| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Special Features |
Diabetes Center and General Clinical Research Center, Massachusetts General Hospital, and Harvard Medical School, Boston, Massachusetts 02114
Address all correspondence and requests for reprints to: Dr. David M. Nathan, Diabetes Unit, Massachusetts General Hospital, Bulfinch 408, Boston, Massachusetts 02114.
Abstract
The shift from empiricism and observational studies to experimental methods as the basis for advancing the treatment of human disease has occurred only recently. The modern age of clinical trials, born with the British Medical Research Council’s study of streptomycin treatment of tuberculosis in 1948 (1), has provided the substrate for evidence-based medicine. Despite the relatively brief period during which clinical trials have held sway, no one would contest the primacy of such studies in directing the development, evaluation, and implementation of new therapies in the past 50 yr. Clinical trials have been credited with "three of the seven years of increased life expectancy over that time and an average of five additional years of partial or complete relief from the poor quality of life associated with chronic disease" (2). The enormous expansion of clinical trials has paralleled the development of new therapies. With the large number of drugs in development and the requirement for controlled clinical trials for approval, the expanding number of clinical trials is not expected. Although the majority of clinical trials have focused on new drugs, a substantial number are performed to evaluate devices, noninvasive and minimally invasive procedures, diagnostic methods, and nonpharmacological interventions.
Despite the obvious benefits of bringing scientific methods to the evaluation of therapies, the need for clinical trials has recently been challenged (3, 4). Proponents of epidemiological studies as substitutes for clinical trials have suggested that conclusions from observational studies, including meta-analyses, often give similar answers as clinical trials. Although empirical data may provide useful information regarding established therapies, it should be obvious that they play a very different role from clinical trials. For example, new and as yet unapproved drugs cannot be tested observationally. Moreover, recent controlled clinical trials of nonpharmacological interventions (5, 6) that disproved therapies commonly accepted on the basis of epidemiological studies (7, 8) belie the notion that controlled clinical trials are no longer needed. Finally, although observational studies may arguably provide comparable estimates of effects as clinical trials, they cannot provide the causal link between interventions and outcomes or precise estimates of risks. Without clinical trials, risk/benefit analyses, which are critical in the selection of acceptable therapies, cannot be performed.
Diabetes mellitus is the most common chronic endocrine disorder, affecting an estimated 5–10% of the adult population in industrial westernized countries and an increasing fraction of populous countries in Asia, Africa, and Central and South America where agrarian, often subsistence, economies are giving way to industrialization (9, 10, 11). The accompanying shift in life-style to more sedentary activity with higher fat, lower fiber diets and resultant obesity, apparently underlies much of the increased prevalence of type 2 diabetes (11).
Given the increasing importance of diabetes mellitus, and especially type 2 diabetes, as a public health problem, the recognition of its myriad long-term complications (12), and the increasing number of potential therapies to treat the metabolic disorder(s) and the attendant complications, it should come as no surprise that clinical trials have played a major role. Therapies established as effective in clinical trials have had a major salutary effect on the life span and quality of life for persons with diabetes. This review will examine the clinical trials that have had the greatest impact on type 1 and type 2 diabetes, the two major forms, and their complications, addressing the specific questions that have been answered, clinical issues that remain unsettled, and some studies in progress. Although there is no entirely satisfactory way of organizing such a discussion, I have chosen to classify the trials according to primary prevention, secondary intervention, and tertiary intervention (Table 1
and Fig. 1).
|
|
Type 1 diabetes: diabetes-specific complications (retinopathy, nephropathy, neuropathy).
The most significant advance in diabetes treatment since the description of the disease by Arateus almost 2000 yr ago (13) was the discovery and use of insulin by Banting and Best in 1922 (14). Insulin transformed a uniformly fatal disease in which "life was short, disgusting, and painful" (13) to one in which long-term complications prevailed (12). Even clinical trial enthusiasts recognize that controlled clinical trials were not necessary, or were perhaps even unethical, in testing a drug for a uniformly fatal disease whose natural history was well known and for which no effective therapy existed.
