Diabetes and Endothelial Dysfunction: A Clinical Perspective

Transcription

1 X/01/$03.00/0 Endocrine Reviews 22(1): Copyright 2001 by The Endocrine Society Printed in U.S.A. Diabetes and Endothelial Dysfunction: A Clinical Perspective JORGE CALLES-ESCANDON AND MARILYN CIPOLLA Departments of Internal Medicine (J.C.-E.) and Obstetrics and Gynecology (M.C.), College of Medicine, University of Vermont, Burlington, Vermont ABSTRACT The main etiology for mortality and a great percent of morbidity in patients with diabetes mellitus is atherosclerosis. A hypothesis for the initial lesion of atherosclerosis is endothelial dysfunction, defined pragmatically as changes in the concentration of the chemical messengers produced by the endothelial cell and/or by blunting of the nitric oxide-dependent vasodilatory response to acetylcholine or hyperemia. Endothelial dysfunction has been documented in patients with diabetes and in individuals with insulin resistance or at high risk for developing type 2 diabetes. Factors associated with endothelial dysfunction in diabetes include activation of protein kinase C, overexpression of growth factors and/or cytokines, and oxidative stress. Several therapeutic interventions have been tested in clinical trials aimed at improving endothelial function in patients with diabetes. Insulin sensitizers may have a beneficial effect in the short term, but the virtual absence of trials with cardiovascular end-points preclude any definitive conclusion. Two trials offer optimism that treatment with ACE inhibitors may have a positive impact on the progression of atherosclerosis. Although widely used, the effect of hypolipidemic agents on endothelial function in diabetes is not clear. The role of antioxidant therapy is controversial. No data have been published regarding the effects of hormonal replacement therapy on endothelial dysfunction in postmenopausal women with type 2 diabetes. (Endocrine Reviews 22: 36 52, 2001) I. Introduction II. Endothelial Cell Dysfunction A. Normal endothelial cell function B. Endothelial dysfunction III. Endothelial Dysfunction and Diabetes A. Insulin effects on the vasculature B. Endothelial dysfunction in type 1 diabetes C. Endothelial dysfunction in type 2 diabetes IV. Reversal of Endothelial Dysfunction: Lessons from Human Clinical Trials A. Insulin sensitizers B. ACE inhibitors C. Hypolipidemic therapy D. Arginine supplementation and antioxidants V. Summary and Conclusions I. Introduction THE MAIN etiology for death and for a great percent of morbidity in patients with diabetes (type 1 or type 2) is vascular disease (1, 2). Type 2 diabetes affects small (microangiopathy) or large vessels (macroangiopathy). Microvascular disease is the hallmark of retinopathy, neuropathy, and nephropathy, whereas macroangiopathy in diabetes is manifested by accelerated atherosclerosis, which affects vital organs (heart and brain). Atherosclerosis in patients with type 2 diabetes is multifactorial and includes a very complex interaction including hyperglycemia, hyperlipidemia, oxidative stress, accelerated aging, hyperinsulinemia and/or hyperproinsulinemia, and alterations in coagulation and fibrinolysis (3). A current hypothesis for the initial lesion of atherosclerosis involves changes in endothelial cell (EC) function (4). Endothelial dysfunction has been documented in patients with type 2 diabetes (5 10) and also in individuals with type 1 diabetes especially when there is clinically manifest microalbuminuria (11 17). Recent data demonstrate that endothelial dysfunction may also be present in individuals who have insulin resistance (e.g., obese patients) (18) or who are at high risk for developing type 2 diabetes [i.e., impaired glucose tolerance (IGT), metabolic syndrome] (19) and in patients with former gestational diabetes (20). This review will present a synopsis of our current understanding of endothelial dysfunction in patients with diabetes; special emphasis will be directed to patients with type 2 diabetes. Address reprint requests to: Jorge Calles-Escandon, MD, Associate Professor of Medicine, University of Vermont, College of Medicine, Given Building C-344A, Burlington, Vermont, jcallese@ zoo.uvm.edu II. Endothelial Cell (EC) Dysfunction A. Normal EC function The EC lines the internal lumen of all the vasculature and serves as an interface between circulating blood and vascular smooth muscle cells (VSMC). In addition to serving as a physical barrier between the blood and tissues, the EC facilitates a complex array of functions in intimate interaction with the VSMC, as well as cells within the blood compartment as depicted in Fig. 1. The EC is no longer considered a simple barrier. The last two decades of research have established unambiguously that the EC has a critical role in overall homeostasis whose functions are integrated by a complicated system of chemical mediators as indicated in Table 1 (21 23). The system exerts effects on both the surrounding VSMC and the cells in the blood that lead to one or more of the following alterations: 1) vasodilation or vasoconstriction to regulate organ blood, 36

2 February, 2001 DIABETES AND ENDOTHELIAL DYSFUNCTION 37 2) growth and/or changes in the phenotypic characteristics of VSMC, 3) proinflammatory or antiinflammatory changes, and 4) maintenance of fluidity of blood and avoidance of bleeding (22, 24 28). Thus, as summarized in Table 1 the cellular effects of the EC maintain a balance of opposing physiological and molecular effects. It is conceptualized currently as maintaining a balance of opposing forces with the end result of maintaining a proper blood supply to tissues and regulating inflammation and coagulation. 