The purpose of the current investigation was to determine and compare the direct effect of isoflurane on normal resistance coronary arteries and similarly sized coronary arteries in collateral-dependent (CD) circulation. A differential vasomotor effect of isoflurane might contribute to flow redistribution, either adverse or favorable, between normal and CD regions.
The authors used a swine model of chronic coronary occlusion. Six weeks after ameroid occluder placement around the left circumflex artery, the heart was removed. Myocardial blood flow, measured in vivo before the removal, was lower during rapid atrial pacing in the CD region than in the normally perfused region. These in vivo studies established the existence and location of CD circulation. Subepicardial microvessels, measuring approximately 100 microns, were dissected from both the CD region and the normal region. Either with or without preconstriction of the vessels with the thromboxane analogue U46619 1 microM, direct vasomotor effect of isoflurane was examined in vitro by video microscopy. In addition, vasomotor responses to endothelin-1, the endothelium-dependent dilator adenosine 5' diphosphate, and the endothelium-independent dilator sodium nitroprusside were compared between CD and normal resistance vessels.
Isoflurane caused greater concentration-dependent constriction of the normal vessels than of the CD vessels (P < 0.05); this constrictive effect was masked by preconstriction of the vessels. Vasodilation of ADP was less in CD vessels than in normal vessels (P < 0.001), but vasodilation to sodium nitroprusside was not significantly different (P = 0.45). Vasoconstriction to endothelin-1 was greater in CD vessels than in normal vessels (P < 0.01).
The vasoconstrictive effect of isoflurane normally observed in coronary resistance vessels is impaired in resistance vessels that supply a CD area. The basis of this may be related to endothelial dysfunction.
Key words: Anesthetics, volatile: isoflurane. Coronary circulation: collateral-dependent circulation; flow distribution.
REIZ et al. and Buffington et al. suggested that isoflurane may cause adverse redistribution of coronary flow away from collateral-dependent (CD) circulation, resulting in myocardial ischemia (i.e., coronary steal). Such redistribution of flow would be dependent on a differential coronary vasomotor response to isoflurane of normal and CD circulation. However, no information is available on the direct effect of this volatile anesthetic on the resistance vessels in CD circulation.
In the current investigation, we used a swine model of chronic coronary occlusion to address this question. Previous studies, using similar models, demonstrated reduced vasodilation of CD coronary microvessels in response to beta-adrenergic agonists or endothelium-dependent agonists. In addition, the CD vessels may lack functioning alpha-adrenergic receptors [5,6]and demonstrate a hypercontractile response to vasopressin. Enhanced response to vasopressin may be based on endothelial dysfunction (i.e., reduced vasopressin-mediated release of nitric oxide from the endothelium) that tends to modulate vasopressin-mediated constriction. We performed an in vitro investigation of the direct vasomotor effect of isoflurane on both the CD and the normal coronary microcirculation.
Preparation of the Animal
In accordance with the standards of and with approval of the institutional animal care committee, Yorkshire pigs of either gender, weighing 10-15 kg, were anesthetized with 10 mg/kg intramuscular ketamine and halothane and prepared for sterile surgery. The heart was approached via left thoracotomy, and an ameroid constrictor ring (2.77 or 3.0 mm internal diameter) was placed around the proximal left circumflex artery (LCx). The incision was then closed and the animal allowed to recover. Eight pigs were studied 6 weeks after surgery, when they were afebrile and fully active. This porcine model of collateral development was shown to be associated with near normal perfusion in the CD region at rest, but reduced perfusion during exercise if infarction had not occurred. 
Six weeks after ameroid ring placement, each pig was anesthetized with 10 mg/kg intramuscular ketamine, followed with intravenous alpha-chloralose and urethan (3.2 g alpha-chloralose and 20 g urethan in 100 ml normal saline; 60 ml initially and 10-15 ml every 45-60 min as needed). The trachea was intubated and the lungs mechanically ventilated with a Harvard pediatric ventilator (Harvard Apparatus, Cambridge, MA). The right femoral artery was dissected and a 7-F introducing sheath was placed. A 7-F JR4 diagnostic angiography catheter (Cordis, Miami, FL) was introduced over a 0.035-inch J wire. Selective angiography was performed in multiple left and right anterior oblique projections with ionic contrast media (Renograffin, Squibb Diagnostics, Princeton, NJ). Complete occlusion of LCx was confirmed in seven of the eight pigs by coronary angiography, and only these seven were studied further. The region distal to the occluded LCx was, therefore, completely CD at the time of the reported observations.
