Objective: To investigate whether blood flow in normal and post-stenotic coronary arteries is altered by therapeutic doses of the sulfonylurea agent glibenclamide.
Patients: 12 patients with a high grade stenosis of the left anterior descending coronary artery (n = 10) or left circumflex coronary artery (n = 2), and an angiographically normal corresponding left circumflex artery or left anterior descending artery, respectively.
Design: Two Doppler ultrasound wires were positioned in the “normal” and post-stenotic artery for simultaneous measurements of coronary blood flow velocity under baseline conditions and after intravenous glibenclamide, 0.05 mg/kg body weight. Local coronary blood flow was calculated from the average peak velocity and the cross sectional area derived from quantitative coronary angiographic analysis. Coronary flow reserve was determined after intracoronary injection of 30 μg adenosine and 12 mg papaverine.
Results: One hour after glibenclamide, serum insulin increased from (mean (SD)) 7.4 (2.0) to 44.8 (25.5) mU/l (p < 0.005), and C peptide from 1.4 (0.4) to 3.4 (1.2) ng/l (p = 0.005). In normal coronary arteries coronary flow reserve was 2.6 (0.4) after adenosine and 3.0 (0.4) after papaverine, while in post-stenotic arterial segments it was 1.2 (0.3) after adenosine (p = 0.005) and 1.3 (0.3) after papaverine (p = 0.005). There was no significant difference after glibenclamide. In non-stenotic arteries, average peak velocity (18.8 (5.2) cm/s) and calculated coronary blood flow (23.8 (10.7) ml/min) were not altered by glibenclamide (18.3 (5.2) cm/s and 22.8 (10.4) ml/min, respectively). In post-stenotic arteries, baseline average peak velocity was 13.3 (4.9) ml/min and coronary blood flow was 9.1 (3.0) ml/min, without significant change after glibenclamide (13.3 (5.2) cm/s, 9.0 (3.2) ml/min).
Conclusions: Glibenclamide, 0.05 mg/kg intravenously, is effective in increasing serum insulin, suggesting a KATP channel blocking effect in pancreatic β cells. It does not compromise coronary blood flow and vasodilatation in response to adenosine and papaverine in post-stenotic and angiographically normal coronary arteries at rest.
- coronary circulation
- KATP channels
- coronary artery disease
- Doppler flow measurements
Statistics from Altmetric.com
In the USA, approximately six million people suffer from non-insulin dependent diabetes mellitus (85% of all diabetic patients), and are potentially eligible for treatment with sulfonylurea agents.1 Sulfonylurea drugs inhibit potassium efflux through ATP sensitive potassium channels (KATP channels) in pancreatic β cells. This results in increased exocytic insulin release.2 KATP channels have been found in several tissues—for example, cardiac myocytes, vascular smooth muscle cells, in the central nervous system, and in the kidney. Thus complex effects on different organ systems in patients with sulfonylurea treatment need to be considered during follow up.
In several animal models, hypoxic dilatation of coronary arteries has been shown to be mediated by activation of vascular KATP channels.3 The role of these channels in the coronary circulation has been investigated in various experimental settings. It has been suggested that they contribute to the control of basal coronary tone and coronary vasodilatation in response to adenosine.3–5 In dogs, a synergistic role of adenosine and KATP channels in maintaining coronary flow distal to an artificial stenosis has been demonstrated.6 Thus a potential effect of KATP channel blocking agents, such as the antidiabetic sulfonylurea drugs, on the coronary circulation in different physiological and pathophysiological conditions is a matter of debate.7 However, the effects of sulfonylurea drugs on coronary blood flow in patients with coronary artery disease have not yet been clarified.
