Mortality Within 24 Hours of Thrombolysis for Myocardial Infarction

Transcription

1 2658 Mortality Within 24 Hours of Thrombolysis for Myocardial Infarction The Importance of Early Reperfusion Neal S. Kleiman, MD; Harvey D. White, MD; E. Magnus Ohman, MD; Allan M. Ross, MD; Lynn H. Woodlief, MS; Robert M. Califf, MD; David R. Holmes, Jr, MD; Eric Bates, MD; Matthias Pfisterer, MD; Alec Vahanian, MD; Eric J. Topol, MD; for the GUSTO Investigators* Background A paradoxical increased risk of death has been reported during the first 24 hours after thrombolysis for myocardial infarction. The mechanism of this phenomenon is not known, nor is its relation to the success or failure of reperfusion. The present study was a prospectively designed analysis of deaths occurring within the first 24 hours in the GUSTO trial. Methods and Results There were patients enrolled in GUSTO, a randomized comparison of streptokinase with intravenous or subcutaneous heparin, accelerated tissue-type plasminogen activator (TPA), and combination of streptokinase and TPA. An angiographic mechanistic substudy examined reperfusion (using the TIMI flow grading criteria) 90 minutes after the assigned thrombolytic regimen was begun in 1567 patients. There were 1125 deaths (2.8%) within 24 hours ("early deaths") and 1726 additional deaths (4.2%) after 24 hours but within 30 days ("later deaths"). At the time of presentation, the most potent predictors of early death were hypotension and sinus tachycardia. In a multiple logistic regression model, lower systolic blood pressure, shorter height, higher heart rate, and the absence of prior smoking distinguished early death from later death. Reinfarction occurred in 26 patients (2.4%), shock in 572 patients (52%), atrioventricular block in 308 patients (28%), and tamponade in 106 patients (10%) dying early compared with 262 (15%), 788 (46%), 396 (23%), and 74 (4%) respective patients dying later. There were no differences in early mortality among the lacebo-controlled studies of thrombolytic therapy have demonstrated a reduction in mortality after acute myocardial infarction.2-4 This survival benefit may be partially offset by a paradoxical increase in the risk of death during the first 24 hours after thrombolytic therapy is begun.5'6 Nearly half of the deaths among patients receiving thrombolytic drugs occur during this period.5-9 As further advances are made in reperfusion strategies, incremental mortality reductions will require that the mechanisms of death Received May 5, 1994; revision accepted July 6, From the Section of Cardiology (N.S.K.), Department of Medicine, Baylor College of Medicine, Houston, Tex; Cardiology Department (H.D.W.), Green Lane Hospital, Auckland, New Zealand; Division of Cardiology (E.M.O., L.H.W., R.M.C.), Department of Medicine, Duke University Medical Center, Durham, NC; Division of Cardiology (A.M.R.), George Washington University Medical Center, Washington, DC; Department of Internal Medicine (D.R.H.), Mayo Clinic, Rochester, Minn; Division of Cardiology (E.B.), University of Michigan Medical Center, Ann Arbor, Mich; Division of Cardiology (M.P.), University Hospital thrombolytic regimens for the first 6 hours after randomization. By 24 hours, however, mortality was 2.89% for streptokinase recipients, 2.84% for combination therapy recipients, and 2.36% for accelerated TPA recipients (P=.005). There was little difference among patients with differing flow grades in the infarct artery during the first 4 hours, although mortality among patients with grade 2 flow was initially higher. After the fourth hour, there were very few additional deaths during the first day in patients with grade 3 flow. At 24 hours, mortality was 2.35% for patients with flow grade 0 or 1, 2.92% for patients with flow grade 2, and 0.89% for patients with flow grade 3. Conclusions Even with aggressive management regimens, mortality within the first 24 hours accounted for a large proportion of postthrombolytic deaths. Patients dying early were more likely to present with pump failure than were those dying later and were more likely to die of events related to left ventricular dysfunction, although cardiac tamponade also accounted for a significant minority of these deaths. Thus, the severity of the clinical presentation rather than the underlying risk factors predicts early mortality. Based on the angiographic substudy data, it appears that rather than hastening early mortality, successful restoration of complete antegrade flow in the infarct-related artery protects against early death. (Circulation. 1994;90: ) Key Words * thrombolysis * infarction * reperfusion a mortality * tamponade during this period be elucidated. Whether these deaths result from myocardial necrosis already present at the time of clinical presentation, rupture, increased bleeding, reperfusion-induced injury, or failure to restore antegrade flow in the infarct-related artery is of critical importance in designing new therapeutic strategies. Prior studies have been unable to address these questions due to either a lack of angiographic data or a limited sample size. The Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Basel, Basel, Switzerland; Cardiology Service (A.V.), Hopital Tenon, Paris, France; and Department of Cardiology (E.J.T.), Cleveland Clinic Foundation, Cleveland, Ohio. Reprint requests to Neal S. Kleiman, MD, Section of Cardiology, The Methodist Hospital, 6535 Fannin, MS F-905, Houston, TX Presented in part at the American Heart Association 66th Annual Scientific Sessions, Atlanta, Ga, November 8, *A complete listing of GUSTO Investigators can be found in N Engl J Med. 1993;329: American Heart Association, Inc.