A median sternotomy was performed. After measurement of myocardial blood flow (see below), the heart was excised rapidly, and selected tissue samples were placed in cold, modified Krebs buffer solution. 
Myocardial Blood Flow
Colored polystyrene microspheres (CPM; size 15 plus/minus 0.1 micro meter [mean plus/minus SD]; Triton Technology, San Diego, CA) were injected into the left atrium during resting conditions and during rapid atrial pacing (140-160 beats per minute, or approximately 1.5-1.8 times resting heart rate) by the method developed and validated by Kowallik et al. A suitable number of CPM, 1,800,000 for red CPM and 7,000,000 for white, yellow, and blue CPM, preselected to account for their different absorbent characteristics, was diluted with saline (containing 0.02% polysorbate-80) to a total volume of 3 ml and ultrasonicated for 1 min immediately before injection. Colored polystyrene microspheres were injected through a catheter into the left atrium, followed by a saline flush. Reference blood samples were drawn from a femoral arterial cannula starting 10 s before injection of CPM and continuing for 100 s at a rate of 4 ml/min. After excision of the heart, myocardial samples were obtained from the regions subtended by the left anterior descending artery and LCx and cut into blocks that weighed approximately 1 g. Tissue blocks were weighed for their wet weight and then digested in 4 M potassium hydroxide (KOH) and 2% polysorbate-80 for 24 h. Blood samples were similarly digested with 16 M KOH. Colored polystyrene microspheres were reclaimed using vacuum filtering. The dyes were extracted from CPM by vortexing with methylformamide. The dye solution was separated from any remaining particles by centrifugation, to minimize scatter in the subsequent spectrophotometry. Photometric density was measured using a spectrophotometer (HP 8452 A, Hewlett-Packard, Palo Alto, CA). Regional subepicardial blood flow was calculated as follows:
Blood flow (ml *symbol* min sup -1 *symbol* g tissue sup -1) = (blood sample withdrawal rate (ml *symbol* min sup -1)/tissue sample weight [g]) x (photometric density of the tissue sample/photometric density of reference blood sample).
In Vitro Vessel Preparation
Subepicardial coronary arteries of approximately the fourth generation from both the CD LCx and the normal left anterior descending artery region (diameter of approximately 60-150 micro meter) were dissected free of the surrounding tissues. Each vessel was placed in a vessel chamber, cannulated with dual micropipettes (50-75 micro meter in diameter), and secured with 10-0 sutures. The vessel was continuously bathed with modified Krebs buffer, gassed with a 95% Oxygen2-5% CO2mixture, and maintained at 37 degrees Celsius and at a pH of 7.4. The total volume of Krebs buffer circulating in the vessel chamber, buffer reservoir, and the connecting tubing was 100 ml. The partial pressure of oxygen in the vessel chamber exceeded 400 mmHg, and the vessel was studied in a no-flow state. The pressure in the micropipettes was maintained at 40 mmHg to provide vessel distention. The vessel was visualized with an inverted phase-contrast microscope (Olympus IMT-2, Tokyo, Japan) connected to a video camera. The vessel image was projected onto a television screen (Panasonic, Osaka, Japan). The vessel internal lumen diameter was measured with an optical density video detection system (Living Systems Instrumentation, Burlington, VT) as described previously. A schematic of this in vitro preparation was published previously. 
Vasomotor Function of the Collateral-dependent Versus Normal Vessels
After preconstriction with 1 micro Meter of the thromboxane analogue U46619, vasodilation to 10 sup -9 - 10 sup -4 M of the endothelium-dependent dilator ADP and 10 sup -9 - 10 sup -4 M of the endothelium-independent dilator sodium nitroprusside for 6 vessels from the CD and 6 from the normal region were examined. Percent dilation of the vessels in response to a dilator was calculated as follows: % dilation = 100 * (diameter after the dilator - U46619-constricted diameter)/(baseline diameter before preconstriction - U46619-constricted diameter).
In addition, vasoconstrictive response to 10 sup -13 - 10 sup -8 M endothelin-1 of 6 vessels from the CD and 6 from the normal region were examined. Percent constriction of the vessels was calculated at each dose as follows: % constriction = 100 * (1 - diameter after the constrictor/baseline diameter).