Cardiovascular mortality in diabetic patients on sulfonylurea drugs may be increased in comparison with those on insulin treatment.8 In comparison with insulin treated patients, an increased mortality has recently been reported in patients on sulfonylurea drug treatment undergoing primary angioplasty for acute myocardial infarction.9
It is suggested that opening of the cardiac KATP channels is involved in the phenomenon of ischaemic preconditioning, in which a notable reduction in infarct size occurs if the myocardium has been pretreated with brief episodes of ischaemia followed by reperfusion. The increased mortality during treatment with sulfonylurea agents has mainly been attributed to inhibition of “preconditioning-like” effects, resulting in a reduced cardioprotective potential.10 In human patients, however, this has not yet been proven. Theoretically, the blockade of coronary KATP channels by sulfonylurea drugs might compromise coronary blood flow. For example, it has been shown that the forearm vasodilator response to brief periods of arterial occlusion is reduced after pharmacological blockade of KATP channels.11 This possible interaction between sulfonylurea agents and vascular KATP channels is expected to occur at therapeutic doses of the drug. Although the plasma concentrations of sulfonylurea required to block cardiac and vascular KATP channels are higher than those inducing pancreatic insulin release, concentrations of glibenclamide that are effective at pancreatic β cells could also affect a proportion of the vascular KATP channels.12
Our aim in the present study was to investigate the effects of therapeutic doses of glibenclamide on coronary blood flow in patients with coronary artery disease. We hypothesised that a therapeutic dose of glibenclamide could alter post-stenotic coronary blood flow or vasodilatation in response to adenosine. We studied the influence of glibenclamide on coronary flow in angiographically normal coronary arteries and arteries with a high grade stenosis using intracoronary Doppler flow measurements and quantitative coronary angiography.
Written informed consent was obtained from all the patients 24 hours before they were included in the study. The study protocol was approved by the local institutional ethics review board.
We studied 12 patients (mean (SD) age, 54 (6) years; six female, six male) with a high grade stenosis (> 90% diameter stenosis) of the left anterior descending coronary artery or the left circumflex artery (or a major branch). After diagnostic coronary angiography the patients were scheduled for angioplasty because of significant ST segment depression or terminal T wave inversion during or after bicycle testing (four patients), objective signs of ischaemia on myocardial scintigraphy (five patients), or reproducible symptoms of angina during exercise testing (seven patients).
Patients were included in the study if there was a high grade stenosis of the left anterior descending coronary artery (10 patients) with an angiographically normal left circumflex coronary artery, or a high grade stenosis of the left circumflex artery (two patients) with a “normal” left anterior descending artery.
Exclusion criteria were unstable angina, acute myocardial infarction (less than four weeks before), atrial fibrillation, second or third degree atrioventricular block, left or right bundle branch block, valvar heart disease, cardiomyopathy, chronic renal failure (serum creatinine > 177 μmol/l), a history of allergic reactions to sulfonylurea agents or sulfonamides, previous treatment with sulfonylureas, and oral anticoagulation treatment.
All cardiovascular drugs, including β receptor blockers, calcium antagonists, long acting nitrates, and angiotensin converting enzyme inhibitors, were withheld 24 hours before the procedure. Where there were anginal symptoms, patients were treated with short acting nitrates.
Coronary angiography and quantitative diameter measurements
Cardiac catheterisation was performed routinely by the Judkins technique with 7 or 8 French catheters through a femoral sheath (contrast agent: iopromide, Ultravist-370, Schering AG, Berlin, Germany). Quantitative measurements of the coronary artery diameters and left ventricular ejection fraction were obtained from digitised cineangiograms, using the coronary angiography analysis system II (Pie Medical, Maastricht, Netherlands). Details of this edge detection based method have been published previously.13 No intracoronary glyceryl trinitrate was given, as this might interfere with the measured vessel diameters and blood flow velocities. The percentage diameter stenosis and the average lumen diameters in two orthogonal projections 5 mm distal to the tip of the Doppler wire were measured, using the guiding catheter as a scaling device along with a user defined reference lumen diameter of an angiographically normal coronary artery segment.
