INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

VENTRICULAR FIBRILLATION (VF) occurring in the absence of heart failure or hypotension is a common complication of acute myocardial infarction (4, 6, 32). Most of the studies that have analyzed the influence of primary VF on the prognosis after an acute myocardial infarction (3, 4, 6, 29, 32) have shown an adverse outcome associated with VF, despite its adequate detection and treatment (3, 4, 6, 32). The reasons for this adverse outcome are far from being fully understood, and available data on infarct size or recurrent ischemia in patients with and without VF are inconclusive. Some data suggest that the poorer prognosis in patients with primary VF could be explained, at least in part, by lower patency rate in the infarct-related artery. In contrast with animal models, in which VF is frequently brought about by coronary reperfusion (22), primary VF is uncommon in patients with acute myocardial infarction receiving thrombolytic therapy (4, 15, 18, 32), and there is no evidence of more successful reperfusion in patients presenting it. On the contrary, a delayed creatine kinase peak has been found in these patients (5), and in the Thrombolysis in Myocardial Infarction Phase II (TIMI-II) Trial, an angiographic examination performed 18-48 h after thrombolytic treatment in 2,546 patients found an occluded infarct-related artery more frequently in the patients who developed primary malignant arrhythmias during the first 24 h after study entry than in those patients without arrhythmias (4). These findings suggest that such arrhythmias pinpoint either failed reperfusion or early coronary reocclusion.

Reocclusion of infarct-related coronary arteries may occur within the first few hours after successful thrombolysis despite appropriate antithrombotic and anticoagulant treatment and is associated with a worse prognosis (24, 27, 31). The ability of early coronary angiography to predict subsequent reocclusion from culprit lesion characteristics (shape, residual stenosis, or thrombosis, TIMI grade flow) has widely ranged among several studies (14, 24, 25, 27, 30), which suggests that other factors may influence reocclusion.

The present study tested the hypothesis that primary VF during acute, transient myocardial ischemia favors the occurrence of spontaneous coronary reocclusion in the following hours, and we sought to assess the possible mechanisms of this influence. The association between VF during ischemia and coronary reocclusion was analyzed in a series of pigs submitted to prolonged coronary ligation followed by reperfusion, in which the effects of coronary intimal injury and aspirin on infarct size were investigated (2).

	METHODS

Top Abstract Introduction Methods Results Discussion References

Animal preparation and study design. Fifty-two farm pigs of either sex (25-35 kg) received 10 mg/kg im azaperone (Stresnil, Janssen Pharmaceutica) followed by a 10 mg/kg bolus of thiopental sodium and a continuous intravenous infusion to maintain anesthesia. The animals were intubated and connected to a mechanical ventilator (Monaghan 228 Ventilator, Littleton, CO) using room air. One mammary vein and the right femoral artery were catheterized, the right carotid artery was dissected, a midline sternotomy was performed, and the pericardium was opened. The left anterior descending coronary artery (LAD) was dissected free at its midpoint and surrounded by an elastic snare as previously described (12). Two pairs of ultrasonic crystals 1 mm in diameter were inserted into the inner third of the left ventricular wall. The crystals of each pair were placed 1-2 cm apart along a plane perpendicular to the long axis of the left ventricle. One pair was implanted in the myocardium to be made ischemic, and the other pair was inserted in the lateral wall of the left ventricle.

