascular outcomes, but the data are inconsistent and fraught with interpretative challenges.96-99 The results of prospective outcome studies underway will be necessary in order to determine whether these effects yield measurable clinical benefits and indeed improve the macrovascular complications of type 2 diabetes.
Recent reports also suggest that TZDs may “preserve”β-cell function. The most convincing data come from the Troglitazone in the Prevention of Diabetes study100 of diabetes prevention that tested troglitazone or placebo in relatively young women with a history of gestational diabetes mellitus. Progression to type 2 diabetes was reduced by > 50% in women on active treatment, likely a reflection of improved β-cell function that accompanied increased insulin sensitivity. Whether such preservation of insulin secretory capacity occurs in patients once diabetes is established is less clear. To date, small,short-term studies suggest benefit on markers of β-cell function.101-103 Convincing data from long-term clinical trials with adequate methodology are still lacking.
Published trials since our original report have confirmed that the A1C-lowering effect of the TZDs is equivalent104 and typically in the same range as that achieved by the SUs or metformin, in both placebo-controlled and head-to-head studies39, 105-108 (Tables 1 and 2). These agents are also approved as monotherapy and in combination with most other agents, including metformin—a combination that is increasingly popular and now available in a single proprietary product (rosiglitazone/metformin)(Table 3).
Side effects include weight gain and edema, which have precluded their widespread use for patients with heart failure. Recently, more concern has arisen regarding the potential effect of TZDs in heart failure patients. A consensus statement from the American Diabetes Association and the American Heart Association addressed this issue and endorsed the FDA's current recommendation that the drugs not be used in patients with advanced heart failure symptoms (class III or IV New York Heart Association classification).109 Caution was also advised in patients with less severe heart failure. In a recent retrospective study of Medicare beneficiaries, decreased mortality was observed in diabetic patients prescribed a TZD after a hospitalization for heart failure.68
Randomized studies are needed to confirm these data before any change in practice is considered. Troglitazone, the TZD primarily associated with idiosyncratic hepatocellular injury, has been off the market for several years. Although the remaining agents have not been shown to pose a similar risk, recommendations still exist regarding periodic surveillance of liver function for patients on TZDs.
α-Glucosidase inhibitors (AGIs) act by inhibiting an enzyme on the enterocyte brush border that breaks down complex starches, delaying intestinal absorption of carbohydrate and particularly attenuating postprandial blood glucose elevations.110,111 Current members of this drug class include acarbose and miglitol.
In placebo-controlled trials, AGIs have usually been shown to reduce A1C by only 0.5-1%112-122 and are therefore generally considered less efficacious than other classes. Additional studies since our last report continue to confirm this trend, both in placebo-controlled and head-to-head trials18,58,123-126 (Tables 1 and 2). AGIs are approved for use as monotherapy and in combination with sulfonylureas and metformin(Table 3).
Side effects include abdominal bloating and cramping, frequently leading to cessation of drug use. The AGIs' more modest efficacy and higher incidence of side effects have limited their widespread use in the United States, although,interestingly, they remain very popular in other countries, particularly Germany and Japan.
In post hoc analysis of data from the Study to Prevent Non-Insulin Dependent Diabetes Mellitus trial, acarbose was observed to have an impressive effect on the risk of myocardial infarction (hazard ratio =0.09).127 These data support the view, based on epidemiological studies, that postprandial hyperglycemia has a greater influence on cardiovascular outcomes than does fasting glucose. No long-term outcomes data are available on vascular end points in type 2 diabetic patients, however.
Clinical trial research published since our original report does not compel any change in the prevailing view that most of the available oral agents are appropriate as initial therapy, barring, of course, any contraindications that might exist in specific patient circumstances. Most classes of drugs are equally efficacious in reducing A1C, with the exception of the AGIs and nateglinide. This conclusion is now garnered from newer studies, both when a specific agent is compared to placebo(Table 1) or when two drugs are compared to each other (Table 2).
Actual medication choice should incorporate not only consideration of glucose-lowering efficacy and contraindications, but also the myriad of other clinical features of individual patients. These include comorbidities, the capacities and tolerances of the patient, anticipated side effects, the degree of glucose control desired, concurrent drug therapy, dosing frequency, and cost.
Most endocrinologists continue to prefer metformin as the optimal first-line agent, particularly in obese patients, as long as no contraindications are present. First-line therapy with TZDs is becoming increasingly popular, but in the absence of convincing outcomes data and in light of side effects and cost, such a choice cannot yet be considered evidence based. Cardiovascular outcomes studies and investigations exploring the effects of TZDs on β-cell function should be available over the next 1-2 years. The results of these may indeed alter recommendations regarding the optimal initial approach to this disease.
