INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

SINCE THE PHENOMENON of "ischemic preconditioning" was described in 1986 (46), it has been established that the heart is capable of rapid adaptation after brief ischemic episodes. This provides a marked increase in resistance to lethal myocyte injury after a prolonged ischemic challenge (29, 37, 47). Both the warm-up phenomenon, which refers to the improved performance during a second exercise test compared to the first one, and the better outcome after myocardial infarction, if it is preceded by angina, are ascribed to this endogenous protective phenomenon (3, 34, 42, 49, 60, 61). The cardioprotective mechanisms underlying ischemic preconditioning are not yet clarified. However, evidence exists that adenosine is one of the triggers to preconditioning in several species, including humans (5, 38, 44, 57, 64).

Thromboembolism in the coronary microcirculation is common in patients with ischemic heart disease (11, 13). Because coronary embolization with microspheres has been shown to increase coronary blood flow (CBF) due to a massive release of adenosine (23, 24), coronary microembolization may constitute a protective phenomenon in patients with ischemic heart disease. Because reversible microembolization can be achieved by biodegradable microspheres, and this approach may be a tool to protect the ischemic heart, the present study was designed to clarify if such a protective phenomenon really exists. Furthermore, injection of microspheres is often performed in experimental cardiovascular research to measure myocardial blood flow. Whether embolization by such microspheres influences ischemic tolerance has not yet been investigated. Additionally, it is unknown whether microspheres, in amounts presently used to measure tissue perfusion, by themselves affect CBF.

In anesthetized open-chest pigs, we examined the relationship between the amount of injected microspheres and the effect on CBF. We also examined whether pretreatment with microspheres, in amounts that significantly increased CBF but without causing myocardial necrosis (25) or heart failure, affects myocardial ischemic tolerance against infarction and arrhythmias.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Animals used in the present study were maintained and housed in accordance with the conditions set by the Norwegian Council for Animal Research. The investigation conformed with the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (NIH Publication No. 85-23, Revised 1985).

Experimental design and procedure. Twenty domestic pigs of either sex, weighing between 27 and 34 kg, were used in this study, which was performed mainly in two parts (Fig. 1). The first part was designed to establish the relationship between the amount of injected microspheres and the effect on CBF. The amounts of microspheres suited to establish this relationship were determined in preliminary experiments. In six pigs 0, 5, 10, 15, 30, and 40 × 10⁶ polystyrene microspheres (15 ± 1 µm, means ± SD, density 1.05 g/ml, Dynospheres, Dyno Particles, Lillestrøm, Norway), suspended in 6 ml of 0.02% Tween 80 in isotonic saline at 37°C, were injected into the left atrium in randomized order. Additionally, 60 × 10⁶ microspheres were injected after the randomized injections. Each injection was performed over 30 s, followed immediately by a slow flush with 3 ml of isotonic saline at the same temperature, and at intervals of 25 min. Before injections, the microspheres were ultrasonicated and thoroughly dispersed by vortex mixing. Microscopic examination of these suspensions showed no aggregation of microspheres. Blood samplings for metabolic measurements (see below) were performed before and 1.5 and 20 min after the start of the injections. The withdrawn blood was replaced with Macrodex (Dextran 70, 60 mg/ml; NaCl 154 mmol/l). Both Tween 80 and Macrodex were administered before any injection of microsphere suspensions to desensitize animals (19).

View larger version (10K): [in this window] [in a new window]   Fig. 1.   Time diagram illustrating different protocols used. All pigs were allowed a 30-min stabilization period after we completed the surgery. Downward arrows, randomized injection of either 0, 5, 10, 15, 30, or 40 × 106 microspheres. Bold downward arrow, injection of 60 × 106 microspheres. , Injection of 40 × 106 microspheres. Periods with coronary artery occlusion are shown as solid black. See text for descriptions of Parts 1 and 2.