As insulins with different activity profiles were developed, mostly for patient convenience, clinical studies to determine which were the most effective insulins and combinations of insulin were necessary. The initial goal of these studies was to determine which insulins would provide better control of blood glucose levels. The development of the first longer-acting insulin, neutral protamine Hagedorn or NPH, in 1935 was aimed at sparing patients from the inconvenience of three to five daily injections with the rapid-acting, shorter duration insulin that was available at the time.
The observation that many patients with type 1 diabetes were surviving only to develop vision-threatening retinopathy, nephropathy ending in renal failure and death, and neuropathy spurred a 50 yr debate regarding the meaning of better glucose control (15, 16, 17, 18). The debate focused on secondary and tertiary intervention and whether near-normal glycemia would prevent the development or delay the progression of long-term diabetic complications (the glucose hypothesis). The development of accurate, convenient methods for self monitoring of blood glucose (19), the glycosylated hemoglobin assay to measure long-term glycemia objectively and accurately (20), multiple daily insulin regimens and continuous sc insulin infusion with external pumps to achieve near-normal glycemia (21), and quantitative methods to measure long-term complications all set the stage for testing the glucose hypothesis.
The Diabetes Control and Complications Trial (DCCT) and Stockholm Diabetes Study were large, carefully designed, controlled clinical trials, with 1441 and 102 subjects, respectively, to determine whether intensive diabetes management aimed at achieving and maintaining glycemic levels as close to the nondiabetic range as possible would prevent the development and delay the progression of long-term complications in type 1 diabetes (22, 23). With its 1441 subjects, 99% of whom completed the study, 6.5 yr of average follow-up, 98% compliance to assigned therapy, and numerous carefully measured end points, the DCCT was large and long enough to provide definitive, conclusive answers to the glucose hypothesis (Fig. 2A and Table 2). Retinopathy was the primary study end point in the DCCT, with nephropathy, neuropathy, cardiovascular disease (CVD) and its risk factors, adverse effects of therapy, and the economic consequences of therapy being other important end points. Two separate cohorts based on retinopathy and albuminuria at baseline were recruited: the primary prevention cohort had no retinopathy, less than 40 mg/24 h albuminuria, and 1- to 5-yr duration, and the secondary intervention cohort had at least one microaneurysm in either eye (but not more than moderate nonproliferative retinopathy), no more than 200 mg/24 h albuminuria, and diabetes duration of 1–15 yr.
|
|
, demonstrated a durable, uniform, and major effect of intensive therapy on all three diabetes-specific long-term complications. These results were obtained in the setting of maintaining hemoglobin A1c (HbA1c) approximately 2% lower in the intensive treatment group than that in the conventional treatment group (Fig. 2A). Intensive therapy was also associated with a 3-fold increase in hypoglycemia, including severe episodes, and an increased risk for weight gain compared with conventional treatment (24). Subsequent analyses demonstrated the association of lower glycemia with lower risk of retinopathy and nephropathy (25); the benefits of intensive therapy during pregnancy (26, 27); preservation of residual endogenous insulin secretion with intensive therapy, which translated into lower glycemia with lower doses of insulin, decreased risk for hypoglycemia, and lower risk for retinopathy (28); and the cost/benefit of intensive therapy (29). The DCCT Research Group concluded that most patients with type 1 diabetes should be treated as early as possible with intensive therapy with the aim of maintaining glycemia as close to the normal range as safely possible. These recommendations were widely endorsed (30). The benefits of intensive therapy have been reinforced by even longer term follow-up of the DCCT cohort (31). After an additional 4 yr of follow-up, intensive therapy continues to manifest a more than 75% reduction in long-term complications compared with conventional therapy (Table 2). CVD.
