Inhaled Anesthetics Alter the Determinants of Coronary Collateral Blood Flow in the Dog

An ameroid constrictor was used to enhance collateral vessel development in the area supplied by the left anterior descending coronary artery (LAD). Twelve adult mongrel dogs of either sex were sedated with pentobarbital and anesthetized with halothane. The trachea was intubated, and the lungs were ventilated with Oxygen²-enriched room air. The left thorax was entered via the fifth intercostal space, and a short incision was made in the pericardium. The LAD proximal to the first diagonal branch was freed from surrounding tissue, and an ameroid constrictor was placed around the vessel. Care was taken to avoid stretching or kinking the vessel. The ameroid was 8 mm in diameter and surrounded by a steel band. The lumen size ranged from 2.3-2.8 mm and was chosen to provide a snug fit. The pericardium was reapproximated but not closed tightly, and the chest wall was closed in layers. The dog received buprenorphine (0.3 mg, intravenously) for postoperative pain relief and then aspirin (80 mg, orally) daily during the 4-5-week recovery period. Aspirin was given to prevent early thrombotic occlusion of the narrowed LAD. The dog was permitted to exercise daily, in part to stimulate the formation of collateral vessels. [2]Experimental work conformed to the standards of the American Physiological Society, and was approved by the Animal Care and Use Committee of the University of Pittsburgh.

Twelve dogs were studied 4-5 weeks after surgery when they were afebrile and active. Approximately 1 h after sedation with morphine sulfate (2.5 mg *symbol* kg sup -1 subcutaneously), each dog was anesthetized with an initial injection of alpha-chloralose (100 mg *symbol* kg sup -1 intravenously). Anesthesia was maintained with a continuous infusion of alpha-chloralose (10 mg *symbol* kg sup -1 *symbol* h sup -1 intravenously). NaHCO³(150 mM, 5 ml *symbol* kg sup -1 *symbol* h sup -1 intravenously) was given to counteract the metabolic acidosis that occurs during anesthesia with this agent in dogs. The lungs were ventilated with Oxygen²-enriched room air at 15 ml *symbol* kg sup -1. The respiratory rate was adjusted to maintain arterial CO²tension at approximately 40 mmHg, and end-tidal CO²in the trachea was monitored (Datex, Helsinki, Finland). Blood gas tensions and pH were measured at intervals during the experiment (Radiometer, Copenhagen, NV, Denmark).

Halothane (1%) was administered through a non rebreathing circuit, and the left thorax was entered through the fifth intercostal space. Adhesions were divided by cautery, and the heart was suspended in a pericardial cradle. Bupivacaine (0.25%, 1-2 ml at each site) was injected at the fourth, fifth, and sixth intercostal spaces posterior to the incision to obtain intercostal blocks. Formalin (0.1 ml) was injected into the area of the atrioventricular node to cause a complete heart block. Occasionally, several injections were necessary to achieve a block, but no more than 0.5 ml of formalin was used. Digoxin (50 micro gram, intravenously) was given to prevent nodal arrhythmias. The heart was paced throughout the experiment at 100 beats/min with a lead sutured to the apex of the right ventricle. A ground electrode was connected to the chest wall. If ventricular dysrhythmias occurred during or shortly after embolization (see below), they were treated with bolus lidocaine injections or, if necessary, direct-current cardioversion. Arterial blood pressure was measured with a catheter introduced into the thoracic aorta via the right femoral artery. A solid state, catheter-tip transducer (Millar, Houston, TX) inserted through the left atrium (LA) was used to measure left ventricular (LV) pressure. The first derivative of LV pressure with respect to time was obtained with an analog circuit. A small tube with a side hole and an end hole was inserted 2 cm into the LA through a purse-string suture in the appendage for measurement of LA pressure. Halothane was discontinued for at least 30 min before measurements were made (see below) unless it was the first anesthetic studied.