Intracoronary Doppler flow velocity measurements
Intracoronary Doppler flow profiles were obtained using a 0.014 inch (0.36 mm) flexible Doppler guide wire (12 MHz, FloWire, Cardiometrics Inc, Mountain View, California, USA). After stable placement of the guiding catheter in the left coronary ostium, one Doppler guide wire was positioned in the distal left anterior descending coronary artery and the other in the distal left circumflex coronary artery for simultaneous measurements of coronary flow velocity. A switch box allowed measurement of Doppler flow profiles in both coronary arteries without disconnecting the wires. The wire in the coronary artery with the high grade stenosis was positioned distal to the stenosis (fig 1). Data were obtained after we had confirmed that the position of both wires was stable by fluoroscopic imaging and flow measurements over a period of three minutes. Throughout the procedure, average peak velocity, average systolic peak velocity, and average diastolic peak velocity (all in cm/s) were recorded in the coronary artery segments.
In both coronary arteries, coronary flow reserve was determined by intracoronary injection of 30 μg of adenosine and 12 mg of papaverine. Coronary flow reserve was calculated as the maximum average peak velocity (after adenosine or papaverine) divided by the baseline average peak velocity.
Quantitative calculation of local coronary blood flow
During the experimental procedure the average peak flow velocity was recorded on-line in both coronary vessels. When Doppler flow profiles were obtained, biplane coronary angiography was done in two orthograde projections. Vessel diameters were measured 5 mm distal to the tip of the wire. Vessel cross sectional area (CSA) and local coronary blood flow (CBF) were calculated as follows14,15: CSA = 0.25 × DRAO × DLAO × π, where DRAO, DLAO = mean vessel diameter in right anterior oblique and left anterior oblique projections; and CBF = CSA × 0.5 × average peak flow velocity.
Measurements of serum insulin, C peptide, and blood glucose
Before the catheterisation procedure and one hour after infusion of glibenclamide (0.05 mg/kg body weight), serum samples were obtained for analysis of insulin and C peptide. The samples were analysed by immunometric assay and competitive immunoassay (Immulite Insulin/Immulite C Peptide, Immulite Analyser, DPC Diagnostic Products Corp, Los Angeles, California, USA). The central 95% reference range for insulin was 6–27 mU/l, and for C peptide, 0.3–1.3 nmol/l.
Blood glucose was measured by bedside testing (Glucometer Elite 3906, Bayer Diagnostics, Munich, Germany) directly before the infusion of glibenclamide and then at 30 minute intervals.
The following drugs were used during the catheterisation procedure:
glibenclamide (HB 419) (lyophilisate for parenteral use), a gift from Hoechst AG, Frankfurt, Germany; the drug was dissolved in 15 ml sterile 0.9% NaCl directly before infusion
papaverine (Paveron, Linden, Heuchelheim, Germany)
adenosine (Adrekar, Sanofi Winthrop, Munich, Germany).
Four hours before the procedure, an intravenous infusion of 5% glucose at a rate of 60 ml/h was initiated and continued for 24 hours. After insertion of the femoral sheath catheter and intra-arterial administration of 140 IU heparin/kg body weight, biplane coronary angiography of the left coronary artery was performed for selection of optimal fluoroscopic projections. Under fluoroscopic control, one Doppler wire was placed in the left anterior descending coronary artery and a second was placed in the left circumflex coronary artery. In the coronary artery with the high grade stenosis, the wire was placed distal to the stenosis. After recording the baseline Doppler indices (three independent measurements over a period of three minutes) in both coronary arteries, coronary angiography was done in two orthogonal projections. Thereafter, coronary flow reserve was measured after an intracoronary injection of 30 μg adenosine and 12 mg papaverine.
After a second recording of baseline Doppler flow profiles (three measurements) with subsequent angiography, glibenclamide 0.05 mg/kg body weight was given intravenously over a period of 15 minutes. Thereafter, measurements of Doppler indices (three independent measurements) with subsequent coronary angiography were repeated, followed by recording of coronary flow reserve after adenosine and papaverine and a further recording of Doppler flow profiles with subsequent coronary angiography.
