Towards an Understanding of the Role of Rescue Angioplasty for Failed Fibrinolysis: Comparison of the MERLIN, RESCUE and REACT T

Early restoration of infarct-related artery (IRA) blood flow and myocardial perfusion in patients with acute ST-segment elevation myocardial infarction (MI) is linked with preservation of left ventricular function, myocardial salvage and a reduction in mortality.¹ Although primary percutaneous coronary artery intervention (PPCI) is superior to hospitaldelivered thrombolytic treatment, the latter remains the firstline therapy in 30–70% of cases worldwide.^1–3 However, thrombolytic therapy fails to restore full patency of the IRA (TIMI grade 3 flow) in up to 54% of those who receive it. A further 5–10% of patients experience reocclusion after initial successful fibrinolysis.^4,5 Patients with an occluded IRA (TIMI flow grade 0–1) or suboptimal flow (TIMI flow grade 2) 90 minutes after thrombolytic therapy have increased early (within 30 days) and late (30 days to 2 years) mortality compared to patients with normal flow (TIMI flow grade 3).^6,7 This has led to a considerable debate on the optimal treatment strategy in patients with failed fibrinolysis.

Rescue angioplasty (rPCI) is often used to restore TIMI 3 flow when thrombolytic therapy fails. Only three major randomized studies have addressed this issue (Table 1). Some have interpreted these trials as showing variable results, but is this really the case? To assess this, we compared the protocols, demographics and outcomes of the MERLIN (Middlesbrough Early Revascularization to Limit INfarction) trial⁸ with the RESCUE I (Randomized Evaluation of Salvage angioplasty with Combined Utilization of Endpoints),⁹ and REACT (REscue Angioplasty vs. Conservative treatment or repeat Thrombolysis)¹⁰ trials. We also compared 30 day clinical outcomes for RESCUE I trial patients with the anterior MI subgroup patients from the MERLIN trial. Six month clinical endpoints for the MERLIN and REACT trials were also compared.

Comparison of Protocols (Tables 1 and 2)

Eligibility and exclusion criteria

a) Chest pain onset time and randomization. In the MERLIN trial, for patients randomized to the rescue arm, angioplasty was required within 12 hours of chest pain onset and allowing up to 2 hours for transfer to the lab (including inter-hospital transfers), patients were randomized up to 10 hours of chest pain. In the RESCUE I and REACT trials, patients were randomized within 6 hours of chestpain. However, in REACT, for the relevant patients, rescue PCI had to be performed within 12 hours of chest pain, so there was no mandate to perform PCI as soon as possible after randomization.

b) Criteria for failed reperfusion. Whereas in RESCUE I, failed reperfusion was defined at angiography, electrocardiographic (ECG) criteria were used in the MERLIN and REACT trials. Both used a criterion of < 50% ST-segment resolution in the definition of failed reperfusion, the diagnostic ECG being performed earlier (60 minutes in the MERLIN trial than in the REACT trial (90 minutes).

c) Age limit. There was no age limit for enrollment in the MERLIN trial. However, there were age limits for enrollment in both the RESCUE I trial (21–79 years) and the REACT trial (2–85 years), although the initial protocol of REACT excluded patients > 75 years, older patients were enrolled after a protocol amendment.

d) Other features. In RESCUE I, only patients suffering their first anterior infarction were enrolled, whereas the other two trials included all STEMI’s fulfilling the ECG entry criteria. RESCUE I also excluded patients presenting < 3 hours from chest pain onset as well as those with a left main stem stenosis ≥ 50%. There were no specific exclusion criteria (other than significant comorbidities) for enrollment in the MERLIN trial. Because of concerns about bleeding risks in the repeat thrombolysis arm, patients with body weight < 65 kg were excluded from the REACT trial. Stents and glycoprotein (GP) IIb/IIIa inhibitors were not used in the RESCUE I trial.

Treatment crossover. In the RESCUE I trial, angioplasty or bypass surgery was not allowed for 72 hours in the conservative group. Early crossover (in the first 12 hours) from the conservative arm to the rescue arm was not allowed unless the patient was in cardiogenic shock in the MERLIN trial. In the REACT trial, crossover to rescue PCI was allowed for early reinfarction, whereas clinical intervention for reinfarction beyond 12 hours was classified as an end-point in the MERLIN trial.

