Effects of Treatment With Sulfonylurea Drugs or Insulin on Ischemia-Induced Myocardial Dysfunction in Type 2 Diabetes

We studied 19 consecutive type 2 diabetic patients (15 men, 4 women, aged 61 ± 2 years [mean ± SD]) treated with diet plus oral sulfonylurea drugs. The main demographic and clinical characteristics of the study population during glibenclamide or insulin therapy are summarized in Table 1.

All patients had a clinical history of stable angina pectoris (Canadian Cardiovascular Society class 1 or 2) and evidence of significant coronary vessel obstruction (≥70% lumenal diameter narrowing) by coronary angiography (12 patients had two-vessel and 7 had three-vessel disease). Five patients had histories and electrocardiogram (ECG) evidence of previous myocardial infarction (four anterior and one inferior). Patients were excluded for any of the following reasons: 1) previous or ongoing insulin therapy; 2) severe diabetes; 3) diabetic proliferative retinopathy assessed by an experienced ophthalmologist; 4) diabetic neuropathy assessed by autonomic neuropathy tests or diabetic nephropathy (creatinine >1.4 mg/dl); 5) potential sources of gastrointestinal bleeding, anemias, and other endocrine or major organ diseases; 6) arterial systemic hypertension; and 7) inadequate echocardiogram to assess wall motion and thickening in every left ventricular (LV) myocardial segment. Nitrates were the only drugs permitted to control ischemic symptoms in study patients and were withdrawn 12 h before stress echocardiographic studies.

The study protocol was approved by the local ethics committee of the University of Padua School of Medicine, and written informed consent was obtained from all participants after the nature of the procedure was explained.

Before entering the study, all patients underwent a 7-day washout period, during which diet was maintained but sulfonylurea drugs were discontinued. Subsequently, each patient was randomly assigned to either insulin (15–45 IU twice or three times per day; Novo Nordisk, Copenhagen) or glibenclamide (5–15 mg per day) treatment. After 12 weeks of treatment with insulin or glibenclamide, all participants were admitted to the ward in the morning after an overnight fast and were studied at their prevailing plasma glucose value, in stable cardiac conditions. All patients underwent a full medical history and physical examination. Body weight (kg) and height (m) were recorded conventionally, and BMI was calculated. Blood pressure was measured in the sitting position on the same arm by the same observer with a mercury sphygmomanometer (phase V for diastolic value), and the mean value of three consecutive measurements was used. A resting standard 12-lead ECG was recorded in each patient. A blood sample was collected for the plasma determinations of glycated hemoglobin (HbA_1c), glycemia, insulinemia, C-peptide, free fatty acids (FFAs), total cholesterol, HDL cholesterol, LDL cholesterol, triglycerides, and fibrinogen and the blood determination of 3-hydroxybutyrate (3-BOH), lactate, pyruvate, and acetoacetate (AcAc). Finally, a two-dimensional echocardiographic study at rest and during dipyridamole infusion was performed. After completion of these evaluations, each patient was crossed over to either insulin or glibenclamide treatment for an additional 12 weeks. The same demographic, cardiac, and biochemical examinations were performed after the second period of therapy. To avoid any possibility of metabolic control interfering with cardiac performance, during both treatments, we scrupulously sought to keep glycemic control constant by means of outpatient visits at 7-day intervals and home glucose monitoring. Concomitant treatment remained unchanged throughout the study. In addition, we recruited 10 healthy control subjects matched for sex (eight men, two women) and age (60 ± 4 years).

HbA_1c was determined by a chromatographic method. Plasma glucose was measured with a glucose oxidase method on a Beckman glucose analyzer (Fullerton, CA). Plasma free insulin and C-peptide concentrations were determined by standard radioimmunoassays. Plasma FFA concentrations were determined using a microenzymatic technique. Plasma total cholesterol, HDL cholesterol, and triglycerides were determined with enzymatic techniques. LDL cholesterol was calculated with the Friedewald formula. Blood 3-BOH, lactate, pyruvate, and AcAc were determined by fluorimetric techniques. Plasma fibrinogen was determined by a standard laboratory method.