Values are given as mean (SD). For comparison of serum insulin and C peptide concentrations before and after glibenclamide, a paired Student's t test was used. Doppler measurements and haemodynamic and angiographic measurements at different time points were compared by analysis of variance for repeated measurements. Correlation between the grade of the stenosis and coronary flow reserve was determined by the Spearman correlation coefficient. A probability value of p < 0.05 was considered significant.
Twelve patients (mean (SD) age 54 (6) years) were included in the study. Ten patients had a high grade stenosis of the left anterior descending coronary artery and two had a high grade stenosis of the left circumflex coronary artery. Cineventriculographic analysis showed wall motion abnormalities in the corresponding myocardial territory in five patients (moderate hypokinesia in three, severe hypokinesia in two). No patient had a history of myocardial infarction. The ECG did not show Q waves in the corresponding leads. Demographic, clinical, and angiographic data on the patients are shown in table 1. None of the patients was diabetic and therefore none had ever been on treatment with sulfonylurea drugs.
Stability of the measured variables
The mean standard deviation of three independent measurements of the average peak velocity (over a three minute period at baseline) was 2.1 cm/s (range 0.6–4.1 cm/s). Before measurements of coronary flow reserve, recording of the Doppler profile in the post-stenotic artery revealed an average peak velocity of 13.4 (4.8) cm/s. After determination of coronary flow reserve in the post-stenotic artery, baseline average peak velocity was 13.3 (4.9) cm/s (NS v the first baseline recording). In the normal coronary arteries, average peak velocity before and after measurements of coronary flow reserve was also unchanged, at 18.7 (5.2) cm/s and 18.8 (5.2) cm/s, respectively. No intracoronary glyceryl trinitrate was given during the protocol. Stability patterns were similar after administration of glibenclamide.
Determination of vessel diameters before and after measurements of coronary flow using quantitative coronary angiography showed minimal variation in the measured variables—for example, the mean cross sectional area was 2.4 (0.7) mm2, as calculated from the first angiographic image before flow reserve measurements, and 2.4 (0.6) mm2 in the second image after determination of coronary flow reserve.
Coronary flow reserve after adenosine and papaverine
Coronary flow reserve after intracoronary administration of 30 μg adenosine was 1.2 (0.3) in the post-stenotic segment, while in the angiographically normal arteries it was significantly higher, at 2.6 (0.4) (p < 0.005). After administration of glibenclamide, mean coronary flow reserve did not change, remaining at 1.2 (0.2) in the post-stenotic segment and 2.6 (0.3) in the normal coronary artery (table 2). Adenosine induced coronary vasodilatation was not significantly altered after glibenclamide. Coronary flow reserve after administration of 12 mg papaverine tended to be higher than after adenosine induced hyperaemia. Coronary flow reserve determined by papaverine did not change before and after glibenclamide (table 2). There was a significant negative correlation between the degree of maximum diameter stenosis and coronary flow reserve after both adenosine and papaverine (Spearman correlation coefficients, −0.7711 for adenosine (p < 0.05) and −0.8840 for papaverine (p < 0.005)). Adenosine and papaverine cause coronary vasodilatation by different mechanisms, papaverine having a direct relaxing effect on vascular smooth muscle cells. Because of this, the coronary flow reserve was always higher after papaverine than after adenosine.
Doppler profiles, vessel diameters, and coronary angiography
An original Doppler profile, with its typical diastolic increase in flow, is shown in fig 2. In the post-stenotic artery average peak velocity and vessel size were smaller than in the normal coronary artery (table 3). Calculated cross sectional area of the post-stenotic artery was 58% of that in the “normal” coronary artery, as the measurements were performed in the arterial segment distal to the vessel narrowing. In neither the post-stenotic coronary artery nor the normal coronary artery was there a significant change in average peak velocity, cross sectional area, or calculated quantitative coronary blood flow before or after glibenclamide. Heart rate and systemic blood pressure remained constant during the experimental procedure and no significant changes were observed after administration of glibenclamide (table 3).