2 Coronary Arteries (GUSTO) trial afforded a unique opportunity to examine the relation between reperfusion and early mortality because it consisted of randomly assigned groups of patients in whom differing rates of early reperfusion were documented and had adequate statistical power to assess differences in early mortality.1'10 Accordingly, the present study was prospectively designed to test the hypothesis that aggressive thrombolysis would lower mortality at 24 hours. Methods Study Design and Data Collection The methods and results of GUSTO have been published.1 Briefly, patients with ST-segment elevation on the presenting ECG within 6 hours of symptom onset were randomly assigned to receive treatment with 1.5 million U streptokinase (SK) (Kabikinase) with U heparin SC every 12 hours; 1.5 million U streptokinase with 5000-U heparin bolus IV followed by 1000 U/h; accelerated tissue-type plasminogen activator (TPA) (Alteplase, Genentech) 15-mg bolus then 0.75 mg/kg over 30 minutes and then 0.5 mg/kg over 60 minutes, with the same intravenous heparin regimen; or combination TPA plus SK consisting of 1 mg/kg TPA (maximum of 90 mg) over 60 minutes with 10% as a bolus, 1 million U SK over 60 minutes, and 5000-U bolus heparin IV and then 1000 U/h. All patients received 160 mg or more aspirin (Bayer) PO. Atenolol (10 mg IV) (Tenormin, ICI Pharma) was recommended for patients who had no contraindication to 3-blockade. The primary end point of the study was all-cause mortality at 30 days. The study protocol specified that if there were no difference in the primary end point between the SK arms, they would be combined for subsequent analyses. The time to death was defined as the interval between randomization and death. At the time of protocol design, mortality within 24 hours after enrollment was prespecified as a secondary end point. Deaths occurring during this period are referred to as early deaths. Deaths occurring after 24 hours but within 30 days are considered later deaths. Reinfarction was defined as the presence of at least two of the following four criteria: recurrent ischemic symptoms lasting >15 minutes after the resolution of the symptoms of the index infarction, the occurrence of new ST-T-wave changes or new Q waves, a second elevation in cardiac enzymes above the upper limit of normal or by an additional 20% if already elevated, or angiographic reocclusion of a previously documented patent coronary artery. Recurrent ischemia was defined according to the clinical judgment of the individual investigators, but guidelines given on the case report form comprised angina or its equivalent lasting at least 15 minutes. A stroke-related death was defined as any death in a patient who had suffered a stroke. The GUSTO protocol called for computed tomographic scanning or magnetic resonance imaging of the brain in all patients with suspected stroke. This was accomplished in 95% of such patients.1 Bleeding was defined as severe if it caused hemodynamic compromise, whereas moderate bleeding was defined as that requiring transfusion but without causing hemodynamic compromise. Baseline demographic characteristics were collected from the enrollment form; all data were double-entered at the data coordinating centers and were centrally audited for inconsistencies. Source documentation was performed in approximately 12% of cases at each site. An angiographic mechanistic substudy of 2431 patients examined reperfusion at the randomly assigned time points of 90 minutes, 180 minutes, 24 hours, or 5 to 7 days after therapy was begun. The randomization scheme for this substudy was designed so that half of the patients would be assigned to undergo angiography at the 90-minute point. To ensure that baseline characteristics were identical to those of patients enrolled in the main study, centers enrolling patients in this substudy were required to include all of their GUSTO study Kleiman et al Early Mortality After Thrombolysis 2659 TABLE 1. Timing of Mortality in GUSTO No. of Cumulative Deaths (%) Time Overall Mortality Stroke Mortality Nonstroke Mortality 2 h 247 (0.7) 3 (0.01) 267 (0.7) 4 h 520 (1.3) 5 (0.01) 511 (1.2) 6 h 641 (1.6) 7 (0.02) 630 (1.5) 12 h 876 (2.1) 19 (0.05) 847 (2.1) 24 h 1125 (2.7) 59 (0.14) 1056 (2.6) 30 h 2851 (7.0) 244 (0.60) 2591 (6.4) patients.10 Because of these