Which Patients With Suspected Myocardial Ischemia and Left Bundle-Branch Block Should Receive Thrombolytic Agents?

EDITORIAL: Which Patients With Suspected Myocardial Ischemia and Left Bundle-Branch Block Should Receive Thrombolytic Agents? From the Department of Emergency Medicine, Albert Einstein College of Medicine, Bronx, NY. Reprints not available from the author. Address for correspondence: E. John Gallagher, MD, Albert Einstein College of Medicine, Department of Emergency Medicine, 111 East 210th Street, Bronx, NY 10467-2490; 718-920-7459, fax 718-798-6084; E-mail jgallagh@montefiore.org. Copyright 2001 by the American College of Emergency Physicians. 0196-0644/2001/$35.00 + 0 47/1/114761 doi:10.1067/mem.2001.114761 E. John Gallagher, MD See related article, p. 431. [Gallagher EJ. Which patients with suspected myocardial ischemia and left bundle-branch block should receive thrombolytic agents? Ann Emerg Med. May 2001;37:439-444.] Timely administration of thrombolytic agents to patients with ischemic -segment elevation decreases mortality in acute myocardial infarction (AMI). 1 Stratification of this heterogeneous group of AMIs by ECG reveals a steep risk/benefit gradient when expressed as the number of patients needed to treat (NNT) or harm (NNH). As is evident from Figure 1, individuals with bundle-branch block are a particularly important stratum of patients to identify. This is true not only because they have a high baseline mortality and receive the greatest incremental improvement in survival when given thrombolytic agents (NNT = 21; 95% CI, 12 to 72) 1 but also because we tend to undertreat them. 2,3 The reluctance to administer thrombolytic agents to these patients can be attributed, at least in part, to prevailing clinical confusion driven by the ambiguity of the term bundle-branch block (BBB). In the Fibrinolytic Therapy Trialists (FTT) definitive systematic review of 9 major randomized clinical trials of thrombolytic agents versus placebo, no distinction was made between left BBB (LBBB) and right BBB (RBBB). 1 Rather, the 2 were merged into a single category, as shown in Figure 1. Examination of each study entered into the systematic review reveals that 2 trials excluded patients with BBB (AIMS 4 and USIM 5 ), 1 did not code BBB as a distinct ECG finding (ASSET), 6 1 recorded LBBB as a separate entity (GISSI), 7 and the remaining 5 trials simply reported undifferentiated BBB, presumably aggregating LBBB with RBBB (ISAM, 8 ISIS-2, 9 ISIS-3, 10 LATE, 11 and EMERAS 12 ). Until such time as the more than 2,000 ECGs showing BBB entered into the FTT systematic review 1 can be exhumed from the archives of multiple study coordinating MAY 2001 37:5 ANNALS OF EMERGENCY MEDICINE 439

centers, dusted off, and reclassified as left, right, or other, or until further clarifying information on this subgroup is forthcoming from additional sources, no explicit data are available to determine whether the mortality decrement associated with administration of thrombolytic agents to patients with BBB and suspected AMI is distributed equally or differentially between those patients with RBBB versus those with LBBB. In the absence of clear evidence, the logic of cardiac conduction argues that suspected AMIs with LBBB rather than RBBB are likely to receive the greatest therapeutic benefit from thrombolysis. This is because normal conduction traverses the atrioventricular node and depolarizes the septum from left to right, thus activating the left ventricle first. Standard 12-lead ECGs identify ischemia or infarction most readily when it occurs in those areas of the heart depolarized early. Hence, an RBBB, which does not interfere with either the direction of septal depolarization or the order of ventricular depolarization, will not ordinarily obscure the -segment elevation of AMI because it does not greatly alter initial electrical forces. 