Immediately after we administered 1.5 mg/kg iv lidocaine, we performed a 48-min coronary occlusion of the LAD in all animals by tightening the elastic snare. We selected this duration of coronary occlusion according to a regression model of infarct size over the occlusion period calculated from previous studies (13) to obtain incomplete, histochemically detectable infarcts. At the end of the occlusion period, the snare was released and reperfusion was allowed for 6 h. In a random manner (2 × 2 factorial design), the animals were given 250 mg of intravenous aspirin or placebo 90 min before coronary occlusion, and intimal injury of the LAD or no injury immediately before occlusion was applied. Intimal injury was produced with an intracoronary catheter introduced through the right carotid artery. Four animals were excluded due to a ruptured LAD, left main coronary occlusion, failure in the injection of fluorescein (making it impossible to measure the area at risk), and prolonged LAD catheterization maneuvers (causing transient ischemia), respectively, and 48 experiments remained valid. When an animal presented reocclusion, the treatment allocation of this animal was reintroduced in the randomization matrix to obtain the same number of animals without reocclusion in each treatment group at the end of the study. This resulted in 9 pigs receiving aspirin, 14 with LAD injury, 11 with aspirin and LAD injury, and 14 with none of these interventions (2). The experimental methods were approved by the Research Commission of the Hospital General Vall d'Hebron and followed the "Guiding Principles in the Care and Use of Animals."

Study monitoring. Frequent measurements of arterial pH, PO₂, and PCO₂ were performed to adjust the ventilatory parameters to maintain normal blood gases. Serum potassium was measured before we performed the sternotomy. Aortic blood pressure was continuously monitored with a crystal quartz transducer (Coulbourn Instruments, Lehigh Valley, PA). When VF occurred, it was converted to sinus rhythm by internal direct-current shocks. The paddles were placed on the lateral walls of the right and left ventricles, far away from the LAD. The defibrillation shock waveform was sinusoidal, and the strength of the first shock was 10 J, with subsequent shocks of 20 J if defibrillation was not accomplished after the first attempt. The two-segment ultrasonic length signals were analyzed by an ultrasonic dimension system (System 6/200, Triton Technology, San Diego, CA) and monitored with an HM 205-3 oscilloscope (Hameg Instruments, Frankfurt Main, Germany). These signals, along with lead II of the electrocardiogram and aortic pressure, were amplified in a Coulburn Modular Instrument System (Coulburn Instruments) and continuously recorded by a thermic pad recorder (MT-9500, Astro-Med, West Warwick, RI) at a sampling rate of 200 kHz. The conditioned signals were digitized (Tecfen ISC-16E/CR digitization card, RC Electronics, Goleta, CA) at a sampling rate of 100 Hz per channel, and digitized signals were stored on a hard disk.

Animals were carefully monitored during the reperfusion period to detect coronary reocclusion. Coronary occlusion could easily be identified by visual inspection of the LAD-dependent myocardium (in which the epicardial surface adopted a typical bluish color) and of the segment of the LAD distal to the occlusion site, which in the pig model, virtually lacking collaterals (11, 23), consistently collapses a few seconds after coronary occlusion. In most cases reocclusion was accompanied by widening of QRS signal in the local electrogram obtained with the ultrasonic crystals.

Segment length measurements. Mean aortic pressure was calculated by integration of the digital aortic pressure curve in the stored signals, with the aid of appropriate software (Enhanced Graphics Acquisition and Analysis, RC Electronics). Segment length measurements were performed on the paper tracings. End diastole was identified in the crystal tracing of the control segment as the point of initial shortening during isovolumic systole. End systole was identified in the control trace as a break in the shortening curve corresponding to the onset of isovolumic relaxation (12, 28) (Fig. 1). Systolic shortening was calculated as follows: systolic shortening = [end-diastolic length (EDL)  end-systolic length]/EDL. Systolic bulging was defined as the ratio of maximal segment length during systole and EDL of the same beat times 100%. Segment length measurements were performed in normally conducted sinus beats; before, 5, 15, and 47 min after coronary occlusion; and 15, 30 min, and every hour after reperfusion. EDL was expressed as a percentage of values immediately before coronary occlusion. The reduction in the amplitude of passive segment length change in the ischemic region throughout the occlusion period was also measured as an index of myocyte rigor contracture (11). The amplitude of passive segment length change was expressed as a percentage of the value 5 min after coronary occlusion.

