RSS Feeds
The Pathophysiology of Diabetic Complications: How Much Does the Glucose Hypothesis Explain?
- David M. Nathan, MD
- From Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts. For the current author address, see end of text. Note: This article is one of a series of articles comprising an Annals of Internal Medicine supplement entitled “Risks and Benefits of Intensive Management in Non-Insulin-dependent Diabetes Mellitus: The Fifth Regenstrief Conference.” To view a complete list of the articles included in this supplement, please view its Table of Contents. Grant Support: In part, by the Diabetes Control and Complications Trial (NIDDK UO1 DK30643-13) and the Mallinckrodt General Clinical Research Center (RR01066-18). Requests for Reprints: David M. Nathan, MD, Diabetes Unit, Massachusetts General Hospital, Boston, MA 02114.
Abstract
Objective: To examine the putative pathogenetic mechanisms of the long-term, specific complications of diabetes mellitus.
Data Sources: Literature review relevant to long-term diabetic complications and their pathogenesis.
Study Selection: Studies of animal models of diabetes, epidemiologic investigations of diabetes and its long-term complications, and interventional studies examining intensive treatment of diabetes and its effect on the development and progression of complications.
Data Synthesis: Diabetic retinopathy, nephropathy, and neuropathy occur in all clinical forms of diabetes mellitus, regardless of the cause of the diabetes. Hyperglycemia appears to be the major variable shared among these different clinical forms; and epidemiologic data, studies in animal models of diabetes, and the results of recent interventional studies such as the Diabetes Control and Complications Trial, all support an important and perhaps dominant role of hyperglycemia in the pathogenesis of complications. However, the diverse complications may not share the same pathogenesis. Different pathogenetic mechanisms may operate in different types of diabetic complications or at different stages of specific complications, or both.
Conclusions: The level of chronic glycemia is the best established concomitant factor associated with diabetic complications. The mechanism by which hyperglycemia might cause complications remains unknown, and evidence for a uniform pathogenetic mechanism is far from established.
Diabetes mellitus is a metabolic disorder characterized by hyperglycemia and alterations in fat and protein metabolism and by the occurrence of a specific set of long-term microvascular and neurologic complications. The clinical features and hyperglycemia of insulin-dependent diabetes mellitus (IDDM) are sufficiently characteristic so as not to require the occurrence of long-term complications to define it as a nosologic entity. However, the hyperglycemia of non–insulin-dependent diabetes mellitus (NIDDM) is part of a continuum. The glucose cutoffs that define NIDDM were selected on the basis of their association with diabetic complications, especially retinopathy [1]. Patients with modest levels of glucose intolerance not associated with long-term diabetes-specific complications are classified as having impaired glucose tolerance. On the other hand, patients with the more marked hyperglycemia associated with retinopathy are categorized as having diabetes. Thus, the currently accepted definition of diabetes mellitus tacitly recognizes the link between hyperglycemia and diabetic complications.
Despite the close phenomenologic linkage of hyperglycemia with complications, which has given rise to the “glucose hypothesis,” the causal mechanism of diabetic complications is not established. Is hyperglycemia the cause of diabetic complications? If so, what is the mechanism of its effect? In addition, do all long-term complications share similar pathogenetic mechanisms? This article reviews the animal model, epidemiologic, and clinical data that support the glucose hypothesis; the observations that support a common cause of diabetic complications and those that suggest diverse causes; and the potential pathogenetic mechanisms by which hyperglycemia may mediate the development or progression, or both, of long-term complications.