Heparin (750-U *symbol* kg sup -1 bolus plus 250 U *symbol* kg sup -1 *symbol* h sup -1 intravenously) was given. The first or second diagonal branch of the LAD was dissected from the surrounding tissue, ligated, and cannulated with a short 16-G cannula. Blood from a femoral artery provided antegrade perfusion of this area. The circuit pressure just proximal to the cannula was decreased to 70-75 mmHg, about 80% of systemic arterial pressure, by tightening a screw clamp in the extracorporeal circuit. Antegrade flow in the circuit was measured with an electromagnetic flow probe, and blood was mixed with a magnetic stir bar in a small in-line chamber. This circuit was interrupted occasionally with a clamp, and a side port was opened to determine retrograde blood flow (side port open) and distal pressure (side port closed). Latex microspheres (25 micro meter in diameter, 1.2 million/ml; Duke Scientific, Palo Alto, CA) were injected in aliquots of 0.2 ml until antegrade flow stopped to occlude the arteriolar circulation of the diagonal branch. In general, 2-3 million spheres were used. The perfusion pressure in the diagonal branch was held 15-20 mmHg less than arterial pressure to prevent microspheres from passing through the collaterals and embolizing the arterioles of the supplying artery.

After the embolization procedure, retrograde flow from the diagonal branch was assumed to equal total collateral flow. Flow increased 40-100% with embolization, an indirect indication of the magnitude of microvascular collaterals. Retrograde flow was measured as outflow pressure was increased in steps by gradual occlusion of the circuit with a screw clamp. Measurements were made after 10-30 s at each step when a steady state was present. The pressure obtained when flow stopped was taken as the pressure at the origin of the collaterals in the supplying artery (stem pressure [P^stem]). A high-gain electromagnetic flow probe (Zepeda, Seattle, WA) was used to measure retrograde flow. This flow probe was calibrated by timed collection of blood after each experiment. The calibrations showed a linear relation between output voltage and flow.

Blood was supplied to the left circumflex coronary artery (LCx) through a stainless steel cannula (Figure 1). The cannula was inserted into the aorta via the left carotid artery and wedged into the proximal LCx. The seal at the tip was tested by temporarily stopping flow. If the seal was complete, pressure at the cannula tip decreased to 20-25 mmHg, and a brisk reactive hyperemia occurred when flow resumed. Blood from a femoral artery was supplied to the cannula by a servo-controlled roller pump that maintained a constant coronary pressure of 60 mmHg. Flow into the LCx was measured with an electromagnetic flowmeter that was calibrated by timed collection of blood after each experiment.

Halothane and isoflurane were administered from calibrated vaporizers into a nonrebreathing ventilator circuit. Liquid desflurane was injected directly into a semiclosed anesthetic circuit containing a CO²absorbent. End-tidal anesthetic concentration was determined continuously (Datex, Helsinki, Finland). The monitor was calibrated with a reference gas each month. Anesthetics were given for 20-25 min before retrograde flow measurement were taken to ensure equilibration with the vessel-rich organs of the dog and to ensure that most of the previous agent had been eliminated. A small carry-over effect is likely with this approach, but the agents were given in a random order to avoid systematic confounding effects. The concentration of inhaled agent was initially adjusted to give a mean arterial pressure of 63-65 mmHg and then held constant during the measurement of retrograde flow.

Mean arterial pressure was controlled at 60 mmHg throughout the experiment in part by administration of volatile anesthetics and in part by use of a pressurized, 1-l blood reservoir connected to a femoral artery. Blood flowed from the animal into the reservoir if arterial pressure exceeded reservoir pressure, and vice versa. The reservoir contained 500-600 ml blood during control measurements and 100-200 ml during treatment with volatile anesthetics. The heart rate was controlled at 100 beats *symbol* min sup -1 by ventricular pacing. Retrograde pressure-flow curves were determined during control conditions (morphine and chloralose anesthesia) and during halothane, isoflurane, and desflurane administration. The order of treatments was randomized.