View larger version (78K): [in this window] [in a new window]   Fig. 1.   Representative example of electrocardiogram (ECG), hemodynamic, and left anterior descending (LAD) and control segment length signals before and immediately after coronary occlusion (CO, solid arrow). Note increase in end-diastolic length (thin arrows), abolition of systolic shortening, and development of systolic bulging in ischemic region during the first minutes after CO.

Electrocardiogram changes and ventricular arrhythmias. The duration of the QRS complex in the local electrocardiogram was measured at baseline and at several time points throughout the occlusion period from the paper tracings. Five consecutive R-R intervals were measured, and their standard deviation (SD) and coefficient of variation (CV = SD × 100/mean) were calculated as indexes of R-R variability. The number of premature ventricular beats during the occlusion period; the number, time, and duration of VF episodes; and the number of electric shocks applied were also counted.

Regional myocardial blood flow. Regional myocardial blood flow was analyzed in 25 animals by injecting colored microspheres 15.5 µm in diameter (Triton Technology) into the left atrium at baseline and after 30 min and 5 h of reflow (2). Microspheres were not injected during coronary occlusion given the paucity of collateral circulation in the porcine heart (23). Aortic blood was withdrawn at a rate of 12 ml/min (2400-003 infusion pump, Harvard Apparatus) for 10 s before and 90 s after each injection. After we excised the heart, transmural myocardial samples were obtained from the fourth slice. Blood and myocardial samples were digested, and their dye content was quantified by determining their photometric absorption (ultraviolet 160 A, Shimazdu, Kyoto, Japan). Regional myocardial blood flow was calculated as the specific absorbance of each dye per sample multiplied by the withdrawal rate of reference arterial blood and divided by the specific absorbance in reference blood (2).

Postmortem studies. After 6 h of reperfusion, the LAD was reoccluded and 5 ml of 10% fluorescein was injected into the left atrium. The heart was immediately excised, immersed in Ringer solution at 4°C, and cut into 5- to 7-mm slices perpendicular to its long axis. The slices were weighed in a high-precision scale (model AJ50, Mettler instruments, Greifensee, Switzerland), illuminated from the basal side with 356-nm ultraviolet light to delineate the area at risk, and imaged by a Sony TR 705E video camera. Instead of the fourth slice used for blood flow measurements, the apical surface of the fifth slice was imaged. The images, along with a reference scale, were digitized online into 768 × 576 pixel images (Matrox IP8 digitization card, Matrox Electronic Systems, Dorval, Quebec, Canada).

Myocardial samples (50-100 mg) were obtained from the central and lateral thirds of the zone at risk and from the control zone in the apical side of two consecutive slices for water content measurement as previously described (12). The samples were introduced in assay tubes that had been previously weighed. The tubes containing the samples, along with an empty assay tube, were weighed immediately before and after desiccation during 24 h at 100°C. Myocardial water content was calculated as the difference between fresh and dry weight divided by dry weight (expressed as ml of water/100 g of dry tissue).

The myocardial slices were then incubated at 37°C in 1% triphenyltetrazolium chloride, buffered for pH = 7.4 for 5-10 min, and imaged again under white light. The zone at risk and the area of necrosis were measured in the digital images by using commercially available software (Image Pro-Plus, Media Cybernetics, Silver Spring, MD), and the mass of myocardium at risk and infarct size were calculated from these measurements and from the weight of the slices (2, 12).