13,14 In contrast to RBBB, in the presence of an LBBB, conduction traverses the septum in the opposite direction Figure 1. Bars above the horizontal line demarcating benefit from harm indicate the number of patients with suspected AMI needed to treat (NNT) with thrombolytic agents (versus placebo) to save one life, stratified by ECG findings. Note that the smaller the NNT, the greater the incremental survival benefit. Bars below the horizontal line indicate the number of patients with suspected MI needed to harm (NNH), that is, the number who must receive thrombolytic agents before one excess death occurs. Note that those with -segment depression fare worse than those with normal ECGs, although neither group benefits from thrombolysis. Hi-low vertical lines display 95% CIs for the NNT and NNH. Data were calculated from the FIT systematic review of indications for fibrinolytic therapy in suspected AMI. 1 NNT NNH 200 150 100 50 0 50 100 150 BBB 21 28 Anterior 32 Other 120 Inferior ECG findings 72 Normal 144 (from right to left) and depolarizes the ventricles in the reverse order through propagation of an electrical front spreading radially across the left ventricle from the termination of the right bundle. 15 This obscures early vectors that ordinarily inscribe the characteristic ECG signature of AMI, rendering it illegible in the vast majority of patients. The American Heart Association (AHA) and American College of Cardiology (ACC) appear to have followed similar reasoning in formulating their recently updated joint practice guideline. 16 In the executive summary of this extensive and well-referenced document, the Committee on Management of AMI states that Symptoms consistent with acute MI and left bundle branch block (LBBB) should be managed like -segment elevation. 16 The authors go on to comment elsewhere in the guideline that Bundle-branch block (obscuring -segment analysis) and history suggesting AMI [constitutes] a Class I indication for thrombolysis. 16 Other than in the portion of the document dedicated to indications for pacing in AMI, RBBB is not mentioned further in the more than 200 pages of text. For the reasons noted above, this is presumably because an RBBB tends not to obscure -segment analysis. 13-15 Thus, the recommendation of the ACC/AHA appears to favor a logical strategy limiting thrombolysis to those patients with suspected AMI and LBBB, excluding those with RBBB unless they also have evidence of ischemic -segment elevation. Consistent with the notion that identification of AMI in the context of an LBBB is both important and difficult, more than 50 ECG signs have been proposed over the past half century for the detection of ischemia in LBBB. 17 Of these, 2 of the 3 criteria recently proposed by Sgarbossa et al 18 appear to have sufficiently powerful likelihood ratios to be helpful at the bedside: (1) -segment elevation of 1 mm or more in the same direction as the QRS complex and (2) -segment depression of 1 mm or more in leads V 1, V 2, or V 3. For practical use, these criteria can be remembered simply as directional concordance of the segment with the QRS complex immediately preceding it. Under ordinary circumstances, such concordance is unusual because the primary depolarization abnormality of LBBB is followed by a secondary abnormality in repolarization. Thus, in LBBB the T-wave complex will normally move in a direction opposite (discordant) to that of the QRS deflection. When this fails to occur, the resulting QRS- concordance strongly suggests acute coronary occlusion. More specifically, concordance of elevation suggests AMI in almost any lead; concordance of depression is most indicative of AMI when seen in the right precordial leads V 1, V 2, or V 3. 18 440 ANNALS OF EMERGENCY MEDICINE 37:5 MAY 2001

Unfortunately, as Kontos et al 19 convincingly demonstrate in this month s issue of Annals, the criteria of Sgarbossa et al 18 are too insensitive to be used as a screening (rule out) test to determine which patients with an LBBB do not have an MI. Indeed, if one combines the prospective data gathered by Kontos et al with recent earlier retrospective work targeted at the same end point, 20 the 2 Sgarbossa criteria were found in only 3% (95% CI, 2% to 6%) of 372 patients with LBBB and suspected ischemia. Among the subset of 49 patients with LBBB and confirmed AMI, at least one of the 2 criteria was present in only 20% of patients (95% CI, 10% to 34%). The Sgarbossa criteria are, however, highly specific (96%; 95% CI, 94% to 98%) and can be used reliably as a confirmatory test to rule in AMI in patients with LBBB. 19,20 A better sense of the clinical utility of the Sgarbossa criteria 18 can be obtained by converting the traditional test properties of sensitivity and specificity into likelihood ratios (LRs). 