entry requirements, patients in the angiographic substudy were highly representative of patients in the main cohort. Perfusion on the 90-minute angiogram was determined by a core laboratory blinded to the treatment assigned and was graded using the Thrombolysis in Myocardial Ischemia Phase I (TIMI I) flow criteria." Statistical Analysis Descriptive statistics are provided as percentages for categorical variables and mean (25th, 50th, and 75th percentiles) for continuous variables. For categorical variables, statistical testing was performed using the x2 test. For continuous variables, the Wilcoxon rank-sum test was used. Results are also presented as odds ratios with 95% confidence intervals. A logistic regression model was used to assess predictors of death within 24 hours versus death after 24 hours. Candidate predictors included baseline characteristics from which a final model was determined using a backward elimination method (elimination criterion, P<.05). Results Overall Timing of Mortality Among the patients enrolled in GUSTO, complete data concerning survival status at 24 hours are available on , or 99.9%. There were 1125 early deaths (2.8%) and 1726 additional later deaths (4.2%). Thus, 39% of all deaths occurred within 24 hours. More detailed timing of these deaths is shown in Table 1. Six hundred forty-one early deaths (57%) occurred within the first 6 hours after beginning thrombolytic therapy. The median time from randomization to initiation of thrombolytic therapy was 35 minutes. Of the 1125 early deaths, 124 (11%) occurred shortly after the time of study entry but before the assigned treatment could be begun. Baseline Demographic Characteristics and Hospital Course Important baseline characteristics of patients dying early are compared in Table 2 with those surviving the first 24 hours, and odds ratios for death occurring within the early period are given in Table 3. Patients dying early were older and more likely to be female, to have anterior infarction, previous angina, or a history of prior infarction than those surviving 30 days but not more so than those dying later. The median time from the onset of symptoms to beginning therapy was slightly (13 minutes) longer in those patients dying early than in those surviving 30 days. Although the most potent predictors of early death were hypotension and sinus tachycardia, and a larger proportion of patients dying

3 2660 Circulation Vol 90, No 6 December 1994 TABLE 2. Baseline Demographic Characteristics According to Time of Death Characteristic Died <24 h (n=1 125) Died at 24 h to 30 d (n= 1726) Survived 30 d (n=37 979) Age, y 70.2 (64, 72, 78) 70.5 (65, 72, 78) 60.2 (52, 61, 69) Male, n (%) 647 (58) 1039 (60) (76) Weight, kg 73.8 (64, 73, 82) 73.9 (64, 73, 82) 79.8 (70, 79, 89) Height, cm (160,168,175) (160,168,175) (165,172,178) Time to treatment, h 3.31 (2.0, 3.0, 4.3) 3.43 (2.3, 3.2, 4.4) 3.07 (2.0, 2.8, 3.9) Prior angina, n (%) 478 (44) 769 (45) (36) Prior Ml, n (%) 306 (28) 478 (28) 5893 (16) Diabetes, n (%) 220 (20) 416 (24) 5343 (14) Hypercholesterolemia, n (%) 233 (25) 463 (29) (35) Hypertension, n (%) 492 (45) 823 (48) (37) History of smoking, n (%) 488 (49) 944 (56) (71) Anterior Ml, n (%) 616 (55) 961 (56) (38) Killip class 1, n (%) 648 (59) 1109 (65) (87) Killip class 1i, n (%) 254 (23) 439 (26) 4397 (12) Killip class Ill, n (%) 76 (7) 100 (6) 370 (1) Killip class IV,n(%) 117 (10.7) 64 (4) 132 (0.4) BP <100 mm Hg systolic, n (%) 360 (34) 209 (13) 2620 (7) HR >100 bpm, n (%) 237 (23) 331 (20) 3181 (9) Systolic BP, mm Hg (90,110,130) (110,125,141) (113,130,144) HR, bpm 84.4 (66, 82,100) 82 (68, 80, 95) 74.8 (62, 73, 85) Mi indicates myocardial infarction; BP, blood pressure; HR, heart rate, and bpm, beats per minute. Data for continuous variables are presented as mean (25th, 50th, and 75th percentiles). early were in Killip class III or IV than were those surviving, it is noteworthy that 82% were in Killip class I or II at the time of presentation. When the baseline characteristics of patients dying early and those dying later are compared, they appear to be essentially simi- TABLE 3. Odds Ratios for Death Within 24 Hours Odds Ratio 95% of Death Confidence Baseline Characteristic <24 h Limits BP <100 mm Hg systolic , 7.4 Age >70 y , 5.0 Heart rate > 100 bpm , 4.5 Female sex , 2.5 Prior myocardial