The slices, along with a segment of the LAD including the occlusion site, were fixed in 10% formaldehyde. The third slice was dehydrated in graded alcohol and embedded in paraffin. Serial transverse 4-µm-thick sections were obtained from the arterial samples and from the whole slice (Polycut microtome, Reichert Jung Cambridge Instruments, Heidelberg, Germany), stained with hematoxylin and eosin and Schiff's periodic acid, and examined with a Nikon Labphot microscope. Examination of the LAD was performed to confirm reocclusion and to quantify the severity of intimal damage produced by the occluding snare itself. Endothelial denudation was graded according to the following scores: 0, no denudation; 1, denudation of <50% of the endothelial circumference; and 2, denudation of >50% of the circumference. Internal elastic lamina status was graded as follows: 0, intact; 1, focal disruption; and 2, extensive disruption. In the heart sections, the content of polymorphonuclear leukocytes (PMN) and the intensity of red blood cell extravasation were quantified in the subendocardial, midventricular, and subepicardial layers of the area at risk and in control myocardium as indexes of microvascular damage. The content of PMN was graded according to the following scores: 0, absent; 1, scant; 2, intravascular plugs; and 3, intravascular plugs and PMN present in interstitium. Red blood cell extravasation was graded as follows: 0, no extravasation; 1, slight extravasation; 2, extensive infiltration of the interstitial space; and 3, areas of confluent intramyocardial hematoma.

Statistical analysis. Statistical analysis was performed using commercially available software (SPSS/PC+ 4.0). Changes in physiological parameters within the same animal were assessed by analysis of variance for repeated samples and by paired t-tests. Group comparisons were performed either by ²-tests, by t-tests for independent samples, or by nonparametric tests (Mann-Whitney U-test) if variable distribution departed significantly from the normal distribution. The statistical significance of multiple "a posteriori" comparisons was corrected according to the Bonferroni method. The relationship between normally distributed variables was assessed by linear regression analysis. Stepwise logistic regression analysis was performed to identify variables independently associated with either coronary reocclusion or VF. A critical two-tailed P value of 0.05 was used for all tests. Values are expressed as means ± SE or as median (range) if they were not normally distributed. Histological scores are expressed also as means ± SE.

	RESULTS

Top Abstract Introduction Methods Results Discussion References

Laboratory results and functional data. Baseline serum potassium was 3.2 ± 0.1 meq/l. Hemodynamic data are summarized in Table 1. Immediately after occlusion, heart rate increased slightly, whereas mean aortic pressure did not depart from baseline values. After reperfusion, heart rate remained stable and mean aortic pressure showed a slight and transient increase.

In four animals, the quality of the segment length traces was not good enough to allow reliable measurements. Measurements from the remaining animals are summarized in Table 1. In the control zone, EDL and systolic shortening remained stable throughout the experiment. In LAD-dependent myocardium, coronary occlusion was immediately followed by an increase in EDL, a decrease in systolic shortening, and the development of systolic bulging (Fig. 1). Five minutes after coronary occlusion, EDL averaged 111.5 ± 0.9% of the basal value (P < 0.001), systolic shortening 15.6 ± 2.6% of the basal value (P < 0.001), and systolic bulging 2.2 ± 0.3%. The amplitude of passive segment length change decreased progressively after 15-20 min of coronary occlusion, and at the end of the occlusion period the amplitude was 74.0 ± 2.5% of the value 5 min after occlusion (P < 0.001). Reperfusion induced a rapid and marked reduction in EDL and a small recovery of systolic shortening.

Electrocardiogram changes and ventricular arrhythmias. The duration of the QRS complex in the local electrocardiogram increased from 22 ± 2 ms at baseline to 82 ± 7 ms 15 min after coronary occlusion (P < 0.001) and was 76 ± 8 ms at the end of the occlusion period [P = not significant (NS) with respect to the value at 15 min]. The SD of five consecutive R-R intervals was 17.5 ± 4.5 ms at baseline (CV = 1.6 ± 0.3%), 7.0 ± 1.0 ms 15 min after occlusion (0.9 ± 0.1%, P = 0.05 with respect to baseline), and 7.3 ± 1.1 ms (0.9 ± 0.1%) 48 min after occlusion.

