Vasospasm is an important concern in coronary artery

Reversal of Preexisting Vasospasm in Coronary Artery Conduits Jyotirmay Chanda, MD, PhD, and Charles C. Canver, MD Division of Cardiothoracic Surgery, Albany Medical College, Albany, New York Background. To reverse preexisting coronary graft spasm, we investigated the vasodilative effect of the average therapeutic plasma concentration of nitroglycerin () alone and various calcium antagonists in combination with in human arterial and venous conduits. Methods. Vasodilative effects of 2 10 8 mol/l alone and 10 8 mol/l in combination with 2.2 10 7 mol/l diltiazem, 2.8 10 7 mol/l nifedipine, 10 7 mol/l verapamil, or 5.6 10 8 mol/l nicardipine were assessed in human radial artery, internal thoracic artery, and saphenous vein segments precontracted with a mixture of ten times the maximum plasma concentrations of endothelin-1 (8.6 10 13 mol/l), angiotensin II (36 10 11 mol/l), 5-hydroxytryptamine (3.4 10 7 mol/l), and norepinephrine (1.7 10 8 mol/l). The studies were done in organ baths. Results. The therapeutic concentration of alone or nifedipine, verapamil, diltiazem, or nicardipine in combination with caused equal relaxation in a particular group of vascular segments (average vasodilation: radial artery, 83% to 95% [p 0.7608 by analysis of variance]; saphenous vein, 47% to 70% [p 0.3142]; internal thoracic artery, 54% to 79% [p 0.2783]). These combinations were not equally effective when compared between different groups of vascular segments (vasodilation; radial artery > internal thoracic artery > saphenous vein [p < 0.0001 by analysis of variance]). Although not significant, in comparison with alone, in combination with a calcium antagonist caused less vasodilation in any group of vascular segments. Conclusions. Nitroglycerin alone or in combination with nifedipine, verapamil, diltiazem, or nicardipine effectively reverses preexisting vasospasm in coronary artery conduits. (Ann Thorac Surg 2001;72:476 80) 2001 by The Society of Thoracic Surgeons Vasospasm is an important concern in coronary artery bypass grafting (CABG) operations. This phenomenon has led to premature constriction of graft conduits perioperatively and postoperatively. The proposed mechanisms of vasospasm have been elucidated somewhat and include mechanical factors such as surgical manipulation and trauma as well as biochemical and molecular factors such as increased vasoconstrictor receptor expression and vascular smooth muscle hyperactivity [1, 2]. The internal thoracic artery (ITA), saphenous vein (SV), and radial artery (RA) are three conduits that are commonly chosen as grafts in coronary bypass procedures. The development of ITA grafting is the most remarkable achievement in CABG surgery. In the past two decades, numerous studies have confirmed that the patency rate of ITA grafts is excellent and that they result in an improvement in survival of 10% to 30% and greater freedom from major cardiac-related events compared with the rates in patients whose CABG was performed with vein grafts only [3]. However, the presence of vascular smooth muscle in the ITA, as in other arteries, Accepted for publication April 25, 2001. Address reprint requests to Dr Canver, Division of Cardiothoracic Surgery, Mail Code 55, Albany Medical College, 47 New Scotland Ave, Albany, NY 12208; e-mail: canverc@mail.amc.edu. makes this vessel susceptible to vasospasm during surgical manipulation [1, 2]. With demonstration of the dramatic benefits obtainable using SV grafting came recognition of early failure of the operation because of the accelerated stenosis of grafted SV conduits. The relative absence of smooth muscle in the tunica media of the SV makes it more resistant to vasospasms than either the ITA or the RA [2]. Nonetheless, the SV lacks the arterial structure of the ITA and in the long run, succumbs to high coronary arterial pressures resulting in occlusion owing to neointimal formation and smooth muscle hyperplasia [4]. The RA, like other arterial conduits, is susceptible to vasospasm during harvesting and because its media is heavily invested with vascular smooth muscle. This vasospasm is more intense and is often difficult to reverse [1, 2]. Previous studies have shown that the RA has a greater contractile response to