Early Reinfarction After Fibrinolysis
Experience From the Global Utilization of Streptokinase and Tissue Plasminogen Activator (Alteplase) for Occluded Coronary Arteries (GUSTO I) and Global Use of Strategies To Open Occluded Coronary Arteries (GUSTO III) Trials
Background Trials report a 2% to 6% incidence of reinfarction after fibrinolysis for acute myocardial infarction (MI). We combined the Global Utilization of Streptokinase and Tissue plasminogen activator (alteplase) for Occluded coronary arteries (GUSTO I) and Global Use of Strategies To Open occluded coronary arteries (GUSTO III) populations to better define frequency, timing, and clinical predictors of in-hospital reinfarction.
Methods and Results In 55 911 patients with ST-segment elevation myocardial infarction (MI) who were receiving fibrinolysis, we compared baseline characteristics and mortality rate by reinfarction incidence and developed multivariable logistic regression models to predict in-hospital reinfarction and composite of death or reinfarction. Reinfarction occurred in 2258 patients (4.3%) a median of 3.8 days after fibrinolysis; rates did not differ between GUSTO I (4.0%) and GUSTO III (4.2%) or by fibrinolytic assignment (streptokinase, 4.1%; alteplase, 4.3%; reteplase, 4.5%; combined streptokinase and alteplase, 4.4%; P=0.55). Advanced age, shorter time to fibrinolysis, non-US enrollment, nonsmoking status, prior MI or angina, female sex, anterior MI, and lower systolic blood pressure were associated significantly with reinfarction. Patients with reinfarction had higher mortality at 30 days (11.3% versus 3.5% without reinfarction; odds ratio, 3.5; P<0.001) and from 30 days to 1 year (4.7% versus 3.2%; hazard ratio, 1.5; P<0.001). Significant multivariate predictors of in-hospital death or reinfarction included age, Killip class, systolic and diastolic blood pressures, heart rate, anterior MI, smoking status, prior MI, sex, and country of enrollment (all P<0.001).
Conclusion Reinfarction occurs infrequently after fibrinolysis but confers increased risk of 30-day and 1-year mortality. Some predictors of reinfarction differ from known predictors of death after MI. Improved treatment and prevention strategies for reinfarction deserve study.
Received April 25, 2001; revision received July 15, 2001; accepted July 17, 2001.
Fibrinolysis reduces mortality in patients with acute myocardial infarction (MI), as shown conclusively in trials of thousands of patients.1 These trials have characterized the frequency, timing, and clinical predictors of death after ST-segment elevation MI in the fibrinolytic era.2,3 Less attention has focused on incidence, outcomes, and predictors of early reinfarction after fibrinolysis.
Studies report an incidence of reinfarction between 2% and 6% within 4 to 6 weeks after fibrinolysis.4–8 Pooled data from the Thrombolysis and Angioplasty in Myocardial Infarction (TAMI) trials indicate that patients with reinfarction have significantly higher rates of death and heart failure than patients without recurrent ischemic events.9 The much larger Global Utilization of Streptokinase and Tissue plasminogen activator (alteplase) for Occluded coronary arteries (GUSTO I) and Global Use of Strategies To Open occluded coronary arteries (GUSTO III) studies, comprising more than 56 000 patients with MI, provide a unique opportunity (1) to define the incidence and timing of in-hospital reinfarction, (2) to compare baseline characteristics of patients with and without reinfarction, (3) to develop a multivariable model to predict in-hospital reinfarction, (4) to examine the effect of reinfarction on risk of 30-day and 1-year mortality, and (5) to develop a multivariable predictive model for the composite of in-hospital death or reinfarction.
GUSTO I and GUSTO III studies were international fibrinolytic trials that enrolled 56 080 patients with ST-segment elevation MI <6 hours after symptom onset, from 1990 through 1993 and 1995 through 1997, respectively. Entry criteria and treatment protocols for these studies, nearly identical, have been described.2,10
We excluded 164 patients (0.3%) for missing or unknown reinfarction status and 5 for missing mortality information, which left a cohort of 55 911 patients for analyses of 30-day mortality rate and composite of in-hospital death or reinfarction. Another 3254 patients died in hospital and were labeled as having no reinfarction by local investigators. Because these events were not adjudicated, we suspected that some of these patients suffered an unrecognized reinfarction before death. Therefore, we excluded them from analysis of baseline characteristics and reinfarction predictors and tested the final logistic model of in-hospital reinfarction in both the 52 657-patient cohort with known reinfarction status who survived hospitalization and the 55 916-patient cohort with known reinfarction status, regardless of mortality information.