Ameroid closure can result in myocardial infarction if closure occurs rapidly or collateral vessels are inadequate. The presence of infarction was sought at autopsy by visual inspection of the myocardium perfused by the LAD. In addition, thin transmural slices of fresh myocardium were incubated for 10 min in warm triphenyltetrazolium chloride (1.5% solution in 20 mM potassium phosphate buffer). A region of infarction was identified by the absence of the brick-red color produced when triphenyltetrazolium chloride reacts with vital tissue. None of the animals displayed evidence of myocardial infarction.

Hemodynamic variables and flows were recorded on an oscillograph (Gould, Cleveland, OH). The analog flow signals were electronically averaged (2-s time constant). Retrograde flow, outflow pressure, LCx flow, arterial pressure, and LA pressure were determined by averaging the values from six to eight beats at each point. The measured outflow pressure was corrected for a small (1-3-mmHg) pressure decrease across the diagonal branch cannula to obtain the pressure opposing retrograde flow. Plots of retrograde flow versus opposing pressure were constructed for each agent in each animal. These data were fit by a dual-line linear regression technique. [3]The computer program models the data as two straight lines and seeks by iteration the lowest combined sum of squares due to error for both lines (Figure 2). Interpolation was used to determine the retrograde flow at zero pressure. The slope of the upright segment was taken as the (negative) inverse of collateral resistance. The correlation coefficient of the fit for this segment averaged 0.89 plus/minus 0.05 (SD) and ranged from 0.76 to 0.95. The supplying artery was modeled as a voltage divider, with P^stemas the intermediate "voltage." The precollateral resistance LCx was calculated as the difference between proximal LCx pressure and P^stemdivided by LCx flow (taken when retrograde flow was zero). Postcollateral resistance in the LCx was calculated as P^stemminus LA pressure divided by LCx flow.

A standard statistical package (SPSS PC version 1.1) was used to analyze the data. One-way analysis of variance was used to test each dependent variable for treatment effects. If the analysis of variance revealed significance, paired t tests were used to compare the values obtained during halothane, isoflurane, and desflurane with control values. Significance was taken as P < 0.05 and was not corrected for the multiple testing problem because these were a priori hypotheses. Data are presented as the means and SD.

Systemic hemodynamics were stable during the experiment. Heart rate was constant at 100 beats *symbol* min sup -1, and mean arterial pressure averaged 59 mmHg under all conditions (Table 1). The inhaled anesthetics caused myocardial depression, reflected in a 12-18% decrease in LV dP *symbol* dt sup -1 and increase in LA pressure. Blood was transferred from the pressurized reservoir to the dog to hold arterial pressure constant. Blood flow in the LCx increased modestly with halothane and increased further during isoflurane and desflurane, probably as a result of the arteriolar dilating properties of these agents. Coronary venous blood was not sampled for Oxygen²content, so these data provide only indirect evidence for direct dilation by inhaled agents. Arterial hemoglobin concentration, pH, and blood gas values were stable throughout the experiment.

Retrograde flow averaged 9.7 ml *symbol* min sup -1 under control conditions (Table 2) and ranged from 1.34 to 24.9 ml *symbol* min sup -1, reflecting the variability between dogs in collateral capacitance and differences in the size of the bed supplied by the test vessel. Retrograde blood flow decreased by about 20% (P < 0.01) during desflurane but was unchanged during halothane and isoflurane. None of the anesthetics affected collateral segment resistance (R3). All of the anesthetics reduced postcollateral resistance (R2), with isoflurane (-35%, P < 0.001) and desflurane (-45%, P < 0.001) having a greater effect than halothane (-15%, P < 0.05). Precollateral resistance (R1) in the supplying artery was not affected by halothane, but decreased during isoflurane (P < 0.05) and desflurane administration (P < 0.05). P^stemreflects the balance between pre- and postcollateral resistance. It decreased 2-3 mmHg with desflurane (P < 0.02), but was unchanged with halothane and isoflurane.