various vasospasm mediators than does the ITA [5]. This is especially true during surgical manipulation of the vessel prior to anastomosis. However, despite this elevated vasospastic response, the use of the RA as a graft conduit in patients undergoing CABG has provided good long-term patency and survival [6]. Although a fair amount is known about the etiology and the treatment of vasospasm, the comparative efficacy of vasospasm treatment regimens in frequently used vascular conduits such as the RA, ITA, and SV has not been well established to date. 2001 by The Society of Thoracic Surgeons 0003-4975/01/$20.00 Published by Elsevier Science Inc PII S0003-4975(01)02799-0

Ann Thorac Surg CHANDA AND CANVER 2001;72:476 80 CORONARY ARTERY CONDUIT VASODILATION 477 In a recent study, we [7] demonstrated that the vasodilative effect of a calcium antagonist (CA) in combination with nitroglycerin () is superior to either the CA alone or alone in an RA undergoing vasospasm in vitro. In this present study, to simulate perioperative vasospasm in vitro and account for the lack in the in vitro system of other potential vasoconstrictors not involved in the experiments, vascular segments were precontracted with a mixture of ten times the maximal human plasma concentrations of norepinephrine (NE), angiotensin II (AII), endothelin-1 (ET-1), and 5-hydroxytryptamine (5- HT). A therapeutic plasma concentration of alone or a calcium-channel antagonist in combination with was used to reverse the RA, ITA, and SV vasospasm induced in vitro. Material and Methods Discarded segments of RA, ITA, and SV were collected from 21, 25, and 23 patients, respectively, undergoing CABG. Permission to use discarded human RA, ITA, and SV segments was obtained from the Institutional Review Board of Albany Medical College. Experimental Protocol After harvest, the RA and SV were flushed with papaverine hydrochloride solution (60 mg of papaverine in 18 ml of normal saline solution) and preserved in a solution containing 250 ml of normal saline solution in combination with 60 mg of papaverine and 500 units of heparin sodium. After completion of the distal anastomosis, either RA or SV conduit was flushed with a solution of 50% papaverine and 50% heparinized blood. The leftover segment of artery or vein was trimmed off for sizing of the proximal anastomosis and was placed in a solution containing 250 ml of normal saline solution in combination with 60 mg of papaverine and 500 units of heparin until transfer to a container with cold physiologic salt solution (140 mmol/l NaCl, 4.7 mmol/l KCl, 1.2 mmol/l NaH 2 PO 4, 2.0 mmol/l 3-[N-Morpholino]propanesulfonic acid (MOPS), 0.02 mmol/l Na 2 EDTA [disodium ethylenediaminetetraacetic acid] 1.6 mmol/l CaCl 2, 1.2 mmol/l MgSO 4, 1 g/l dextrose, ph 7.4) and taken to the laboratory. The vessels were placed in a dish containing physiologic salt solution, dissected out from their surrounding connective tissue, and cut into rings 3 to 4 mm long. The number of rings from each patient ranged from eight to 12. Vessel rings were mounted on wire hooks (Radnoti Tissue Support; Radnoti Glass Technology Inc, Monrovia, CA) with one end connected to an isometric force-displacement transducer (FT03; Grass Instrument Company, Quincy, MA) and then suspended in physiologic salt solution in water-jacketed organ baths (Radnoti). The force-displacement transducer was linked to a bridge amplifier (Gould Instruments SAF, Ballainvilliers, France) and a thermal recorder (Gould) to give a continuous record of tension in the vessel segment. The solution in the organ baths was aerated with 100% oxygen at 37 C. Every effort was made not to damage the endothelium while dissecting and mounting the vascular rings [7]. Vascular segments (RA, 158 rings; ITA, 130 rings; SV, 196 rings) were progressively stretched and allowed to equilibrate for 90 to 120 minutes. Vasodilative effects of the average therapeutic plasma concentration of (2 10 8 mol/l) (American Regent Laboratories, Inc, Shirley, NY) alone or half of that concentration (10 8 mol/l) in combination with the average plasma concentration of hydrochloride diltiazem (Dil) (2.2 10 7 mol/l) (Abbott Laboratories, North Chicago, IL), nifedipine (Nif) (2.8 10 7 mol/l) (Sigma Chemical Co, St. Louis, MO), verapamil