Serum insulin, C peptide, and blood glucose
Serum insulin increased significantly one hour after administration of glibenclamide, 0.05 mg/kg body weight (p < 0.005) (fig 3). C peptide concentrations also increased significantly over the same time period (fig 3). There was only a slight decrease in blood glucose during the procedure (fig 4), as 5% glucose was given continuously intravenously. Blood glucose concentrations were monitored for 24 hours. No patient had clinical symptoms of hypoglycaemia during follow up.
The main results of this invasive study in patients with coronary artery disease were as follows:
coronary blood flow before and after administration of 0.05 mg/kg glibenclamide was not significantly different in post-stenotic and angiographically normal coronary arteries
intravenous glibenclamide at a dose of 0.05 mg/kg body weight increased serum insulin and C peptide concentrations significantly
the amount of coronary vasodilatation induced by adenosine and papaverine was not significantly different before and after administration of glibenclamide
Coronary flow and KATP channels in animal studies
Recent studies on different animal models have provided a large body of knowledge about the mechanisms responsible for the regulation of coronary blood flow in different conditions, such as hypoxia, ischaemia, coronary autoregulation, and increased metabolic demand.
First described by Daut and colleagues in 1990,3 hypoxic coronary vasodilatation is mediated by opening of the vascular KATP channels in the isolated saline perfused guinea pig heart. Apart from the myogenic and flow induced vascular control of basal coronary tone, it has been suggested that KATP channels play an important role in maintaining basal coronary flow.5,7,12,16 A significant proportion of the vasodilator action of adenosine is also mediated by the activation of KATP channels in different models.3,17 As shown by Duncker and colleagues,6,18 synergism between endogenous adenosine and opening of the KATP channels is important in maintaining coronary flow during exercise both in normal hearts and distal to an artificial coronary stenosis in dogs. Studies of the contribution to coronary flow of KATP channel activation, adenosine, and nitric oxide suggest the nearly complete suppression of nitric oxide production in hypoxia but an important role for KATP channels and adenosine in determining coronary tone during the different phases of hypoxia.4,19 In dogs, adenosine receptor blockade and nitric oxide synthase inhibition do not impair the normal increase in coronary flow during exercise. However, following blockade of the KATP channel system, both adenosine and nitric oxide seemed to contribute to the exercise induced increase in blood flow.20 Other studies have shown that the regulation of coronary tone during metabolic stimulation, pacing, and exercise depends mainly on nitric oxide and not on KATP channel activation.21–23
In the present study, coronary blood flow in the angiographically normal coronary artery can be compared with investigations on basal coronary tone in animal studies. However, a simple transfer of animal models to a clinical situation in human patients is not feasible. Several animal studies have been performed on isolated hearts perfused with saline at constant pressure. In such studies, the role of the KATP channel in the coronary circulation might not be completely comparable because of perfusion pulsatility, the effects of the type of perfusion medium, denervation of the heart, and alterations of neurohumoral interactions. As recently shown in a dog model, glibenclamide lowered regional coronary flow under basal conditions, but with higher perfusion pulsatility no reduction in regional coronary flow was observed during glibenclamide administration.24
Post-stenotic coronary flow in human patients represents a situation of impaired oxygen supply, but might be best described as a chronic adaptation to a reduced perfusion pressure. In a closed chest model in dogs, coronary autoregulation (defined as the ability to maintain constant flow under conditions of changing perfusion pressure) was not affected by glibenclamide in a range between 60–100 mm Hg.25 Animal models of pure hypoxia and increased metabolic stimulation differ from the clinical setting of chronic coronary stenosis with respect to the regulation of coronary flow.