1. Nitric oxide: a key mediator of the EC. During the last decade, a multitude of experimental arguments have led to the concept that nitric oxide (NO) is not only involved in the control of vasomotor tone but also in vascular homeostasis and neuronal and immunological functions. Endogenous NO is produced through the conversion of the amino acid, l-arginine to l-citrulline by the enzyme, NO-synthase (NOS) from which several isoforms have recently been isolated, purified, and cloned. NOS-type I (isolated from brain) and type III FIG. 1. Microanatomy of a small vessel. The endothelium (EC) confers a lining to all the vasculature. It interacts directly with the VSMC of the vessels and with the blood cells as well as with the plasma components. Via several chemical mediators, the EC is in fact a regulator of the VSMC and plays a key role in maintaining hemostasis and blood fluidity. IEL, Internal elastic lamina. (isolated from ECs) are termed constitutive-nos and produce picomolar levels of NO from which only a small fraction elicits physiological responses. These isoforms are regulated by Ca( 2 )-calmodulin with NADPH, flavin adenine dinucleotide/mononucleotide (FAD/FMN), and tetrahydrobiopterin (HB 4 ) as cofactors and reveal a high degree of homology with the amino acid sequence of cytochrome P450 reductase within the C-terminal domain. Functionally, neuronal-nos type I is important in neurotransmission, the central control of vascular homeostasis, and possibly learning and memory. In the peripheral nervous system, NOS appears to be linked to nonadrenergic noncholinergic (NANC) neuronal pathways. Endothelial-NOS (enos) type III is essential for the control of vascular tone in response to several stimuli, including mechanical (e.g., shear stress) and receptor dependent (e.g., acetylcholine) and receptor independent (e.g., calcium ionophore) (29). NO produced by NOS type III in the endothelium diffuses to the vascular smooth muscle (VSM) where it activates the enzyme guanylate cyclase. The concomitant increase in cyclic GMP then induces relaxation of the VSM. Thus, the net effect of an increase in NO is vasodilation (Fig. 2). NO production by the NOS type III is also basally produced and in some circulations (e.g., cerebral), basal NO production is substantial. The continual vasodilation produced by basal NO production has a role in regulation of blood pressure as well. Many studies have demonstrated that systemic infusion of NOS inhibitors elevate blood pressure. NOS type III also contributes to the prevention of abnormal platelet aggregation (30 36). NOS types II and IV (isolated from macrophages) are Ca( 2 )-calmodulin independent and are termed inducible-nos since their activation is only promoted under pathophysiological situations in which macrophages exert cytotoxic effects in response to cytokines (e.g., sepsis). 2. Measurement of NO-mediated vasodilation. Typically, NOdependent vasodilatation is probed by the vasodilatory response to infusion of a compound (e.g., acetylcholine or TABLE 1. Endothelial cell functions Functional targets of the endothelial cell Specific cellular or physiological action (mediators are listed in italics) Lumen Vasoconstriction Vasodilation Endothelin NO Angiotensin II Bradykinin ET-1 Hyperpolarizing factor Thromboxane A 2 PGH 2 Growth Stimulation Inhibition Platelet growth-derived factor NO Fibroblast Growth Factor PGI2 IGF-1 TGF Endothelin Angiotensin II Inflammation Proinflammatory Antiinflammatory Adhesion molecules ELAM, VCAM, ICAM Hemostasis Prothrombotic Antithrombotic PAI-1 Prostacyclin TPA TPA, Tissue plasminogen activator.

3 38 CALLES-ESCANDON AND CIPOLLA Vol. 22, No. 1 are regulated by the activity of ACE. ACE breaks down bradykinin into inactive peptides (52, 55). Hence, high ACE concentrations will antagonize NO activity not only by increasing ANG-II generation but also, and possibly most importantly, by decreasing concentrations of bradykinin. A model of regulation of vascular tone (and lumen regulation) in which ACE plays a key role has emerged in recent years (Fig. 3). This model predicts that high ACE activity will result in vasoconstriction because of a decrease in NO generation and increased generation of ANG-II. This results in contraction of VSMCs and decreased lumen diameter. Moreover, sustained activity of this enzyme will presumably be associated with an increase in the growth, proliferation, and differentiation of the VSMC as well as a decrease in the antiproliferative action of NO coupled with a decrease in local fibrinolysis and an increase in platelet aggregation. FIG. 2. Endothelial cell as a regulator of the smooth muscle cells. The EC produces NO, gas that diffuses into the VSMC and activates the enzyme guanylate cyclase which produces cyclic GMP. The latter induces muscle relaxation, which is physiologically translated into vasodilation. The immediate precursor of NO is the amino acid arginine and the key enzyme in its production is NOS. methacholine), which increases the synthesis and release of NO via a receptor-mediated response that is calcium dependent (9, 29, 37 39) or in response to reactive hyperemia which stimulates shear stress-induced NO production. This response is compared with the vasodilation evoked by specific chemical compounds that directly act on VSMC (e.g., sodium nitroprusside). The difference in vasodilation observed between the two conditions can be considered endotheliumdependent vasodilation. In addition, specific inhibitors of NOS [e.g., nitro-l-arginine (l-nna)] have been used to further probe EC function in vivo (13, 33, 40 42). 3. Angiotensin II (ANG-II). The EC also produces mediators that induce vasoconstriction, including endothelin (43 45), prostaglandins (46, 47), and ANG-II (48 51) and regulates vascular tone by maintaining a balance between vasodilation (NO production) and vasoconstriction (e.g., A-II generation). ANG-II is produced in local tissues by the EC (52, 53). The key enzyme that regulates the local generation of ANG-II is angiotensin converting enzyme (ACE). This proteolytic enzyme is synthesized by the EC, expressed in the surface of the EC, and exerts activity upon the blood-borne angiotensin I. Angiotensin I is produced by cleavage of a precursor macromolecule (angiotensinogen) effected by plasma renin, another proteolytic enzyme produced in the kidney. ANG-II binds to and regulates VSMC tone via specific angiotensin (ANG) receptors. Depending upon the specific receptor activated, ANG-II can exert regulatory effects upon several VSMC functional activities including contraction (i.e., vasoconstriction) and growth, proliferation, and differentiation. Overall, the actions of ANG-II oppose those of NO. As reviewed above, NO is a product of the enzyme NOS, which responds to specific activators and inhibitors. NOS also is regulated by local concentrations of bradykinin (54). This peptide acts with b2 receptors on the EC cell surface membrane, increasing the generation of NO via NOS activation. Interestingly, the local concentrations of bradykinin 4. The EC as a regulator of hemostasis. Functions of the EC extend beyond those pertaining to vascular tone. The EC has a prominent role in maintaining blood fluidity and restoration of