Effect of Isoflurane
Each vessel was equilibrated for at least 30 min in the vessel chamber, and a baseline internal diameter was measured. The vessel was then randomized to be either (1) preconstricted with 1 micro Meter of the thromboxane analogue U46619 for 5 min or (2) not preconstricted. The vessel was then subjected to increasing concentrations of isoflurane, 0.5%, 1%, 2%, and 3%, for 10 min each, by adding the anesthetic to the 95% Oxygen2-5% CO2mixture bubbling into the Krebs buffer solution, using an inline Vernitrol bubble-through vaporizer (Ohio Medical Products, Madison, WI). In a preliminary experiment, it was determined, by gas chromatography, that it took less than 10 min for isoflurane to reach a steady-state concentration after it was introduced in the vessel chamber, and that the millimolar concentrations and partial pressures of isoflurane in the vessel chamber remained consistently proportional to the concentration of isoflurane in the gas mixture bubbled into the buffer solution. The anesthetic content in the gas mixture was continuously monitored, using a Rascal II Gas Analyzer (Ohmeda, Salt Lake City, UT) calibrated with industrial standards. At each concentration of isoflurane, the internal diameter of the vessel was measured. At the end of each experiment, the anesthetic was discontinued. The vessel chamber was flushed with fresh Krebs buffer and the vessel reequilibrated at 37 degrees Celsius. Potassium chloride was then added to achieve a concentration of 100 mM in the buffer and vessel chamber. Adenosine 5' diphosphate (10 micro Meter) was then added, to test for endothelial integrity. Only those vessels that constricted by at least 15% to potassium chloride and significantly dilated to ADP were considered still viable and included for data analysis.
Each animal contributed no more than one vessel to any one experimental group, but could contribute to multiple groups. Therefore, n for each group represents the number of animals as well as the number of vessels. A comparison of blood flows between the normal and CD regions was performed by Student's t test (2-tailed). Vasomotor responses to ADP, sodium nitroprusside, or endothelin-1 of the normal and CD vessels were compared by two-way analysis of variance (ANOVA), with a repeated measures factor. When the initial ANOVA yielded a significant P value, stratified z tests were performed to identify the concentrations of the vasomotor agents that resulted in different responses. Whether any experimental group of vessels demonstrated a concentration-dependent response to isoflurane was examined by one-way ANOVA (Scheffe's linear contrast). A comparison of two (or more) concentration-response curves to isoflurane was made by two-way ANOVA with a repeated measures factor, with post-hoc Student-Neumann-Keuls test for between-groups comparison and stratified z tests, when the initial ANOVA yielded a significant P value. P < 0.05 was considered significant. All statistics were calculated using True Epistat software (Epistat Services, Richardson, TX). All data are presented as mean plus/minus SD.
In Vivo Flows in the Collateral-dependent Versus Normal Microcirculation
There was no significant difference in baseline blood flow between the CD and normal regions (Table 1). However, with rapid atrial pacing, flow in the normal region became significantly greater (P < 0.05) than in the CD region.
Vasomotor Responses of the Collateral-dependent Versus Normal Vessels
After preconstriction with U46619, relaxation response to the endothelium-independent dilator sodium nitroprusside was not significantly different between the CD (n = 6, baseline diameter 131 plus/minus 15 micro meter [mean plus/minus SD], range 110-150 micro meter) and the normal vessels (n = 6, diameter 123 plus/minus 17 micro meter, range 101-142 micro meter; P = 0.45; Figure 1(a)). Conversely, after preconstriction with U46619, the normal vessels (n = 6, diameter 123 plus/minus 16 micro meter, range 102-143 micro meter) dilated to the endothelium-dependent dilator ADP significantly more than the CD vessels (n = 6, diameter 132 plus/minus 16 micro meter, range 110-151 micro meter; P < 0.001; Figure 1(b)). During viability testing, after constriction with potassium chloride, the normal vessels dilated to 10 micro Meter ADP significantly more (% relaxation = 87 plus/minus 6%) than the CD vessels (59 plus/minus 10%; P < 0.05).