Echocardiographic examinations were performed with Hewlett-Packard Sonos 5500 echocardiographic equipment (Andover, MA). Echocardiographic studies were coded and read by two independent observers in a blinded manner, with no knowledge of the patient’s identity or experimental condition. Echocardiographic analysis was performed using the digitized cine loop method (PreVue III System; Nova MicroSonics, Mahwah, NJ). Wall motion and myocardial thickening were analyzed by dividing the left ventricle into 16 segments, and a semiquantitative score system (1 = normal, 2 = hypokinesia, 3 = akinesia, and 4 = dyskinesia) was used (16). Wall motion was detected by examining echocardiographic images of the left ventricle obtained in the apical four- and two-chamber views and in the parasternal long- and short-axis views. LV wall motion was analyzed by repeated viewing. Rest wall motion was considered abnormal when asynergy was detected in at least two contiguous segments. A wall motion score index (WMSI) was derived for rest and dipyridamole echocardiograms for each patient. The wall motion score index was acquired by summation of individual segment scores divided by the number of interpreted segments. Agreement of interobserver analysis for segmental asynergy was seen in 98% of the visualized segments. Discrepancies were resolved by consensus.

LV volumes were calculated by an ellipsoid biplane area-length method (17). LV ejection fraction (LVEF) was derived as end-diastolic volume minus end-systolic volume divided by the end-diastolic volume. LV endocardial echocardiograms in the apical four- and two-chamber views, for a minimum of two to four cardiac cycles, were digitized at end-diastole (R wave peak) and end-systole (time of smallest cavity area) by two independent observers. A discrepancy of >10 ml for LV volume required the analysis of the echocardiographic tracing by a third observer. Agreement was achieved by consensus. Interobserver and intraobserver variability for LV area (r = 0.96 and r = 0.98, respectively) and for LV length (r = 0.95 and r = 0.95, respectively) was acceptable. The interstudy variability of LVEF was assessed by echocardiographic examination twice within a half-hour and calculated by averaging the absolute differences between the two measurements. Changes in LVEF were considered significant if they were outside the 95% confidence limit of interstudy variability.

Two-dimensional echocardiographic and 12-lead ECG monitoring were performed in combination with a dipyridamole infusion of 0.56 mg/kg body wt over 4 min followed by 4 min of observation, then an infusion of 0.28 mg/kg over 2 min. The cumulative dose was therefore 0.84 mg/kg over 10 min. Two-dimensional echocardiograms were obtained continuously during and up to 10 min after dipyridamole administration. Test results were considered positive when the wall motion score increased by one grade or more at peak stress (e.g., a normal segment becoming hypokinetic, akinetic, or dyskinetic or a hypokinetic segment becoming akinetic or dyskinetic). Akinesia becoming dyskinesia was not considered a positive result, however, because the change may have been due to a passive stretching phenomenon rather than active ischemia.

Results are expressed as means ± SD. Multiple comparisons were performed by a two-way repeated-measurement ANOVA, followed by the Fisher protected least significant difference test. A P value ≤0.05 by the two-tailed test was considered to indicate statistical significance.

Baseline characteristics of study patients are shown in Table 1. Measurements of LV end-diastolic volume index (LVEDVI) (64 ± 9 ml/m², range 54–74) and LVEF (60 ± 9%, range 57–70) in the control group were used to determine the means and 95% confidence limits of normal values. Resting values of LV dimensions and function are listed in Table 2. In diabetic patients, no differences were detected in baseline LVEDVI (glibenclamide, 109 ± 20 ml/m²; insulin, 109 ± 19 ml/m²; NS) and LVEF (glibenclamide, 43 ± 7%; insulin, 46 ± 8%; NS) during both treatments; the baseline LVEDVI values were higher (P < 0.005) and LVEF lower (P < 0.005) than those of control subjects, regardless of treatment group. Heart rate and systolic and diastolic blood pressures did not differ between diabetic patients and control subjects, regardless of treatment group. Regional impairment of myocardial contractility in diabetic patients did not differ during treatment with glibenclamide or insulin; in fact, baseline LV WMSI was similar in diabetic patients during both treatment strategies (glibenclamide, 1.40 ± 0.29; insulin, 1.40 ± 0.28; NS).