infarction , 2.3 Anterior myocardial infarction , 2.2 Diabetes ,1.7 Prior angina , 1.5 Prior CABG ,1.7 Hypercholesterolemia , 0.72 Family history , 0.71 Current smoker , 0.54 Killip class IV , 28.1 Treatment <2 h ,1.03 BP indicates blood pressure; bpm, beats CABG, coronary artery bypass graft surgery. per minute; and lar, except in the proportions of patients with Killip class IV heart failure and with a history of smoking among patients dying later. The following characteristics in the regression model predicted death within 24 hours versus death after 24 hours: decreased systolic blood pressure, decreased height, increased heart rate, and absence of prior smoking history. The complications that occurred during hospitalization are reported in Table 4. Ventricular arrhythmias occurred in 45% of patients dying early, and atrioventricular block occurred in 28%. Recurrent ischemia and reinfarction were rarer events among patients dying early than among those surviving. Cardiac tamponade occurred in 10% of patients dying early, in 4% of patients dying later, and in 0.4% of patients who survived 30 days. Compared with the patients who died within 24 hours but in whom tamponade was not reported, patients with tamponade were likely to be female (62% versus 40%, P<.0001) and to have received thrombolytic therapy within the first 2 hours of symptoms (87% versus 77%, P=.015) but were less likely to have had a previous infarction (8.7% versus 30%, P<.0001). Mean systolic and diastolic blood pressures were higher in the former group of patients (122 versus 110 mm Hg, P<.0001, and 75 versus 69 mm Hg, P=.0019, respectively). There were no differences in age, the proportion of patients with anterior infarctions, or the proportion of patients receiving any of the four assigned thrombolytic regimens. As depicted in Table 5, intra-aortic balloon counterpulsation was performed in 9% of the patients who died early. Fewer than half the

4 TABLE 4. Complications and Clinical Events Died Died >24 h Survived Complication/.24 h and <30 d 30 d Event, n (%) (n=1125) (n=1726) (n=37 979) Reinfarction 26 (2.4) 262 (15.3) 1336 (3.5) Ischemia 91 (8.3) 539 (32) 7469 (20) Shock 572 (52) 788 (46) 1084 (3) Anaphylaxis 5 (0.5) 7 (0.4) 199 (0.5) Hypotension 724 (65) 910 (53) 3238 (9) AV block 308 (28) 396 (23) 2672 (7) Sustained ventricular tachycardia 299 (26) 448 (26) 1775 (5) Ventricular fibrillation 401 (36) 478 (28) 1858 (5) Tamponade 106 (10) 74 (4) 132 (0.4) Atrial fibrillation/flutter 106 (10) 440 (26) 3273 (9) Acute MR 20 (2) 98 (6) 452 (1) VSD 24 (2) 55 (3) 115 (0.3) Bleeding Severe 25 (2) 123 (7) 346 (1) Moderate 48 (4) 358 (21) 4240 (11) AV indicates atrioventricular; MR, mitral VSD, ventricular septal defect. regurgitation; and patients dying early were placed on mechanical ventilators, and 256 (23%) had temporary pacemakers inserted. Cardioversion or defibrillation was performed in 485 (44%). Coronary angiography was performed in 112 patients (10%) dying early, and percutaneous transluminal coronary angioplasty and coronary artery bypass graft surgery were performed considerably less frequently. In 34 patients, the angiogram was required by the angiographic substudy protocol, whereas in 78 it was performed at physician discretion. Of the 78 patients (7%) undergoing angiograms not mandated by the angiographic substudy, data regarding patency of the infarctrelated artery were available in 58 (74%). In these 58 patients, the investigators reported that 4 (7%) had TIMI grade 2 flow and 3 (5%) had TIMI grade 3 flow, whereas the artery was reported to be "patent" (actual TABLE 5. Treatments During Hospitalization Died Died >24 h Survived <24 h and <30 d 30 d Procedure, n (%) (n=1125) (n=1726) (n=37 979) Coronary angiogram 112 (10) 550 (32) (58) PTCA 61 (5) 191 (11) 8644(23) CABG 7 (0.6) 137 (8) 3375 (9) IABP 97 (9) 284 (17) 1099 (3) Temporary pacemaker 256 (23) 337 (20) 2272 (6) Mechanical ventilation 460 (41) 729 (42) 3479 (9) Intravenous 3-blocker 215 (19) 621 (36) (45) PTCA indicates percutaneous transluminal coronary angioplasty; CABG, coronary artery bypass graft surgery; and IABP, intra-aortic balloon pump. Kleiman et al Early Mortality After Thrombolysis 2661 flow grade unspecified) in 4 (7%). Thus, 47 patients (81%) were reported to have a totally occluded artery. Among the hour survivors