The findings of this study confirm and explain those of previous investigations. The resistance of the coronary circulation distal to the origin of the collateral vessels was reduced by isoflurane (-35%), desflurane (-45%), and by halothane (-15%), indicating dilation of arteriolar vessels. Isoflurane and desflurane (but not halothane) reduced resistance in the arterial segment proximal to the origin of the collaterals. None of the inhaled agents affected collateral vessel resistance. Desflurane reduced retrograde flow by about 20% by reducing collateral perfusion pressure.

We injected 25-micro meter diameter latex spheres into the test artery during antegrade perfusion to obliterate the arteriolar bed. Embolization prevents diversion of collateral flow into the native circulation. This technique increased retrograde flow by 50-100% and raised P^stema corresponding amount (Figure 3). The magnitude of flow and pressure increases observed in our study is similar to that found by other investigators. [4-6]We assumed that embolization completely stopped antegrade flow. If not, then P^stemwould underestimate the pressure at the origin of the collaterals, and retrograde flow measurements would underestimate true collateral flow. We also assumed that few spheres passed through collaterals during embolization to lodge in the tissue surrounding the test zone. We deliberately set proximal pressure in the artery being embolized 15-20 mmHg below systemic pressure to minimize this problem. The paired nature of our experimental design should diminish the influence of these factors on the conclusions.

Retrograde flow decreased by an average of 52% (range 17-83%) when LCx flow was temporarily stopped. This finding indicates that retrograde flow reached the test zone from sources other than the LCx artery. Both mean arterial pressure and mean LCx pressure were held constant at 60 mmHg, so the resistance calculations should be valid. That the LCx supplied only a fraction of retrograde flow may also explain the "negative" results of other investigators seeking to demonstrate intercoronary steal using a model involving only an LCx stenosis. [7,8] 

Some inaccuracies are inherent in the calculation of segmental resistances. [1]For example, P^stemis measured with retrograde flow stopped and falls slightly when retrograde flow starts because of resistance in the precollateral segment of the supplying artery. This is likely a small effect so long as retrograde flow is a small fraction of total LCx flow, however, it may account for some of the curvilinearity of the "collateral conductance" segment of the pressure-flow curve. We recognize that the measured P^stemis actually some combination of multiple pressures because collateral vessels arise from different levels of the supplying artery and thus have different pressures at their origin. For this reason, the calculated segmental resistances should be considered "lumped" parameters.

We assumed that the pressure-flow curve had two straight segments and an inflection point based on our results and those of other investigators. [5,9]Eng and Kirk [5]provided clear evidence of a break point at about 20 mmHg when mean arterial pressure was 85 mmHg. Other investigators have found a pressure-flow curve without a definable breakpoint. [10,11]The presence of a breakpoint is not crucial to the analysis or interpretation of our data.

We assume that the "basal" anesthetic (morphine and chloralose with bupivacaine intercostal blocks) did not affect the results. This anesthetic technique preserves cardiovascular reflexes that could have caused sympathetic coronary vasoconstriction during control measurements involving hemorrhage into the pressurized blood reservoir. If this occurred, then the inhaled anesthetics may have released the vasoconstriction and caused an "artificial" dilation. We did not measure the Oxygen²tension of coronary venous blood in this experiment, and thus cannot assess whether or not a state of relative vasoconstriction existed before anesthetic administration.

The ameroid constrictor used in these studies had a small central hole and probably occluded the LAD within 1 week. The dogs were studied 4-5 weeks after ameroid implantation, and a wide variability in collateral capacity was found, probably due to differences in the number and capacity of "native" collateral vessels. A longitudinal study of the histologic appearance of collateral vessels after ameroid implantation revealed that the veinlike native collateral vessels were enlarged and the internal elastic membrane was fractured at 4-5 weeks after ameroid closure. [12]Endothelial cells were present, and the intima was hypertrophied. Smooth muscle cells were found in a disorderly arrangement with evidence of active cell division. Perivascular inflammation was observed, but had waned considerably from the early period after ameroid closure.