21 Use of positive and negative predictive values is misleading because they fluctuate directly as a function of the prevalence of the target disorder in the population to which the criteria are applied. 22 If one defines the presence of either concordant criterion as a positive test result and the absence of both criteria as a negative test result, the positive LR is 22 (95% CI, 7 to 72) and the negative LR is 0.8 (95% CI, 0.7 to 0.9). A positive LR reflects the odds that a patient with a positive test result has the target disorder, in this case AMI, as opposed to the odds that a patient with the same positive test result does not have AMI: Positive LR = Sensitivity/(1 Specificity). In contrast, a negative LR reflects the odds that a patient with a negative test result has the target disorder, AMI, as opposed to the odds that a patient with the same negative test result does not have an AMI: Negative LR = (1 Sensitivity)/Specificity. The degree to which the LR of a test deviates from one (positive LR >1; negative LR <1) is indicative of the power of that test to revise disease probability. Thus, the effect of a negative test result (ie, absence of both criteria) will reduce the posttest odds of AMI only marginally by a factor of 0.8. Therefore, lack of concordance of segments with the QRS complex in a patient with LBBB and suspected ischemia is of no help diagnostically and should not be used in clinical decisionmaking to rule out AMI. In contrast, the presence of -segment and QRS concordance increases the odds of AMI in the setting of LBBB by about 22-fold, essentially confirming the diagnosis anytime either of the concordant Sgarbossa criteria are seen. Kontos et al 19 use the low sensitivity (poor negative LR) of the Sgarbossa criteria to support their argument that too few people with AMI would be treated if these criteria were used to identify thrombolytic candidates with LBBB. These authors also point out that the low prevalence of AMI among patients with LBBB would unnecessarily expose too many people to the risks of thrombolytic agents in return for insufficient benefit. Although their former position seems squarely evidence based, the latter is at odds with the ACC/AHA s summary recommendation to treat all patients with LBBB with suspected AMI 16 and therefore warrants closer examination. One means of explicitly weighing the quantitative implications of competing therapeutic recommendations is through use of decision analysis. Shlipak et al 23 performed such an analysis recently, concluding that treatment of all patients with LBBB and suspected AMI would result in about 10 more lives saved (without the complication of, -free) per 1,000 patients than either use of the Sgarbossa ECG criteria 18 to select patients for thrombolysis or a strategy of declaring all patients with LBBB ineligible for thrombolytic agents. 23 However, this analysis was based on a prevalence of AMI in LBBB that was nearly twice that found by Kontos et al 19 and Li et al, 20 who reported an identical prevalence of 13% (95% CI for combined cohort, 10% to 17%). Shlipak et al 23 also did not have information on the age of the LBBB to incorporate into his decision analysis, which both Li et al 20 and Kontos et al 19 have recently shown may have some predictive utility in identifying patients at particularly high risk for AMI. Stratification by age of the LBBB, again combining data from Li et al and Kontos et al (provided by the lead authors of each article on request), allows calculation of interval likelihood ratios for the odds of AMI among patients with new (LR, 1.7; 95% CI, 1.1 to 2.7), indeterminate (LR, 1.1; 95% CI, 0.8 to 1.5), and old LBBB (LR, 0.5; 95% CI, 0.3 to 0.8). Although a slight gradient is evident, the LR associated with the distinction between old versus new or indeterminate LBBB appears to be the only LR that might be clinically useful. However, even this is a relatively weak test, indicating that the presence of an LBBB known to be old reduces the odds of AMI by only about one half. Further consolidation of these 3 categories of LBBB into 2 groups (new/indeterminate/presumably new versus known to be old) is consistent with the ACC/AHA s practice guideline, in which they state that Thrombolytic therapy is highly effective in patients with elevation or presumably new LBBB (which obscures the electrocardiographic diagnosis of MI). 