Relationship Between Insulin Resistance and an Endogenous Nitric Oxide Synthase Inhibitor

Context Increased levels of asymmetric dimethylarginine (ADMA) are associated with endothelial dysfunction and increased risk of cardiovascular disease. Several cardiovascular risk factors are associated with reduced sensitivity to insulin, but elevated ADMA concentrations have not been fully linked to the metabolic syndrome.

Objective To evaluate the relationship between insulin sensitivity and plasma ADMA concentrations, and to determine whether a pharmacological treatment that increases insulin sensitivity would also modulate ADMA concentrations.

Design, Setting, and Subjects Cross-sectional study, containing a nonrandomized controlled trial component, of 64 healthy volunteers without diabetes (42 women, 22 men; 48 with normal blood pressure and 16 with hypertension), which was conducted at a university medical center between October 2000 and July 2001.

Intervention Rosiglitazone (4 mg/d for 4 weeks and then 4 mg twice daily for 8 weeks), an insulin-sensitizing agent, was given to 7 insulin-resistant subjects with hypertension. These subjects were studied before and after 12-week treatment.

Main Outcome Measures Insulin sensitivity measured by the insulin suppression test, and fasting plasma levels of low-density lipoprotein cholesterol, triglycerides, high-density lipoprotein cholesterol, glucose, insulin, and ADMA concentrations.

Results Plasma ADMA concentrations were positively correlated with impairment of insulin-mediated glucose disposal in nondiabetic, normotensive subjects (r = 0.73; P<.001). Consistent with the metabolic syndrome, ADMA levels were also positively correlated with fasting triglyceride levels (r = 0.52; P<.001) but not with low-density lipoprotein cholesterol levels (r = 0.19; P = .20). Plasma ADMA concentrations increased in insulin-resistant subjects independent of hypertension. Pharmacological treatment improved insulin sensitivity and reduced mean (SD) plasma ADMA concentrations from 1.50 (0.30) to 1.05 (0.33) µmol/L (P = .001).

Conclusion A significant relationship exists between insulin resistance and plasma concentrations of ADMA. Pharmacological intervention with rosiglitazone enhanced insulin sensitivity and reduced ADMA levels. Increases in plasma ADMA concentrations may contribute to the endothelial dysfunction observed in insulin-resistant patients.

In its most recent adult treatment panel report, the National Cholesterol Education Program recognized the metabolic syndrome (syndrome X; insulin resistance syndrome) as a new target of risk-reduction therapy.¹ The metabolic syndrome is a cluster of closely associated and interdependent abnormalities, including insulin resistance, compensatory hyperinsulinemia, hyperuricemia, dyslipidemia, and hypertension,² and predisposes individuals to type 2 diabetes, hypertension, and coronary heart disease (CHD).^3,4

In patients at high risk of CHD, endothelial dysfunction is observed in morphologically intact vessels before the onset of clinically manifested vascular disease.⁵ Indeed, there are several lines of evidence that indicate that endothelial function is compromised in situations with reduced sensitivity to endogenous insulin. For example, the increase of blood flow in the legs in response to methacholine (a measure of endothelium-dependent vasorelaxation) is reduced in nondiabetic insulin-resistant individuals.⁶ In addition, nitric oxide–dependent, flow-mediated dilatation of the brachial artery is impaired in hypertensive⁷ and normotensive⁸ subjects with insulin resistance. Moreover, the endothelium also has been shown to modulate several other processes important in the development of CHD, including inflammation and thrombosis. Thus, the findings that plasma concentrations of plasminogen activator inhibitor 1 and endothelin 1 are elevated in the metabolic syndrome may indicate a more generalized endothelial dysfunction beyond the regulation of local blood flow.^9,10 Along the same lines, we recently found that plasma concentrations of soluble adhesion molecules are increased in proportion to the degree of insulin resistance in healthy volunteers.¹¹ This observation may partially explain the finding that adhesiveness of circulating mononuclear cells isolated from nondiabetic individuals to cultured endothelium is closely correlated with their degree of insulin resistance.¹² Given the importance of endothelial function and monocyte adhesion in the early stages of atherogenesis,¹³ it is not unreasonable to speculate that these alterations may be part of the link between insulin resistance or compensatory hyperinsulinemia and CHD. With this rationale, we propose that loss of the homeostatic functions of the endothelium may be added to the cluster of abnormalities that make up the metabolic syndrome.