Vascular KATP channels in humans
Despite intense animal research, the role of vascular KATP channels in the human coronary circulation is still under debate. The forearm vasodilator response to diazoxide (a KATP channel opener)—as measured by venous occlusion plethysmography in healthy volunteers—was inhibited by glibenclamide.26 It was concluded that therapeutic concentrations of glibenclamide result in a significant blockade of vascular KATP channels in humans. Similarly, the vasodilator response to ischaemia in the forearm was reduced by glibenclamide.11 However, there is only limited information about the role of KATP channels in the human coronary circulation. In a combined angiographic and intracoronary Doppler study, post-stenotic blood flow was increased after intracoronary application of nicorandil, a KATP channel opener with additional effects similar to those of nitrates.27
In the present study, the influence of a therapeutic dose of glibenclamide on coronary flow was investigated in patients with coronary artery disease. Measurements of serum insulin and C peptide concentrations after administration of glibenclamide revealed a pronounced effect on insulin release, suggesting an inhibitory effect on pancreatic KATP channels which results in enhanced exocytic insulin release—as expected from the pharmakokinetic data on glibenclamide.28,29 Concentrations of sulfonylurea required to block vascular KATP channels may be greater than those effective at the pancreatic β cells.30 Therapeutic doses of glibenclamide possibly inhibit a fraction of KATP channels in coronary smooth muscle cells and might impair hypoxic coronary vasodilatation, as suggested by the apparent dissociation constants of glibenclamide at vascular and pancreatic KATP channels in guinea pigs.5 Our present study suggests that coronary basal tone in angiographically normal coronary arteries, post-stenotic coronary flow at rest, and the hyperaemic response to adenosine are not significantly altered after therapeutic doses of glibenclamide.
Coronary flow assessed by Doppler flow profiles and quantitative angiography
Determination of coronary flow characteristics using intracoronary Doppler measurements of coronary flow velocity in combination with quantitative angiographic analysis is a well established method of investigating coronary blood flow in vivo. Validation studies have shown a high correlation in experimental animal studies14 between 133Xe perfusion studies,31,15O-H2O myocardial positron emission tomography,32 and continuity equation calculations.33 The estimated mean blood flow velocity was calculated as 0.5 × average peak velocity by assuming a time averaged parabolic velocity profile across the vessel.15 Using a combination of Doppler flow measurements and cross sectional intravascular ultrasound imaging, other groups have suggested a better correlation with a calculated mean flow velocity of 0.47 × average peak velocity.34 Assessment of coronary flow reserve from average peak velocity after papaverine and adenosine is a well established method, but it might underestimate the hyperaemic response, as changes of vessel diameters are not taken into account by this method.35,36
Sulfonylurea drugs and coronary artery disease
The potential risk of sulfonylurea drug treatment in diabetic patients with coronary artery disease is still under debate. In a prospective trial in 3867 newly diagnosed diabetic patients, there was no difference in the clinical outcome after 10 years between sulfonylurea treatment and insulin treatment.37 On the other hand, some clinical studies have suggested that there is an increased cardiovascular mortality in patients on sulfonylurea drugs.8,9,38 These effects have mainly been attributed to the inhibitory actions of sulfonylurea drugs on preconditioning-like cardioprotective effects. In the human model of ischaemic preconditioning during angioplasty, the degree of ST segment shift during balloon inflation is reduced during the second balloon inflation compared with the first. This preconditioning-like effect is prevented by glibenclamide and thus could be an expression of the same effect that is responsible for the increased cardiovascular mortality in sulfonylurea drug treatment.39,40
A deterioration in coronary blood supply is an alternative explanation but has not so far been investigated in humans. In our present study no significant impairment of coronary blood flow was observed after administration of a therapeutic dose of glibenclamide. Nevertheless, significant effects on coronary haemodynamics and collateral circulation under conditions of exercise or electrical pacing, acute myocardial ischaemia, left ventricular hypertrophy, or in diabetic microvascular disease with impairment of nitric oxide mechanisms cannot be excluded. The results of the present study do not necessarily apply to diabetic patients, to patients with acute myocardial ischaemia, or in particular to patients with microvascular disease. The coronary flow reserve in the angiographically normal artery was above the level of 2.0 in all patients, suggesting no significant microvascular disease in this population, and no patient was diabetic. These are major limitations of the study that warrant further investigation.
Alternatives to sulfonylurea drugs should be considered carefully in the treatment of diabetic patients with coronary artery disease, despite there being no deterioration in coronary flow at rest under the influence of glibenclamide in this population.1,30