The major risk for CVD in the setting of type 1 diabetes is associated with the development of nephropathy (32). Glycemia itself has been suggested to be a risk factor for CVD (33). Although it may be logical to assume that the prevention or delay of nephropathy with intensive therapy or lower glycemia would eventually result in a decrease in the development of CVD, no study of type 1 diabetes has been of long enough duration to demonstrate this result. The DCCT population was relatively young and generally healthy and, not surprisingly, did not have very many severe CVD events during the course of the study (22, 34). Although there were numerically fewer CVD events in the intensive than the conventional treatment group, the difference did not achieve statistical significance. Long-term follow-up of the DCCT cohort will examine differences between the original treatment groups in atherosclerosis and CVD (31).
The DCCT with its extraordinary degree of study completion and compliance with an extremely demanding protocol may have set the gold standard for controlled clinical trials. It successfully achieved all of its goals, ending a more than 50-yr debate regarding the importance of glycemic control in the treatment of type 1 diabetes mellitus. In doing so, it established justifiable, rational metabolic goals for type 1 diabetes as well as the risk/benefit ratio of currently available therapies. Despite its many successes, the DCCT left unanswered whether the heavily subsidized, resource-rich intensive therapy performed in motivated research volunteers in the DCCT could be effectively translated into the nonresearch, clinical setting. [Recent reports suggest that the interventions and glycemic goals recommended by the DCCT are being implemented and achieved with some degree of success (35).] Moreover, although the DCCT was very successful in demonstrating how to reduce disease in type 1 diabetes over time, the burden of intensive care has been placed on patients with type 1 diabetes. Finally, whether CVD, a nonspecific complication of type 1 diabetes, is decreased by intensive glycemic therapy is unknown.
Type 2 diabetes
Type 2 diabetes does not have the long pedigree of type 1 diabetes due to the relatively recent recognition of this form of diabetes associated with older age, obesity, and not requiring insulin for survival (36); however, by virtue of the similar spectrum of complications as in type 1 diabetes, Type 2 diabetes has been embroiled in the same debate regarding control and complications. The first large multicenter trial to determine whether more intensive therapy would affect outcome, the University Group Diabetes Program, failed to demonstrate that more intensive therapy had a beneficial outcome (37). A larger, more powerful study, the United Kingdom Prospective Diabetes Study (UKPDS) was initiated in 1977 and completed in 1997 (38). Recruiting approximately 5000 recent-onset type 2 diabetic subjects, the UKPDS sought to determine whether an active intervention aimed at achieving normal glycemia would result in better aggregate diabetes outcomes, including diabetes-specific and nonspecific cardiovascular outcomes, than a control intervention (diet). The second question addressed was whether any particular hypoglycemic regimen was advantageous. After an initial 3-month dietary run-in period, eligible UKPDS volunteers were randomly assigned to diet control, sulfonylurea, or insulin therapy. In addition, overweight subjects could be randomly assigned to metformin. If glycemic control was not being maintained, additional therapy was added. Unfortunately, glycemic control proved very difficult to maintain in all of the intervention groups, with HbA1c levels exceeding baseline levels in the active intervention groups by 5 yr and leading to a high level of therapeutic cross-over (Fig. 2B). The stepped intervention design interfered with the ability of the UKPDS to judge differences between assigned therapies (39). Over 10 yr of follow-up, intensive therapy resulted in a 1% (absolute) lower HbA1c value than the conventional therapy (Fig. 2B). The 11% difference in HbA1c was associated with a 12% lower risk in aggregate diabetes outcomes, with most of the reduction predicated on a 25% reduction in retinopathy and nephropathy (Table 3). CVD, although more frequent than in type 1 diabetes in the DCCT, was not significantly reduced with intensive diabetes therapy. Although comparison of the different therapies was problematic, as noted above, metformin treatment was associated with less weight gain than insulin and sulfonylurea therapies.
|
Nonmetabolic therapies: secondary and tertiary intervention
Retinopathy.