Collateral vessels in dogs are commonly said to be epicardial in location; however, Scheel et al. found that cauterization of the epicardium between branches of the coronary arteries reduced collateral flow by only about 50%, demonstrating that significant flow occurs through vascular connections within the ventricular wall. [11]Similar results were obtained by other investigators using a microsphere method. [13]Both experiments were carried out in dogs with native collaterals, but it is likely that a similar distribution of vessels also exists after ameroid stimulation. [14] 

Mean arterial pressure and heart rate were carefully controlled in the current experiment because these hemodynamic variables affect the determinants of retrograde flow. [15-17]Retrograde flow is very sensitive to changes in arterial pressure (Figure 4). [18-21]As a result, the decreased arterial pressure that commonly accompanies inhalational anesthesia is likely a greater threat to collateral flow than minor changes produced by direct effects of the agents on segmental coronary resistance. Increases in heart rate cause metabolic dilation in the supplying artery, decrease P^stem, and thus decrease retrograde flow. [17] 

The driving pressure for retrograde flow is less than aortic pressure because of the resistance of the precollateral segment. Our P^stemvalues were about 70% of aortic pressure, suggesting that collaterals in the current study arose at the 50-200-micro meter-diameter level of the supplying artery. Micropuncture studies have demonstrated that arteries larger than 200 micro meter account for about 25% of the pressure drop across the coronary circulation. [22]Use of "voltage divider" analysis allowed us to determine the effects of anesthetics on the resistance of small coronary arteries (as distinct from arterioles), an area that has not been extensively investigated because of technical limitations.

Previous studies of isoflurane in ameroid models provide evidence for vasodilator reserve in the circulation supplying collateral-dependent areas. [18,23,24]This reserve might be located in the arteriolar bed of the test zone or reflect changes in segmental resistance of the supplying vascular network. The current results argue against a direct effect of inhaled anesthetics on collateral vessel resistance even though these agents affect other segments of the coronary circulation. Thus, by deduction, vasodilator reserve in the collateral circulation most likely resides in the arterioles of the collateral-dependent zone. In our previous study, [23]it was necessary to decrease total coronary flow by about 20% to exhaust this vasodilator reserve and make flow into the collateral circulation pressure-dependent. Retained arteriolar reserve in the test bed may thus confound the interpretation of several previous studies. [7,24,25] 

Isoflurane reduced postcollateral resistance by about 35%, a finding in line with other studies indicating that this agent causes moderate coronary arteriolar dilation. [7,18,24,26,27]The results are midway between studies demonstrating minimal arteriolar dilation from isoflurane [25,28]and other studies demonstrating that isoflurane is a powerful dilator. [29,30]The reason for discrepancies between animal models is not clear. A moderate dilation in humans anesthetized with isoflurane has been documented by measurement of increased coronary sinus Oxygen²content. [31]Desflurane decreased coronary arteriolar resistance by about 45%, confirming the findings of Merin et al. [32]but conflicting with those of Hartman et al. [33]who found no increase in flow to normal myocardium with desflurane (12.5%, end-tidal) when blood pressure and heart rate were controlled. Measurements of myocardial Oxygen sub 2 extraction in humans during desflurane anesthesia have not been reported. In the current study, halothane reduced arteriolar resistance by about 10%, a predictable effect. [23,34] 

Both isoflurane and desflurane reduced the pressure at the origin of the collaterals (P^stem) by dilating the postcollateral arterioles and increasing flow and the pressure gradient across the precollateral segment. This is the mechanism of intercoronary steal. The decrease in P^stemwas small, on the order of 2-3 mmHg, and retrograde flow decreased only with desflurane. Larger effects on P^stemand thus on retrograde flow would likely occur in the presence of an epicardial stenosis of the supplying artery. [35] 