The second part of the study was designed to determine whether ischemic tolerance is affected by microembolization that increases CBF. Five pigs were pretreated by two injections, each of 40 × 10⁶ microspheres at 30-min intervals. Thirty minutes after the second injection, the left anterior descending coronary artery (LAD) was occluded for 45 min before 2 h of reperfusion. Six pigs served as controls and underwent no pretreatment before a 45-min LAD occlusion period followed by 2 h of reperfusion.

To verify that the hyperemic flow response induced by microembolization is attributed to the vasodilatory effect of adenosine, 40 × 10⁶ microspheres were injected both before and after pretreatment of the LAD area with the adenosine receptor blocker 8-p-sulfophenyl theophylline (8-SPT) (Research Biochemicals, Natick, MA) in three additional pigs. 8-SPT was given at a dose of 1-1.5 mg/kg (by slow injection), which reduced the vasodilatatory effect of intracoronary adenosine in doses up to 110 µg/min by >70%. In these pigs the coronary flow reserve before and after the first injection of 40 × 10⁶ microspheres was also examined by using 20-s LAD occlusions. Furthermore, to evaluate whether microembolization by 15-µm microspheres causes necrosis per se, one of these three pigs was kept alive for 6 h after the second injection of microspheres.

If ventricular fibrillation occurred, one or if necessary more DC countershocks (15-30 J) were given to the heart. If electroconversion was successfully achieved within 1 min, the animal was allowed to continue the experimental protocol. Throughout the experimental protocols, body temperature was monitored by a rectal thermometer and kept within a narrow range by wrappings and a homeothermic blanket system unit (50-7103 Harvard Homeothermic Blanket System, South Natick, MA) to minimize temperature-induced variability in infarct size (7, 9). At the end of the experiment, the animals were killed by intracardial KCl injection.

Animal preparation. All animals were fasted overnight and anesthetized with pentobarbital sodium, initially 40 mg/kg body wt intraperitoneally followed by a sustaining intravenous infusion of 5-20 mg · kg¹ · h¹ according to the depth of anesthesia. The pigs were ventilated through a tracheostoma with a 50-50% mixture of oxygen and air with a volume-regulated ventilator (model 101; New England Medical Instruments, Medway, MA). A positive end-expiratory pressure of 50 mm of water was established. Ventilation frequency and volume were adjusted to keep PCO₂ and pH within normal ranges. Polyethylene catheters were placed in the right femoral vein for administration of drugs and fluids and in the right femoral artery for blood sampling and pressure recordings.

The heart was exposed through a midsternal split and suspended in a pericardial cradle. In the first part of the study a 4- to 5-mm long segment of one or both main branches of the left coronary artery was carefully dissected free to allow placement of flow probes for measurement of CBF by transit time flowmetry. Furthermore, these animals were equipped with a polyethylene catheter in the coronary sinus, advanced through the left azygos vein that was occluded around the catheter, for withdrawal of myocardial venous blood samples. According to Andersen et al. (2), more than 90% of the blood in the coronary sinus originates from the left ventricular (LV) myocardium in pigs when the azygos vein is occluded. To prevent clotting of blood in this catheter, heparin (300 IU/kg) was given as an intravenous bolus immediately after preparation of these animals, followed by a continuous infusion of 67 IU · kg¹ · h¹.

In the second part of the study, a 4- to 5-mm-long segment of LAD, distal to the first major branch, was dissected free to facilitate occlusion with a Mayfield clip. When the clip was not applied, LAD flow was measured by a transit time flow probe at this place.

In the three pigs where adenosine blockade should be obtained, LAD was catheterized proximal to a flow probe for intracoronary administration of 8-SPT. These pigs were also equipped with a snare around LAD, proximal to the flow probe, for brief occlusions to evaluate CBF reserve.

For injection of microspheres, pigs were instrumented with a catheter placed in the left atrium through the atrial appendage. Urine was drained continuously through a cystostoma, and after the surgical preparation was completed, all pigs were allowed a stabilization period of 30 min.