The present study was initiated to extend further the link between insulin resistance and endothelial dysfunction. More specifically, we hypothesized that changes in the concentration of asymmetric dimethylarginine (ADMA) may play a role in this relationship. Asymmetric dimethylarginine is an endogenous inhibitor of nitric oxide synthase (NOS),¹⁴ and plasma concentrations of ADMA are elevated in clinical syndromes associated with increased risk of vascular disease.¹⁵ Moreover, there is evidence that ADMA correlates closely with nitric oxide–mediated vasorelaxation¹⁶ and with adherence of mononuclear cells to the endothelium.¹⁷

The study population consisted of 64 nondiabetic individuals: 48 with normal blood pressure and 16 with hypertension. The participants were recruited from the San Francisco Bay area by advertisements in local newspapers indicating our interest in studying the relationship between insulin resistance and risk factors for CHD in healthy volunteers and patients with hypertension. The metabolic studies were performed at the General Clinical Research Center of Stanford University Medical Center. The study protocol was approved by the Stanford Human Subjects Committee and all participants gave written informed consent. None of the volunteers were paid for participation in the study. All subjects were in good general health, with no past history or current symptoms of atherosclerotic disease. They had normal findings on physical examination (with the exception of hypertension) and chemical screening battery, and were nondiabetic by the criteria of the American Diabetes Association.¹⁸ Degree of obesity was estimated by body mass index, and hypertension was defined as at least 2 resting blood pressure measurements greater than 140/90 mm Hg or a history of taking antihypertensive medication. Thirteen of the 16 patients with hypertension were being treated with 1 or more of the following antihypertenesive agents: α-receptor (n = 1) or β-receptor (n = 3) antagonists, calcium channel blockers (n = 3), angiotensin-converting enzyme inhibitors (n = 5), or diuretics (n = 4). Other than the α-blockers, β-blockers, and diuretics, subjects were not taking any drugs that might affect carbohydrate or lipoprotein metabolism. Following admission to the General Clinical Research Center, blood was drawn after an overnight fast for measurements of plasma ADMA as well as insulin, glucose, triglyceride, high-density lipoprotein (HDL), and low-density lipoprotein (LDL) cholesterol concentrations as described previously.¹⁹

Insulin-mediated glucose disposal was estimated by a modification²⁰ of the insulin suppression test as introduced and validated earlier by our research group.²¹ After an overnight fast, each patient had an intravenous catheter placed in each arm. Blood was sampled from one arm for measurement of plasma glucose and insulin concentration, and the other arm was used for administration of test substances. Octreotide acetate, a somatostatin analogue, was administered at a rate of 0.27 µg/m² per minute to inhibit endogenous insulin secretion. Simultaneously, insulin and glucose were infused at rates of 32 mU/m² per minute and 267 mg/m² per minute, respectively. Blood was sampled every 30 minutes until 150 minutes into the study, then every 10 minutes until 180 minutes had elapsed. The 4 values obtained between 150 and 180 minutes were averaged to calculate the steady-state plasma insulin and steady-state plasma glucose (SSPG) concentrations for each individual. Because steady-state plasma insulin concentrations are similar for all individuals, the SSPG concentration provides a direct measure of the ability of insulin to mediate disposal of an infused glucose load: the higher the SSPG concentration, the more insulin resistant the individual.

Fasting plasma ADMA concentrations were measured by high-performance liquid chromatography (HPLC) with precolumn derivatization with o-phthaldialdehyde (OPA) using a modification of a previously described method.¹⁶ Briefly, 0.5 mL of sample was spiked with 10 µmol/L of L-homoarginine as an internal standard and ADMA was isolated from plasma by solid-phase extraction with a cation-exchange column (Bond Elute SCX 50 mg, Varian Inc, Palo Alto, Calif) according to Pettersson et al²² after protein precipitation. The eluates were evaporated to dryness at 50°C under nitrogen and resuspended in double-distilled water. Chromatography was carried out on a computer-controlled chromatography system (Varian Star) consisting of an HPLC pump (Varian 9010), an automatic injector with sample-reagent mixing capabilities (Varian 9100) and a fluorescence detector (Varian Fluorichrome II). The samples were incubated for exactly 1 minute with OPA reagent (5.4 mg/mL of OPA in a borate buffer [pH = 8.4] containing 0.4% β-mercaptoethanol) before automatic injection into the HPLC system. The OPA derivatives of L-arginine, ADMA, symmetric dimethylarginine, and the internal standard, L-homoarginine, were separated on a 250 × 4.5-mm (internal diameter) 7-µm nucleosil phenyl HPLC column (Supelco Inc, Bellafonte, Pa) with the fluorescence detector set at 340 nm excitation and 450 nm emission. Amino acids were eluted from the column with an isocratic gradient of 50 mM potassium phosphate buffer (pH = 6.6)/90% methanol (80:20) at a flow rate of 1 mL/min. Concentrations of ADMA were calculated by comparing the ADMA/homoarginine ratio with standards of known concentrations. The recovery rate for ADMA was 85% and the intrasample variation was 6%. The detection limit of the assay was 0.1 µM.