End Points and Definitions
Reinfarction was identified prospectively by local investigators. No protocol-defined algorithm for ECG or cardiac-marker monitoring for recurrent ischemia or revascularization existed. In GUSTO I, reinfarction was defined as a second MI during hospitalization for the first MI, which satisfied ≥2 clinical criteria: (1) recurrent ischemic symptoms for >15 minutes after resolution of symptoms from initial MI, (2) new ST-T-wave changes or new Q waves; (3) reelevation in creatine kinase (CK) or CK-MB to higher levels than normal (or by another 20% if already higher than normal); and (4) angiographic reocclusion of previously documented patent infarct-related artery. In GUSTO III, reinfarction was determined (1) by new, significant Q waves in ≥2 leads different from those affected during initial MI; (2) by reelevation of CK-MB to higher levels than normal (or by another 50% if already higher than normal); or (3) by reelevation of CK-MB to >3 or >5 times upper limit of normal after angioplasty or bypass surgery, respectively.
Case-report forms were used to collect primary data in both studies. International coordinating centers entered and queried for missing data. Missing mortality data were sought by telephone and registered-mail inquiries, locator services, and national registries. Thirty-day mortality data were 99.6% and 99.8% complete for GUSTO I and GUSTO III, respectively, whereas 1-year mortality data were available for 96% and 97% of patients, respectively. Data quality was verified by randomly auditing 10% of the GUSTO I forms against medical records, revealing an error and omission rate of 1.6% for all reinfarction entries.
Continuous data are summarized as medians with interquartile ranges. Prospectively selected baseline and treatment variables were compared between patients with reinfarction and surviving patients without reinfarction, with χ2 tests for discrete variables and Wilcoxon 2-sample rank sum test for continuous variables.
We created multivariable predictive models for in-hospital reinfarction and the composite of in-hospital death or reinfarction by use of logistic regression modeling techniques. We defined prospectively certain baseline factors as candidate variables: hypertension, diabetes, smoking status, hyperlipidemia, prior MI, prior cerebrovascular disease, prior angina, prior bypass surgery, prior angioplasty, enrollment region (United States versus elsewhere), sex, race, Killip class, MI location, trial enrollment (GUSTO I versus GUSTO III), age, time from symptom onset to fibrinolysis, heart rate, diastolic and systolic blood pressures, weight, height, and treatment assignment (alteplase, reteplase, streptokinase, or combination alteplase-streptokinase).
Shape of the relation between each continuous variable and outcome was evaluated by flexible model-fitting methods (restricted cubic-spline transformations),11 which allow graphical and statistical evaluation of assumed linear association. When the relation appeared not to follow a straight line, appropriate transformations were made. Stepwise regression was used to determine the set of main effects and interaction terms of multivariable significance. The final model for reinfarction was tested, and all in-hospital deaths were included and excluded to assess possible bias from misclassification of reinfarctions in nonsurvivors. Two models have been established to predict mortality in the GUSTO I population: 1 for mortality at 30 days, and 1 for mortality from 30 days to 1 year in 30-day survivors.3,12 Before and after adjustment for these models, we evaluated effects of reinfarction on 30-day and 1-year mortality (assuming 30-day survival). We used logistic regression to test 30-day mortality, and Cox proportional-hazard modeling to assess 1-year mortality.
Overall, 2258 of the 52 662 patients (4.3%) had reinfarction (1628 of 38 475 GUSTO I patients [4.2%] and 630 of 14 187 GUSTO III patients [4.4%]) at a median of 3.8 (2.0 to 6.1) days. Table 1 presents baseline characteristics by reinfarction status. Reinfarction incidence did not differ by fibrinolytic assignment (P=0.55; Table 2). Table 2 also shows reinfarction rates associated with other in-hospital medications. Anticoagulation intensity, measured by 24-hour activated partial thromboplastin time, was significantly greater in patients with reinfarction (58 [44 to 85] s versus 55 [42 to 77] s for no reinfarction; P<0.001).