Changes in heart rate and systolic blood pressure during dipyridamole infusion were similar in both treatment groups of diabetic patients (Table 2). During glibenclamide treatment, peak stress LVEF decreased (43 ± 7 vs. 37 ± 12%; P < 0.005) (Fig. 1) and WMSI increased significantly (1.4 ± 0.28 vs. 1.98 ± 0.24; P < 0.001) (Fig. 2) in comparison to the baseline values; in patients treated with insulin, peak stress LVEF (46 ± 8 vs. 45 ± 11%; NS) (Fig. 1) and WMSI (1.4 ± 0.29 vs. 1.6 ± 0.4; NS) (Fig. 2) did not differ from baseline values. It was noted that during insulin treatment, patients had significant changes in peak stress LV function compared with values obtained during glibenclamide therapy: LVEF was higher (45 ± 11 vs. 37 ± 12%; P < 0.001) and WMSI lower (1.6 ± 0.4 vs. 1.98 ± 0.24; P < 0.001).

As shown in Table 2, no changes were observed in short- or long-term metabolic control between the two treatments; furthermore, no changes were observed in intermediary metabolites or plasma lipid levels. Only FFAs were slightly higher during glibenclamide treatment, although the difference was not statistically significant.

The major aim of this study was to assess the effects of either insulin or glibenclamide therapy on susceptibility to ischemia, in terms of extension of left ventricular myocardial dysfunction, during dipyridamole infusion in type 2 diabetic patients with coronary artery disease. Because metabolic control per se may deeply affect several variables of cardiac function such as substrate utilization (18), coronary blood flow reserve (19), and myocardial contractility (20), we meticulously ensured the same metabolic control under the two different treatments. This control was maintained weekly by modifying therapies according to each patient’s home blood glucose monitoring (Table 1). This result allowed us to assess the effect of each treatment on cardiac function without the confounding consequences of different metabolic control levels.

In patients with type 2 diabetes, many factors contribute to increased morbidity and mortality by ischemic heart disease. One contributor may be the use of sulfonylurea drugs. Such a possibility, which first arose several years ago, has recently resurfaced after the discovery that sulfonylureas act by inhibiting K_ATP channels. In the heart, inhibition of these channels prevents ischemic preconditioning, an endogenous mechanism that protects the heart from ischemic injury. Furthermore, experimental data (21) suggest that the myocardium in streptozotocin-induced diabetic rat hearts may benefit more from preconditioning than normal myocardium, possibly as a result of the reduced production of glycolytic metabolites during sustained ischemia and the concomitant attenuation of intracellular acidosis.

The difficulty of translating these findings to the outcome of patients is demonstrated by some U.K. Prospective Diabetes Study data (22) that showed no increase in cardiovascular events with sulfonylurea drug use. On the other hand, in a recent prospective study (23), it has been shown that sulfonylurea drugs increased early mortality in patients with diabetes after direct angioplasty for acute myocardial infarction. However, the outcome of these patients may be influenced by several factors that could explain the controversies regarding long-term results; for example, angiotensin II type 1 receptor antagonists and ACE inhibitors are beneficial not only for suppressing ventricular remodeling (24) but also for preserving the preconditioning mechanism (25).