vessel wall integrity (when injured) to avoid bleeding. Broadly speaking, the systems that maintain hemostasis in the vasculature include: 1) the lumen of the vessel (vasoconstrictor and/or vasodilatory effects); 2) platelets; 3) coagulation; and 4) fibrinolysis. The EC plays a key role in the balance between the coagulation and fibrinolytic systems (Fig. 4). The coagulation cascade will not be detailed in this review. In essence, the coagulation explosion has the ultimate function of generating active thrombin (56). Thrombin is a proteolytic enzyme, and fibrinogen is its natural and most abundant substrate. Upon activation of thrombin, fibrinogen is transformed into fibrin with the release of fibrinopeptides A and B. Fibrin then undergoes polymerization and cross-linking, creating a stable clot. Thereafter, the clot is dissolved upon the action of another proteolytic enzyme, plasmin, which is the main effector of the fibrinolytic system. The transformation of the plasmin precursor, plasminogen, to plasmin results from specific activators. Physiologically (as well as pharmacologically) the most important activator of the conversion of plasminogen to plasmin is tissue plasminogen activator (t-pa). This peptide has a critical role in the dissolution of clots and maintenance of vessel lumen and has been used therapeutically in the treatment of events in which acute occlusion by thrombi is a precipitating event of life-threatening disease states (i.e., myocardial infarction, stroke, massive pulmonary embolism). Several positive and negative activators regulate t-pa activity. Physiologically, the most important regulator of t-pa is the peptide, plasminogen activator inhibitor (PAI) (57). There are four types of the PAI, of which type 1 (PAI-1) seems to play the most preeminent role. 5. The EC as a mediator of VSMC growth and inflammation. The EC also plays a key role in growth and differentiation of the VSMC through the release of either promoters of growth and/or inhibitors of growth and differentiation and, as such, has an impact on vascular remodeling (58). A large number of peptides have been proposed as the main messengers for growth signals [insulin-like growth factor 1 (IGF-1), PGF, basic fibroblast growth factor (bfgf), etc.]; however, strong

4 February, 2001 DIABETES AND ENDOTHELIAL DYSFUNCTION 39 FIG. 4. Local coagulation and fibrinolysis are regulated by the endothelial cell. The main inhibitor of the fibrinolytic system is PAI-1, which has been documented to be elevated in disease states with insulin resistance (e.g., obesity, diabetes). FIG. 3. The role of ACE in endothelial cell function. The EC membrane holds the ACE which, when overactive or overexpressed, produces a large amount of ANG-II. The latter acts directly on the VSMC by attaching to specific receptors located on the membrane of the VSMC. Many of the actions of ANG-II are antagonistic to those of NO as depicted on the figure. ACE activation also leads into faster catabolism of bradykinin. evidence suggests that promotion of VSM growth is mediated by local production of PGF and ANG-II (59, 60). Two key mediators are proposed to be antagonists of the growthpromoting actions of ANG-II: NO and prostacyclin (PGI 2 ). The EC is also involved in the production of specific molecules that have a regulatory role in inflammation (61). The most important are LAM, intracellular adhesion molecule (ICAM), and vascular cell adhesion molecule (VCAM). These molecules are denominated adhesion molecules and function to attract and anchor those cells involved in the inflammatory reaction. Very recently it has been demonstrated that the atherosclerotic process is associated with an increased blood level of inflammation (i.e., acute phase proteins) markers (62). B. Endothelial dysfunction Since the actions of the EC are multiple and involve several systems, alterations in EC function may affect one or more of these systems, either simultaneously or at distinct time periods. Thus, no single definition of EC dysfunction covers the whole array of possible disruption in normal function. In consequence, endothelial dysfunction has been defined pragmatically. It basically involves either an increase (or a decrease) in any of the EC-related chemical messenger and/or by alteration in any of the functional changes listed earlier (Table 1). Some examples of EC dysfunction include an increased permeation of macromolecules (22, 63, 64), increased or decreased production of vasoactive factors producing abnormal vasoconstriction/vasodilation (22, 30, 43, 65), and increased prothrombotic and/or procoagulant activity (66). However, the most commonly accepted EC dysfunction alteration pertains to abnormalities in the regulation of the lumen of vessels. In this context, EC dysfunction has been defined by blunting of the vasodilatory response to acetylcholine or hyperemia, both of which are known to produce NO-dependent vasodilation. In some specific circumstances, endothelial dysfunction has been defined by a paradoxical vasoconstrictive response to acetylcholine or similar pharmacological agents (i.e., metacholine). At the heart of the definition of EC dysfunction is the measurement of EC function. This review will focus on the methods that are available for measurement of EC function in vivo in humans. The methodology available for in vitro measurement is beyond the scope of this manuscript. 1. How do we measure in vivo EC function in humans? As explained before, the range of EC function(s) may differ according to the type of vessel(s) affected as well as the tissue or organ perfused. EC action may affect one or several functions, either simultaneously or in a temporal sequence and thus cannot be considered a single, discrete, and uniquely defined entity. In consequence, there is no gold standard for measurement of EC dysfunction. In general, EC function is measured experimentally by 1) methods that assess the functional consequences of EC activity, alone or complemented by 2) measurement of the concentration of those chemical mediators that mediate EC function. The approach to measuring EC function in vivo stems from the fact that the most widely recognized function of the EC pertains to its effects on vascular tone. The EC produces chemical mediators that may induce contraction or relaxation of the adjacent VSM (i.e., vasoconstriction or vasodilation). The vasodilatory molecules include several peptides, hormones as well as NO (30 33, 52, 