Hemodynamic measurements. A microtip pressure transducer catheter (Millar Instruments, Houston, TX) was introduced into the left ventricle through the right carotid artery for measurements of LV pressure and the contractility parameter LV dP/dt_max (maximal positive value of the first derivative of LV pressure). An electromagnetic flow probe was placed on the ascending aorta and connected to a square-wave electromagnetic flowmeter (model 376; Nycotron, Drammen, Norway). Arterial blood pressure was measured by a Statham pressure transducer (model P23 Gb, Gould Instruments, Hato Rey, Puerto Rico). CBF was measured by transit time flowmetry (T208, Transonic Systems, NY). Hemodynamic variables were continuously recorded on an eight-channel galvanometric recorder (model 7758 B, Hewlett-Packard, Medical Products Group, Andover, MA). In sampling periods when higher resolution of data was required, the output of the recorder was sampled at 100 Hz, and the signals were transformed by an analog-to-digital converter and stored on floppy disks. Computer samplings of hemodynamic variables were obtained at end expiration, and values from four to six consecutive beats were averaged by the computer. Additionally, mean CBF was continuously recorded on large-scale paper (model MC6621, Multicorder, Graphtec, Tokyo, Japan).

Metabolic measurements. Blood samples were drawn anaerobically and simultaneously from the carotid artery and the coronary sinus into heparinized syringes. Hemoglobin concentrations and oxygen saturation were analyzed on a hemoximeter (OSM3, Radiometer, Copenhagen, Denmark). Partial pressures of oxygen, PCO₂, and pH were analyzed on an automatic blood gas analyzer (model 945, AVL Biochemical Instruments, Graz, Austria). Myocardial oxygen consumption (MO₂) was calculated by the formula: MO₂ (µmol/min) = [(Sa_O2  Sv_O2) × 62.1 × Hb/100 + (Pa_O2  Pv_O2) × 0.00141] × CBF (in ml/min), where Sa_O2 and Sv_O2 are oxygen saturation (in percent) in arterial blood and in blood from the coronary sinus, respectively; the constant 62.1 is the oxygen binding capacity of hemoglobin (in µmol/g); Hb is hemoglobin concentration (in µg/ml); Pa_O2 and Pv_O2 are partial oxygen pressures (in mmHg) in arterial blood and in blood from the coronary sinus, respectively; and the constant 0.00141 is the solubility of oxygen in blood at 37°C (in µmol · ml¹ · mmHg¹). Lactate concentrations were measured enzymatically (54), and the regional arteriovenous differences in lactate concentrations were multiplied by the CBF to give regional lactate extraction (µmol/min).

Arrhythmias. In the second part of the study Holter monitoring was performed with a two-channel tape recorder (Oxford Medilog 4500). The total number of ventricular extra systoles (VES) were counted by using the Oxford Medilog Exel 2.0 device (Holter Managment System, Oxford Instruments, UK). The analysis of arrhythmias was restricted to the first 10 min of ischemia, because myocardium subjected to 10 min of ischemia does not develop irreversible cell injury (50). Accordingly, infarction as a covariate in the analysis of arrhythmias was avoided.