hydrochloride (Vp) (10 7 mol/l) (American Regent Laboratories, Inc), or nicardipine hydrochloride (Nic) (5.6 10 8 mol/l) (Wyeth-Ayerst Laboratories, Inc, St. Davids, PA) were assessed in RA, ITA, and SV segments precontracted with a mixture of ten times the maximum plasma concentrations of NE (1.7 10 8 mol/l) (Abbott Laboratories), AII (36 10 11 mol/l (Sigma), ET-1 (8.6 10 13 mol/l) (American Peptide Co, Sunnyvale, CA), and 5-HT (3.4 10 7 mol/l) (Sigma). The maximum human plasma concentrations of NE [8], AII [9], ET-1 [10], and 5-HT [11] reported to date are 1.7 10 9 mol/l (36 10 11 mol/l, 8.6 10 14 mol/l, and 3.4 10 8 mol/l, respectively. The average therapeutic plasma concentrations of [12], Nif [13], Vp [13], Dil [13], and Nic [14] are 2 10 8 mol/l, 2.8 10 7 mol/l, 10 7 mol/l, 2.2 10 7 mol/l, and 5.6 10 8 mol/l, respectively. Statistical Analysis The data were reported as the mean the standard deviation of the mean. Values were compared by repeated-measures analysis of variance complex compact variable techniques (StatView, version 5.0.1; SAS Institute Inc, San Francisco, CA). Multiple comparisons were carried out with Fisher s protected least significant difference test. A p value of less than 0.05 was considered significant. Results Ten times the maximum plasma concentrations of ET-1 AII 5-HT NE caused contractions of 5.74 3.4 g (median, 5 g), 3.06 1.6 g (median, 2.8 g), and 2.3 2.6 g (median, 1.4 g) in RA, ITA, and SV, respectively (Fig 1). The vasodilative effect of 2 10 8 mol/l alone or half of this concentration (10 8 mol/l) in combination with a CA in RA, ITA, and SV segments precontracted with a mixture of ten times the maximum plasma concentrations of ET-1 AII 5-HT NE is shown in Figure 2. More than 90% vasodilation was achieved with in combination with a CA in RA segments (see Fig 2). All combinations of a CA with or alone caused equal relaxation in a particular group of vascular segments (RA, p 0.7608 by one-way analysis of variance; SV, p 0.3142; ITA, p 0.2783) (Table 1). These combinations were not equally effective when compared between different groups of vascular segments (vasodilation: RA ITA SV; p 0.0001 by analysis of

478 CHANDA AND CANVER Ann Thorac Surg CORONARY ARTERY CONDUIT VASODILATION 2001;72:476 80 Table 1. Average Vasodilative Effects of Nitroglycerin Alone or in Combination With Calcium Antagonist a c Vessel Vasodilator Vasodilation (%) p Value Fig 1. Contraction of human radial artery (RA), internal thoracic artery (ITA), and saphenous vein (SV) segments to a mixture of ten times the maximum plasma concentrations of endothelin-1 (8.6 10 13 mol/l), angiotensin II (36 10 11 mol/l), 5-hydroxytryptamine (36 10 11 mol/l), and norepinephrine (1.7 10 8 mol/l). Data are shown as mean standard error. variance). The vasodilative effect of some combinations of a CA with in RA and ITA segments was significantly higher compared with that of some combinations of a CA with in SV and ITA sections (Table 2). Comment Fig 2. Vasodilative effects of 2 10 8 mol/l nitroglycerin () alone and of 10 8 mol/l in combination with 2.2 10 7 mol/l diltiazem (Dil ), 5.6 10 8 mol/l nicardipine (Nic ), 2.8 10 7 mol/l nifedipine (Nif ), or 10 7 mol/l verapamil (Vp ) in human radial artery (RA), internal thoracic artery (ITA), and saphenous vein (SV) segments precontracted with a mixture of ten times the maximum plasma concentrations of endothelin-1 (8.6 10 13 mol/l), angiotensin II (36 10 11 mol/l), 5-hydroxytryptamine (3.4 10 7 mol/l), and norepinephrine (1.7 10 8 mol/l). Data are shown as mean standard error. ITA Dil 76 33.2 Nic 70 28.5 Nif 79 33 0.2783 Vp 72 33 53.6 7.1 SV Dil 70 28.7 Nic 52.3 25 Nif 64.5 26.5 0.3142 Vp 63.4 30.7 47.3 20.9 RA Dil 92.4 28.5 Nic 95.2 29.6 Nif 91.6 30 0.7608 Vp 93.7 35 82.7 10.2 a The following agents and concentrations were used: nitroglycerin alone (2 10 8 mol/l) or nitroglycerin (10 8 mol/l) in combination with diltiazem (2.2 10 7 mol/l), nicardipine (5.6 10 8 mol/l), nifedipine (2.8 10 7 mol/l), or verapamil (10 7 mol/l) in human internal thoracic artery saphenous vein, and radial artery segments precontracted with a mixture of ten times the maximum plasma concentrations of endothelin-1 (8.6 10 13 mol/l), angiotensin II (36 10 11 mol/l), 5-hydroxytryptamine (3.4 10 7 mol/l), and norepinephrine (1.7 10 8 mol/ L). b Data are shown as the mean the standard deviation. c A p value of less than 0.05 by one-way analysis of variance was significant. Dil diltiazem; ITA internal thoracic artery; Nic nicardipine; Nif nifedipine; nitroglycerin; RA radial artery; SV saphenous vein; Vp verapamil. The findings in this study demonstrate that the therapeutic plasma concentration of alone or that of Nif, Dil, Vp, or Nic in combination with half the therapeutic concentration of is equally effective in relaxing RA, ITA, and SV undergoing vasospasm in vitro. In addition, the reversal of vasospasm in precontracted vascular segments using this treatment combination was most effective in the RA, and the overall efficacy when compared across all three vessel types was, in order of decreasing efficacy, RA, ITA, and SV. Various vasoactive substances are implicated in the genesis of vasospasm including endothelium-derived contracting factors such as ET-1 and thromboxane A 2 [15], circulating sympathomimetic substances such as epinephrine and norepinephrine [16], platelet-derived contracting substances such as 5-HT [17], renin-angiotensin system related substances like AII [18], and arachidonic acid metabolites like leukotrienes C4, D4, and E4 [19]. For the purpose of this study, the conduit vasospasm that can occur postoperatively in patients undergoing CABG was simulated in vitro using ten times the maximum plasma concentrations of ET-1, AII, 5-HT, and NE. In our opinion, such concentrations produce the maximum feasible constriction of grafted vessels. The purpose of doing this was twofold: using such high concentrations eliminated the normal variability in vasospasm mediator concentration that exists in individuals, and higher concentrations of these mediators accounted for the lack in an in vitro system of other possible vasoconstrictors that normally may be involved in the genesis of a vasospasm. Such mediators include thromboxane A 2, epinephrine, and leukotrienes C4, D4, and E4. The maximal vasospasm induced in this study was presumably ideal for the assessment of the vasodilative

Ann Thorac Surg CHANDA AND CANVER 2001;72:476 80 CORONARY ARTERY CONDUIT VASODILATION 479 Table 2. Significant Differences in Average Vasodilative Effects of Nitroglycerin Alone and in Combination With Various Calcium Antagonists a,b Significant p Values SV ITA Arterial Segment Dil Nic Nif Vp Dil Nic Nif Vp RA Dil 0.0473 0.0005 0.0142 0.007 0.0001... 0.0362...... 0.0003 Nic 0.0307 0.0003 0.009 0.0043 0.0001... 0.023... 0.0328 0.0002 Nif... 0.0009 0.021 0.0114 0.0002............... Vp 0.0379 0.0004 0.0111 0.0054 0.0001... 0.0286... 0.0407 0.0003... 0.0109...... 0.0031............ 0.0101 ITA Dil... 0.0461...... 0.0159... 0.0467......... Nic.............................. Nif... 0.0271...... 0.0088............ 0.0267 Vp............ 0.0433............................................. a The following agents and concentrations were used: nitroglycerin alone (2 10 8 mol/l) or nitroglycerin (10 8 mol/l) in combination with diltiazem (2.2 10 7 mol/l), nicardipine (5.6 10 8 mol/l), nifedipine (2.8 10 7 mol/l), or verapamil (10 7 mol/l) in human radial artery, internal thoracic artery, and saphenous vein segments precontracted with a mixture of ten times the maximum plasma concentrations of endothelin-1 (8.6 10 13 mol/l), angiotensin II (36 10 11 mol/l), 5-hydroxytryptamine (3.4 10 7 mol/l), and norepinephrine (1.7 10 8 mol/l). b Multiple comparisons were carried out with Fisher s protected least significant difference test. A p value of less than 0.05 by one-way analysis of variance was significant. Dil diltiazem; ITA internal thoracic artery; Nic nicardipine; Nif nifedipine; nitroglycerin; RA radial artery; SV saphenous vein; Vp verapamil. efficacy of a CA in combination with in counteracting it. Similar to other arterial conduits, the RA is susceptible to spasm during harvesting. Because the media of the RA is substantially thicker than the media of other arterial conduits [20], the spasm is more intense and more difficult to reverse. The RA has a greater contractile response to norepinephrine and serotonin than the ITA [5]. Serotonin contracts vascular smooth muscle directly through 5-HT 2 receptors. When endothelium is lost and perhaps also when it is damaged, platelets aggregate in the area where endothelium is denuded and release substances such as 5-HT that strongly contract smooth muscle. 