Diabetes Control, Long-Term Complications, and Large Vessel Disease
The historical documentation of type 1 diabetes includes at least 5000 years of testimony to its dramatic clinical onset with the apparent “melting of flesh into urine,” followed by starvation, inanition, and certain death.1 However, it wasn’t until the near miraculous cure of diabetes, with the introduction of insulin less than a century ago, that long-term complications began to be observed with any frequency.2 With longer survival of children and adolescents with what is now called type 1 diabetes, we began to see manifestations of diabetes that had never been seen before. Diabetic retinopathy and nephropathy were first described in the 1930s.3,4 Insulin therapy transformed type 1 diabetes from a disease that was generally fatal in the first 6 to 24 months of its appearance, to a chronic degenerative disorder with a host of long-term complications that affected the eye, kidney, nervous system, and heart.
The development of these complications, which ultimately affected most patients afflicted with type 1 diabetes, had a severe toll. By the time insulin had been in use for 50 years, diabetes had become a major cause of blindness, kidney failure, and amputations and a contributor to cardiovascular disease.5 With the epidemic of type 2 diabetes, diabetes had become by 1990 the greatest single cause of these disorders. It is not surprising that some early clinicians/investigators considered the pathogenesis of these complications to be iatrogenic, with insulin, directly or indirectly, causing the complications. Still others thought that the complications were co-inherited but independent from the metabolic perturbations of the disease. Finally, a small but vocal group insisted that the nonphysiologic control of glucose levels was the cause of long-term complications.6–9
Unfortunately, the passionate proponents of these theories had no proof to support or refute the hypotheses, and the theories remained open to often rancorous debate well into the 1970s. Beginning in the mid-1970s, a series of animal studies investigated what became known as the “glucose hypothesis” and demonstrated in animal models that glucose control was linked to the risk for developing diabetic eye and kidney disease.10–15 The subsequent development of a number of critical clinical research tools—including an objective means of measuring long-term glycemia (glycated hemoglobin [HbA1c] assay),16 glucose monitoring, insulin delivery algorithms, devices that could achieve near normal glycemia,17 and objective methods to assess the development and progression of complications—allowed the organization and implementation of clinical trials to examine the glucose hypothesis.18 In 1993, after almost 10 years of study, the most important of these trials, the Diabetes Control and Complications Trial (DCCT), provided definitive answers to the questions that had dominated the diabetes debate for longer than 60 years.19 The DCCT established the primacy of intensive therapy aimed at achieving near-normal glycemia to prevent and delay the specific long-term complications of diabetes mellitus. In doing so, the DCCT19 and other clinical trials in type 1 and type 2 diabetes have helped identify the pathogenesis of the long-term complications of diabetes.20–22 This chapter focuses on our understanding of the risk factors and, in particular, the role of glycemia in the pathogenesis of the long-term complications of diabetes mellitus.
Animal Studies
The animal models of diabetes and its complications are sufficiently different from human models of type 1 and type 2 diabetes and provide only suggestive evidence regarding the pathogenesis of diabetic complications. Nevertheless, they uniformly support the role of therapies that normalize blood glucose levels to prevent and/or delay the progression of retinopathy, nephropathy, and neuropathy. Three major animal models have been studied. In one model, animals with chemically induced (alloxan or streptozotocin) diabetes are treated with insulin with the goal of achieving tight or loose control of blood glucose levels.10,11 In another model, pancreatectomized animals are treated with pancreatic or isolated islet cell replacement.12,13 Finally, animals with genetic diabetes (with models of autoimmune diabetes such as the NOD mouse or BB rat, or with models of type 2 diabetes) and various degrees of glycemia have been studied.14,15 Most studies have demonstrated primary prevention of complications with intensive therapy aimed at maintaining glucose levels close to the physiologic range. Secondary intervention studies, that is, prevention of progress of established complications, have been much less common.