Laser photocoagulation was established as a means of delaying the progression of severe retinopathy more than 20 yr before the role of intensive diabetes therapy in secondary and tertiary intervention was demonstrated (40, 41). Taking advantage of the generally symmetric nature of diabetic retinopathy, the Diabetes Retinopathy Study (DRS) randomly assigned the eyes of patients with type 1 and type 2 diabetes that were affected by proliferative or severe nonproliferative retinopathy to receive pan-retinal argon or xenon arc laser therapy or no therapy (40). By 2 yr, the DRS demonstrated a 50% reduction in severe loss of vision (acuity worse than 5/200) in the treated compared with untreated eyes (6.4 vs. 15.9%; Fig. 3A). Although almost all laser-treated eyes benefited from therapy, the side-effects of laser therapy (e.g. decreased peripheral fields and night vision) dictated that the most acceptable benefit/risk ratio occurred with photocoagulation of high risk eyes, defined post-hoc as having the greatest risk for visual loss. High risk characteristics included 1) proliferation within one disk diameter of the optic disk and exceeding one quarter of the disk area in size, 2) any proliferation within one disk diameter in association with vitreous or preretinal hemorrhage, or 3) proliferation elsewhere at least half the disk area in size and associated with hemorrhage.
|
). The ETDRS had several important negative findings. Treatment of less severe retinopathy than was addressed in the DRS with either focal or scatter photocoagulation did not reduce the rate of moderate or severe vision loss compared with deferred treatment. In addition, aspirin proved not to have a beneficial effect on retinopathy; however, it did not appreciably increase the risk of retinal bleeding and appeared to decrease the risk of CVD mortality. The DRS and ETDRS were elegantly designed and carefully performed studies that established means to decrease the vision loss associated with diabetic retinopathy. Widespread implementation of intensive therapy should decrease the need for laser therapy over time; DCCT long-term follow-up demonstrated 74–77% reductions in macular edema, proliferative retinopathy, and laser therapy with intensive therapy (31).
The Diabetes Vitrectomy Study (DVS) cohort included patients with more severe retinopathy than those in the DRS or ETDRS (42). Eyes with severe proliferative retinopathy were randomly assigned to receive vitrectomy or deferral of surgery for 1 yr. The DVS belied, to some extent, the notion that clinical trials cannot include random assignment to surgery vs. no surgery. The potential adverse outcomes of vitrectomy, including worsening of vision, made careful evaluation of its benefits mandatory. Early vitrectomy resulted in a substantial improvement in visual acuity, especially in patients with type 1 diabetes.
Other therapies for retinopathy have been evaluated in clinical trials (41, 43). Intensive blood pressure management has proved to be the most effective nonglycemic approach to prevent and or delay progression of retinopathy. One of the components of the UKPDS was a study of "tight" blood pressure control aimed at achieving systolic and diastolic blood pressures less than 150 and 85 mm Hg, respectively (43). Subjects randomly assigned to tight control were assigned to either atenolol or captopril. Tight blood pressure control with either agent resulted in significant differences in blood pressure over time (mean systolic blood pressure of approximately 144 and 154 mm Hg and diastolic of 82 and 87 mm Hg in tight and less tight treatment groups, respectively). Intensive blood pressure therapy with either atenolol or captopril reduced the risk of a two-step progression of retinopathy by 34% and of deterioration of visual acuity by 47%.