Since prevalence of insulin resistance is increased in patients with essential hypertension² and since plasma ADMA concentrations have been reported to be elevated in this syndrome,¹⁵ we measured plasma ADMA concentrations in 16 hypertensive patients with SSPG concentration values in the upper and lower tertiles of insulin resistance distribution.²³ Eight of these patients were classified as being insulin resistant (SSPG >167 mg/dL [9.25 mmol/L]), and 8 as insulin sensitive (SSPG <113 mg/dL [6.25 mmol/L]). Sixteen normotensive individuals, selected from the sample of 48 healthy volunteers, were matched for age, body mass index, and degree of insulin sensitivity with the 16 patients with hypertension and were similarly divided into an insulin-resistant (SSPG >167 mg/dL [9.25 mmol/L]) and an insulin-sensitive (SSPG <113 mg/dL [6.25 mmol/L]) group. In order to further evaluate the relationship between insulin resistance and plasma ADMA concentrations, 7 insulin-resistant, hypertensive volunteers were restudied after receiving rosiglitazone for 3 months (4 mg/d for 4 weeks, followed by 4 mg, twice daily, for 8 weeks).

Summary statistics are expressed as mean (SD) and range. Pearson correlation coefficients were calculated, first between plasma ADMA concentration and SSPG concentration, and then between plasma ADMA concentration and age, body mass index, systolic blood pressure, diastolic blood pressure, LDL cholesterol, HDL cholesterol, triglyceride, fasting glucose, and insulin concentrations, in the 48 normotensive individuals. A multiple regression analysis was performed to further quantify the relationships between plasma ADMA concentration and the above-mentioned risk factors. Specifically, the dependent variable, plasma ADMA concentration, was regressed for age, sex, body mass index, systolic blood pressure, diastolic blood pressure, LDL cholesterol, HDL cholesterol, triglyceride, fasting glucose, fasting and insulin, and SSPG concentrations. All variables were entered in the model simultaneously. The α was set at .05.

Plasma ADMA concentrations and the metabolic parameters in the identified 16 hypertensive and 16 normotensive individuals were subsequently compared using an unpaired t test. The effect of rosiglitazone treatment on plasma ADMA and SSPG concentrations in the subpopulation of insulin-resistant, hypertensive subjects was compared using a paired t test. All statistical analyses were performed using Systat version 10.01 (SPSS Science, Chicago, Ill).

The demographic and metabolic characteristics of the 48 normotensive individuals and 16 patients with hypertension are summarized in Table 1. Although the mean values for all the clinical and metabolic variables were within conventionally accepted normal limits, the wide SD indicates considerable interindividual variability.

The results in Figure 1 illustrate the significant relationship that existed between insulin resistance, as quantified by the SSPG concentration, and plasma ADMA concentration in the 48 normotensive, healthy volunteers studied (r = 0.73, P<.001). The Pearson correlation coefficients between plasma ADMA concentrations and the other CHD risk factors measured are shown in Table 2. The results demonstrate that systolic blood pressure and plasma triglyceride concentrations were significantly correlated with plasma ADMA levels, but the degree of relationships noted were of lesser magnitude than the one between plasma ADMA and SSPG concentrations. It should be noted that the relationship between plasma insulin concentration, which is often used as a surrogate marker of insulin resistance, and SSPG concentration was weaker (r = 0.52, P<.001) than that between SSPG and ADMA concentrations. Furthermore, there was no correlation between plasma ADMA concentrations and serum creatinine levels (r = −0.16, P = .28).

Multiple regression analysis was performed to define the independent relationship between plasma ADMA concentrations and the CHD risk factors listed in Table 2. The results of this analysis are seen in Table 3 and indicate that the only statistically independent relationship was between ADMA and SSPG concentrations.