67) and an unidentified endothelium-dependent hyperpolarizing factor (EDHF) (68). From the viewpoint of lumen regulation, NO occupies a prominent role and is considered one of the most important molecules that the EC produces to regulate vascular tone. From the clinical perspective, EC function has been estimated by measuring invasively (55, 69 71) (i.e., coronary catheterization) or noninvasively (72, 73) (i.e., ultrasound) changes in blood flow. Thus, physiologically, in vivo, EC function is defined in humans as an increase in blood flow or in the diameter of a vessel in response to agents that increase the concentration of NO. This can be correlated with a decrease in blood flow evoked by a decrease in the local concentrations of NO after production of the latter is blocked by L-NNA. The response must be attributable to the EC and not to dysfunction of the SMC. The latter is probed by measuring the blood flow response (or the diameter of the vessel) in response to chemical agents that act directly on the SMC

5 40 CALLES-ESCANDON AND CIPOLLA Vol. 22, No. 1 (e.g., nitroprusside). Blood flow and/or vessel diameter can be measured in human beings using a wide array of techniques. A discussion of pros and cons of these techniques is beyond the scope of this review, but a summary of those currently available is presented in Table 2. Endothelial function can be further evaluated by using a physiological measurement of blood flow coupled with blood level determination of selected compounds thought to reflect EC function. Such compounds include endothelin (43, 45, 74 76), Von-Willebrand factor (77 80), thrombomodulin (81, 82), selectin (83), adhesion molecules (83, 84) (VCAM, ICAM), and t-pa as well as its inhibitor, PAI-1 (85, 86). Caution in interpreting the results of these studies must be exerted since high plasma levels in endothelial-derived compounds may reflect increased synthesis, decreased clearance, or a combination of both. Moreover, the precise cellular origin of some of these compounds is still not defined. For example, PAI-1 may be produced not only by the EC but also by VSMC, hepatocytes, and adipose cells. III. Endothelial Dysfunction and Diabetes Diminished capacity of NOS to generate NO has been demonstrated experimentally when ECs are exposed either in vitro or in vivo to a diabetic environment (75, 87 94). The EC is then a target of the diabetic milieu and endothelial dysfunction is thought to play an important role in the vasculopathy of this disease state. A large body of evidence in humans indicates that endothelial dysfunction is closely associated to microangiopathy and atherosclerosis in both types 1 and 2 diabetes mellitus (11). This association is particularly true in those patients with type 1 diabetes who have either early (microalbuminuria) or late (macroalbuminuria) nephropathy. In these patients, a great variety of markers indicate endothelial dysfunction: poor EC-dependent vasodilation, increased blood levels of von Willebrand factor (vwf), thrombomodulin, selectin, PAI-1, type IV collagen, and t-pa (11 13, ). Once established, EC dysfunction can, in turn, induce alterations in vessels that worsen vasculopathy and progress disease. Of note is that arteries and arterioles are not considered commonly as target tissues/ organs of insulin action. However, in recent years a body of evidence has accumulated that supports the hypothesis that vessels are insulin responsive. A. Insulin effects on the vasculature Several years ago Jialal and colleagues (101) described the presence of receptors for insulin, IGF-I, and IGF-II on cells from micro- and macrovessels with binding characteristics that are similar to those in other cells. Interestingly, these investigators suggested that the finding of large numbers of IGF-I and IGF-II receptors on ECs supported a physiological role for these growth factors and proposed the hypothesis that they may be involved in vascular complications associated with diabetes. In a previous publication, this same group (102) had demonstrated that the concentrations of receptors for aortic smooth muscle cells were 10-fold fewer than other cell types. Moreover, insulin stimulated glucose incorporation into glycogen and stimulated DNA replication in retinal ECs and pericytes and aortic smooth muscle cells, but had no effect on aortic endothelium. These data suggested that a differential response to insulin may exist between endothelium of micro- and macrovasculature and that retinal capillary endothelium and retinal pericytes are both very insulin-sensitive tissues. Insulin deficiency and chronic hyperglycemia can increase the concentration of the membrane-bound protein kinase C (PKC) and total diacylglycerol (DAG) levels. Insulin administration and consequently euglycemia can prevent the increase in PKC activities and DAG levels (103). Significant information regarding insulin signaling in the vascular tissues has emerged recently. Insulin signaling on the phosphatidylinositol 3-kinase (PI 3-kinase) and mitogen-activated protein (MAP) kinase pathways were compared in vascular tissues of lean and obese Zucker rats (104). As anticipated, insulin-stimulated tyrosine phosphorylation of insulin receptor -subunits in microvessels of obese rats was significantly decreased compared with lean rats as well as insulin-induced tyrosine phosphorylation of insulin receptor substrates 1 and 2 (IRS-1 and IRS-2). TABLE 2. Methods for measurement of blood flow in humans Catheter Fick principle (chemical dilution, thermodilution) Gold standard, used extensively, validated Ultrasound Doppler principle Noninvasive, simple, widely available, accurate measurement of vessel diameter Invasive, costly, available only in small number of centers Requires expertise and poses risks Wide variability, observer dependent, blood flow estimated, not measured directly Applicable only to macrocirculation PET scan Tracer distribution, measures microcirculation Accurate, precise, tissue specific Model-dependent, expensive, costly, not widely available Laser Doppler flowmetry Measures microcirculation Noninvasive, simple, cheap Needs further validation Plethysmography Changes in electrical impedance using strain gauge mesh due to changes in volume of the forearm or leg PET, Positron emission tomography. Simple, noninvasive, cheap, easy to implement Very dependent upon local expertise, high variability, poor correlation with gold standard