in whom nonprotocol-mandated angiograms were done during the first day, data regarding patency were available in 1527 patients (72%). TIMI grades 2 and 3 flow were reported in 296 (19%) and 431 (28%) of these 1527 patients, respectively, whereas the artery was reported to be "patent" in an additional 77 patients (5%). Early Mortality, Thrombolytic Treatment Assignment, and Patency Fig 1 shows the timing of mortality according to the thrombolytic regimen assigned. There were no differences in 24-hour mortality among the SK-treated groups (SK with intravenous heparin, 2.94%; with subcutaneous heparin, 2.85%; P=.693); therefore, they are combined in this analysis. There was very little difference in early mortality among patients receiving the assigned thrombolytic regimens until the sixth hour, when the curves began to diverge. From this point on, mortality progressed at a slower rate in the accelerated TPA-treated patients. Odds ratios and confidence limits for mortality according to treatment are shown in Fig 2. As shown in Table 6, patients dying within the first 6 hours were more likely to have diabetes, previous angina, or prior myocardial infarction and were more likely at the time of clinical presentation to be either hypotensive or in cardiogenic shock than were those dying between hours 6 and 24. Stroke-related mortality (Fig lb) accounted for very few early deaths in any of the treatment groups. In Fig 3, early mortality according to patency status is shown among patients enrolled in the angiographic substudy and assigned to undergo a protocol-directed angiogram 90 minutes after beginning thrombolytic therapy. There was little difference among patients with differing flow grades in the infarct artery during the first 4 hours, although mortality among patients with grade 2 flow was initially higher. However, after the fourth hour, there were very few additional deaths during the first day in patients with grade 3 flow. At 24 hours, mortality in this group was approximately one fourth of that in patients with grade 0, 1, or 2 flow. Discussion Even with modern aggressive reperfusion strategies, mortality during the early hours following acute myocardial infarction accounts for more than one third of the postinfarction deaths, but it is reduced by the restoration of antegrade blood flow in the infarctrelated artery. Our data show that rather than being responsible for an "early hazard," early reperfusion appears to protect against it. This principle is consistent with the finding that, as a rule, myocardial infarctions in this study were more likely to be associated with pretreatment evidence of left ventricular dysfunction and pump failure in patients who died early than in those surviving the first 24 hours. At least two distinct etiologies are likely to be responsible for the mechanism of these early deaths: pump dysfunction and myocardial rupture. First, a large body of evidence suggests that pump failure played a central role in the early deaths in this report. Hemodynamic evidence of left ventricular compromise was consider- ably more common among patients dying within the first

5 2662 Circulation Vol 90, No 6 December p a ISK t-pa + SK b -it-pa...sk - t-pa + SK time from Randomization (hours) FiG 1. a, Plot of mortality according to time from randomization for each treatment arm in GUSTO. In this and subsequent figures, arrowheads are used on the abscissa to represent the median time to beginning thrombolytic therapy. t-pa indicates tissue-type plasminogen activator (TPA); SK, streptokinase. b, Plot of stroke-related and nonstroke-related mortality for each treatment arm in GUSTO. Uk 1.5 Non-stroke Mortality a Stroke mortality Time from RandomizatlQn (hours) 24 hours, and in the multivariate model of mortality, systolic blood pressure and Killip class were the strongest predictors distinguishing the deaths occurring within the first 24 hours from those that occurred within the remainder of the first month after infarction. We have previously shown concordant advantages in both survival and early patency for the accelerated TPA regimen compared with the other thrombolytic regimens tested'10 as well as improved left ventricular function and survival at 30 days in patients with complete restoration of antegrade flow (TIMI grade 3) in the infarct artery.'0 In the