Outcomes of Reinfarction
Among all enrolled patients, 30-day mortality rate more than tripled among those with reinfarction versus no reinfarction (11.3% versus 3.5%; odds ratio, 3.50; 95% confidence interval [CI], 3.14 to 3.90; P<0.001). After adjustment for other predictors of mortality, reinfarction remained significantly associated with increased 30-day mortality (odds ratio, 3.11; 95% CI, 2.78 to 3.49). Median time to death (to 1 year) after reinfarction was 1.99 (0.21 to 11.12) days, with 40% of the deaths occurring within 24 hours of reinfarction. Earlier reinfarction was associated with worse survival; median time to reinfarction was 2.9 (1.3 to 6.0) days for patients who died within 30 days versus 3.8 (2.1 to 5.9) days for 30-day reinfarction survivors (P=0.003). Unadjusted mortality from 30 days to 1 year was 50% higher in patients with reinfarction (4.7% versus 3.2%; hazard ratio, 1.50; 95% CI, 1.21 to 1.88; P<0.001). After adjustment for baseline and hospital factors, a trend remained toward increased mortality during this period in patients with in-hospital reinfarction (hazard ratio, 1.25; 95% CI, 0.97 to 1.61). Table 3 shows relations between reinfarction and other nonfatal, in-hospital complications.
Predictors of Reinfarction
Several baseline factors were important multivariable predictors of in-hospital reinfarction (Figure 1). C-index value (area under the receiver-operator characteristic curve) was 0.64 for the predictive model of reinfarction. Advanced age, shorter symptom duration before fibrinolysis, enrollment outside the United States, and nonsmoking status were the most important predictors of reinfarction. Several predictors had a nonlinear association with reinfarction; for example, relation of heart rate to reinfarction differed for low versus high heart rates. Increased heart rate was associated with a reduced reinfarction risk if ≤90 bpm. At >90 bpm, increases in heart rate were associated with increased reinfarction risk. Symptom duration before fibrinolysis showed a similar, complex relationship, with a steep decrease in reinfarction as time to treatment increased between 0 and 3 hours and a gradual increase in reinfarction risk with increases in time thereafter.
Replication of the model in the cohort of all patients regardless of mortality status changed its predictive ability only slightly. C-index value decreased minimally to 0.63, and all factors remained significantly associated with reinfarction except previous bypass surgery, which was no longer significant when all deaths were included in the analyses.
Predictors of Death or Reinfarction
Composite 30-day incidence of death or in-hospital reinfarction was 9.7% (Figure 2). Multivariable logistic regression model for this end point had a C index of 0.73; Figure 3 gives predictive factors. Age, Killip class, systolic blood pressure, and heart rate were the most important predictors of the composite end point. As expected, increased Killip class was associated significantly with increased rate of death or infarction: 8.1%, 16.5%, 34.5%, and 58.8% for Killip classes I, II, III, and IV, respectively (P<0.001). Figure 4 shows nonlinear significant relations of heart rate and systolic blood pressure with death or reinfarction. Incidence of death or reinfarction significantly increased with both increases in systolic pressures >120 mm Hg and decreases in systolic pressures <120 mm Hg. Relation of heart rate to this end point was J-shaped, with a nadir incidence of the end point at a heart rate of 60 bpm, modest increases at lower heart rates, and marked increases at higher heart rates.
The chief finding of this analysis is that in-hospital reinfarction after fibrinolysis, although an uncommon complication, is associated with increased mortality. Unadjusted mortality was 3 times higher at 30 days and 1.5 times higher from 30 days to 1 year in patients with early reinfarction versus those without reinfarction. After adjustment for other known predictors of mortality, reinfarction remained significantly associated with 30-day mortality. Patients who died after reinfarction did so soon after the event, with half of the deaths occurring within 48 hours. Our data are consistent with MI studies that have shown early reinfarction to be associated with cardiac mortality. In pooled results from the TAMI studies, more patients with reinfarction died during initial hospitalization than did patients without reinfarction or recurrent ischemia (21% versus 4%; P=0.0001).9 Similarly, TIMI II investigators reported an increased relative risk of death at 3-year follow-up in patients with early reinfarction versus patients without nonfatal reinfarction (relative risk, 1.9; 99% CI, 1.1 to 3.2).13 This high associated mortality and narrow therapeutic window suggest that prevention of reinfarction should be an important goal of MI management.