6 February, 2001 DIABETES AND ENDOTHELIAL DYSFUNCTION 41 Moreover, the association of the p85 subunit to the IRS proteins and the IRS-associated PI 3-kinase activities stimulated by insulin in the aorta of obese rats were significantly decreased compared with the lean rats as was the serine phosphorylation of Akt. These in vitro results were comparable to in vivo studies using the euglycemic clamp technique. In marked contrast, these investigators found that insulin-stimulated tyrosine phosphorylation of MAP kinase (ERK-1/2) was equal in isolated microvessels of lean and obese rats, although basal tyrosine phosphorylation of ERK-1/2 was higher in the obese rats. Thus, insulin has a direct effect on vascular tissues using signaling similar to other tissues. A selective resistance to PI 3-kinase (but not to MAP kinase pathway) in the vascular tissues of obese Zucker rats was found which the investigators hypothesize may be of importance in the vascular disease of insulin resistance states. More recently, using physiological concentrations of insulin, King s group (105) found that the hormone increased the levels of enos mrna, protein, and its activity by 2-fold after 2 8 h of incubation of ECs. Interestingly, this effect of insulin was seen in microvessels isolated from Zucker lean insulinsensitive rats but not from insulin-resistant Zucker fatty rats. PKC activators inhibited both the activation by insulin of PI-3 kinase and enos mrna levels. These investigators concluded that insulin may have not only an acute vasodilatory effect but also chronically modulate vascular tone. Moreover, they postulated that activation of PKC in the vascular tissues may be a primary event that leads to endothelial dysfunction in insulin-resistant states. B. Endothelial dysfunction in type 1 diabetes It is not clear from the literature whether EC dysfunction is the consequence of the diabetic milieu in type 1 diabetes or a marker of vascular damage. In a experimental model of type 1 diabetes, endothelial function was evaluated directly in rats (106) with chronically implanted flow probes. The responses to acetylcholine and sodium nitroprusside were not altered significantly; moreover, neither endotheliummediated vasodilation nor responsiveness to NO was impaired, and hyperglycemia did not directly or significantly impair endothelium-mediated relaxation in this model of insulin-dependent diabetes mellitus. In a recent and well designed study (13), endothelium-dependent and endothelium-independent vasodilatation, endothelium-dependent hemostatic factors, and vasoconstrictor responses were determined in type 1 diabetic patients during euglycemia with and without microvascular complications. Forearm endothelium-dependent and endothelium-independent vasodilatation and adrenergic responsiveness were unaltered in type 1 diabetic patients with and without microvascular complications. Relative to healthy control subjects, endothelium-dependent vasodilatation was depressed during a repeated ACh challenge (with l-arginine coinfusion) in the diabetic patients without complications or with microalbuminuria. In contrast, this vasodilatation was enhanced in the patients with retinopathy. Elevation of the endothelium-derived tissue factor plasminogen inhibitor was the most consistent marker of endothelial damage of all the endothelial markers measured in these group of patients but showed no correlation with the presence or absence of microvascular complications. Clarkson et al. (107) compared 80 young adults with insulin-dependent diabetes with 80 matched nondiabetic control subjects. Flow-mediated dilation was significantly impaired in diabetic subjects ( % vs % in control subjects, P 0.001). Using multivariate analysis, flow-mediated dilation was inversely related to duration of diabetes (r 0.26, P 0.05) and low density lipoprotein cholesterol (LDL-C) levels (r 0.38, P 0.005). Thus, in this study, EC dysfunction was found as an early manifestation of vascular disease but late in the course of type 1 diabetes. One cross-sectional study assessed endothelial adhesion molecules in patients with type 1 diabetes (n 70), with varying degrees of metabolic control and status of diabetic late complications, and compared them to those in agematched healthy subjects (n 70) (108). Concentrations of cicam-1 and cvcam-1 were elevated in insulin-dependent diabetes mellitus (IDDM) whereas celam-1 did not differ between the groups. The levels of cvcam-1 were more markedly elevated in IDDM patients with diabetic retinopathy than in those patients with micro- or macroalbuminuria, whereas no difference in cicam-1 and celam-1 was apparent regarding the clinical status of diabetic microangiopathy. No correlations were found between hemoglobin A1c and cams. This relation to the duration of diabetes was not found by Johnstone et al. (14), who measured vascular reactivity in the forearm resistance vessels of 15 patients with IDDM and 16 age-matched normal subjects. These investigators stated that no patient had hypertension or dyslipidemia. The vasodilative response to methacholine was less in diabetic than in normal subjects, but forearm blood flow (FBF) responses to nitroprusside and verapamil and the forearm vasoconstrictor responses to phenylephrine were similar in diabetic and healthy subjects. In diabetic subjects, endothelium-dependent vasodilation correlated inversely with serum insulin concentration but not with glucose concentration, glycosylated hemoglobin, or duration of diabetes. In another study (108a), the plasma fibronectin 30-kDa domain was measured in 44 type 1 diabetic patients and in 20 healthy subjects. A significantly raised mean concentration of a free N-terminal fibronectin 30-kDa domain was found in plasma of diabetic patients with proliferative retinopathy as compared with healthy persons, and a positive correlation was observed between free N-terminal fibronectin and vwf in plasma of all examined subjects (r 0.62). A similar correlation was present between fibronectin and the degree of albuminuria (r 0.56). However, no relationship was found between fibronectin and the degree of control of diabetes. Thus, in these two cross-sectional studies, type 1 diabetes has been associated more with the presence of microvascular disease than with the diabetic milieu. It is the general consensus that the occurrence of EC dysfunction in type 1 diabetes signifies a very high risk of microand macroangiopathy and, although the diabetic state predisposes to EC dysfunction in this disease, is not sufficient to cause it. More likely, other agents (genes, environment) are likely to play a role in determining those patients that will develop aggressive angiopathy and hence EC dysfunction. Irrespective of whether EC dysfunction is a cause or a con-