Nephropathy
Diabetic nephropathy, progressing from the earliest stages of glomerular hyperfiltration and hypertrophy to microalbuminuria (urinary albumin excretion, 30–300 mg/24 h), clinical proteinuria (>300 mg albumin excretion/24 h), and inexorably to end-stage renal failure, is the diabetes-specific complication that carries the most morbidity, mortality, and expense. As such, it has attracted much attention and a large number of clinical trials. In addition to controlled clinical trials of antihypertensive therapy, the use of angiotensin-converting enzyme inhibitors (ACE inhibitors) has dominated the field. The UKPDS 9-yr study failed to demonstrate a consistent and durable effect of intensive blood pressure control with either atenolol or captopril on diabetic nephropathy, including microalbuminuria, proteinuria, and creatinine levels (Table 3) (43). Numerous, generally short-term clinical trials have shown a beneficial effect of antihypertensive therapies (44), including a variety of ACE inhibitors (but, interestingly, at this date not the angiotensin receptor blockers), on discrete stages of diabetic nephropathy, such as the development of microalbuminuria (45, 46), the progression of microalbuminuria (47, 48), and the development of worsening renal failure in type 1 and type 2 diabetes (49). None of these studies has been "patient" enough to demonstrate the ultimate, long-term benefit of such therapies, i.e. to decrease the occurrence of renal failure in patients starting with the earliest stage of nephropathy. However, piecing together individual studies, each of which has demonstrated benefits at specific stages of diabetic nephropathy, creates a relatively convincing tapestry of long-term improvements in diabetic nephropathy. (The UKPDS data are sobering in their failure to demonstrate a significant impact after 9 yr of tight blood pressure therapy.) The most convincing clinical data arise from the study by Lewis et al., in which progression to severe renal failure was slowed with captopril (49) (Fig. 4). In concert with intensive glycemic control, therapy with ACE inhibitors is projected to decrease the occurrence of nephropathy substantially. Some (50), but not all (51), studies have suggested that the beneficial effects of these therapies are already evident. In addition to intensive glycemic and blood pressure therapy, diet therapy with protein restriction has been studied with controlled clinical trials (52, 53).
|
Intensive diabetes management has been shown to decrease the development and slow the progression of the myriad forms of diabetic neuropathy that contribute to the occurrence of diabetic foot ulcers and amputations and manifestations of autonomic neuropathy (22, 54). The putative role of sugar alcohols, such as sorbitol, in the pathogenesis of diabetic neuropathy has spurred the development of aldose reductase inhibitors and the conduct of clinical trials to determine whether inhibition of sorbitol production would decrease the development of neuropathy. Despite optimistic results in animal models of diabetes, human clinical trials have played an important, albeit sobering, role in failing to demonstrate any convincing clinical effects (55). Further development of such agents has slowed, if not stopped. Clinical trials have also played a critical role in the development of effective therapies for symptomatic somatosensory and autonomic neuropathies (56, 57, 58). Appropriate placebo controls have been particularly important in these studies, in which subjective reporting of symptoms is often the end point.
CVD
As noted above, intensive glycemic therapy has not as yet been demonstrated to reduce CVD morbidity or mortality in long-term interventional studies in type 1 or type 2 diabetes, although one interesting controlled clinical trial demonstrated decreased mortality at 1 and 3 yr after an acute myocardial infarction when intensive insulin therapy was used during the immediate postinfarct period (59). On the other hand, recent controlled clinical trials of nonglycemic therapies have suggested beneficial effects on CVD in type 2 diabetes similar to those in nondiabetics. Aggressive therapy of hypertension reduced the rate of cerebrovascular events in the UKPDS (43), and diuretic therapy was effective in the primary prevention of severe CVD in type 2 diabetic patients in the Systolic Hypertension in the Elderly Program (60). Aggressive therapy of hypercholesterolemia with HMG coenzyme A reductase inhibitors has been effective in secondary intervention of CVD similar to results seen in nondiabetics (61, 62). In addition, gemfibrozil treatment of patients with low levels of high density lipoprotein in the setting of modest hypertriglyceridemia, the most common lipoprotein phenotype in type 2 diabetes, decreased the occurrence or recurrence of major CVD events (63). Finally, aspirin therapy is probably effective in preventing myocardial infarctions in diabetes, similar to its effect in nondiabetics (64, 65). Of note, and characteristic of most controlled clinical trials of CVD in diabetes, is that despite a generally similar relative efficacy of interventions in diabetics and nondiabetics, the absolute rate of CVD events remains higher in treated diabetics than in nontreated (control) nondiabetics. Thus, the risk for CVD contributed by diabetes and the comorbid conditions that accompany it, such as hypertension, dyslipidemia, and obesity, is only partially reversed by aggressive interventions. This observation suggests that there are risk factors for CVD in diabetes as yet to be discovered or addressed and/or that conventional therapies must be applied more aggressively than in nondiabetics to improve substantially the long-term prospects of people with diabetes.