Area at risk and infarct size. After a 2-h reperfusion period, the pigs in the second part of the study were killed, and the heart from each pig was excised. Thereafter, the aorta was cannulated and the coronary vasculature perfused with 800 ml of 0.9% saline to remove the blood in the vascular tree. The LAD was then reoccluded with a ligature, and 100 ml of a suspension with 2 mg zinc-cadmium sulfide particles (1-10 µm in diameter; Duke Scientific, Palo Alto, CA) per milliliter isotonic saline were infused into the aortic root at a pressure of 100 mmHg. Both atria were then removed and the ventricles molded in 2% agarose at 37°C and cooled in a refrigerator. After the agarose gelled, the ventricles were cut parallel to the atrioventricular groove into ~7-mm thick slices. The slices were cleaned for agarose, weighed, and stained by incubation for 20-30 min at 37°C in 0.1 mol/l sodium phosphate buffer containing 2% (wt/vol) triphenyltetrazolium chloride. They were then immersed in 10% Formalin to enhance the contrast of the stain, after which they were placed between two glass slides to a uniform thickness. The areas at risk were demarcated by the absence of fluorescence under ultraviolet light, and the triphenyltetrazolium chloride-negative regions representing irreversibly injured myocardium were indicated by the absence of red formazan precipitate. The regions on both sides of the slices corresponding to the left ventricle, both ventricles, LV area at risk, total area at risk, and LV area at risk infarcted were traced directly on acetate sheets and planimetered on a digitizing tablet (Videoplan 2, Kontron electronics, Germany). The weights of the ventricular areas, areas at risk, and infarcted areas were then calculated by multiplying the average fractions of ventricular areas, areas at risk, and infarcted areas for each slice with the slice weight and summing the products. Total weight of LV area at risk was expressed as a percentage of total LV weight, and the LV infarct size was expressed as percentage of the LV area at risk infarcted. Because the size of ischemic area may influence occurrence of arrhythmias (63), total area at risk, expressed as percentage of both ventricles, was also calculated.

Because body temperature is a major determinant of infarct size (9), infarct size values were also normalized to the normal body temperature of the pig (38.5°C) (21) by the relation presented by Duncker et al. (9): normalized infarct size at 38.5°C = measured infarct size + [(38.5  body temperature during ischemia) × 20].

Histological examination. Tissue samples from the pig that was kept alive for 6 h after microembolization were taken from the myocardium supplied by the left circumflex coronary artery. The samples were fixed in 2% glutaraldehyde, postfixed in osmium tetroxide, dehydrated in graded etanols, and embedded in Epon. Ultrathin sections (120 nm) were cut on LKB-ultratome III, contrasted with uranyl acetate and lead citrate, and examined in a Philips EM 301S electron microscope for ultrastructural evidence of necrosis.

Tissue samples from three of the hearts that were pretreated before LAD occlusion and had been fixed in 10% Formalin were also examined for evidence of necrosis. These samples were studied by immunohistochemical techniques and by conventional histology. Immunohistochemical techniques were used to examine the myocardial distribution of fibrinogen and albumin, proteins that appear to be specific for indicating irreversible injury if present within myocardial fibers (15, 31, 32). Immunostaining was performed with antisera against human albumin and fibrinogen (Dakopatts, Glostrup, Denmark). Sections were automatically processed in a Ventana immunostaining machine (Ventana Medical Systems, Tucson, AZ) using the Ventana diaminobenzidine-detection kit. Histological evaluation for coagulation necrosis was performed by light microscopy on routinely processed, paraffin-embedded, hematoxylin-eosin-stained, 5-µm thick sections. In addition, both the distribution and the density of microspheres within tissue samples from three of the hearts that were pretreated before LAD occlusion were examined. This examination was performed by light microscopy on hematoxylin-eosin-stained, 10-µm thick sections.

Estimation of entrapped microspheres. The density of microspheres in pretreated hearts was estimated by the following formulas. where and Statistical analysis. Data are expressed as group means ± SE. A one-way repeated-measures ANOVA was applied on normally distributed data obtained before and after injection of microspheres. When the normality of data distribution was questionable, Friedman repeated-measures ANOVA on ranks was applied. If ANOVA showed an overall difference, Dunnett's test was used to compare data obtained after injection with preinjection data. An unpaired t-test was applied to compare infarct size, temperature, VES, and hemodynamics between the two groups of pigs. Comparison of the incidence of ventricular fibrillation was performed using Fisher's exact test. To test the relationship between the number of injected microspheres and the maximal change in CBF, Pearson product moment correlation coefficient (r) was calculated. A value of P < 0.05 was considered statistically significant.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Adverse effects. Tween 80 induced a biphasic coronary flow response in 3 of 11 pigs, whereas Macrodex induced hypotension and decrease in CBF in 1 of 6 pigs. These adverse responses waned by repetitive injections of the agents.