The mean number of premature ventricular beats during the occlusion period was 78 ± 14 beats. One animal had VF 5 and 7 min after occlusion, which was converted to sinus rhythm by 9 shocks. Eleven animals (including this one with immediate VF) had a median of 1 (1-7) episode of delayed ischemic VF, 27.2 ± 1.8 min after occlusion, which was converted by a median of 3 (1-9) shocks. The duration of VF during coronary occlusion averaged 52 ± 9 s. Although ischemic VF was always preceded by premature ventricular contractions (16 ± 5 in the 5-min period before the first episode of VF), it was not associated with a higher premature ventricular beat density during coronary occlusion (70 ± 16 vs. 83 ± 21 premature beats during the 48-min occlusion period, respectively, in animals presenting and not presenting ischemic VF, P = NS, or 21 ± 12 vs. 18 ± 7 premature beats, respectively, during the first 20 min of occlusion, P = NS). Immediately after reperfusion, all animals had repetitive bursts of idioventricular rhythm, and 21 animals had VF (most of them during the first minute after reflow), which was reverted by a median of 2 (1-8) shocks. The duration of postreperfusion VF was 34 ± 4 s. In total, 26 animals had VF either during the occlusion period or immediately after reperfusion. There was no association between ischemic VF and postreperfusion VF.

Coronary reocclusion. Reocclusion of the LAD was detected in 12 animals. It occurred in the first hour after reperfusion in 7 animals, in the second hour in 4 animals, and after 4 h in 1 animal. Reocclusion was transient in 7 animals and permanent in 5.

Regional myocardial blood flow. Regional myocardial blood flow did not change in the control region. In the area at risk, blood flow increased in the animals without reocclusion from 1.34 ± 0.17 ml · min¹ · g¹ before coronary occlusion to 2.23 ± 0.23 ml · min¹ · g¹ 30 min after reflow (P = 0.02) and returned to values similar to baseline (1.35 ± 0.12 ml · min¹ · g¹) after 5 h of reflow.

Postmortem studies. The area at risk averaged 14.5 ± 0.6 g, which represented 10.3 ± 0.5% of total ventricular mass. Infarct size represented 68.1 ± 6.7% of the area at risk in the animals with coronary reocclusion and 22.3 ± 3.3% in those animals without reocclusion (P < 0.001). Myocardial water content was 394 ± 5 ml/100 g dry tissue in the control region. In the area at risk it was 492 ± 2 ml/100 g in the animals with coronary reocclusion and 472 ± 11 ml/100 g in those animals without reocclusion (P = NS).

Histological examination of the LAD at the occlusion site showed local endothelial and internal elastic lamina damage in most animals. Intravascular thrombosis ranging from small mural platelet aggregates to occlusive thrombi was observed in all but in six animals. Occlusive thrombi were present in all animals with permanent reocclusion (Fig. 2) and in none of the remaining animals. Endothelial denudation at the occlusion site was of similar severity in animals with and without coronary reocclusion (0-2 score 1.6 ± 0.2 and 1.7 ± 0.1, respectively, P = NS). The damage of the internal elastic lamina at the occlusion site was also similar in animals with and without reocclusion (0-2 score 0.9 ± 0.3 and 1.2 ± 0.1, respectively, P = NS). Histological analysis of myocardial tissue did not detect PMN or red blood cell extravasation in control myocardium. In the area at risk, the mean histological 0-3 score for PMN content was 1.6 ± 0.3 in animals with reocclusion and 1.3 ± 0.2 in those without reocclusion (P = NS). The mean 0-3 scores for red blood cell extravasation were 1.1 ± 0.3 and 0.6 ± 0.1, respectively (P = NS).

View larger version (158K): [in this window] [in a new window]   Fig. 2.   Histological section of LAD coronary artery at occlusion site in one animal with coronary reocclusion. A thrombus rich in platelets completely occludes the lumen. Hematoxylin and eosin, original magnification ×100.