Primary prevention
The overwhelming burden of diabetes and its treatments, including the human and economic costs of the long-term complications, has stimulated studies of prevention. The autoimmune basis of type 1 diabetes, the ability to identify affected persons early in their clinical or preclinical course, and the potent environmental factors that influence the development of type 2 diabetes, which may be modifiable, lend themselves to prevention. Numerous small, generally underpowered studies of immune modulation to try to prevent type 1 diabetes from developing in persons at high risk or to ameliorate metabolic abnormalities in persons with new onset disease have been performed (66, 67). Two larger, more robust, controlled clinical trials are ongoing (68, 69).
Controlled clinical trials in type 2 diabetes focusing for the most part on modifiable environmental factors, such as diet and exercise, have been performed (70, 71). Early results from Chinese and Finnish studies suggest that life-style interventions aimed at increasing physical activity and decreasing body weight may be effective in preventing diabetes in high risk populations. An ongoing U.S. study, the Diabetes Prevention Program, is testing whether an intensive life-style intervention or metformin therapy will prevent or delay the development of type 2 diabetes in a high risk impaired glucose tolerance population (72).
Translation of clinical trial results
Ultimately, the results of clinical trials remain academic unless they are implemented. Several factors can interfere with the translation of interventions demonstrated as effective in clinical trials. To the extent that a study population is highly selected and not representative of the nonstudy disease population, the study results may not be generalizable. In addition, study therapies that cannot be implemented in the nonstudy setting owing to their complexity, expense, or other factors will be of limited utility. On the other hand, if study populations and interventions are carefully chosen, balancing research needs and clinical relevance, effective therapies can be promulgated. The usual routes for disseminating information regarding new effective therapies include publication in peer-reviewed journals, presentation of results at medical meetings, guidelines prepared by professional organizations and governmental agencies, and promotional advertising by the pharmaceutical industry. Investigators often lose control of their study "message" during the bedside translation; however, their participation in the educational outreach programs and formulation of guidelines that often follow large, successful, clinical trials should serve to preserve the scientific and clinical integrity of the study results. Investigators should be encouraged to participate in this important translational phase of clinical research.
Summary
Clinical trials have played a critical role in defining effective therapies for diabetes mellitus and for different stages of its myriad complications. Equally as important, ineffective therapies have been discarded on the basis of controlled clinical trials. The net effect of controlled clinical trials has been an expansion of life span and an improvement in quality of life for persons afflicted with this chronic degenerative disorder. Future studies may provide effective prevention strategies and cures.
Footnotes
Abbreviations: ACE inhibitors, Angiotensin-converting enzyme inhibitors; CVD, cardiovascular disease; CSME, clinically significant macular edema; DCCT, Diabetes Control and Complications Trial; DVS, Diabetes Vitrectomy Study; ETDRS, Early Treatment Diabetic Retinopathy Study; HbA1c, hemoglobin A1c; UKPDS, United Kingdom Prospective Diabetes Study.
Received January 22, 2001.
Accepted March 26, 2001.
References
This article has been cited by other articles:
 |
|
 |
|
  L. C. Tapsell, L. J. Gillen, C. S. Patch, M. Batterham, A. Owen, M. Bare, and M. Kennedy Including Walnuts in a Low-Fat/Modified-Fat Diet Improves HDL Cholesterol-to-Total Cholesterol Ratios in Patients With Type 2 Diabetes Diabetes Care, December 1, 2004; 27(12): 2777 - 2783. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. K. Plante and J. L. Nadler Diabetes and Vascular Disease Seminars in Cardiothoracic and Vascular Anesthesia, September 1, 2003; 7(3): 295 - 310. [Abstract] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