Effects on CBF by microembolization. Injection of microspheres into the left atrium induced an increase in CBF. The magnitude of the increase was strongly dependent on the number of microspheres, because a significant correlation was found between the number of injected microspheres and the maximal change in CBF (r = 0.84; P < 0.001) (Fig. 2). By linear regression this relation is described as: maximal increase in CBF (%) = 2.8 ± 1.5 + (5.8 ± 0.7 × 10⁷ × number of injected microspheres). Maximal change in CBF occurred at 95 ± 11 s after the start of the randomized injections. A final injection of 60 × 10⁶ microspheres increased CBF to 133 ± 6% of preinjection flow 107 ± 45 s after the start of the injection.

View larger version (15K): [in this window] [in a new window]   Fig. 2.   Maximal increase in coronary blood flow (CBFmax) as percentage of preinjection flow after injection of 0, 5, 10, 15, 30, and 40 × 106 microspheres. First-order regression line for all data on plot is shown.

Table 1 shows CBF at different intervals after injection of different amounts of microspheres. By injections of 30 × 10⁶ microspheres or more, the accompanying flow response lasted for at least 20 min. Twenty minutes after the start of the injection with 30 × 10⁶ microspheres, CBF was still 7 ± 2% above the preinjection flow (P < 0.05). However, in the second part of the study, no cumulative effect on the acute CBF responses by two injections of 40 × 10⁶ microspheres was observed. The first injection of 40 × 10⁶ microspheres increased CBF to 125 ± 7% of preinjection flow, whereas the second injection increased CBF to 123 ± 10% of preinjection flow (P = not significant). Figure 3 shows a typical tracing after the injection of 40 × 10⁶ microspheres. No change of statistical significance was found in any hemodynamic variable by injections of the different doses of microspheres (Table 2).

View larger version (17K): [in this window] [in a new window]   Fig. 3.   Tracing from one experiment showing change in CBF to injection of 40 × 106 microspheres into left atrium.

Effects on CBF by microembolization after intracoronary administration of 8-SPT. In the three pigs that were pretreated with 8-SPT, hyperemic flow response was not observed by the injection of 40 × 10⁶ microspheres.

Metabolic alterations by microembolization. Injection of microspheres that altered CBF induced a concomitant, but inverse, change in arteriovenous oxygen difference (Fig. 4). Accordingly, no significant changes were found in MO₂ by microembolization (data not shown). No change of statistical significance was found for either lactate extraction, arteriovenous difference in pH, or PCO₂ by injection of the different doses of microspheres (data not shown).

View larger version (16K): [in this window] [in a new window]   Fig. 4.   A: number of microspheres injected into left atrium plotted vs. CBF in percentage of preinjection flow, 1.5 min after start of injection (CBF1.5 min). B: number of microspheres injected into left atrium plotted vs. arteriovenous difference in hemoglobin oxygen saturation in percentage of preinjection value, 1.5 min after start of injection (a-v HbO2sat1.5 min). First-order regression lines for all data on either plot are shown (in means ± SE): CBF1.5 min = 102.1 ± 1.0 + (0.26 ± 0.05 × number of injected microspheres); a-v HbO2sat1.5 min = 100.6 ± 0.7  (0.19 ± 0.03 × number of injected microspheres).