Influence of catheter-induced coronary intimal injury and aspirin. Neither LAD injury nor aspirin treatment were associated with any detectable difference in hemodynamic or segment length changes throughout the experiments nor with differences in the size of the area at risk. Serum potassium was 3.4 ± 0.1 and 3.1 ± 0.1 meq/l in animals receiving and not receiving aspirin, respectively (P = 0.02). The QRS duration, the changes in R-R intervals, and the number of premature ventricular beats were not modified by intimal injury or aspirin. The incidence of VF during coronary occlusion was also similar in animals with and without LAD injury (24 vs. 22%, respectively, P = NS) and receiving or not receiving aspirin (15 vs. 29%, respectively, P = NS); the incidence of postreperfusion VF was also similar (36 vs. 52% and 45 vs. 43%, respectively, both P = NS). Reocclusion rate was also similar in animals with and without catheter-induced coronary injury (28 and 22%, respectively, P = NS), and aspirin-treated animals showed a strong trend to lower reocclusion rate with respect to those receiving placebo (10 vs. 36%, respectively, P = 0.09).

Predictors of coronary reocclusion. The number and time of premature ventricular beats during coronary occlusion were similar in animals with and without subsequent reocclusion (Fig. 3). VF during the occlusion period, but not postreperfusion VF, was strongly associated with coronary reocclusion (Fig. 4 and Table 2). Reocclusion occurred in 6 of 11 (54%) animals that had VF during the ischemic period and in 6 of 37 (16%) remaining animals (P = 0.029). Among animals presenting VF, the total number of electric shocks applied to terminate the arrhythmia in each animal showed a near-significant positive association with reocclusion (P = 0.06). Among hemodynamic and segment length variables, a larger increase in EDL in the ischemic region induced by coronary occlusion was also associated with reocclusion, and this association remained highly significant after we applied the Bonferroni correction for multiple comparisons (Table 2). When the variables listed in Table 2, along with the presence or absence of LAD injury and aspirin treatment, were included in a logistic regression analysis, the only variable independently associated with reocclusion was the early postischemic increase in EDL (P = 0.019). The duration of VF during the ischemic period showed a nearly significant association with coronary reocclusion (adjusted P = 0.069).

View larger version (22K): [in this window] [in a new window]   Fig. 3.   Distribution of premature ventricular beats during CO in animals with (solid bars) and without (open bars) subsequent reocclusion.

View larger version (43K): [in this window] [in a new window]   Fig. 4.   Distribution of episodes of ventricular fibrillation (VF, dark squares) during CO and in the first minutes after reperfusion (Rep). VF during CO was more frequent in the animals that subsequently had coronary reocclusion (top) than in those without reocclusion (bottom). Each horizontal line denotes one case. Rectangles represent multiple episodes of VF close in time.

In animals without reocclusion, ischemic VF was not associated with regional myocardial blood flow or with infarct size, myocardial water content, PMN content, or red blood cell extravasation in the area at risk, as shown in Table 3 (animals with coronary reocclusion were not analyzed due to the potential influence of reocclusion itself on these variables). The increase in EDL induced by coronary occlusion was not correlated with the reduction in the amplitude of passive segment length change during the ischemic period (r = 0.17, P = NS). In animals without reocclusion, EDL during ischemia was not correlated with regional blood flow at 30 min (r = 0.31) or 5 h (r = 0.21) after reflow or with infarct size (r = 0.01), myocardial water content (r = 0.35), PMN content (r = 0.14), or blood extravasation (r = 0.06) in the area at risk.

Predictors of VF during coronary occlusion. The potential association between regional ventricular expansion during acute ischemia, as expressed by the increase in EDL in the ischemic region and VF, was analyzed. VF was strongly associated with the increase in EDL: 15 min after coronary occlusion EDL was 116.3 ± 2.0% of baseline in animals that had subsequent VF during the occlusion period and 111.2 ± 0.9% in those without VF (P = 0.01). VF occurred in 1 of 19 animals in which EDL 15 min after occlusion was below 110% of baseline, in 7 of 24 animals in which it was between 110 and 120%, and in 3 of 5 animals in which it was over 120% (Fig. 5). By contrast, neither the potassium levels nor the hemodynamic changes, the changes in systolic shortening, the magnitude of systolic bulging, or the size of the area at risk were associated with ischemic VF (Table 3). A logistic regression analysis, including the increase in EDL after coronary occlusion, along with the size of the area at risk, serum potassium, allocation to coronary injury and/or aspirin, and the hemodynamic and segment length variables listed in Table 3, was performed. This analysis retained the increase in EDL as the only variable independently associated with ischemic VF (P = 0.022). There was no correlation among EDL and the number of premature ventricular beats (r = 0.16, P = NS), the amplitude of the QRS signal, or R-R variability during ischemia, and none of these parameters was associated with ischemic VF, although a nonsignificant (P = 0.16) trend toward a lower R-R interval variability in animals subsequently presenting ischemic VF was observed (Table 3). In contrast with ischemic VF, postreperfusion VF was not associated with the increase in EDL during ischemia.