The decrease in precollateral resistance during desflurane and isoflurane is surprising because in general these agents are considered only weak dilators of large coronary arteries. [36-39]Adding to the confusion, halothane did not decrease precollateral resistance despite a study showing it to be a potent dilator of epicardial vessels. [36]One possible explanation is that the increased flow and decreased pressure resulting from arteriolar dilation by isoflurane and desflurane produced "ascending vasodilation" of the proximal segments. [40]This hypothesis is supported by microscopic data from beating hearts showing that isoflurane increased the diameter of 100-500-micro meter coronary vessels, whereas halothane had a smaller effect. [41]Adenosine [1]and reactive hyperemia [42,43]are examples of prominent arteriolar dilators that also reduce precollateral resistance. In vitro studies of coronary microvessels done with static distending pressure in the absence of flow cannot duplicate these important aspects of the control of the coronary circulation. [44]This model provides an interesting approach to the study of the interactions of anesthetic agents with myogenic and flow-related phenomena in the coronary circulation.

Whatever the cause of reduced precollateral resistance with isoflurane and desflurane, the effect is to raise P^stemand oppose coronary steal. It is possible that this mechanism accounts for the negative results obtained with isoflurane in models in which no stenosis of the supplying artery was used. [24,45] 

Retrograde flow derived from collaterals decreased about 20% during desflurane because of a decrease in driving pressure across the collateral segment. Arteriolar dilation decreased pressure at the origin of the collaterals, and it is likely that increased LV pressure during diastole (reflected by increased LA pressure) also contributed to the decrease in perfusion pressure. Previous studies have shown that retrograde flow occurs primarily during diastole [16]and diminishes with increased LV diastolic pressure. [46] 

The study was done in dogs, and the results may not apply to humans. The study was done approximately 4 weeks after ameroid closure when histologic studies show that collateral vessels are still developing into small arteries. Perhaps different results would be obtained in more mature collateral vessels. Because of embolization, no potential beneficial effects of the anesthetics (such as reduced Oxygen²demand in the collateral-dependent zone) could be documented in this study. We studied only one concentration of each anesthetic, and it is possible that different concentrations would produce qualitative as well as quantitative differences.

In summary, inhaled anesthetics have multiple effects on the hemodynamic determinants of coronary collateral blood flow. The importance of an adequate arterial pressure to drive flow through the high-resistance collateral network was confirmed. Arteriolar resistance in the artery supplying collaterals was reduced by desflurane (-45%), isoflurane (-35%), and halothane (-15%), effects in line with previous studies. Surprisingly, the resistance of the supplying artery proximal to the origin of the collaterals was also reduce by desflurane (-21%) and isoflurane (-22%) but not halothane. This effect on small coronary arteries has not been reported before and may be due to ascending vasodilation. The reduction in precollateral resistance offset the decrease in pressure at the origin of the collaterals caused by arteriolar vasodilation--an "anti-steal" effect. None of the agents directly affected collateral segment resistance in these animals with moderately well-developed collateral circulations. Desflurane reduced retrograde flow, an index of collateral flow, by 20% because it reduced collateral perfusion pressure. In contrast, neither halothane nor isoflurane affected retrograde flow. Experimental models that include a proximal stenosis of the supplying artery would be more likely to demonstrate a steal phenomenon with these agents. Potential beneficial effects of desflurane on Oxygen²demand in the collateral-dependent zone were not measured. These findings help explain conflicting results from previous studies and provide a useful model for investigation of the effects of inhaled agents on small coronary arteries.

This study could not have been done without the expert technical assistance of Sue Dase and Marc Wallace. Carolyn Cuba provided secretarial assistance, and Francie Siegfried, editorial review.