16 From this grouping, it should then be possible to develop a decision analysis comparing performance of the combined criteria of either MAY 2001 37:5 ANNALS OF EMERGENCY MEDICINE 441

concordant Sgarbossa ECG criteria 18 OR a new/indeterminate LBBB as a positive test result indicating need for thrombolysis with the simpler strategy of administering thrombolysis to all patients with LBBB and suspected AMI. Consistent with this model, a negative test result would then be defined by the presence of BOTH Sgarbossa concordant criteria AND an LBBB known to be old. Given the high positive LR of the concordant Sgarbossa ECG criteria, providing thrombolytic agents to no patients with LBBB and suspected AMI is not a clinically sensible strategy and will not be included as a third branch in the decision tree. The range of probabilities used for sensitivity analyses at each decision node is based on the 95% CIs surrounding the point estimates, as shown in Table 1. 1,19,20,23,24 The prevalence of AMI in patients with LBBB and the test properties of the concordant Sgarbossa ECG criteria are drawn from the combined data of Kontos et al 19 and Li et al, 20 as presented previously. The frequency of and death among patients with AMI and LBBB who receive or do not receive thrombolytic agents is taken from the FTT definitive systematic review. 1 Mortality among untreated patients without MI is derived from data on outcomes of patients hospitalized for suspected ischemia who neither had an MI nor received thrombolytic agents. 24 By definition, among patients with MI, the sensitivity of the combined criteria will equal the probability of receiving thrombolysis according to the algorithm displayed in Figure 2. Similarly, in patients without MI, the specificity of the Table 1. Probability estimates for decision analysis. combined criteria will equal the probability of not receiving thrombolytic agents according to the same algorithm. Comparison of these combined criteria with a strategy of thrombolysis for all patients with LBBB and suspected ischemia according to a standard decision analysis (Figure 2) reveals that the outcomes of the 2 strategies are remarkably similar (Table 2). This appears to be true with respect to mortality, as well as -free survival. For quality-of-life estimates, dividing those patients who had a into those with major and minor disability 1 and assigning to major a weight equivalent to death does not materially alter the conclusion that an algorithm incorporating the concordant Sgarbossa criteria combined with the age of the LBBB performs no better than the much simpler strategy of thrombolysis for all suspected AMIs in patients with LBBB. Why might this be? Kontos et al 19 and Li et al 20 have already shown that the Sgarbossa criteria perform poorly as screening tests because most patients with LBBB and AMI do not have either concordant -segment elevation or depression. With respect to the age of the LBBB, one plausible reason for finding such poor discrimination, as was suggested by the marginal interval LRs estimated above, might be the decidedly different clinical implications of an LBBB not previously noted (and therefore characterized as presumably new ) versus one that develops acutely on serial ECGs in real time. Most clinicians would agree that the latter, if not rate related or toxic-metabolic, is strong evidence of ischemia. In contrast, an LBBB that appears to be new rather than acute could, in fact, be quite old, if only because the age of the LBBB is arbitrarily defined by