Our results reflect different effects of insulin and glibenclamide treatment on myocardial contractile function in patients with type 2 diabetes during acute ischemia provoked by dipyridamole infusion, and they should not be translated to other sulfonylurea drugs without further evidence. The diversities of sulfonylurea drugs relate to their potency and half-lives. However, there is increasing data to support that they may differ in their cardiovascular effects, possibly because of a distinct affinity for cardiac and vascular K_ATP channels. For example, the half-maximal inhibitory concentration (IC₅₀) of glimepiride in blood vessels has been reported to be 31.6 nmol compared with an IC₅₀ of 6.8 nmol for glibenclamide. It should be noted, however, that these IC₅₀ values are several orders of magnitude greater than the IC₅₀ values for K_ATP channels in pancreatic islet cells. Plasma levels of glibenclamide in patients treated for type 2 diabetes range between 0 and 1.5 mmol/l. Because glibenclamide concentrations as low as 0.01 mmol/l have been shown to inhibit vascular relaxation induced by K_ATP channel–opening drugs, it is quite possible that significant vascular effects occur during glibenclamide treatment in type 2 diabetic patients (26). Moreover, in experimental and pharmacological studies, glibenclamide is used as a classical K_ATP channel inhibitor (27), and its deleterious effect on ischemic preconditioning has been widely investigated, suggesting a possible negative role in myocardial function in diabetic patients with coronary artery disease (27,28). Furthermore, glibenclamide is by far the most commonly used sulfonylurea drug for type 2 diabetes in Italy. For these reasons, and because of a more homogeneous effect using exclusively one instead of several different oral hypoglycemic agents, we decided to compare insulin versus glibenclamide. Further studies are needed to assess possible differences on the in vivo cardiac effects of other sulfonylurea drugs.

The results of this study show that ischemic myocardial dysfunction induced by dipyridamole infusion is more severe during treatment with sulfonylurea drugs than during insulin treatment of type 2 diabetic patients with documented ischemic heart disease. In fact, during glibenclamide treatment, peak stress LV global pump function (evaluated as ejection fraction) decreases and is significantly less than during insulin treatment. This impairment in LV systolic function is the result of an extension in ischemic myocardial dysfunction induced by dipyridamole, as demonstrated by the increase of LV WMSI. On the other hand, patients treated with insulin do not show a worsening of myocardial function during dipyridamole stress, as demonstrated by the absence of a significant increase in LV WMSI.

It is reasonable to suggest that the restitution of a preconditioning mechanism in insulin-treated patients is able to reduce the severity of myocardial dysfunction associated with ischemia induced by dipyridamole stress. Other possible mechanisms may be a locally higher plasma insulin concentration during insulin treatment causing 1) an increase in coronary blood flow by a primary reduction in coronary tone secondary to insulin-mediated elaboration of vasoactive and/or paracrine factors within the coronary circulation (29) and 2) a direct positive inotropic effect on postischemic heart, reducing the extension of myocardial dysfunction (30).

In this study, dipyridamole stress has been chosen for induction of myocardial ischemia on the basis of several considerations. It has been shown that dipyridamole is more effective, or at least no less effective, than dobutamine at producing flow disparity between myocardial regions supplied by normal and stenotic arteries (31–34). Moreover, minor and major complications appear more frequently during dobutamine than dipyridamole tests (33,34). On the other hand, the mechanism of action of dipyridamole renders this pharmacological stress particularly useful in the diabetic heart characterized by severe abnormalities in the coronary microcirculation and flow reserve along with the atherosclerotic occlusion of epicardial vessels. In fact, the standard dose of dipyridamole can induce maximal or near-maximal coronary vasodilatation in most patients (35), whereas dobutamine produces a significantly smaller increase in myocardial blood flow (36,37). Finally, the mechanism of action of dobutamine involves stimulation of cardiac adrenergic receptors. In the diabetic heart, a defective cardiac response to catecholamines occurs. This has been explained by an impairment of the cardiac adrenergic signaling system mainly due to the decrease in β-adrenergic receptor concentration (38–40).

Thus, this randomized crossover design study may help to define the potential negative action of glibenclamide in type 2 diabetic patients with coronary artery disease. The results demonstrate that ischemic myocardial dysfunction is easily inducible by dipyridamole in type 2 diabetic patients with ischemic heart disease. Insulin treatment instead of glibenclamide therapy significantly prevents this phenomenon. Although the precise underlying pathophysiological mechanism is not completely known, these data may eventually be important in designing optimal therapy for patients with type 2 diabetes.