7 42 CALLES-ESCANDON AND CIPOLLA Vol. 22, No. 1 sequence of vascular injury in type 1 diabetes, therapeutic efforts aimed at restoring EC to normal will more likely have an affect on the natural history of vasculopathy in type 1 diabetes. C. Endothelial dysfunction in type 2 diabetes The role of endothelial dysfunction in type 2 diabetes is more complicated than that for type 1. The effects of aging, hyperlipidemia, hypertension, and other factors add to the complexity of the problem. In contrast to patients with type 1 diabetes, endothelial dysfunction can also occur in patients with type 2 diabetes even when the patients have normal urinary albumin excretion. In fact, markers of endothelial dysfunction are often elevated years before any evidence of microangiopathy becomes evident (11, ). A major pathophysiological alteration of type 2 diabetes is insulin resistance. As a result, a great research effort has been focused on defining the possible contribution of insulin resistance to endothelial dysfunction. 1. Endothelial dysfunction and insulin resistance. There is a growing body of evidence accumulating to demonstrate the coexistence of insulin resistance and endothelial dysfunction. Insulin-induced vasodilation, which is partially mediated by NO (116) release, is impaired in obese individuals who do not have type 2 diabetes but who display insulin resistance (18). Moreover, the obese state, a model of human insulin resistance, is associated with high levels of endothelin in plasma (117). PAI-1 concentrations in blood also are high in patients with otherwise uncomplicated obesity; a drastic decrease in PAI-1 levels has been noted by our laboratory in response to moderate weight loss (118). Recently, it has been demonstrated that women with previous gestational diabetes have evidence of endothelial dysfunction (20, 119, 120). In women with polycystic ovary syndrome, high levels of PAI-1 have been found which improve with any therapeutic intervention that improves insulin sensitivity ( ). Further, evidence suggests that endothelial dysfunction occurs in a concomitant manner with insulin resistance and antedates overt hyperglycemia in patients with type 2 diabetes. Steinberg et al. (125) recently demonstrated that elevated free fatty acid levels in blood (produced in normal individuals by simultaneous infusion of triglycerides and heparin) induced endothelial dysfunction. FFA are classically elevated in obese patients, patients with type 2 diabetes, and, in general, in those individuals who display features of the syndrome of insulin resistance. Thus, data support the hypothesis that the metabolic abnormalities of insulin resistance may lead to endothelial dysfunction. More recently, Caballero et al. (126) demonstrated early abnormalities in vascular reactivity and biochemical markers of EC activation in individuals at risk of developing type 2 diabetes. These investigators measured the increase in blood flow in the microcirculation (laser Doppler flowmetry) and in the macrocirculation (ultrasound) in four groups of individuals: 1) healthy normoglycemic subjects with no history of type 2 diabetes in a first-degree relative (controls); 2) healthy normoglycemic subjects with a history of type 2 diabetes in one or both parents (relatives); 3) subjects with IGT; and 4) patients with type 2 diabetes without vascular complications. Moreover, these investigators measured plasma concentrations of endothelin-1 (ET-1), vwf, soluble intercellular adhesion molecule (sicam), and soluble vascular cell adhesion molecule (svcam) were also measured as indicators of EC activation. The vasodilatory responses to acetylcholine were reduced in groups 2, 3, and 4 compared with controls. The plasma levels of ET-1 were significantly higher in these three groups. On stepwise multivariate analysis, age, sex, fasting plasma glucose, and body mass index (BMI) were the most important contributing factors to the variation of vascular reactivity. However, all clinical and biochemical measures explained only 32 37% of the variation in vascular reactivity. These results suggest that abnormalities in vascular reactivity and biochemical markers of EC activation are present early in individuals at risk of developing type 2 diabetes, even at a stage when normal glucose tolerance exists and that factors in addition to insulin resistance may be instrumental in the EC dysfunction of individuals at high risk of developing type 2 diabetes. The insulin resistance syndrome encompasses more than a subnormal response to insulin-mediated glucose disposal. Patients with this syndrome also frequently display elevated blood pressure, hyperlipidemia, and dysfibinolysis, even without any clinically demonstrable alteration in plasma glucose concentrations. Of note, endothelial dysfunction also has been demonstrated in patients with hypertension (85, ), which is one of the features of the syndrome of insulin resistance. It is tempting to speculate that loss of endothelial-dependent vasodilation and increased vasoconstrictors might be etiological factors of hypertension. Moreover, loss of activity and/or quantity of endothelium-bound protein lipase activity may contribute to hyperlipidemia, which is typical of the insulin resistance syndrome. A synergistic interaction and vicious cycle may exist in which endothelial dysfunction contributes to insulin resistance and vice versa. 2. Endothelial dysfunction, dysfibinolysis, and insulin resistance. Under normal conditions, the blood is constantly in a balance between a basal on-going activation of coagulation and compensatory fibrinolysis. A current hypothesis states that, in patients with endothelial dysfunction, PAI-1 levels are elevated, which in turn inhibits dissolution of fibrin deposits on the luminal side of the vessel wall. Several investigators have suggested that PAI-1 plays a major role in the generation and/or progression of atherosclerosis. Our laboratory and many others have found high levels of PAI-1 in disease states in which insulin resistance is a prominent pathophysiological feature (122, ). Examples of this are type 2 diabetes, upper body obesity, and polycystic ovary syndrome. These disease states also are associated with accelerated atherosclerosis, which supports the hypothesis that high levels of PAI-1 may play a role in initiation and/or progression of macrovascular disease. It has been proposed that the hyper (pro)-insulinemia of insulin resistance might be implicated as an etiological alteration for the high blood levels of PAI-1 (140). Several in vitro, in vivo, and animal models are strongly supportive of this hypothesis ( ). However, other studies have found that infusion of insulin