whether it was previously noted in the medical record. Range for Probability in Patients Sensitivity With LBBB Estimate (%) Analysis * (%) AMI 19,20 13 10 17 : ( ) thrombolysis, ( ) AMI 23,24 2 1 3 : ( ) thrombolysis, (+) AMI 1 23.6 21.0 26.2 : (+) thrombolysis, ( ) AMI 23 2.2 0.9 3.1 : (+) thrombolysis, (+) AMI 1 18.7 16.3 21.1 Stroke: (+) thrombolysis 1 2.1 1.3 3.2 Stroke: ( ) thrombolysis 1 1.1 0.5 1.9 Sensitivity of combined criteria 19,20 82 68 91 Specificity of combined criteria 19,20 38 32 43 * Range for sensitivity analyses are based on 95% CIs of estimate. Range 23 and estimate 24 are based on different sources. Therefore, 95% CIs were not used for sensitivity analysis of patients who neither had AMI nor received thrombolytic agents but were hospitalized for suspected myocardial ischemia. Table 2. Outcome of decision analysis comparing 2 strategies for administration of thrombolytic agents to 1,000 patients with LBBB and suspected AMI. Application of Algorithm * Thrombolysis Outcome for All (per 1,000) for All (per 1,000) Mortality 43 44 43 44 Stroke-free survival 935 944 935 944 * The thrombolysis algorithm shown in Figure 2 is defined as thrombolytic agents administered to any patient with either of the 2 concordant Sgarbossa criteria 18 or an LBBB of new/indeterminate age, that is, not known to be old. Thrombolysis was withheld from any patient with neither concordant Sgarbossa criteria nor an LBBB known to be old, that is, noted before current presentation. 442 ANNALS OF EMERGENCY MEDICINE 37:5 MAY 2001

Thus, the poor performance of the age of the LBBB in a decision analysis, even when combined with the concordant Sgarbossa criteria, 18 might simply reflect the inadequacy of our operational definitions of new/indeterminate versus old LBBB and the confounding of these designations by random circumstance (eg, when an individual last happened to have an ECG at that particular institution). Kontos et al 19 are entirely correct in concluding that the Sgarbossa criteria identify only a small minority of patients and that administering thrombolytic agents to all patients with LBBB and suspected ischemia would result in treatment of a significant number of patients without AMI. Unfortunately, quantitative interpretation of a large body of evidence, 1,19,20,23,24 with the aid of decision analysis, 18 suggests that combining the Figure 2. Decision analysis comparing 2 strategies for administering thrombolytic agents to patients with LBBB and suspected AMI: application of thrombolytic algorithm (see text for details) versus thrombolysis for all suspected AMIs with LBBB. Probabilities and ranges for sensitivity analyses at each decision node are listed in Table 1. Outcomes are displayed in Table 2. LBBB with suspected AMI Thrombolysis algorithm Thrombolysis for all (+) MI ( ) MI ( ) MI (+) MI (+) Algorithm => thrombolysis ( ) Algorithm => no thrombolysis (+) Algorithm => thrombolysis ( ) Algorithm => no thrombolysis MAY 2001 37:5 ANNALS OF EMERGENCY MEDICINE 443

Sgarbossa criteria with the presumed age of the LBBB to create an algorithm confers no improved mortality or quality-of-life benefit when compared with administration of thrombolytic agents to all suspected AMIs in patients with LBBB. Although this may seem both unsatisfactory and counterintuitive, it does appear to be evidence based. Until such time as contravening data become available, the best answer to the question of which patients with suspected AMI and L-BBB should receive thrombolysis is also the simplest: All of them. REFERENCES 1. 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. 2. Barron HV, Bowlby LJ, Breen T, et al. Use of reperfusion treatment for acute myocardial infarction in the United States. Circulation. 1998;97:1150-1156. 3. Krumholz HM, Murillo JE, Chen J, et al. Thrombolytic therapy for eligible elderly patients with acute myocardial infarction. JAMA. 1997;277:1683-1688. 4. AIMS (APSAC Intervention Mortality Study) Trial Study Group. Effects of intravenous APSAC on mortality after acute myocardial infarction: preliminary report of a placebo-controlled clinical trial. Lancet. 1988;1:545-549. 