The Glucose Hypothesis
The potential connection between the level of glycemia and the development of long-term complications was suggested many years ago and supported by some but not all early epidemiologic studies [2, 3]. The glucose hypothesis that evolved proposed that hyperglycemia was either directly or indirectly related to the development or progression, or both, of diabetes-specific complications, including retinopathy, nephropathy, and neuropathy. The consensus was that hyperglycemia was a necessary but insufficient element in the pathogenesis of the complications and that interventions decreasing glycemic levels would ameliorate complications. In the absence of reliable measures of long-term glycemia or complications or of animal models of diabetes, the glucose hypothesis could not be tested. The advent of glycosylated protein assays [4] and quantitative, precise measures of long-term complications led to more reliable epidemiologic studies and further support of the relation between glucose levels and retinopathy in both NIDDM [5, 6] and IDDM [7]. It has been more difficult to establish an association between nephropathy and neuropathy and objective, reliably measured levels of chronic glycemia [8]. Finally, the association between glycemic levels and the nonspecific macrovascular complications that often accompany diabetes has been particularly difficult to establish, not an unexpected finding considering the multifactorial pathogenesis of macrovascular disease [9]. A recent study has shown an association between prevalent cardiovascular disease and glycosylated hemoglobin level across the entire spectrum of HbA1c levels [10]. In addition to the human epidemiologic data, interventional studies in animal models of diabetes have shown that more stringent measures to obtain near-normal glycemia invariably result in improved outcome [11, 12]. The combined epidemiologic and animal model studies, spanning a diverse range of diabetes and its complications, vigorously support the glucose hypothesis.
The recent demonstration of a beneficial effect of intensive therapy designed to lower glycemia on long-term complications in IDDM [13, 14] has boosted confidence in the glucose hypothesis, providing a final leg to the fulfillment of a variation of Koch's postulates for chronic diseases. Moreover, the relatively uniform salutary effects of intensive therapy on retinopathy, nephropathy, and neuropathy suggest that hyperglycemia may play a similar role in the pathogenesis of all three major diabetic complications. In the Diabetes Control and Complications Trial, intensive therapy reduced the development and progression of all complications by approximately 50% [14].
Shared and Unshared Characteristics of Long-Term Complications
If hyperglycemia causes long-term complications, does it cause all of the complications? In addition, are all phases of complications—for example, initiation of nonproliferative versus progression to proliferative retinopathy—related to hyperglycemia? The most obvious shared features of retinopathy, nephropathy, and neuropathy include their dependence on duration of disease [2]; the occurrence of complications in tissues in which glucose or its metabolites accumulate based on their plasma concentration and relatively independent of insulin [15-17]; and the involvement of relatively small caliber vessels with abnormal basement membranes (the retinal capillaries, vascular elements of the glomeruli, and vasa nervorum). Moreover, the abnormalities in retinal vessels, including early loss of the supportive pericytes, and in basement membrane [18] may be homologous to the mesangial accumulation and basement membrane thickening of the glomerulus [19, 20]. Finally, the association between glucose levels and the development of complications has been established for the complications, and, as noted above, the development and progression of all three diabetes-specific complications decrease with treatments that lower glucose levels.
Despite the shared elements, many features are not shared. For example, although the retinal and glomerular lesions are in the microvasculature, the most common neuropathic complication—peripheral somatosensory neuropathy—is more likely to be secondary to a metabolic derangement in neurons rather than a focal vascular problem [21]. Other forms of neuropathy (for example, the focal mononeuropathies) are probably secondary to obstruction of the vasa nervorum with ischemia [22, 23]. Although most patients with IDDM develop retinopathy over time, only 30% to 50% of patients with IDDM and no more than 15% of patients with NIDDM develop nephropathy to any degree [24, 25]. Similarly, many experimental animal models of diabetes develop retinopathy, but few develop renal disease. The development and progression of some complications, including nephropathy [26, 27] and perhaps retinopathy [28], are clearly influenced by other variables such as blood pressure, whereas neuropathy is not.
Finally, different stages of complications may have different pathogenetic mechanisms. The more advanced stage of retinopathy is characterized by the development of new vessels, so-called proliferative retinopathy, which appears to develop in the setting of widespread vascular obstruction or dropout with resultant ischemia [29]. The proliferative changes, but not the earlier background retinopathy, are associated with angiogenic factors that probably have no direct relation to glycemia [30].
Potential Glucose-Related Mechanisms of Long-Term Complications
Assuming that glucose levels are directly related to the development of some if not all diabetes-specific complications, what are the mechanisms? The two putative pathogenetic mechanisms that have received the most attention are the aldose reductase-sorbitol and the glycation models. Each of these, but especially the former, has been advanced through the availability of pharmacologic probes. Aldose reductase inhibitors have been available for more than two decades and have been tested in animal models and in human diabetes [31]. Glycation inhibitors such as aminoguanidine have been tested in animal models, and studies in humans are under way [32].