View larger version (13K): [in this window] [in a new window]   Fig. 5.   Individual (circles) and mean (horizontal lines) values of end-diastolic segment length (EDL) in ischemic region 15 min after CO. Increase in EDL over baseline values was significantly greater in animals with subsequent VF during ischemic period than in those without VF. Animals with coronary reocclusion are represented by filled circles.

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

This study demonstrates that VF occurring during coronary occlusion is strongly associated with an increased rate of spontaneous reocclusion and that this association cannot be explained as a trivial consequence of more extensive or more severe ischemia. Marked expansion of the ischemic region early after coronary occlusion appears as a prominent and independent determinant of both ischemic VF and coronary reocclusion.

Association of VF and coronary reocclusion. Primary VF might favor coronary reocclusion by several mechanisms. First, it causes transient hemodynamic collapse, with subsequent slowed blood flow and stagnation, which can facilitate thrombus formation and clot propagation in a coronary segment with damaged endothelium (26). This thrombus can cause residual coronary stenosis during reperfusion, which is a potent stimulus for subsequent thrombus growth and reocclusion (8). In this study, the duration of VF during the ischemic period showed a nearly significant independent association with coronary reocclusion, suggesting that blood stagnation during VF may have a role in favoring reocclusion.

Second, the electric shocks applied to convert the arrhythmia could increase tissue injury and favor thrombosis (33). However, the lack of an independent association of reocclusion with the number of shocks applied to restore sinus rhythm is against a role of electrical injury.

Finally, an increased myocardial damage secondary to ischemia might be responsible for both malignant arrhythmias (20) and less effective reflow after release of coronary occlusion (7), favoring reocclusion. However, in the present study primary VF was not associated with larger infarct size or with more pronounced myocardial edema, hemorrhage, or PMN infiltration. This suggests that an increased tissue damage secondary to ischemia does not account for the link between VF and reocclusion. Moreover, the virtually complete lack of native collaterals in the pig model (23) makes the existence of marked individual differences in the severity of ischemic insult among the animals included in this study unlikely.

We failed to detect any association of the generally accepted risk factors for ischemic VF (20) with coronary reocclusion. By contrast, the increase in EDL in the ischemic region during coronary occlusion, reflecting early regional expansion, was shown to be strongly associated with ischemic VF and emerged as the best predictor of reocclusion. The mechanism by which ischemic expansion is associated with coronary reocclusion cannot be elucidated by this study. The increase in EDL during ischemia was not correlated neither with myocardial blood flow nor with infarct size or with other parameters reflecting tissue damage, which suggests that the association between expansion and coronary reocclusion is not explained by increased myocardial injury. Whatever the mechanism of this influence is, the results suggest that factors related with the precedent ischemia may affect reocclusion. This is in agreement with the reported difficulty to predict reocclusion merely from culprit lesion characteristics (25).