5. Rossi P, Bolognese L, on behalf of Urochinasi per via Sistemica nell Infarto Miocardico (USIM) Collaborative Group. Comparison of intravenous urokinase plus heparin versus heparin alone in acute myocardial infarction. Am J Cardiol. 1991;68:585-592. 6. Wilcox RG, von der Lippe G, Olsson CG, et al. Trial of tissue plasminogen activator for mortality reduction in acute myocardial infarction: Anglo-Scandinavian Study of Early Thrombolysis (ASSET). Lancet. 1988;2:525-530. 7. Gruppo Italiano per lo Studio della Sopravvivenza nell Infarto Miocardico (GISSI). Effectiveness of intravenous thrombolytic treatment in acute myocardial infarction. Lancet. 1986;1:397-401. 8. ISAM (Intravenous streptokinase in Acute Myocardial Infarction) Study Group. A prospective trial of intravenous streptokinase in acute myocardial infarction (ISAM); mortality, morbidity, and infarct size at 21 days. N Engl J Med. 1986;314:1465-1471. 9. ISIS-2 (Second International Study of Infarct Survival Collaborative Group. Randomised trial of intravenous streptokinase, oral aspirin, both, or neither among 17,187 cases of suspected acute myocardial infarction. Lancet. 1988;2:349-360. 10. ISIS-3 (Third International Study of Infarct Survival) Collaborative Group. 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: ISIS-3. Lancet. 1992;339:753-770. 11. Late Assessment of Thrombolytic Efficacy (LATE) study with alteplase 6-24 hours after onset of acute myocardial infarction. Lancet. 1993;342:759-766. 12. EMERAS (Estudio Multicentrico Estreptoquinasa Republicas de America del Sur) Collaborative Group. Randomised trial of late thrombolysis in patients with suspected acute myocardial infarction. Lancet. 1993;342:767-772. 13. Gussak I, Zhou SH, Rautaharju P, et al. Right bundle branch block as a cause of falsenegative ECG classification of inferior myocardial infarction. J Electrocardiol. 1999;32:279-284. 14. Ortega-Carnicer J, Gomez-Grande ML, Ambros A. Right bundle branch block-induced Q waves simulating anterior myocardial infarction extension. J Electrocardiol. 2000;33:387-391. 15. Wellens HJJ. Acute myocardial infarction and left bundle-branch block can we lift the veil? N Engl J Med. 1996;334:528-529. 16. Ryan TJ, Antman EM, Brooks NH, et al. ACC/AHA guidelines for the management of patients with acute myocardial infarction: 1999 update: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Committee on Management of Acute Myocardial Infarction). Available at: http://www.acc.org/clinical/guidelines and http://www.americanheart.org. Accessed on January 17, 2000. 17. American College of Emergency Physicians. Clinical policy: critical issues in the evaluation and management of adult patients presenting with suspected acute myocardial infarction or unstable angina. Ann Emerg Med. 2000;35:521-544. 18. Sgarbossa EB, Pinski SL, Barbagelata A, et al. Electrocardiographic diagnosis of evolving acute myocardial infarction in the presence of left bundle-branch block. GUO-I (Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries) Investigators. N Engl J Med. 1996;334:481-487. 19. Kontos MC, McQueen RH, Jesse RL, et al. Can myocardial infarction be rapidly identified in emergency department patients who have left bundle-block branch block. Ann Emerg Med. 2001;37:431-438. 20. Li SF, Walden PL, Marcilla O, et al. Electrocardiographic diagnosis of myocardial infarction in patients with left bundle branch block. Ann Emerg Med. 2000;36:561-565. 21. Gallagher EJ. Application of likelihood ratios to clinical decision rules: defining the limits of clinical expertise. Ann Emerg Med. 1999;34:664-667. 22. Feinstein AR. Clinical Epidemiology. The Architecture of Clinical Research. Philadelphia, PA: W.B. Saunder; 1985:434-437. 23. Shlipak MG, Lyons WL, Go AS, et al. Should the electrocardiogram be used to guide therapy for patients with left bundle-branch block and suspected myocardial infarction? JAMA. 1999;281:714-719. 24. Krumholz HM, Pasternak RC, Weinstein MC, et al. Cost effectiveness of thrombolytic therapy with streptokinase in elderly patients with suspected acute myocardial infarction. N Engl J Med. 1992;327:7-13. 444 ANNALS OF EMERGENCY MEDICINE 37:5 MAY 2001