Reinfarction and Fibrinolytic Agents
The 4.3% incidence of reinfarction in this large population is consistent with the 2.1% to 6.1% incidence reported from other fibrinolytic trials.4,5–7 Early placebo-controlled trials showed reinfarction rates of 2.8% and 4.1% with streptokinase, not significantly higher than the rate with placebo treatment.14 In the TIMI II trial, reinfarction occurred in 6.1% of patients within 42 days of alteplase treatment.5 The International Study Group reported no significant difference in reinfarction incidence after alteplase versus streptokinase treatment (2.6% versus 3.0%, respectively; relative risk, 0.87; 95% CI, 0.74 to 1.03).15 In the 41 299-patient International Study of Infarct Survival (ISIS) 3 trial, reinfarction occurred in 2.9%, 3.5%, and 3.6% of alteplase-, streptokinase-, and anistreplase-treated patients, respectively (P<0.02 for alteplase versus streptokinase; P=NS for anistreplase versus streptokinase).7
In our present study, reinfarction rate did not differ significantly by fibrinolytic assignment: 4.3% for alteplase, 4.5% for reteplase, 4.1% for streptokinase, and 4.4% for combination alteplase-streptokinase (P=0.55). Rates also varied little with the use of intravenous versus subcutaneous heparin (4.5% versus 4.2%, respectively). Our results differ from those of the International Study Group and ISIS 3, which showed a trend toward fewer reinfarctions after alteplase treatment. Although reinfarction criteria and assessment differ in various studies and settings, our large sample size, variety of adjunctive agents, and wide enrollment time span suggest that no fibrinolytic is superior for prevention of reinfarction.
Reinfarction was associated with increased in-hospital complications. Nearly 1 in 6 reinfarction patients suffered cardiogenic shock, and 1 in 3 had congestive heart failure. Similar increases in inpatient complications have been noted with recurrent ischemia in GUSTO I and with reocclusion of a successfully reperfused infarct-related artery in the TAMI studies.16,17 These increased complications affirm the clinical usefulness of our criteria for reinfarction and show the resource burden associated with early reinfarction.
Patients who suffered reinfarction were older, more often female or nonsmokers, more often had prior coronary or cerebrovascular disease and anterior MI at presentation, and had shorter symptom duration before fibrinolysis. Aspirin and β-blocker use were associated with lower reinfarction incidence, whereas treatment with angiotensin-converting-enzyme inhibitors, calcium channel blockers, and nitrates conferred no protection from reinfarction. However, causality should not be inferred, because these therapies were nonrandomized and probably given after reinfarction onset in many patients. Consistent with prior studies, patients suffering reinfarction had higher activated partial thromboplastin times, which suggests that inadequate anticoagulation does not predict reinfarction.18,19 The 4 variables most related to reinfarction in our multivariable model were advanced age, short times to fibrinolysis, enrollment outside the United States, and nonsmoking status (Figure 1). Differences in infarct-artery patency, length of hospital stay, and use of angiography or revascularization may explain the time-to-treatment and geographic associations; unfortunately, our analysis did not incorporate early angiographic data or other postrandomization covariates and procedures.
In our present analysis, current smokers had a significantly reduced incidence of reinfarction. Current smoking status remained a strong and independent negative predictor of reinfarction after adjustment for differences in age, sex, prior MI, and other comorbidities that have accounted for the paradoxical benefit of smoking on survival after MI in other studies.20,21 The influence of smoking on reinfarction also was examined in the International Tissue Plasminogen Activator-Streptokinase Mortality Trial. Barbash and colleagues22 reported that active smokers had significantly lower rates of in-hospital reinfarction compared with exsmokers and nonsmokers (2.7% versus 5.0% versus 4.7%, respectively; P<0.001). Similarly, the TIMI II investigators have shown that current nonsmoking was the only independent predictor of nonfatal reinfarction occurring within 3 years of the index MI (relative risk, 1.3; 99% CI, 1.0 to 1.8).13 However, this “protective” effect of smoking may be lost quickly. Our present study was not designed to capture postdischarge reinfarction events, but other investigators have shown a 4-fold higher rate of reinfarction at 1 year in patients who continue to smoke versus those who stop smoking.20
Death or Reinfarction
Reinfarction after fibrinolysis has been understudied because of uncertainty over its clinical and prognostic meaning. By showing that reinfarction is a strong, independent predictor of mortality, our analysis supports ongoing clinical trials of strategies to prevent early reinfarction. In our logistic model, advanced age, higher Killip class, systolic blood pressure, heart rate, and anterior MI location were most strongly associated with the composite outcome of death or reinfarction. Advanced age was the single most predictive factor of reinfarction and the composite event of death or reinfarction, which affirms its powerful role in risk stratification after MI. Models for reinfarction and death or reinfarction differed somewhat. Hemodynamic factors (Killip class, heart rate, and blood pressure) and anterior MI location were more strongly associated with survival and the composite outcome than with reinfarction. We propose that the variables above be considered in future analyses of this composite end point, with the hope that our models may allow physicians to better risk-stratify patients and compare new therapies.