Association of regional expansion during ischemia and VF. Neither the amplitude of the QRS signal nor the changes in R-R variability, the size of the area at risk, or other previously described determinants of ischemic VF (20) such as serum potassium levels or heart rate at baseline or throughout the ischemic period were significantly associated with this arrhythmia, although it cannot be ruled out that the trend toward less R-R variability in pigs developing VF could have reached statistical significance with a larger sample size. Ischemic VF also was not associated with infarct size. The fact that in most animals VF occurred relatively early after coronary occlusion, before irreversible myocyte injury can be reasonably expected to take place (13), is in agreement with this lack of association. Previous studies do not convincingly demonstrate the existence of a relationship between the incidence of VF during transient coronary occlusion and the magnitude of infarct size as a fraction of the area at risk (10, 17).

By contrast, the magnitude of the increase in EDL induced by coronary occlusion showed a strong association with VF, suggesting that regional myocardial stretch caused by this expansion plays a role in the development of ischemic VF. The increase in EDL was not correlated, however, with the number of premature ventricular contractions during coronary occlusion. The lack of an association between the density of premature contractions and VF observed in this study may help to explain this apparent contradiction. The experimental induction of myocardial stretch by changing mechanical loading conditions has been demonstrated to cause several electrophysiological changes and ventricular arrhythmias (9, 16, 19, 21). This mechanoelectrical feedback has been claimed to account for the occurrence of malignant arrhythmias in patients with chronic left ventricular dilatation as a consequence of a myocardial infarction (16). It has also been hypothesized that spontaneous regional stretch during acute myocardial ischemia could favor the development of malignant arrhythmias in the same manner as global experimentally induced stretch does in the absence of ischemia (9). To our knowledge, the results of the present study provide the first direct, observational evidence in support of this hypothesis.

Methodological considerations and study limitations. All animals received lidocaine, which could have affected the occurrence of arrhythmias (1), but it is highly improbable that lidocaine interfered with the association of VF and coronary reocclusion: all animals received the same dose of the drug; at that dose lidocaine does not have relevant effects on thrombogenicity and does not modify coronary flow or hemodynamics. In this study, histological examination of the LAD at the site of coronary occlusion confirmed the presence of intimal damage caused by the occluding snare, which may explain the lack of effect of additional extensive coronary damage on reocclusion. The lack of continuous monitoring of LAD flow in this study prevented us to assess the possible occurrence of periodic flow reductions preceding coronary reocclusion. However, because our aim was to analyze coronary reocclusion, we did not place a flow probe around the LAD to prevent any possible distortion of the artery caused by the probe itself. Aspirin could have affected the association between VF and reocclusion, but the potential influence of this effect was accounted for by means of adequate multivariant analysis. Moreover, the overall results were similar, with less statistical power, if analyses were made after exclusion of the 20 animals receiving aspirin [VF during coronary occlusion occurred in 60 vs. 11% of the animals with or without the presence of reocclusion (P = 0.02), and logistic regression analysis retained EDL 15 min after coronary occlusion as the only predictor of ischemic VF (P = 0.06) and of reocclusion (P = 0.04)]. Finally, the conclusion on the association between the increase in EDL and ischemic VF would had been strengthened if the study had initially been designed specifically to analyze this relationship. However, it is highly improbable that the finding of this association in this study was due to chance, because it was tested "a priori" in view of the arrhythmogenic effects of stretch demonstrated in previously published studies.

Implications. This study demonstrates that factors related with the preceding myocardial ischemia, in addition to those related with the culprit lesion, can influence coronary reocclusion, and establishes a clear association among early ischemic myocardial expansion, ischemic VF, and coronary reocclusion. Although it is tempting to speculate that myocardial stretch favors VF during ischemia and that thrombotic activation induced by cardiac arrest favors reocclusion, the cause-and-effect relationships among myocardial ischemic expansion, ischemic VF, and reocclusion cannot be elucidated from the present study. However, whatever these relationships are, the present results could help explain the observation that primary VF in patients with acute myocardial infarction is associated with a lower patency rate of the infarct-related artery, with a poorer prognosis. This could have therapeutic implications and supports the notion that a more aggressive management might be justified in these patients. Finally, the results stress the deleterious effect of early regional expansion during myocardial ischemia and the potentially beneficial effect of measures aimed to prevent this expansion.