Endpoint definitions. Rescue PCI was considered successful if final TIMI flow grade was ≥ 2 and the residual stenosis was ≤ 50% in the IRA in RESCUE I compared with TIMI flow grade ≥ 2 and < 50% in the MERLIN trial, but TIMI flow grade 3 and < 50% stenosis in the REACT trial. The primary end-point in the RESCUE I trial was the 30- day ejection fraction, but in MERLIN it was 30-day mortality with a composite of clinical events being a secondary endpoint (and patients were subsequently followed up to 3 years). The REACT trial had a composite primary endpoint at 6 months. Different definitions of heart failure were used in the three trials.

Power calculation. In the RESCUE I trial, it was esti-mated that 138 patients would be required to detect a 4 ± 12% difference in ejection fraction with a two-sided p-value of 0.05. In the MERLIN trial, 150 patients were required in each arm to detect a 30-day mortality difference between the conservative group (estimated rate 18%) and the rPCI group (estimated at 6%), with 90% power and a p-value of 0.05. Initially in the REACT trial, it was estimated that 1,200 patients were needed to detect a difference in the composite endpoint (death, reinfarction, cerebrovascular events and severe heart failure), which was estimated at 20% in the conservative group and 12% in the rPCI group, with 80% power and a p-value of 0.05. During the study, the sample size was recalculated on the basis of a reduction in the rates of death or reinfarction from 29% in the conservative group to 15% in the rPCI group, requiring 150 patients in each group.

Comparison of Patients

Recruitment (Figure 1). Recruitment in each participating center was much higher in the MERLIN trial than the others, and the enrollment of first anterior infarcts was also higher compared with the RESCUE I trial. The MERLIN trial patients were enrolled from only 3 centers in just over 3 years; more than 70% of the eligible patients at the revascularization center were entered in the trial. Following the amended power calculation, the REACT trial ended after enrollment of 427 of the planned 450 patients (over nearly 4 years). Recruitment was from 35 centers, more than half recruiting fewer than 10 patients. The differing study designs resulted in equivalent numbers of patients undergoing conservative therapy or rPCI in the MERLIN and REACT trials.

Demographics (Tables 3 and 4). Patients in the MERLIN trial were older than the patients enrolled in either the REACT or the RESCUE I trials, and fewer were current smokers. In the REACT trial, the number of patients who had experienced a previous MI was lower in the rPCI group (9.8%) compared to both the conservative (12.1%) and repeat lysis groups (16.2%). In the rPCI arms, more patients with anterior MI were enrolled in the MERLIN trial compared to the REACT trial (48.4% vs. 42.7%).

Time-to-treatment (Tables 3 and 4). The times given in the trials were variably given as mean or median times, but nevertheless, pain-tothrombolysis time was probably shorter in the REACT trial than in MERLIN, but pain-to-laboratory time was longer in the rPCI arm of REACT than in MERLIN. Pain-to-laboratory time was lowest in the RESCUE I trial. In the rescue arms, the time from thrombolysis-tolaboratory was notably longer in the REACT trial.

Comparison of Treatments

Type of lytic used (Table 4). Streptokinase was the thrombolytic agent used in 96% (97% in the conservative group) of patients enrolled in the MERLIN trial compared to 60% in the REACT trial. Streptokinase, urokinase or tPA were used in the RESCUE I trial, and patients who had received a fibrin-specific lytic agent received additional streptokinase (500,000 U) or urokinase (1 x 10⁶ U).

Stents and GP IIb/IIIa inhibitors. Stents and GP IIb/IIIa inhibitors were not used in the RESCUE I trial. The MERLIN trial started recruitment earlier than the REACT trial, and this probably explains the less frequent use of stents (50.3% vs. 68.5%, respectively). GP IIb/IIIa inhibitors were used rarely in the MERLIN trial (3.3% vs. 43.4% in the REACT trial).

Treatment crossovers. In the REACT trial, 12.5% (18/144) of patients randomized to rPCI did not undergo angiography, 14 crossed over to conservative treatment and 2 received repeat thrombolysis. Specific reasons for crossover were not given. In the MERLIN trial, 2.6% (4/153) of patients randomized to rPCI did not undergo angiography for clinical or technical reasons. Analysis for both studies was carried out on an intention-to-treat basis.