8 February, 2001 DIABETES AND ENDOTHELIAL DYSFUNCTION 43 directly in humans for up to 6 h is not associated with an increase in the blood levels of PAI-1 (150, 151). Recently, we found that simulation of the diabetic environment in normal individuals by exogenous insulin infusions, resulting in hyperinsulinemia, hyperglycemia, and hyperlipidemia, was associated with an increase in the blood concentrations of PAI-1 (152). We concluded from that study that the dysfibrinolysis of type 2 diabetes is most likely a multifactorial process, which includes changes in hormonal (insulin) and substrate (lipids, glucose) concentrations. 3. Cellular and molecular basis for EC dysfunction in diabetes and insulin resistance. The biochemical and cellular factors that are associated with endothelial dysfunction in diabetes are listed in Table 3 and summarized as follows: 1. NADPH is required for proper NO generation. Hyperglycemia may lead to intracellular changes in the redox state with activation of PKC resulting in depletion of the cellular NADPH pool (153). 2. Overexpression of growth factors (154, 155) has also been implicated as a link between diabetes and proliferation of both ECs and VSM, possibly promoting neovascularization. Levels of these growth factors are increased in animal diabetic models, but the temporal sequence is not well defined and, therefore, these issues require further investigation. 3. Chronic hyperglycemia (156) leads to nonenzymatic glycation of proteins and macromolecules and, hence, this biochemical reaction has been implicated in many of the chronic complications of diabetes. Changes in the properties of protein and DNA as well as antigenic changes have been demonstrated to occur as a consequence of nonenzymatic glycation. Independent of chronic effects of hyperglycemia, acute glucose exposure dilates arteries with intrinsic tone and impairs cerebrovascular reactivity to changes in intravascular pressure via an endothelium-mediated mechanism that involves NO and prostaglandins (157). 4. Hyperglycemia increases the flux of glucose through the glycolytic pathway, increasing de novo synthesis of DAG ( ). Increased DAG has been shown to occur in both EC and VSM, leading to increased PKC activity. Both DAG and PKC are important intracellular signaling molecules involved in wide variety of cellular responses, including modulating vasoconstriction (94). Increased PKC-induced contraction has been demonstrated in rat mesenteric arteries exposed to elevated glucose (94). 5. The EC is very susceptible to damage by oxidative stress. The diabetic state is typified by an increased tendency for oxidative stress (109, 115, ) and high levels of oxidized lipoproteins, especially the so-called small, dense LDL-C. High levels of fatty acids and hyperglycemia have both been shown to induce an increased level of oxidation of phospholipids as well as proteins. A proposed hypothesis suggests that this might be one of the etiological factors in inducing endothelial dysfunction in type 2 diabetes. 6. The diabetic state in humans is associated with a prothrombotic tendency as well as increased platelet aggregation as already discussed above. This may be related to several factors, including diminished NO production (10) and decreased fibrinolytic activity related to high levels of PAI-1 levels found in the blood of patients with this disease (166). Remarkably, this defect may be an acquired one. As described before, we have recently demonstrated that mimicking the diabetic environment in normal individuals by simultaneous infusion of glucose and intravenous fat emulsion induces an increase in the blood concentrations of PAI-1. Assuming that the latter represents a market for endothelial dysfunction, we may conclude that the metabolic abnormalities characteristic of type 2 diabetes induce endothelial dysfunction. In addition to decreased fibrinolysis, the diabetic state is also associated with an increase in the activation of the coagulation cascade by various mechanisms such as nonenzymatic glycation, formation of advanced glycosylation end-products (AGE) (156, 167, 168), and decreased heparan sulfate synthesis. Although there is no direct link between activation of the coagulation cascade and endothelial dysfunction in humans, it is possible to speculate that repeated activation of the coagulation cascade may cause overstimulation of ECs and induce endothelial dysfunction. 7. Tumor necrosis factor (TNF) has been implicated as a link between insulin resistance, diabetes, and endothelial dysfunction (169). Increased expression of this factor in human obesity supports the hypothesis that elevated TNF induces insulin resistance. TNF also can induce the synthesis of other cytokines, which alone or in concert with others, may alter endothelial function. 8. In type 2 diabetes, insulin levels tend to be either normal or elevated. The effects of hyperinsulinemia per se on endothelial function, however, have not been extensively studied. The hypothesis has been advanced in recent years that insulin and/or insulin precursors may be atherogenic; however, the data from the recently published United Kingdom TABLE 3. Cellular and molecular basis for endothelial dysfunction in diabetes Molecular defect Increased activation of PKC Overexpression of growth factors (endothelin, ANG-II) Nonenzymatic glycation of proteins and other molecules (DNA) Hyperglycemia induced increase in synthesis of DAG Impaired insulin activation of PIP-3 kinase but normal MAP-kinase response Increased production of PAI-1 Oxidative stress Result Increased proliferation of vessels, altered contraction, altered signal transduction Increased growth and phenotypic change of SMC Change in antigenicity with consequent immune mediated damage Impaired vasodilation and enhanced proliferation of VSMC Increased growth and proliferation of vessels in response to hyperinsulinemia Decreased fibrinolysis, prothrombotic tendency Decreased production of NO, hyperreactivity of SMVC to vasoconstrictive stimuli, increase in proinflammatory, adhesion molecules (ICAM, ELAM, VCAM)