The CD vessels (n = 6, diameter 119 plus/minus 9 micro meter, range 111-135 micro meter) constricted to endothelin-1 more than the normal vessels (n = 6, diameter 122 plus/minus 16 micro meter, range 101-141 micro meter; P < 0.01; Figure 1(c)).
Preconstriction of the normal and the CD vessels with U46619 was not significantly different: 27.9 plus/minus 3.7% versus 25.6 plus/minus 2.5% (P = 0.24). During viability testing, constriction to potassium chloride was comparable between the normal (31.3 plus/minus 6.6%) and the CD vessels (27.2 plus/minus 6.8%; P = 0.15).
Effect of Isoflurane
After preconstriction with the thromboxane analogue U46619, neither the normal vessels (n = 6, diameter 95 plus/minus 17 micro meter, range 61-115 micro meter) nor the CD vessels (n = 6, diameter 93 plus/minus 13 micro meter, range 72-109 micro meter) demonstrated any significant concentration-dependent vasomotion to isoflurane (P > 0.9 and = 0.49, respectively; Figure 2). In contrast, in the absence of preconstriction, normal vessels (n = 6, diameter 89 plus/minus 17 micro meter, range 64-117 micro meter) demonstrated a mild concentration-dependent constriction to isoflurane (P < 0.01), which was significantly greater (P < 0.05) than for the unpreconstricted CD vessels (n = 6, diameter 93 plus/minus 18 micro meter, range 72-118 micro meter; Figure 3). Constrictive response of the unpreconstricted CD vessels to isoflurane was statistically significant (P < 0.05), but very small in magnitude.
The principal finding of the current study is that, in pigs, isoflurane causes a mild constriction of normal-resistance coronary arteries in vitro, while having a statistically significant, yet physiologically insignificant small vasomotor effect on the CD resistance coronary arteries. In addition, the CD porcine resistance coronary arteries do not dilate to the endothelium-dependent vasodilator ADP as well as normal arteries, while demonstrating an enhanced contractile response to endothelin-1. Our in vivo flow data confirmed the existence of CD region and validated our model of chronic coronary occlusion.
We previously reported isoflurane-mediated constriction of resistance coronary arteries in vitro in two other species--rabbits and rats. As we demonstrated in rats, isoflurane-mediated constriction of resistance coronary arteries depends on preexisting tone of the vessels in pigs. Preconstriction of the vessels masks any vasoconstrictive effect of isoflurane. The magnitude of constriction in pigs appears less than in smaller animals. [8,11]Our data document qualitative species similarities, in that pigs, rabbits, and rats all demonstrate vasoconstriction of resistance coronary arteries to isoflurane. In addition, they indicate species difference(s) in the magnitude of isoflurane-mediated vasomotion.
Perhaps the most interesting finding of the current study is the differential vasomotor effect of isoflurane: Isoflurane constricts the normal vessels more than vessels from a CD region. Previously, isoflurane-mediated constriction of the resistance coronary arteries was found to be endothelium-dependent. This suggests that the CD vessels may have impaired endothelium-dependent constriction. In addition, the response of CD vessels to the endothelium-dependent vasodilator ADP is impaired, relative to the normal vessels of the same size, while dilating equally well to the endothelium-independent dilator sodium nitroprusside. This finding is in agreement with previous reports from our laboratory, [4,12]in which the CD vessels have attenuated responses to receptor-mediated endothelium-dependent dilators such as ADP, serotonin, and bradykinin. Enhanced constriction of the CD vessels to endothelin-1 also may be based on endothelial dysfunction (i.e., reduced endothelin-1-mediated release of nitric oxide from the endothelium) that tends to modulate endothelin-1-mediated constriction. The CD vessels may, therefore, have generalized endothelial dysfunction that hampers both endothelium-dependent vasodilation and endothelium-dependent vasoconstriction (as exemplified by decreased constriction to isoflurane). Similar endothelial dysfunction has been noted after ischemia-reperfusion, after mechanical endothelial injury such as from angioplasty, and in patients with syndrome X. In contrast, the vascular smooth muscle of the CD vessels appears to be functionally preserved, because their responses to potassium chloride and U46619, as well as to sodium nitroprusside, are comparable to the responses of the normal vessels.