Comparison of Outcomes

Complications. RESCUE I did not report bleeding or transfusion rates, but in the MERLIN trial, blood transfusionwas required more often in the rPCI arm (11.1% vs. 1.3% in the conservative arm; p < 0.001). In REACT, there was no difference in the rate of major bleeding between the groups (repeat lysis: 5% vs. conservative: 3% vs. rPCI: 0; p = 0.65). However, those undergoing rPCI experienced significantly more minor bleeds (repeat lysis: 3% vs. conservative: 0 vs. rPCI: 28%; p < 0.001), most being related to the arterial sheath. The requirement for a blood transfusion was not included in the definition of major or minor bleeding in REACT. There was a higher rate of cerebrovascular events in the rPCI arm in the MERLIN trial (4.6% vs. 0.6% in the conservative group; p = 0.03). The rates of cerebrovascular events at 6 months in REACT were also higher in the rPCI arm, although not significantly different from the other groups (repeat lysis: 0.7% vs. conservative: 0.7% vs. rPCI: 2.1%; p = 0.63).

Angiographic results. In the rPCI arm of the MERLIN trial, 97.4% of the patients underwent coronary angiography and 76% had multivessel disease. In 77 (94%) of the 82 patients undergoing intervention for TIMI flow grade < 3, PCI was successful. A total of 130 patients in the rPCI arm (85%) left the catheterization lab with TIMI flow grade 3 in the IRA. In comparison, only 89% underwent coronary angiography in the REACT trial and 115 (80%) underwent rPCI; this was successful in 106 (92%) patients. In the RESCUE I trial, PCI was deemed successful in 92% of patients.

Clinical outcomes (Table 5). None of the trials individually have shown a benefit in terms of a reduction in mortality. Thirty-day mortality in the RESCUE I trial was 5.5% in the rPCI arm, and 9.6% in the conservative arm (not significant). Thirty-day mortality was 9.8% in the rPCI arm of MERLIN compared to 11% in the conservative arm (p = 0.7), with no difference at 6 months; mortality in patients with TIMI flow grade 3 after rPCI was 9.2%. Of note, the mortality of those with persistent ST-segment elevation after rPCI (20.3%) in the MERLIN trial was higher than the mortality rate of those with persistent ST-segment elevation in the conservative arm (15.8%). This was not statistically significant. Mortality rates in REACT at 6 months were 6.2% in the rPCI arm and 12.8% in the conservative arm (p = 0.06). Despite the different protocols, 6-month mortality in the conservative arms of both the MERLIN and REACT trials was remarkably similar.

In each of the trials, the composite endpoints were lower in the rPCI arm. In the RESCUE I trial, there was no significant difference between the two treatment groups for the primary endpoint of resting left ventricular ejection fraction measured by radionuclide imaging at 30 days. However, the (non-prespecified) secondary combined endpoint of death or severe heart failure (6% vs. 17%; p = 0.05) was lower in the rPCI group. In the MERLIN trial, event-free survival was greater in the rPCI arm (62.7% vs. 50%; p = 0.02), almost completely due to a reduction in subsequent revascularization (6.5% vs. 20.1%; p < 0.01). In the REACT trial, the primary combined endpoint of death, reinfarction, cerebrovascular events and severe heart failure at 6 months was significantly lower in patients randomized to the rescue angioplasty group when compared with the conservative group (15.3% vs. 29.8%, respectively; p < 0.01).

Meta-Analyses (Figures 2–5)

Three meta-analyses^11–13 have been published, each with slightly different inclusion criteria but all with the same conclusions. All three analyses suggest that rescue PCI reduces mortality and clinical heart failure at 30 days, but at the expense of an increased risk in strokes. Our own meta-analysis¹⁴ (including the results of the much smaller studies by Belenkie et al and the RESCUE II group,^15,16 suggests a reduction in 30-day mortality from 11% in the conservative group (n = 395) to 7% in the rPCI group (n = 405) (no heterogeneity seen). There is also less reinfarction after rPCI compared with conservative therapy, but more risk of bleeding.

Discussion

These three trials have each evaluated the outcomes of patients randomized to different treatment strategies following failed reperfusion with thrombolysis. The study designs are different in a number of ways, and these differences have to be taken into account when interpreting the results. Several of these differences can be highlighted.

Chest pain onset and randomization. The longer time allowed from chest pain onset to randomization in the MERLIN study may have led to the enrollment of patients at a later stage of infarction, with less potential benefit from an aggressive strategy. However, both MERLIN and REACT aimed to perform angiography within 12 hours of chest pain onset. The timings from the studies actually reveal that patients in MERLIN underwent angiography earlier than those in REACT (this was erroneously documented in the ESC guidelines in 2005).¹⁷ Earlier intervention might havebeen expected to offer an advantage (patients in RESCUE I had the shortest times-to-treatment with good outcomes), but this potential benefit in the MERLIN trial compared with the REACT trial might have been countered in part by the protocol allowing patients to be enrolled beyond 6 hours.