5-Hydroxytryptamine has been suggested to be an important spasmogen in coronary artery spasm even when endothelium is present [21]. Pharmacological treatment of SV with a solution containing and Vp during harvesting allows the use of a lower distention pressure, reduces the breakdown of high-energy phosphates, and improves endothelial coverage in the vein wall [22]. With the introduction of CAs, vasospasm of the RA is less frequently encountered, and the early results have been markedly improved [6]. Complete vasodilation with 3 10 5 mol/l Vp in combination with 3 10 5 mol/l in RA precontracted with 25 mmol/l KCl has been reported [23]. Diltiazem, a calcium-channel blocker, has been empirically selected by most surgeons to minimize RA vasospasm after CABG [6]. However, to date, Dil, Nif, or Vp alone is not capable of eliminating postoperative RA spasm [7]. The RA, though providing the strongest vasospasm, also proved to be the most easily treated conduit when using CAs in combination with. It should be noted, however, that the combination of Nif and provided a median level of relaxation of 100% in the ITA, whereas the other CAs when used in combination with provided much lower levels of relaxation, thereby contributing to the increased level of residual spasm in this conduit. The reason for the increased efficacy of Nif over other CAs in the case of the ITA is unknown. Finally, the average level of residual spasm in the SV was 30.1% of the level of vasospasm before dilation. This implies that the SV was least susceptible to treatment of vasospasm using CAs with and is consistent with the notion that the SV is least susceptible to vasospasm and vasorelaxation because of the relative absence of vascular smooth muscle in its histological structure. The average plasma concentration of [12], Nif [13], Vp [13], Dil [13], and Nic [14] is about 2 10 8 mol/l, 2.8 10 7 mol/l, 10 7 mol/l, 2.2 10 7 mol/l, and 5.6 10 8 mol/l, respectively, after administration of the following:, 15 to 20 g/min intravenously; Nif, 10 to 20 mg every 8 hours orally or 5 to 15 g kg 1 min 1 intravenously; Up, 80 to 160 mg every 8 hours orally or 150 g kg 1 min 1 intravenously; Dil, 60 to 90 mg every 8 hours orally or 75 to 150 g kg 1 min 1 intravenously; and NR, 1 g kg 1 min 1, intravenously. The concentration of used in combination with the CA (10 8 mol/l) was arbitrarily chosen and was half the therapeutic plasma concentration (about 2 10 8 mol/l). Although not significant, alone is less effective in reversing vasospasm than in combination with a CA. This study simulates the setting of existing spasm in coronary bypass grafts and evaluates the efficacy of the therapeutic combination to reverse that spasm. However, the study does not elucidate the preventive effect of the proposed therapeutic agents on prophylaxis of coronary graft spasm. It is known that the hemodynamic and anti-ischemic efficacy of organic nitrates is rapidly blunted because of the development of nitrate toler-

480 CHANDA AND CANVER Ann Thorac Surg CORONARY ARTERY CONDUIT VASODILATION 2001;72:476 80 ance, and long-term treatment has been shown to be associated with cross tolerance to endotheliumdependent vasodilators [24]. As it is difficult to demonstrate the development of tolerance in this setup, we have investigated the efficacy of a half dose of in combination with a full dose of a CA as a substitute to a full dose of alone to reverse the vasospasm. However, this might not be the choice of treatment when the risk of hypotension and rhythm disturbances would overweigh the risk of graft spasm. Even though the RA is vulnerable to the most intense vasospasm of all vessels in this study, it is the most efficaciously treated using any CA in this study in combination with, thus virtually eliminating any worry of vasospastic activity. Nevertheless, it should be noted that other vasodilative agents may provide different results in the RA and may, in fact, lead to similar levels of vasorelaxation after vasospasm in the SV and the ITA as were observed in the RA in this study. Also, the effect of a CA with in counteracting the effects of other mediators involved in coronary artery graft vasospasm, though accounted for in this study, remains unknown. We are grateful for the help of John M. Kelly, MD, and Linda Altieri, RN. References 1. Rosenfeldt FL, He G-W, Buxton BF, Angus JA. Pharmacology of coronary artery bypass grafts. Ann Thorac Surg 1999;67: 878 88. 