present study, a persistent survival advantage for accelerated TPA became apparent several hours after the initiation of therapy. In parallel, at approximately this time, a distinct survival advantage also became evident for patients in the angiographic study who, regardless of thrombolytic strategy, had grade 3 flow in the infarct-related artery on an angiogram performed 90 minutes after beginning thrombolytic therapy. Similarly, few patients in the main study who underwent angiography and who died within the first day had patent infarct-related arteries. In combination, these findings indicate very strongly that failure of reperfusion is a critical factor contributing to early as well as later mortality. On the other hand, the surprising finding that TIMI grade 2 flow was associated with a higher mortality during the first 4 hours than TIMI grade 0 or 1 flow may indicate that partial reperfusion has a detrimental effect on the myocardium. Alternately, suboptimal flow despite lysis of an occlusive coronary thrombus may identify a group of patients in whom myocardial necrosis has occurred rapidly with localized edema and microvascular stasis and in whom early mortality is likely to be high. However, the numbers of patients with grade 0 or 1 and grade 2 flow are not adequate to allow clear distinction between these groups. The early survival advantage for patients with TIMI grade 3 flow also suggests that the pump failure responsible for most of the early deaths is a consequence of myocardial ischemia rather than preexisting necrosis. Accordingly, the time to beginning thrombolytic therapy was only marginally greater in patients who died early, and the proportion with a prior infarction was also only slightly higher. The finding that early mortality is lower rather than higher for patients with early grade 3 flow and for those assigned to treatment with accelerated TPA also indicates strongly that reperfusion-induced

6 Kkeiman et al Early Mortality After Thrombolysis 2663 Odds Ratio 1.5 I FiG 2. Odds ratios with 95% confidence limits for death after randomization to treatment with strep. tokinase (SK) (with intravenous or subcutaneous heparin) versus accelerated tissuetype plasminogen activator (t-pa [TPAJ). ( a Time from Randomization (hours) necrosis is unlikely to play an important role in early mortality. Were this true, mortality would also have been expected to be higher rather than lower in patients with early reperfusion. The presence of cardiac tamponade in 10% of patients dying early suggests that in some patients, rupture of the myocardium may have been the mechanism of death. Given the clinical difficulty of recognizing myocardial rupture, it is possible that this figure underestimates the true frequency of rupture. That shock was reported to develop in only 52% of patients who died early and hypotension was reported in 65% suggests that many of the patients who died early were unexpectedly noted either to be in extremis or to have arrested. Previous authors have reported a biphasic frequency of distribution of cardiorrhexis slitlike ruptures complicating - small infarctions during the first day after infarction and larger ruptures occurring approximately 1 week after the onset of large infarctions. In the first ISIS study, the majority of early deaths in a large subgroup of TABLE 6. Baseline Characteristics of Patients Dying Within 24 Hours Characteristic Died <6 h (n=641) Died at 6 to 24 h (n=484) P Age, y 69 (62, 71, 77) 72 (67, 72, 78) <.001 Age >70 y, n (%) 344 (54) 398 (64) <.001 Male, n (%) 239 (37) 236 (49) <.001 Treated <2 h, n (%) 142 (26) 100 (22).185 Anterior Mi, n (%) 341 (53) 275 (57).182 Killip class, n (%) (56) 303 (64) (23) 110 (23) ill 34 (7) 42 (7) IV 90 (14) 27 (6) <.001 BP <100 mm Hg systolic, n (%) 264 (45) 96 (21) <.001 HR >100 bpm, n (%) 141 (24) 96 (21).217 Prior angina, n (%) 295 (48) 183 (39).003 Prior Mi, n (%) 192 (31) 114 (24).011 Diabetes, n (%) 142 (23) 78 (17).008 Hypertension, n (%) 276 (45) 216 (46).827 Smoker, n (%) 273 (51) 215 (47).262 Weight, kg 75 (65, 75, 83) 72 (63, 72, 80).034 Height, in 169 (160, 170, 175) 167 (160, 168, 175).048 Mi indicates myocardial infarction; BP, blood pressure; HR, heart rate; and bpm, beats per minute. Data for continuous varables are presented as mean (25th, 50th, and 75th percentiles).