The role of aldose reductase in catalyzing the intracellular accumulation of sugar alcohols such as sorbitol in tissues that develop complications has been recognized for almost 30 years [15, 33]. The accumulation of sorbitol in the lens of the eye, pericytes, and neural tissue has many effects, including changes in osmotic gradients, decreased myoinositol concentrations, and perturbation of energy metabolism [21, 33, 34]. Despite the early enthusiasm for the sorbitol hypothesis as an explanation for diabetic complications, especially neuropathy, and the promising results of animal studies using aldose reductase inhibitors [35, 36], the results in studies of human diabetes have not been convincing [37-39]. Whether the failure to improve neuropathy with various relatively potent aldose reductase inhibitors is secondary to inadequate treatment periods, poor penetration of drugs into clinically relevant tissues, an incorrect hypothesis, or to other factors is not known.
Glycation was first recognized by identifying minor hemoglobin fractions that could be separated from adult hemoglobin on the basis of charge. These minor hemoglobin fractions were increased in diabetic patients and eventually identified as products of nonenzymatic glycation [40, 41]. The major clinical utility of glycation products has been in the monitoring of chronic glycemia; glycated hemoglobin measurements reflect the average, integrated level of glucose control during the previous 2 to 3 months and are widely used in clinical practice [4]. Glycation occurs in virtually all proteins exposed to glucose, and the potential for glycation to cause complications has been espoused for more than a decade. The most convincing manifestation of glycation in tissues has been with regard to diabetic “cheiropathy,” including the development of thickened, stiff periarticular structures resulting in the “prayer sign” (inability to extend fully the fingers and appose them) [42]. The same mechanism probably contributes to the increased frequency of adhesive capsulitis and DuPuytren contractures in diabetes. The mechanism for these changes involves the nonenzymatic attachment of glucose to collagen with subsequent rearrangements, condensations, and crosslinking that result in “advanced glycation end products.” The physical characteristics of collagen change with glycation, providing a convincing explanation for the observed clinical changes. Whether similar changes occur in collagen (type IV) in basement membrane or in other proteins and account for the abnormalities observed is being vigorously investigated. The protean manifestations of glycation that have been shown to date may or may not explain some or all of the complications [32, 43]. In animal studies, relatively specific blockers of the formation of advanced glycation end products, such as aminoguanidine have resulted in improvement of some diabetic complications [44].
Other Potential Causes of Long-Term Complications
Although a convincing argument can be made for a direct role of hyperglycemia in the pathogenesis of long-term complications, other potential causes should at least be mentioned. It is possible, for example, that some other metabolic manifestation of diabetes, for example, altered fatty-acid or protein metabolism or relative or absolute insulin deficiency, is the cause of the complications. The similar occurrence of complications in different clinical forms of diabetes, which do not always share metabolic profiles other than hyperglycemia, and the absence of complications in disorders other than diabetes tend to support a primary role of hyperglycemia. Even if hyperglycemia is a central factor in the pathogenesis of long-term complications, it may act indirectly. Hemodynamic and hemorrheologic explanations for the development of microvascular and neurologic complications have been proposed [45, 46]. Vasodilatation and altered vascular reactivity, hyperglycemic pseudohypoxia on the basis of altered NAD+:NADH ratios, changes in nitric oxide metabolism, and abnormalities in other vasoactive substances such as prostaglandins and endothelial-derived relaxation factors have been observed in diabetes and may play some role in the pathogenesis of the microvascular complications [47-51]. Finally, as noted previously, growth factors such as vascular angiogenic growth factor have been observed in the setting of proliferative retinopathy and may mediate its development or progression [30].
Conclusions
The recently reported results of the Diabetes Control and Complications Trial and other studies [13, 14] provide the final and most convincing evidence that hyperglycemia is a critical, albeit not the only, factor involved in the pathogenesis of long-term diabetic complications. Other environmental and genetic factors almost certainly play an additional role in the development of complications. Although the contributing factors to the development of each complication may not be the same, glycemia seems to play a role in all. Identification of the specific pathogenic mechanisms associated with hyperglycemia is under way. To the extent that these distal mechanisms can be inhibited by pharmacologic interventions, we may be able to decrease the development and progression of complications independent of the level of glycemic control.
- Copyright ©2004 by the American College of Physicians