2. Reardon MJ, Conklin LD, Reardon PR, Baldwin JC. Coronary artery bypass conduits: review of current status. J Cardiovasc Surg (Torino) 1997;38:201 9. 3. Loop FD. Internal-thoracic-artery grafts. Biologically better coronary arteries. N Engl J Med 1996;334:263 5. 4. Motwani JG, Topol EJ. Aortocoronary saphenous vein graft disease: pathogenesis, predisposition, and prevention. Circulation 1998;97:916 31. 5. Chardigny C, Jebara VA, Acar C, et al. Vasoreactivity of the radial artery. Comparison with the internal mammary and gastroepiploic arteries with implications for coronary artery surgery. Circulation 1993;88(Suppl 2):115 27. 6. Acar C, Jebara V, Pagny JY, et al. Five-year results of coronary artery bypass grafting using the radial artery. J Thorac Cardiovasc Surg 1998;116:981 9. 7. Chanda J, Brichkov I, Canver CC. Prevention of radial artery graft vasospasm after coronary bypass. Ann Thorac Surg 2000;70:2070 4. 8. McCance AJ, Forfar JC. Plasma noradrenaline as an index of sympathetic tone in coronary arterial disease: the confounding influence of clearance of noradrenaline. Int J Cardiol 1990;26:335 42. 9. Schulz O, Mitrovic V, Schonburg M, Thormann J. High-dose enoximone to evaluate reversibility of pulmonary hypertension: is there a diagnostic value of neurohormonal measurements? Am Heart J 1999;137:887 94. 10. Hülsmann M, Stanek B, Frey B, et al. Value of cardiopulmonary exercise testing and big endothelin plasma levels to predict short-term prognosis of patients with chronic heart failure. J Am Coll Cardiol 1998;32:1695 700. 11. Malyszko J, Urano T, Knofler R, et al. Daily variations of platelet aggregation in relation to blood and plasma serotonin in diabetes. Thromb Res 1994;75:569 76. 12. Armstrong PW, Armstrong JA, Marks GS. Pharmacokinetichemodynamic studies of intravenous nitroglycerin in congestive cardiac failure. Circulation 1980;62:160 6. 13. Henry PD. Comparative pharmacology of calcium antagonists: nifedipine, verapamil and diltiazem. Am J Cardiol 1980;46:1047 58. 14. Saadjian A, Philip-Joet F, Paganelli F, Saadjian M, Levy S. Acute effects of nicardipine on the vascular reactivity of oxygen in patients with respiratory insufficiency and pulmonary hypertension. Arch Mal Coeur Vaiss 1993;86:899 906. 15. Noll G, Luscher TF. The endothelium in acute coronary syndromes. Eur Heart J 1998;19:C30 8. 16. Robertson RM, Bernard Y, Robertson D. Arterial and coronary sinus catecholamines in the course of spontaneous coronary artery spasm. Am Heart J 1983;105:901 6. 17. Nilsson T, Longmore J, Shaw D, et al. Characterisation of 5-HT receptors in human coronary arteries by molecular and pharmacological techniques. Eur J Pharmacol 1999;372: 49 56. 18. Romero JC, Reckelhoff JF. State-of-the-Art lecture. Role of angiotensin and oxidative stress in essential hypertension. Hypertension 1999;34:943 9. 19. Takase B, Maruyama T, Kurita A, et al. Arachidonic acid metabolites in acute myocardial infarction. Angiology 1996; 47:649 61. 20. Van Son JA, Smedts F, Vincent JG, Van Lier HJ, Kubat K. Comparative anatomic studies of various arterial conduits for myocardial revascularization. J Thorac Cardiovasc Surg 1990;99:703 7. 21. Fukai T, Egashira K, Hata H, et al. Serotonin-induced coronary spasm in a swine model. A minor role of defective endothelium-derived relaxing factor. Circulation 1993;88: 1922 30. 22. Roubos N, Rosenfeldt FL, Richards SM, Conyers RA, Davis BB. Improved preservation of saphenous vein grafts by the use of glyceryl trinitrate verapamil solution during harvesting. Circulation 1995;92(Suppl 2):31 6. 23. He G-W. Verapamil plus nitroglycerin solution maximally preserves endothelial function of the radial artery: comparison with papaverine solution. J Thorac Cardiovasc Surg 1998;115:1321 7. 24. Münzel T, Li H, Mollnau H, et al. Effects of long-term nitroglycerin treatment on endothelial nitric oxide synthase (NOS III) gene expression, NOS III mediated superoxide production, and vascular NO bioavailability. Circ Res 2000; 86:7 12.