7 2664 Circulation Vol 90, No 6 December TIMI Grade 3 TIMI Grade TIMIGradeOorl ~ 2-1- O_ o 0.r A i. d 71, '~~~~~~~~~~~ i 1 1-1~ 7 7 7,,, 7_,~~~~~~~~~~~~~~ Time from Randomization (hours) FIG 3. Plot of mortality according to TIMI flow grade on the angiogram performed at 90 minutes in the angiographic study. Data are taken according to flow grade regardless of treatment arm. patients undergoing autopsy were secondary to rupture of the ventricle.12 However, in that study of early intravenous fl-blocker therapy, it is likely that the frequency of fatal rupture would be increased because patients with incipient heart failure would be less likely to be enrolled, thus decreasing the proportion of patients likely to die of early severe pump failure. In the second phase of the TIMI II study,7 the majority of deaths during the first 18 hours after therapy was begun with accelerated TPA were due to pump failure, whereas the incidence of fatal cardiac rupture during the first 18 hours after therapy was 16%. However, among the 50% of patients in that study who underwent autopsy, evidence of rupture was found in 25%, whereas it was judged to be present in 3% of patients in whom an autopsy was not performed.8 It thus is likely that a modest proportion of early deaths in GUSTO was caused by myocardial rupture. Pathological study of ventricles in which rupture has occurred often demonstrates intramyocardial hemorrhage surrounding the rupture tract. Hemorrhage into the myocardium has also been reported to be more common in patients who have received thrombolytic therapy,13 and it has been hypothesized that intramyocardial hemorrhage predisposes patients to rupture of the heart.14 At least one report has suggested that patients in whom rupture occurs may have a stuttering course.15 It is possible that thrombolytic therapy hastens this progression by dissolving the thrombus that temporarily seals the rupture site, by facilitating the dissection by blood along necrotic tissue planes, or by allowing recovery of myocardial contractility, thus increasing the shear forces that lead to disruption of the myocardium. Were rupture alone responsible for the majority of early deaths, one would not have expected to see clear evidence of a survival advantage for patients with complete restoration of flow and for assignment to treatment with accelerated TPA. It is also possible that by limiting infarct size, reperfusion with thrombolytic therapy decreases the substrate in which myocardial rupture can occur. Thus, there may be a dynamic balance between those factors favoring and those preventing rupture in patients receiving thrombolytic therapy. However, without a control group of patients who did not receive thrombolytics, these possibilities remain speculative. What is more puzzling is the similarity in mortality for the treatment regimens and for the TIMI flow grades during the first 6 hours after treatment was begun. The median time required for reperfusion to occur following treatment with both SK and accelerated TPA has previously been reported to be slightly less than 1 hour.11,16 In addition, a small number of patients in GUSTO died before thrombolytic therapy could be begun. Thus, it would be unlikely for differences in mortality to be seen during and shortly after the first hour. The time period extending up to the sixth hour may represent the time required for sufficient recovery of ventricular muscle function after ischemia ("stunning")17 to prevent or reverse pump failure, or it may represent a period during which a detrimental effect of reperfusion is balanced by the benefit of reperfusion. As an alternative, "reperfusion dysrhythmias"'18 during this extremely early period may be responsible for a number of deaths. In the TIMI II experience, the frequency of lethal dysrhythmias did not differ between patients dying within 4 hours and those dying between 4 and 18 hours; however, in that study, there were small numbers of patients in each of the two groups.5 In the present study, however, there was more evidence of both severe pump dysfunction at the time of clinical presentation and of preexisting cardiovascular disease (Table 6) in those patients who died within the first 6 hours. Thus, it appears that patients who died in this very early period had larger, more hemodynamically significant infarctions or had more preexisting myocardial damage and were less able to survive the initial hours until adequate recovery of left ventricular function could occur. The high "mortality density" within the early hours after reperfusion therapy is in agreement with data from other large-scale studies of patients with myocardial infarction.2-4 The present data are unique, however, in that they very strongly establish a link between early restoration of antegrade coronary arterial blood flow and a reduction of early mortality, thus suggesting that at least a large proportion of the deaths during this period results from a failure of reperfusion to occur. It is also quite clear that earlier reperfusion is associated