The most convincing animal studies were done by Engerman and colleagues with alloxan-induced diabetes in dogs.10,11 The diabetic dogs developed retinopathy with microaneurysms and pericyte loss similar to those seen in diabetic humans. Therapy with two daily injections of NPH insulin with the goal of aglycosuria (“good control”) was initiated soon after the dogs were made diabetic.10 Good control was shown to be associated with fewer microaneurysms than occurred with one daily injection of NPH insulin (“poor control”) over a 5-year period. A later study demonstrated that if dogs with alloxan-induced diabetes were treated with the poor-control regimen for 2.5 years, followed by the good-control regimen for 2.5 years, they developed an intermediate number of microaneurysms.11 This prescient study suggested that secondary intervention was not as effective as primary prevention, and it forecast the results of human studies that would follow almost 10 years later. Of note, severe hypoglycemia resulted in the deaths of several dogs in “good control.”
Although glomerular lesions in several animal models of diabetes are similar to those seen in diabetic nephropathy in humans, the time course of the development of the lesions is difficult to compare with that seen in human diabetes. In addition, animal models of renal disease have several limitations. First, diabetic rats develop a renal lesion (mesangial expansion) that differs from the early lesion of human nephropathy (glomerular basement membrane expansion) and do not develop end-stage renal failure. Second, other potentially important variables that might predict or influence development of nephropathy in humans (e.g., hypertension) cannot be easily studied in animal models. Third, rats in which transplants of pancreatic islet cells do not succeed in correcting glucose levels also show improvement in renal results.12 As with retinopathy, studies of nephropathy in animal models can lend support to, but cannot prove, the glucose hypothesis.
Studies in animal models appear to demonstrate that nephropathy can be prevented or even reversed when diabetic animals are treated with pancreatic transplantation or with intensive insulin therapy. Rats with streptozotocin-induced diabetes develop mesangial thickening with immunoglobulin deposition within 6 to 9 months of diabetes onset.12 Successful islet transplantation prevented the development of such lesions or led to stabilization and some improvement in established lesions concurrent with normalization of glucose levels.12 Studies in other animal models such as the BB/W spontaneously diabetic rat15 and the uninephrectomized, alloxan-treated dog23 support the role of glucose control in the genesis of nephropathy.
Animal models of diabetic neuropathy are also limited; however, the ability to measure nerve conduction and to perform nerve biopsies may compensate for difficulty in determining specific sensory and motor deficits. Studies of glycemic control and neuropathy have also supported the glucose hypothesis.24
Human Studies
Retinopathy
Human studies conducted before 1964 were hampered by the absence of quantitative methods to evaluate long-term glucose control and complications and by a poor appreciation of clinical trial methods.25 Beginning in the 1970s, nondilated ophthalmoscopy gave way to seven-field stereoscopic fundus photography and fluorescein angiography, and sporadic blood glucose measurements and semiquantitative measures of glycosuria were supplanted by assays for glycated hemoglobin (HbA1 or HbA1c).18
Although lacking in these modern innovations, the longitudinal study of Pirart deserves mention, if only for its magnitude.26 Pirart followed 4398 patients with diabetes for as long as 25 years, although relatively few were followed for longer than 15 years. He noted that retinopathy, nephropathy, and neuropathy were more common in patients with a higher glycemic index, a value derived from intermittent measurements of blood and urine glucose levels and other factors. The high attrition rate over time, the lack of objective measures of complications and glycemia, and the possibility that complications would lead to worsened glucose control, rather than vice versa, detract from this study.
In the modern era, the population-based, observational Wisconsin Epidemiologic Study of Diabetic Retinopathy (WESDR) examined Wisconsin residents with diabetes using glycated hemoglobin measurements and seven-field stereoscopic fundus photography.27–29 Follow-up over 4 years revealed a striking association between the level of glycated hemoglobin at baseline and the incidence of any retinopathy, progression of retinopathy, or progression to proliferative retinopathy,27 macular edema,28 and vision loss.29 The relationship between levels of glycated hemoglobin and retinopathy was continuous; no threshold for glycated hemoglobin with regard to risk for retinopathy was noted. Associations among glycemia and diabetic complications remained after controlling for duration of diabetes, age, and baseline retinopathy. Other observational studies have confirmed the association of glycemia with retinopathy in selected type 1 diabetic populations30–33 and have suggested that higher glycemic levels are a risk factor for the development of proliferative retinopathy.31,33,34
Although WESDR subjects were not strictly categorized as having type 1 or type 2 diabetes, the separation by age at onset (<30 years vs. ≥30 years) effectively provided type 1 and type 2 populations. WESDR35 and other studies36 have demonstrated an association between retinopathy and glycemia in type 2 diabetes, similar to that seen in type 1 diabetes (Fig. 25-1). Other risk factors for diabetic retinopathy, in addition to the level of glycemia, include hypertension,37 pregnancy,38 hyperlipidemia,39 and a family history of diabetic retinopathy,40,41 but not smoking.42 Despite the familial clustering of retinopathy and the severity of retinopathy, candidate gene and genome-wide association–directed studies have not identified any dominant genetic mediators of retinopathy or retinopathy progression.