Infarct size. Pretreatment with microspheres increased infarct size of the left ventricle from 60 ± 3% of area at risk in the control group to 84 ± 6% of area at risk in the pretreated group (P < 0.01) (Fig. 5). Although not of statistical significance, a trend toward a higher body temperature during ischemia was found in the pretreated group compared with the control group (38.7 ± 0.3°C vs. 38.2 ± 0.2°C). However, when the infarct sizes from the pigs are normalized to an equal temperature, according to the relation described by Duncker et al. (9), there is still a significant difference between the pigs pretreated and not pretreated with microspheres (at 38.5°C: 80 ± 5% and 65 ± 5%, respectively; P < 0.05).

View larger version (31K): [in this window] [in a new window]   Fig. 5.   Area at risk as percentage of left ventricular weight (LVW) and infarct size as percentage of area at risk in 6 control pigs (solid columns) and in 5 pigs that were pretreated with microspheres (hatched columns). Columns represent means ± SE; * P < 0.01 vs. control.

No significant difference was found in the area at risk of the left ventricle between pigs pretreated and not pretreated with microspheres, either when analyzed in grams (26 ± 3 g and 31 ± 6 g, respectively) or as percentage of LV weight (22 ± 3% and 26 ± 4%, respectively) (Fig. 5). Furthermore, no differences of statistical significance were found in any hemodynamic variables before LAD occlusion between pigs pretreated and not pretreated with microspheres (Table 3).

Arrhythmias. Electrocardiogram recordings were successfully obtained in all control pigs and in four of five pretreated pigs. No significant difference was found in the total number of VES during the last 10 min before LAD occlusion between pigs pretreated and not pretreated with microspheres (3 ± 2 and 4 ± 3, respectively). Neither was any significant difference found in the total number of VES during the initial 10 min of ischemia between pigs pretreated and not pretreated with microspheres (14 ± 8 and 41 ± 10, respectively). Furthermore, no significant difference was found in the occurrence of ventricular fibrillation between the two groups. During the initial 10 min of the LAD occlusion, one of the pretreated pigs fibrillated compared with none of the control pigs. Between 10 and 45 min of ischemia, ventricular fibrillation occurred in two of the control pigs and in one of the pretreated pigs.

No significant difference was found in the total area at risk between pigs pretreated and not pretreated with microspheres, either when analyzed in grams (29 ± 4 and 3 ± 57 g, respectively) or as a percentage of ventricular weight (19 ± 3% and 24 ±3%, respectively).

Histological examination. No ultrastructural evidence of necrosis was found by electron microscopy of tissue samples from the pig that was kept alive for 6 h after microembolization. Light microscopic examination of myocardial tissue samples from the three pigs pretreated with microspheres before LAD occlusion showed distinct infarction within the area at risk, but no signs of infarction were found within pretreated myocardium that did not undergo total ischemia. Furthermore, no accumulation of fibrinogen and albumin was found within pretreated myocardium that did not undergo total ischemia. Microscopic examination of myocardial tissue samples from the three pigs pretreated with microspheres before LAD occlusion demonstrated that microspheres were spread throughout the ventricular wall. The density of microspheres within the tissue was 2.0 ± 0.2 × 10⁴ microspheres/g myocardium. This value corresponds well with the calculated value of 2.0 ± 0.2 × 10⁴ microspheres/g myocardium.

Effects on CBF reserve by microembolization. No change in peak reactive hyperemic flow, induced by a 20-s LAD occlusion, was found by injection of 40 × 10⁶ microspheres (510 ± 13% before injection vs. 503 ± 9% after injection, P = not signifcant).

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The present study demonstrates a linear relationship between the number of 15-µm microspheres injected into the left atrium and the subsequent increase in CBF. Furthermore, it demonstrates that embolization of coronary vessels with such microspheres, in amounts that increase myocardial blood flow due to adenosine release, reduces ischemic tolerance against infarction.

Effects of microembolization on CBF. In the present study, we show that injection of as little as 5 × 10⁶ microspheres into the left atrium induces a significant increase in CBF. Although it is known that coronary microembolization may increase CBF (22, 24, 43, 65), this is the first study reporting that such a small amount as 5 × 10⁶ of 15-µm microspheres does affect CBF. When larger amounts of microspheres were injected, the CBF response was proportionally greater, and the maximal increase in CBF was linearly related to the number of microspheres injected.