Table 4 summarizes the data from the 4 subgroups being compared for blood pressure and insulin sensitivity. As expected, the SSPG concentrations were approximately 3-fold higher in the insulin-resistant individuals, both hypertensive and normotensive (P<.001). Plasma ADMA concentrations are seen in the second row of Table 4, and the values are significantly higher (P<.01) in the insulin-resistant individuals, irrespective of blood pressure category. This significant relationship between plasma ADMA and insulin resistance in the hypertensive individuals is also highlighted in Figure 1 (r = 0.70, P<.003). Although the mean HDL cholesterol concentrations were lower and the triglyceride and fasting insulin concentrations were higher in the insulin-resistant individuals compared with their respective insulin-sensitive counterparts, there was a great deal of variability in these individuals and only fasting insulin concentrations were significantly higher in the insulin-resistant hypertensive group. Based on these statistical considerations, it seems evident that the differences in plasma ADMA concentrations between insulin-sensitive and insulin-resistant individuals are present in both hypertensive and normotensive individuals.

Figure 2 illustrates the changes in SSPG and plasma ADMA concentrations following the administration of rosiglitazone to 7 of the 8 subjects with high blood pressure who were also insulin resistant. As expected, treatment with rosiglitazone for 3 months resulted in enhanced insulin sensitivity as demonstrated by reduced mean (SD) SSPG concentrations (263 [52] vs 168 [81] mg/dL [14.61 {2.90} vs 9.33 {4.47} mmol/L]; P = .005). Reduction of insulin resistance was also associated with a significant fall in mean plasma ADMA concentrations (1.50 (0.30) vs 1.05 (0.33) µmol/L; P = .001). However, the mean blood pressure before (142/81 mm Hg) and after (149/79 mm Hg) rosiglitazone treatment was essentially unchanged.

The results of this study demonstrate a significant relationship between direct measures of insulin-mediated glucose disposal (SSPG concentration) and plasma ADMA levels in a population of healthy, normotensive, nondiabetic volunteers. Indeed, the association between plasma ADMA concentrations and insulin resistance was of greater magnitude than that between SSPG concentration (the specific determinant of insulin resistance) and fasting plasma insulin concentration (a commonly used surrogate estimate of insulin resistance).²³ In addition, multiple regression analysis revealed that SSPG concentrations were the strongest predictor of ADMA concentrations and that the relationship between insulin resistance and ADMA concentrations was independent of other factors associated with insulin resistance and increased CHD risk.

The observation that circulating ADMA concentrations were so closely related to insulin resistance may serve to provide a more general explanation for the reports of elevated plasma ADMA concentrations in patients with type 2 diabetes,²⁴ essential hypertension,²⁵ and renal failure.²⁶ An increase in the prevalence of insulin resistance is well documented in patients with type 2 diabetes²⁷ and essential hypertension.² In the present study, plasma ADMA concentrations were elevated in insulin-resistant individuals, whether or not they were hypertensive, when compared with insulin-sensitive individuals with similar blood pressures. To state it more explicitly, plasma ADMA concentrations were not increased in hypertensive patients unless they were also insulin resistant. Moreover, when a subgroup of insulin-resistant individuals with hypertension was treated with pharmacological therapy to increase insulin sensitivity, ADMA levels fell with no alterations in blood pressure. The situation is certainly more complicated in patients with renal failure, but insulin resistance is present in these patients²⁸ and may contribute to the reported elevations of plasma ADMA.¹⁴ The importance of ADMA has recently been highlighted by Zoccali and colleagues,²⁹ who found that plasma ADMA concentration was an independent risk factor for both overall mortality and cardiovascular events in patients with end-stage renal disease.

Although our results raise the possibility that insulin resistance and plasma ADMA concentrations are associated in a cause-and-effect manner, at this time we can only speculate as to the nature of this relationship. However, there are published observations that can serve as a framework for how ADMA regulation may be altered in the setting of insulin resistance. Several lines of evidence indicate that ADMA is formed from the degradation of methylated proteins rather than from the methylation of free L-arginine. Boger and colleagues³⁰ demonstrated that inhibition of the important methylating enzyme, S-adenosylmethionine–dependent methyl transferase, results in reduced endothelial formation of ADMA. Furthermore, expression and activity of several protein arginine N-methyltransferases is upregulated by native or oxidized LDL cholesterol, offering a putative mechanism for elevated ADMA levels associated with hyperlipidemia.