FIGURE 25-1 Association between mean glucose level as measured by level of glycated hemoglobin (HbA1c) and presence of retinopathy in older patients (55 to 75 years) with type 2 diabetes (n = 185). (From Nathan DM, Singer DE, Godine JE, et al: Retinopathy in older Type II diabetics: association with glucose control. Diabetes 35:797–801, 1986.)
Nephropathy
The natural history of diabetic nephropathy, although duration dependent, extends over many more years than retinopathy before clinical expression becomes evident.43,44 A minimum of 12 years, and more often 15 to 18 years, of type 1 diabetes is required before the development of clinical-grade (dipstick positive, i.e., ≥500 mg/24 hours) proteinuria, the first incontrovertible sign of end-stage renal disease. After the development of clinical-grade proteinuria, creatinine clearance declines over 5 to 10 years, terminating in end-stage renal disease.45
The reluctance to perform kidney biopsies early in the course of diabetes for documentation of microscopic changes in the glomerulus and the less than perfect correlation between microscopic changes and clinical course have led to reliance on surrogate markers of evolving nephropathy. “Incipient” nephropathy, as demonstrated by microalbuminuria (generally >20 mg to 30 mg, and <300 mg, of urinary albumin per 24 hours, or >30 µg per mg creatinine in a spot urine test), has been identified as a predictor or marker for the development of end-stage renal disease in retrospective studies of type 146–49 and type 2 diabetes.50 Unfortunately, microalbuminuria can vary considerably in individuals over time, with levels fluctuating from abnormal to normal values. Therefore, abnormal urinary albumin excretion, as defined above, in at least two of three urine collections within a 6-month period has been suggested as a definition of “persistent” microalbuminuria.51 Although the presence of or changes in microalbuminuria have been used as renal end points in many observational studies and clinical trials, a recent study has demonstrated that microalbuminuria may revert to normoalbuminuria more commonly than was previously appreciated, even in the absence of specific interventions.52
For several reasons, investigators have suspected that the association between glucose control and nephropathy may be more complex than that observed with retinopathy. The occurrence of nephropathy in no more than 40% of patients with type 1 diabetes and 25% of patients with type 2 diabetes suggests that variables other than glycemia are operant. Hypertension and family history of hypertension are mediators of nephropathy.53,54 In addition, candidate and genome-wide approaches have identified genetic risk factors for diabetic nephropathy.55,56
The association between levels of glycemia and nephropathy has been more difficult to establish than for retinopathy. Potential reasons for the difficulty in establishing a relationship between glycemia and nephropathy include the following: (1) the development of renal failure may influence glycemic control (e.g., alterations in sensitivity to insulin with development of hypertension and effects of antihypertensive medications on glycemia); (2) uremia, anemia, and transfusions may interfere with, or influence, the accuracy of measurements of glycated hemoglobin; and (3) given the long duration of diabetes before the development of renal failure, infrequent measurements of glycated hemoglobin, representing a relatively brief period of exposure, may not be predictive of the development of nephropathy. Even with these potential problems, studies have demonstrated an association between the derived glycemic index and an increase in creatinine level over time,26 or between mean levels of glycated hemoglobin, measured over 7 years, and risk of microalbuminuria in type 1 diabetes.57
Neuropathy
Quantitative electrophysiologic measures of nerve conduction have been available for more than 40 years and should have contributed to the examination of the association between glycemia and neuropathy. Unfortunately, the complex relationship between neurophysiologic studies and symptomatic clinical diabetic neuropathy has complicated the study of glucose control and neuropathy. For example, the early observation that insulin treatment of new-onset type 1 diabetes reversed slowed motor nerve conduction within 6 weeks in asymptomatic patients supported an acute effect of hyperglycemia on nerve conduction and cast doubt on the role of electrophysiologic testing.58 The absence of histologic data from peripheral nerve biopsies has been a major impediment to our understanding of diabetic neuropathy. A modest association between glycemia and motor and sensory nerve conduction has been documented in type 159 and type 260 diabetes.