Primary therapy with secretagogues is no longer as popular. In certain patients, particularly those in whom there appears to be a greater degree of pancreatic dysfunction as opposed to insulin resistance, or in those with contraindications for the other agents (e.g., advanced heart failure), their use as initial therapy is logical. Patients with erratic meal schedules and those with marked postprandial glucose excursions may do best with the rapid-acting non-SU secretagogues. The AGIs may best benefit those patients with mild hyperglycemia, particularly those with demonstrable postprandial excursions who are able to tolerate the significant side effect profile of drugs in this class.
As discussed above, diabetes is a complex disorder that involves multiple pathophysiological defects. Data from the UKPDS suggested that a ∼ 50%loss of β-cell function was already present in newly diagnosed type 2 diabetic patients.128 As the disease progresses, further functional decline in β-cell output is apparent. As a result, only 50% of patients were adequately controlled on monotherapy 3 years after diagnosis; by 9 years, this figure had fallen to 25%. Thus, combination therapy involving agents with complementary mechanisms of action is not only logical but frequently necessary to achieve control.
Published trials since our original report confirm the additive beneficial effects on glucose control of agents from different therapeutic classes19,33,39,58-60,129-145 (Table 3). Typically, the A1C reduction resembles the effect of the added individual agent when used as monotherapy. Few studies, however, suggest an actual “synergistic”effect. Precisely how various regimens function together metabolically remains incompletely understood but is an area of great interest that warrants further inquiry.
Over the past several years, the availability of several combination products incorporating SUs with metformin or metformin with a TZD have been marketed. These convenient formulations may enhance compliance. Their availability raises the potential for starting patients at the outset with two drugs. Such an approach is logical and will likely result in quicker achievement of target glucose levels, particularly in those with the greatest degree of baseline glycemia.
Additional pharmacological agents will likely soon become available for the management of patients with type 2 diabetes. These include other PPAR-agonists with additional effects on PPAR-α and PPAR-δ, and consequently better lipid effects than current TZDs.146 Several agents are in late-phase trials, although several others have been dropped at this stage because of toxicity concerns.
Modulation of the incretin system is another area of active investigation by several pharmaceutical companies. Incretin mimetics include glucagon-like peptide-1 agonists and the dipeptidyl peptidase-IV inhibitors,147,148 which augment endogenous incretin levels. These drugs improve glucose-dependent insulin secretion while simultaneously suppressing glucagon secretion, delaying gastric emptying, and decreasing appetite. Modest decreases in body weight are described with their use.
Obesity, the principle cause of type 2 diabetes, remains an important target for possible drug therapy. Available antiobesity drugs have limited effectiveness on body weight; clearly, newer therapeutic options are needed. One such agent,rimonabant,149 modulates the endogenous cannabinoid system and appears to be furthest along in development. We suspect that future weight loss agents with more substantive effects on body weight will likely play an increasingly important role in the future therapy of obese type 2 diabetic patients.
During the past 5 years, oral agent options for patients with type 2 diabetes have remained relatively static, and there is a paucity of new information from diabetes clinical trials that would significantly affect the way we should prescribe these drugs. In contrast, over the next several years,as the results of key clinical trials are revealed, the optimal therapeutic approach will likely be better defined, specifically regarding the best initial therapy for drug-naive patients. Such a choice may arise from studies exploring the cardiovascular and β-cell impact of various agents,particularly the insulin sensitizers. Other emerging concepts being addressed by ongoing investigations involve the notion of earlier treatment, perhaps even in the prediabetic state, more aggressive progression to combination strategies, and more liberal use of insulin sooner in the disease course.
It is unlikely, however, that any study result will alter the realization that the ideal drug choice for a specific individual is a complex decision that needs to be made by each practitioner, taking into account the risks and benefits of each agent and the requirements, capacities, and unique clinical features of each patient. Moreover, the actual selection may not be as important as an overall comprehensive approach to care that involves not only glycemic management, but also aggressive modification of other cardiovascular risk factors. To what extent emerging drug classes will affect this therapeutic approach to type 2 diabetic patients remains unclear.
Bonnie Kimmel, MD, is a senior resident in general internal medicine at the Yale Primary Care Residency Program in Waterbury and New Haven, Conn. Silvio E. Inzucchi, MD, is a professor of medicine and clinical director of the Section of Endocrinology at Yale University School of Medicine and director of the Yale Diabetes Center at Yale-New Haven Hospital in New Haven,Conn.
Note of disclosure: Dr. Inzucchi has served on advisory boards for Takeda, Pfizer, and Novartis. He has received honoraria for speaking engagements from Takeda, GlaxoSmithKline, and Bristol-Myers Squibb. These companies market oral pharmaceutical products for the treatment of diabetes.