High concentrations of ADMA can result not only from increased ADMA synthesis, but also from reduced degradation. A selective pathway for the metabolism of ADMA by an enzyme, dimethylarginine dimethylaminohydrolase (DDAH), has recently been described.³¹ Two isoforms of DDAH have been isolated, with DDAH I typically found in tissues expressing nitric oxide synthase I (neuronal NOS), whereas DDAH II predominates in tissues containing NOS III (endothelial NOS).³² DDAH selectively hydrolyzes ADMA to L-citrulline and dimethylamine. Indeed, ADMA concentrations appear to be inversely related to DDAH activity.³³ Furthermore, there is at least indirect evidence that increased oxidative stress, a change common in situations where ADMA concentrations are shown to be elevated (hypertension, hyperglycemia, hypercholesterolemia, and hyperhomocysteinemia), will reduce DDAH activity.^31,34 ADMA is also cleared in the urine, which may partially explain the increase in plasma ADMA levels in patients with renal insufficiency. Each of these mechanisms can potentially alter ADMA concentrations but further research is necessary to delineate which ones come into play in the metabolic syndrome.

Less speculative is the relationship between ADMA and endothelial function. Vallance et al¹⁴ demonstrated that plasma ADMA concentrations were elevated in patients with renal failure and were the first to demonstrate that endogenous ADMA antagonized endothelium-dependent vasodilatation.³⁵ Subsequently, plasma ADMA concentrations have been found to be elevated in patients with associated risk factors for endothelial dysfunction and atherosclerosis.¹⁵ For example, we recently observed a 2-fold elevation of ADMA in patients with hypercholesterolemia,¹⁶ where plasma ADMA concentrations correlated better with endothelial dysfunction than did LDL cholesterol in these patients. In the same study we were also able to show that endothelial vasodilator dysfunction associated with elevated plasma ADMA concentrations was reversible by administration of L-arginine, providing physiological evidence that ADMA is a competitive inhibitor of NOS.

While it is clear that elevations of LDL cholesterol and compromised renal function can have dramatic effects on plasma ADMA concentrations and endothelial function, these mechanisms cannot explain the striking correlations between ADMA and SSPG concentrations observed in the current study. There was no significant correlation between LDL cholesterol and ADMA in the normotensive or hypertensive individuals recruited. Moreover, all study participants had normal renal function as determined by creatinine levels. It is interesting to note that the 2-fold elevations in ADMA observed in the insulin-resistant population are similar to those in individuals with other known risk factors.^15,16,36 Since these concentrations have been shown to impair vasorelaxation in humans,^16,35,37,38 and since ADMA levels can predict risk for cardiovascular events,^29,39 the elevations in ADMA observed in the present investigation should be sufficient to have pathophysiological effects. Consistent with this notion is the observation that basal nitric oxide production is reduced in insulin-resistant individuals.⁴⁰ Furthermore, nitric oxide–dependent, but not–independent, vasorelaxation is impaired in obese insulin-resistant patients⁶ as well as in normotensive, first-degree relatives of those with type 2 diabetes.⁸ Most importantly, a significant correlation between nitric oxide–dependent vasorelaxation and insulin sensitivity was found in all of these studies. Finally, it should be emphasized that the ADMA concentrations reported in the current study were derived from fasting plasma. Fard et al⁴¹ reported similar ADMA values in patients with type 2 diabetes who are, by definition, insulin resistant. However, elevated ADMA concentrations were accentuated after ingestion of a high-fat meal, but not after a low-fat meal. Alterations in ADMA concentrations after the high-fat meal were accompanied by a decline in endothelial function, as monitored by brachial artery vasodilation that persisted for several hours. Thus, fasting levels may not adequately reflect the effect of ADMA on NOS activity, especially considering modern dietary habits as well as the fact that many individuals spend the majority of the day in the postprandial state.

Our previous attempts to reverse the physiological effects of elevated ADMA have used dietary L-arginine supplementation. While this resulted in normalization of ratios of L-arginine to ADMA, there was no effect on ADMA concentrations. In some studies, but not all, hemodialysis appears to be effective in reducing ADMA concentrations in patients with renal failure, implicating renal function in the clearance of dimethylamines.^42-44 However, rosiglitazone, used in this study to enhance insulin sensitivity, represents the first pharmacological intervention resulting in reduced ADMA concentrations in humans. It would, therefore, be interesting to test the effectiveness of insulin-sensitizing compounds in reducing ADMA levels in patients with other risk factors associated with insulin resistance.

In summary, the current study indicates that plasma ADMA concentrations are increased in insulin-resistant normotensive and hypertensive individuals. It can be speculated that this phenomenon plays a significant role in the endothelial dysfunction described in clinical syndromes characterized by insulin resistance.