Cardiovascular Disease
Cardiovascular disease (CVD), including atherosclerosis-based coronary artery, cerebrovascular, and peripheral artery disease resulting in myocardial infarctions, stroke, and foot ulcers that may require amputations, is a less specific consequence of diabetes than are the microvascular and neuropathic complications already discussed. Nevertheless, the risk for CVD is increased by twofold to fivefold in diabetic versus nondiabetic men, and by even more in diabetic versus nondiabetic women.61,62 The relatively greater impact of diabetes on CVD in women than in men may be due to the relatively greater impact of diabetes on CVD risk factors in women than in men.63,64 Both type 1 and type 2 diabetes are associated with increased risk for CVD, with the former imposing a greater relative risk than the latter. The absolute risk for CVD is far greater in persons with type 2 than type 1 diabetes, owing to their greater age and higher coincidence of other CVD risk factors, including hypertension, dyslipidemia (typically a high level of triglycerides and low high-density lipoprotein [HDL]-cholesterol level), and obesity. Autonomic neuropathy and especially diabetic nephropathy increase the risk for CVD in all persons with diabetes. 65
Interventional-Clinical Trials
At best, observational studies can only indicate associations between glycemic control (and other confounders) and complications. The implementation of randomized, controlled clinical trials has facilitated our understanding of cause and effect in the pathogenesis of diabetic complications. Treatments designed to achieve near-normal glucose control (“intensive therapy”) were compared with conventional therapies, and their differential effects on the development and progression of complications were studied. In type 1 diabetes, intensive treatment regimens took advantage of the introduction and refinement of methods for self-monitoring of blood glucose levels and of improved methods of physiologic replacement of insulin, such as continuous subcutaneous insulin infusion (CSII) with pumps and multiple daily injection (MDI) regimens.17 Four well-designed randomized studies,21,66–68 set the stage for the larger and more comprehensive DCCT.19,69 All of these preliminary trials were secondary intervention studies that included only subjects with retinal lesions at baseline and a relatively long mean duration of diabetes. The duration of these trials ranged from 8 to 60 months, and included 30 to 100 subjects. (By contrast, the DCCT studied 1441 subjects with a mean follow-up of 6.5 years.) The total number of patient-years of study was less than 800 in the four previous secondary intervention trials combined. (The total number of patient-years for the secondary intervention component of the DCCT was almost 5000 at study end in 1993.) Except for the Oslo study,68 which included two intensive treatment groups, the studies compared type 1 diabetic patients randomly assigned to conventional treatment versus patients randomly assigned to CSII66,67 or MDI.21 The results of the Kroc,66 Steno,67,70 and Oslo,68,71 studies were similar with regard to retinopathy. In the first 6 to 12 months, a transient worsening of retinopathy occurred among patients receiving intensive treatment. Of the early trials, only the Stockholm Diabetes Study demonstrated a beneficial effect of intensive therapy over time.21
Diabetes Control and Complications Trial: In 1993, the DCCT ended the 60-year debate regarding the relationship between metabolic control and long-term complications. DCCT investigators reported consistent, unequivocal salutary effects of intensive diabetes management on the development and progression of the microvascular and neurologic complications of type 1 diabetes mellitus.19
Design.: The DCCT, initiated in 1983, was designed to answer definitively whether intensive diabetes management would affect the development and/or progression of long-term complications in type 1 diabetes, and at what cost.69 The DCCT addressed primary prevention and secondary intervention of chronic complications by including two parallel studies. The primary prevention study determined whether intensive therapy aimed at achieving glycemic levels as close to the nondiabetic range as possible would prevent the development or slow the progression of complications in type 1 diabetic patients aged 13 to 39 with 1 to 5 years of diabetes duration and no evidence of retinopathy or nephropathy. The secondary intervention study determined whether intensive therapy would prevent the progression of complications in type 1 diabetic patients with 1 to 15 years of diabetes duration and at least one microaneurysm but no more than moderate nonproliferative retinopathy. They could have as much as 200 mg of albumin excretion per 24 hours (although only a small fraction had this level of albuminuria at baseline). The baseline characteristics of the two study cohorts are shown in Table 25-1. Study patients also were selected based on an assessment that they would accept random assignment of therapy, and that they were likely to continue to participate in a long-term study. On average, these patients were probably more motivated than the usual patient with type 1 diabetes. The DCCT also examined the costs, both financial and adverse events, associated with intensive compared with conventional therapy.