The linear relation between the number of injected microspheres and the maximal increase in CBF following microsphere injection indicates that the microspheres were homogeneously scattered throughout the myocardium and in far less amounts than those causing maximal coronary embolization. This was also verified by the histological findings. Experiments in dogs with 15-µm microspheres have shown that resting CBF increases dose dependently up to ~37% of maximal embolization (~2.0 × 10⁵ microspheres/g myocardium), and thereafter decreases almost linearly until maximal embolization (~5.0 × 10⁵ microspheres/g myocardium) (23, 24).

One implication of the present study is that the CBF response has to be taken into account when microspheres are used to measure myocardial blood flow. A prerequisite for correct tissue flow measurement by injecting microspheres is that the particles introduced into the circulation mix uniformly in the blood and are distributed in proportion to blood flow to the various tissues. However, the fact that spheres are discrete particles implies stochastic errors with the technique. The size of such errors depends among other things on the number of microspheres injected. By increasing the number of injected microspheres, the precision level of the flow measurements will increase according to equations presented by Buckberg et al. (6) and Dole et al. (8). However, according to our results, the number of injected microspheres should at least be kept below 2,500 microspheres/ml aortic flow to avoid biased results. Our findings are of particular importance when colored microspheres, combined with light intensity absorption spectrophotometry (36), are used to determine blood flow. When compared with radiolabeled microspheres, two to three times more colored spheres are required due to a lower signal-to-noise ratio in the detection process (36, 55). Kowallik et al. (36), who introduced the use of colored microspheres combined with light intensity absorption spectrophotometry, had to inject as much as 5.0 × 10⁵ colored microspheres (when white or blue spheres were used) into the LAD of pigs to be able to measure myocardial blood flow. In that study, the pigs were of the same size as in the present study and the size of the myocardial tissue samples in which the amounts of dyes were determined, averaged 1.1 g. From the data given by White and Bloor (66) and results obtained in the present study, 5.0 × 10⁵ microspheres into the LAD corresponds to an atrial injection of about 30 × 10⁶ microspheres, which increases CBF by as much as ~20%. The number of colored microspheres, needed for precise flow measurements, can to some extent be reduced by increasing the size of the tissue samples or by using microspheres dyed with colors with suitable absorbance characteristics.

In the present study, we also confirmed an observation previously reported that the microsphere-suspending agent Tween 80 may induce an adverse coronary biphasic flow response (19). This occurred in 27% of the pigs. Because the CBF responses occurred without any significant changes in systemic hemodynamics, it seems necessary to record CBF continuously to avoid erroneous data when myocardial tissue flow is measured by microspheres.

Ischemic tolerance after microembolization. A plethora of studies have described cardioprotective properties of adenosine in different settings of ischemia-reperfusion injury (12, 17, 44). Accordingly, we hypothesized that coronary embolization by microspheres in amounts that do not induce myocardial necrosis itself, but increase CBF due to adenosine release, can increase ischemic tolerance against infarction. Although we increased CBF by microembolization, a hyperemic response that could be blocked by 8-SPT, ischemic tolerance against infarction did not increase but actually decreased.

Our finding of reduced ischemic tolerance against infarction in microembolized tissue cannot be ascribed to variations in known determinants of infarct size as occlusion time (27, 51, 56), body temperature (7, 9), collateral blood flow (30, 51, 53), size of area at risk (30, 39), or systemic hemodynamic variables (18, 41, 48, 52). Control and pretreated pigs were subjected to exactly the same occlusion period. Corrections for variations in body temperature between the pigs were performed according to the relationship described by Duncker et al. (9). Because pigs have extremely few native collaterals (10, 58, 66), it is not necessary to incorporate collateral flow as a covariate in the analysis of infarct size. Although there are few native collaterals in pigs, the relation between the area at risk and the infarcted area has, according to Koning et al. (35), a positive intercept through the area at risk axis. Accordingly, the reduced ischemic tolerance in microembolized myocardium may actually be more pronounced than that reported in the present study because there was a trend, although not statistically significant, toward a larger area at risk in control animals. With regard to systemic hemodynamic variables, no differences of statistical significance were found.