8 with an early survival advantage, and these data strongly suggest that future attention be directed toward more reliable means of ensuring rapid reperfusion, even in those patients who do not have pump dysfunction at the time of initial presentation. Acknowledgments This study was supported by a combined grant from Bayer (New York, NY), CIBA-Corning (Medfield, Mass), Genentech (South San Francisco, Calif), ICI Pharmaceuticals (Wilmington, Del), and Sanofi Pharmaceuticals (Paris, France). References 1. The GUSTO Investigators. An international randomized trial comparing four thrombolytic strategies for acute myocardial infarction. N Engl J Med. 1993;329: Gruppo Italiano per lo Studio della Streptochinasi nell'infarto Miocardico (GISSI). Effectiveness of intravenous thrombolytic treatment in acute myocardial infarction. Lancet. 1986;1: ISIS-2 (Second International Study of Infarct Survival). Randomized trial of intravenous streptokinase, oral aspirin, both, or neither among 17,187 cases of suspected acute myocardial infarction. ISIS 2. Lancet. 1988;2: Wilcox R, Olsson C, Skene A, Von Der Lippe G, Jensen G, Hampton J. Trial of tissue plasminogen activator for mortality reduction in acute myocardial infarction: Anglo Scandinavian Study of Early Thrombolysis (ASSET). Lancet. 1988;2: Mauri F, DeBiase A, Franzosi M, Pampallona S, Foresti A, Gasparini M. Analisi dele casue di morte intraospedaliera. G Ital Cardiol. 1987;17: Fibrinolytic Therapy Trialists' (FTT) Collaborative Group. Indications for fibrinolytic therapy in suspected acute myocardial infarction: collaborative overview of early mortality and major morbidity results from all randomised trials of more than 1000 patients. Lancet. 1994;343: Kleiman N, Terrin M, Mueller H, Chaitman B, Roberts R, Braunwald E, Knatterud G, Solomon R, McMahan R, and the TIMI Investigators. Mechanisms of early death despite thrombolytic therapy: experience from the Thrombolysis in Myocardial Infarction phase II (TIMI II) study. J Am Coil Cardiol. 1992;19: Maynard C, Weaver D, Litwin P, Martin J, Kudenchuk P, Dewhurst T, Eisenberg M, Hallstrom A, Chambers J, MITI Project Investigators. Hospital mortality in acute myocardial infarction in Kleiman et al Early Mortality After Thrombolysis 2665 the era of reperfusion therapy (the Myocardial Infarction Triage and Intervention Project) (MITI). Am J Cardiol. 1993;72: Ohman E, Topol E, Califf R, Bates E, Ellis S, Kereiakes D, George B, Samaha J, Kline E, Sigmon K, Stack R, and the Thrombolysis and Angioplasty in Myocardial Infarction (TAMI) Study Group. An analysis of the cause of early mortality after administration of thrombolytic therapy. Coron Artery Dis. 1993;4: The GUSTO Angiographic Investigators. The effects of tissue plasminogen activator, streptokinase, or both on coronary-artery patency, ventricular function, and survival after acute myocardial infarction. N Engi J Med. 1993;329: Chesebro J, Knatterud G, Roberts R, Borer J, Cohen L, Dalen J, Dodge H, Francis C, Hillis D, Ludbrook P, Markis J, Mueller H, Passamani E, Powers E, Rao A, Robertson T, Ross A, Ryan T, Sobel B, Willerson J, Williams D, Zaret B, Braunwald E. Thrombolysis in Myocardial Infarction (TIMI) trial, Phase I: a comparison between intravenous tissue plasminogen activator and intravenous streptokinase. Circulation. 1987;76: ISIS-1 (First International Study of Infarct Survival Collaborative Group). Mechanisms for the early mortality reduction produced by beta blockade started early in acute myocardial infarction: ISIS-1. Lancet. 1988;1: Waller B, Rothbaum D, Pinkerton C, Cowley M, Linnemeier T, Orr C, Irons M, Helmuth R, Wills E, Aust C. Status of the myocardium and infarct-related coronary artery in 19 necropsy patients with acute recanalization using pharmacologic (streptokinase, r-tissue plasminogen activator), mechanical (percutaneous transluminal coronary angioplasty) or combined types of reperfusion therapy. JAm Coll Cardiol. 1987;9: Honan M, Harrell F, Reimer K, Califf R, Mark D, Pryor D, Hlatky M. Cardiac rupture, mortality and the timing of thrombolytic therapy: a meta-analysis. JAm Coil Cardiol. 1990;16: Oliva P, Hammill S, Edwards W. Cardiac rupture, a clinically predictable complication of acute myocardial infarction: report of 70 cases with clinicopathologic correlations. J Am Coll Cardiol. 1993;22: Mueller H, Rao A, Forman S, TIMI Investigators. Thrombolysis in Myocardial Infarction (TIMI): comparative studies of coronary reperfusion and systemic fibrinogenolysis with two forms of recombinant tissue-type plasminogen activator. JAm Coll Cardiol. 1987; 10: Braunwald E, Kloner R. Prolonged postischemic ventricular dysfunction. Circulation. 1982;66: Ganz W, Geft I, Shah P, Lew A, Rodriguez L, Wei T, Maddahi J, Berman D, Charuzi Y, Swan H. Intravenous streptokinase in evolving acute myocardial infarction. Am J Cardiol. 1984;53:1209.