Because our in vitro vasomotor study examines only the direct vasomotor effect of isoflurane, one must be cautious in projecting from the differential direct vasomotor effect between the normal and CD vessels to any redistribution of flow between the normal and CD microcirculation. The differential direct effect of isoflurane would be but one variable influencing redistribution of flow. In vivo, various indirect effects mediated via autoregulation, metabolism-flow coupling, and changes in compressive resistance may play a role as well. Given this reservation, however, the differential direct effect of isoflurane may tend to contribute to favorable redistribution for the CD region.
Demonstration of isoflurane-induced maldistribution of flow away from the CD region has been possible principally when indirect effect via autoregulation was allowed to come into play and dilate resistance coronary arteries--either because the systemic and coronary perfusion pressures were decreased [1,17,18]or because of controlled decreases in coronary flow. Studies in which coronary perfusion pressure was maintained and autoregulatory changes in resistance coronary arteries were prevented failed to demonstrate isoflurane-induced maldistribution of coronary blood flow. [19,20]More significantly, more recent studies that allowed the perfusion pressure to decrease with isoflurane also failed to demonstrate isoflurane-induced maldistribution of coronary blood flow. [21-23]If we accept that a decrease in coronary perfusion pressure would be associated with dilation of the normal resistance arteries with autoregulatory reserve, and would, therefore, tend to favor redistribution of flow away from the already (nearly) maximally dilated CD vessels, then there must be other mechanism(s) in effect that counterbalances the indirect effect via autoregulation to prevent the steal phenomenon.
The findings of our current study suggest one such mechanism, namely, greater direct vasoconstriction of normal resistance coronary arteries than CD counterparts, tending to redistribute flow favorably toward the CD region. When the coronary perfusion pressure is normal, the normal resistance coronary arteries would have a significant autoregulatory reserve and tone, and may, therefore, be refractory to any direct vasoconstrictive effect of isoflurane. When the coronary perfusion pressure is decreased sufficiently to deplete the autoregulatory reserve of the normal resistance coronary arteries, the direct vasoconstrictive effect of isoflurane might then become evident in these vessels, providing a counteractive force against maldistribution of flow.
Considering previous studies that demonstrated a vasodilatory effect of isoflurane on epicardial coronary arteries, [8,11,24]we can propose another mechanism whereby isoflurane may have an effect against maldistribution of flow. Although the epicardial conduit arteries account for only a small fraction of overall coronary vascular resistance, and dilation of these arteries by isoflurane by itself may not decrease coronary vascular resistance appreciably, the effect of their dilation may be magnified by flow-mediated release of nitric oxide, [25,26]which can then dilate and/or recruit distal vessels. Such increase in upstream flow can augment perfusion pressure for the collateral vessels and increase flow in the CD circulation. Antiischemic effect of nitroglycerin, another preferential dilator of the large conduit arteries may be based on a similar mechanism.
Comparison of Animal Models
Previous animal studies that demonstrated maldistribution of myocardial blood flow with isoflurane used canine models. [2,17,18]Later studies that did not demonstrate the same effect of isoflurane either with maintained perfusion pressure [19,20]or despite decreased perfusion pressure [21-23]were performed in both canine [19-21,23]and swine models. Whereas dogs possess extensive innate coronary collateral circulation, pigs do not, and, in this respect, resemble humans. Because of poor innate collaterals, pigs tend to develop significant myocardial damage with occlusion of a major vessel, and therefore have been more difficult and less popular to study. Roth et al., however, demonstrated that, with gradual occlusion of a major coronary artery with ameroid constrictor, the pig experiences a significant increase in collateral flow throughout a few weeks, and can be a suitable model of CD coronary circulation. This model is what we chose to use in the current study.
One limitation of the current swine model is that the animals are relatively young and may not correspond to the middle-aged to geriatric human population, in which most cases of coronary artery disease are observed. However, due to impracticality of working with fully aged pigs weighing hundreds of kilograms, use of young pigs has been an accepted practice. [3,7,12,22,29]
In summary, we demonstrated that the direct vasoconstrictive effect of isoflurane is greater in normal resistance coronary arteries than in the CD vessels. This effect may be masked in vessels with high preexisting tone. We speculate that the differential direct vasoconstrictive effect of isoflurane may tend to favor redistribution of flow toward the CD region and help to prevent coronary steal even when the coronary perfusion pressure is decreased.
The authors thank Patrick F. Wouters, M.D., for informal discussion of his article (), and Hang Lee, Ph.D., for assistance with statistical analysis.