9 44 CALLES-ESCANDON AND CIPOLLA Vol. 22, No. 1 Prospective Diabetes Study (UKPDS) suggest that this is not the case (170). IV. Reversal of Endothelial Dysfunction: Lessons from Human Clinical Trials Several therapeutic interventions have been tested in clinical trials aimed at improving endothelial function. We will review those that are most relevant for the patients with diabetes and/or insulin resistance, such as those testing the effects of insulin sensitizers, antioxidants, and ACE inhibitors. It is beyond the scope of this paper to review the large literature regarding the possible effects of hypolipidemic agents since none of those trials have focused directly on the effects of these agents in affecting endothelial function in patients with diabetes. Moreover, estrogen replacement therapy has been shown to improve endothelial function; however, as of the day of writing this review, no specific trial on women with diabetes was found in the literature. A. Insulin sensitizers As reviewed earlier, insulin-resistant states have been found to be associated with endothelial dysfunction. Thus, investigators have tested the possibility that therapeutic agents that increase insulin sensitivity may also improve endothelial function. Several studies using cell preparations provide support for this hypothesis. Pasceri et al. (171) found that troglitazone (activator of the peroxisome proliferator receptor- and also insulin sensitizer) inhibits in vitro the expression of VCAM-1 and ICAM-1 in activated ECs. This drug also significantly reduced monocyte/macrophage homing to atherosclerotic plaques. In separate studies (172, 173) it was found that this agent also reduced in a dose-dependent manner the expression of VCAM-1, ICAM-1, and E-selectin induced by different amounts of oxidized LDL and tumor necrosis factor. These studies and others (174) provided a rationale for testing the hypothesis that insulin sensitizers may have a beneficial effect on endothelial function in patients with diabetes or insulin resistance. In consequence, some clinical trials have been conducted to investigate the nonhypoglycemic effects of this class of new antidiabetic drugs on endothelial function. In a short trial, Murakami et al. (175) reported that administration of troglitazone was associated with a substantial reduction in the frequency of episodes of angina in patients with coronary artery disease and type 2 diabetes. Moreover, these investigators found that the decrease in episodes of pain was correlated with angiographic (coronary) improvement in endothelial function. Avena et al. (176) studied patients with peripheral vascular disease and IGT (but not overt type 2 diabetes; occult diabetes ) and normal controls matched for age and gender. Brachial artery (BA) flow was measured before and after 5 min of BA occlusion during fasting and at 30 min, 1 h, and 2 h after the administration of 75 g of glucose [oral glucose tolerance test (OGTT)]. The evaluation was repeated 2 and 4 months after administration of troglitazone (400 mg/day). The so-called occult diabetic group had an abnormal response to hyperemia before the treatment with troglitazone with almost no change in flow in response to BA occlusion. After 4 months of therapy with insulin sensitizer, BA flow normalized both while fasting and after oral glucose intake during the OGTT. Cominacini et al. (172) tested the effect of troglitazone (200 mg once daily) in a randomized, placebocontrolled, parallel group study in 29 patients with type 2 diabetes. The results of this study strongly supported a beneficial role for troglitazone since it was associated with an increase in the resistance of LDL to be oxidized compared with the group receiving placebo, and the serum of the patients treated with this medication had lower toxic effects on ECs in vitro. Moreover, plasma E-selectin levels decreased from to g/liter (no change in the placebo group, P 0.01). These investigators concluded that in type 2 diabetes, troglitazone may slow down the development of atherosclerosis by modifying LDL-related atherogenic events. However, not all trials have reached the same conclusions. Tack et al. (177) used a randomized, double-blind, cross-over design to study the effects of troglitazone (400 mg/day). Fifteen obese subjects participated in the trial. A very comprehensive vascular evaluation was performed at the end of each period, which included forearm vasodilator responses to intraarterially administered acetylcholine and sodium nitroprusside; insulin sensitivity and insulin-induced vascular and neurohumoral responses (clamp); and vasoconstrictor responses to NC-monomethyl-l-arginine (l-nmma) during hyperinsulinemia. These patients also had ambulatory 24-h blood pressure monitoring. The participants were insulin resistant compared with lean subjects, and troglitazone improved whole-body and forearm glucose uptake (from to mol dl 1 min 1 ), P 0.006). Insulin-induced vasodilatation was blunted in obese subjects [percent increase in FBF in lean %, vs % in obese, P 0.04], but did not improve during troglitazone. Vascular responses to acetylcholine, sodium nitroprusside, and l-nmma did not differ between the obese and lean group, nor between both treatment periods in the obese individuals. These investigators concluded that in insulinresistant obese subjects, endothelial vascular function is normal despite impaired vasodilator responses to insulin. Moreover, these authors found that troglitazone improved insulin sensitivity but it had no effects on endothelium-dependent and -independent vascular responses. In summary, the data regarding effects of insulin sensitizers on endothelial function in patients with diabetes or insulin-resistant states is suggestive of a beneficial effect at least in the short term, accepting that the opinion is not uniform since a well designed trial did not show any link between improvement in insulin sensitivity and vascular function. Unfortunately, we do not have long-term trials with well defined cardiovascular endpoints to draw any definitive conclusion. Therefore, at this stage, we feel that the hypothesis that insulin sensitizers may prevent or delay atherosclerosis in patients with type 2 diabetes or in those with the syndrome of insulin resistance deserves testing.