In the present study a total amount of ~2.0 × 10⁴ microspheres passed into each gram of myocardium in pretreated pigs. According to histological examinations, most of these microspheres were entrapped within the heart. However, this amount of microspheres did not cause any infarction itself; at least no sign of infarction was found by histological examination in pretreated myocardium that did not undergo total ischemia. This is in agreement with results obtained by Hori et al. (25) on dogs. They found that embolization by <3.0 × 10⁵ microspheres (15 µm in diameter) per gram of myocardium did not cause myocardial necrosis (25).

At present we do not know why microembolization reduces ischemic tolerance against infarction. However, there are at least three different mechanisms that may be involved. First, microembolization occurred without any significant decrease in myocardial metabolism but with a substantial increase in CBF. These findings, which are in agreement with previous reports (24, 43), indicate compensatory increased metabolism in nonischemic areas surrounding the patchy ischemic foci with reduced metabolism. Because myocardial metabolic rate at the moment of coronary artery occlusion influences infarct size development (28, 40, 41, 45), nonischemic tissue with increased metabolism may be more prone to ischemic damage. Second, oxygen-derived free radicals are known to cause ischemia-reperfusion injury (1, 16, 33). Because such radicals are generated in patchy ischemic lesions caused by microvascular obstructions (23), they may contribute to the reduced ischemic tolerance in the pretreated pigs. Third, although microembolization can induce myocardial ischemia without causing necrosis (25), it may increase infarction after coronary occlusion by restricting reflow. However, additional studies are required to clarify whether embolization by microspheres actually induces changes in metabolism, free radical generation, and/or reflow, which affect myocardial ischemic tolerance. In the present study, no significant difference was found in the occurrence of ventricular arrhythmias between control pigs and pretreated pigs. However, the variability in VES within groups was too great to reach an acceptable statistical power and, accordingly, the results are not conclusive.

Methodological considerations. Pigs were used because they have few native collaterals (10, 58, 66), and therefore, it is not necessary to incorporate collateral flow as a covariate in the analysis of infarct size and arrhythmias. We chose to use 15-µm polystyrene microspheres for several reasons. They are large enough to be almost completely entrapped within the microcirculation (4, 14, 20, 62). Thereby, a quantitative relation between the number of injected microspheres causing microvascular embolization and the CBF can be obtained. Although they do not cause reversible microembolization, which would probably be preferred in the clinical setting, they are small enough not to occlude wide segments of the microcirculation. Thereby, a substantial degree of embolization can be induced without causing myocardial necrosis. Furthermore, embolization with microspheres of this size has been reported to mimic the syndrome of exertional angina with a normal coronary arteriogram (25). Also, microspheres of 15 µm have been shown to induce a sustained coronary hyperemic response in dogs because of the massive release of adenosine from the ischemic myocardium (24, 26, 59). Finally, microspheres of this size are preferred when blood flow is measured with microspheres (55). Accordingly, findings by use of 15-µm microspheres may have implications for blood flow measurements with microspheres.

Clinical implications. According to our findings, coronary microembolism reduces myocardial ischemic tolerance against infarction in pigs. If this is also the case in the human myocardium, efforts should be made to detect and treat microthrombi in patients prone to coronary occlusion. Furthermore, in such patients, procedures involving microembolization of the myocardium, e.g., use of microbubbles in contrast echocardiography, should be performed with caution. However, the clinical relevance of reduced ischemic tolerance in microembolized myocardium remains to be determined.