Table 25-1
Baseline Characteristics of DCCT Cohort
*HbA1c: nondiabetic range, 4% to 6.05%.
†Based on seven time-point capillary blood samples, before and 2 hours after each meal and before bed.
‡The presence of “confirmed” clinical neuropathy was based on a history or physical examination consistent with somatosensory distal neuropathy plus either abnormal electrophysiologic testing (nerve condition studies) or abnormal autonomic testing results.
From DCCT Research Group. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. N Engl J Med 329:977–986, 1993.
Intensive Treatment and Metabolic Goals.: Primary Prevention and Secondary Intervention cohorts were randomly assigned to conventional therapy (designed to mimic the usual diabetes therapy at that time with one or two daily injections of insulin and daily glucose monitoring) or to intensive therapy (designed to normalize blood glucose control). Conventional therapy had the clinical goals of avoiding any symptoms of hyperglycemia or hypoglycemia, but no specific numeric blood glucose targets. Intensive therapy had the goal of achieving blood glucose control as close to the nondiabetic range as possible, including pre-meal blood glucose levels between 70 and 120 mg/dL (3.9 to 6.7 mMol/L), peak postprandial levels less than 180 mg/dL (10 mMol/L), and hemoglobin HbA1c levels in the nondiabetic range (<6.05%). In order to reach these goals, patients assigned to intensive therapy used three or more insulin injections per day or insulin pump therapy, guided by frequent self-monitoring of blood glucose levels and adjusted based on meal size, composition, and exercise. (See Table 25-2 for a description of the intensive regimen.)
Table 25-2
Diabetes Control and Complications Trial Intensive Therapy
Self-Monitoring of Blood Glucose
Glycated Hemoglobin
HbA1c every 3 months measured centrally with HPLC assay (nondiabetic range, 4% to 6.05%)
Insulin
3 or more injections per day or CSII with external pump
Doses adjusted based on ambient glucose (SMBG), meal size and content, and anticipated exercise
Supervision
Results.: Detailed results of intensive compared with conventional therapy in the DCCT have been reported extensively.19,72–90 The initial report19 summarized the major results, whereas subsequent reports presented expanded analyses of the effects of intensive therapy on long-term complications, including retinopathy,72–74 nephropathy,75 neuropathy,76–78 and macrovascular disease and its risk factors79; the effects of intensive therapy on quality of life,80 neurobehavioral outcome,81 and residual insulin secretion82; the implementation83 and adverse effects of intensive therapy84,85; the cost-benefit analysis of intensive therapy compared with conventional therapy86; the results of intensive therapy on pregnancy87; and the association among glycemia, long-term complications, and other risk factors.88–90 A long-term follow-up study of the DCCT cohort, the Epidemiology of Diabetes Interventions and Complications (EDIC) study, is in its 15th year, as of 2008, and is providing further insight into the long-term consequences of intensive therapy.91–98
Adherence and Metabolic Results.: Over the 6.5-year mean follow-up time of the DCCT (range, 3 to 9 years), compliance was excellent, with more than 99% of the cohort completing the trial.19 In addition, virtually no crossover between assigned treatments was noted. Subjects adhered to their assigned treatment for more than 97% of study time. Intensive therapy decreased HbA1c to a nadir of approximately 6.9% by 6 months and maintained mean HbA1c levels during the remainder of the trial that were approximately 2% lower than with conventional treatment (7% vs. 9%) (Fig. 25-2