Reinfarction was a prespecified secondary end point in both trials, and supporting clinical data were collected prospectively. Nevertheless, the present analysis has several limitations. Detection of reinfarction was suboptimal compared with current trial standards, because investigators had no standard, protocol-defined instructions for collecting cardiac enzymes or ECGs after revascularization or ischemic episodes. Reinfarction events were not adjudicated by an events committee, nor were medical records, cardiac-marker results, or ECGs reviewed for missed reinfarction events. Reliance on investigator reporting of “soft” end points such as MI, rather than standardized adjudication, can make a substantial difference in the reported incidence of the end point.
Another issue was the inability to capture unrecognized reinfarctions in patients who died early. Median time to death in the combined trials was 2.0 (0.09 to 6.0) days. Investigators were encouraged to record reinfarction as a separate end point if it had occurred in these deceased patients. Still, some patients probably died from reinfarction before confirmatory enzyme or ECG data became available or before serum cardiac markers could become elevated. These patients would not have been classified with reinfarction. Thus, we chose to exclude from reinfarction analyses the 6% of the patients who died in hospital to examine more precisely predictors in patients with definite, recorded reinfarctions. We urge caution in considering the relations of reinfarction and other postrandomization factors with each other and with mortality.
In the present study, 4% of patients who underwent fibrinolysis for acute MI suffered in-hospital reinfarction associated with reduced survival. In a related study, we have shown that repeated fibrinolysis and early revascularization are associated with better survival after reinfarction.23 We hope that the findings and predictive models will raise awareness of this important postinfarction complication and lead more clinicians to consider such aggressive therapies. Further work also should examine the effects of early revascularization, intravenous and oral antiplatelet agents, and new antithrombin regimens in prevention of reinfarction.
GUSTO I and GUSTO III trials were supported by Bayer (New York, NY), Ciba-Corning (Medfield, Mass), Genentech (South San Francisco, Calif), ICI Pharmaceuticals (Wilmington, Del), Sanofi Pharmaceuticals (Paris, France), and Boehringer-Mannheim Corp (Gaithersburg, Md).
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: 311–322.
Lee KL, Woodlief LH, Topol EJ, et al. Predictors of 30-day mortality in the era of reperfusion for acute myocardial infarction: results from an international trial of 41 021 patients. Circulation. 1995; 91: 1659–1668.
ISIS-3 (Third International Study of Infarct Survival) Collaborative Group. ISIS-3: a randomised comparison of streptokinase vs tissue plasminogen activator vs anistreplase and of aspirin plus heparin vs aspirin alone among 41,299 cases of suspected acute myocardial infarction. Lancet. 1992; 339: 753–770.
Stone CJ, Koo C. Additive splines in statistics.In: Proceedings of the Statistical Computing Section. Washington, DC: American Statistical Association; 1986: 45–48.
Califf RM, White HD, Van de Werf F, et al. One-year results from the Global Utilization of Streptokinase and TPA for Occluded Coronary Arteries (GUSTO-I) Trial. Circulation. 1996; 94: 1233–1238.
Ohman EM, Califf RM, Topol EJ, et al. Consequences of reocclusion after successful reperfusion therapy in acute myocardial infarction. Circulation. 1990; 82: 781–791.
Granger CB, Hirsch J, Califf RM, et al. Activated partial thromboplastin time and outcome after thrombolytic therapy for acute myocardial infarction: results from the GUSTO-I trial. Circulation. 1996; 93: 870–888.
Barbash GI, Reiner J, White HD, et al. Evaluation of the paradoxical beneficial effects of smoking in patients receiving thrombolytic therapy for acute myocardial infarction: mechanism of the “smoker’s paradox” from the GUSTO-I trial, with angiographic insights. J Am Coll Cardiol. 1995; 26: 1222–1229.
Barbash GI, White HD, Modan M, et al. Significance of smoking in patients receiving thrombolytic therapy for acute myocardial infarction: experience gleaned from the International Tissue Plasminogen Activator/Streptokinase Mortality Trial. Circulation. 1993; 87: 53–58.
Barbash GI, Birnbaum Y, Bogaerts K, et al. Treatment of reinfarction after thrombolytic therapy for acute myocardial infarction: an analysis of outcome and treatment choices in the Global Utilization of Streptokinase and Tissue plasminogen activator for Occluded coronary arteries (GUSTO I) and ASSEssment of the safety of a New Thrombolytic (ASSENT 2) studies. Circulation. 2001; 103: 954–960.