Criteria for diagnosis of failed reperfusion and timing of ECGs. The RESCUE I trial diagnosed failed reperfusion on the basis of an occluded artery identified at angiography, clearly the gold-standard approach, but not practical in multicenter studies recruiting patients from noninterventional centers. Moreover, the pragmatic question that is being asked on CCUs on a daily basis is whether to send a patient with failed thrombolysis on ECG criteria for angiography (with a view to rPCI). ECG ST resolution is a rapid, simple and readily available bedside measure to identify success or failure of thrombolysis. Complete ST resolution is highly predictive of unimpaired flow, both at the epicardial and myocardial levels.^18,19 Considering less than 50% resolution in the ECG, either at 60 or 90 minutes, as an indicator of reperfusion will result in some patients with a blocked artery being missed and some with a patent artery being included. The 60- minute ECG used in the MERLIN trial on the basis of minimizing delays, especially when inter-hospital transfer is involved, could lead to the inclusion of patients who might reperfuse pharmacologically during the time taken to transfer for angiography. This criteria would therefore be expected to enroll a slightly lower-risk group of patients than if a later ECG is used. However, having an ECG too late would result in a cohort with little myocardium to salvage. The results of these studies suggest that diagnosis of failed thrombolysis at 90 minutes might be a more appropriate recommendation.

Timing of rescue angioplasty. In the RESCUE I trial, rPCI was offered immediately after the diagnosis of failed thrombolysis was made, whereas there were inherent delays in the MERLIN and REACT trials, and the delay was longer in REACT than MERLIN. Although the mortality in the rPCI arm in REACT was numerically lower than that seen in MERLIN, it seems counter-intuitive to delay a rescue procedure which we believe should be considered as soon as possible after the diagnosis is made. Conversely, a longer delay before performing rescue angioplasty may result in more patients having a patent vessel after fibrinolysis prior to PCI, and may allow for a greater use of GP IIb/IIIa inhibitors, especially when fibrin-specific agents have been used. Any differences in mortality that might or might not exist between the two larger trials are probably explained by a number of factors over and above these timing differences (e.g., earlier randomization in RESCUE I and REACT, age of patients, differences in other demographic factors such as incidence of previous myocardial infarction).

Recruitment and statistical power. The REACT trial was terminated early due to difficulties with recruitment. It is also notable that an important change was made to the power calculation for the REACT trial after a significant number of patients had been recruited.²⁰ This resulted in the trial recruiting less than half of its initial target number of patients. The differing recruitment rates in the trials suggest a much greater degree of selection in the RESCUE I and REACT trials compared to the MERLIN trial. Although low recruitment from individual centers raises the possibility of selection bias, conversely high recruitment from a small number of centers may not result in a population representative of hospitals worldwide. Therefore, teasing out the differences in patient populations and timings of the various trials becomes important, as there may be a greater benefit from rPCI in some cohorts compared to others.

Baseline characteristics. The RESCUE I trial probably enrolled lower-risk patients than did MERLIN on the basis of younger age, shorter chest pain-to-presentation times, shorter presentation-to-angioplasty time and exclusion of patients with previous MI or left main stem disease. The REACT study enrolled slightly younger patients with fewer patients with anterior MI than in MERLIN. There were more current smokers in the RESCUE I and REACT trials. Current smoking is a factor that increases the likelihood of successful thrombolysis,²¹ and it is possible that it also has a bearing on the outcome after rPCI. The fact that fewer patients in the rPCI arm of REACT had suffered previous infarction might also have influenced the results.

Complications. More transfusions in the MERLIN trial might be attributable to the long half-life of streptokinase in comparison to the fibrin-specific lytic agents. However, the transfusion rate in MERLIN was actually lower than that observed in an early trial of PTCA after streptokinase.²² Early PCI after fibrin-specific thrombolytic agents has been shown to be relatively safe, with improvement in the rates of major adverse cardiac events (MACE) compared with conservative care. However, PCI combined with streptokinase has been shown to increase MACE, especially when abciximab is being considered as adjunctive therapy. In the TIMI 14 trial, rates of major hemorrhage were higher in the streptokinase and abciximab group, at 10%, compared to 6% in the alteplase group, and 7% with alteplase plus abciximab group.²³ The different thrombolytic agents used might have influenced the different stroke rates in MERLIN and REACT (4.6% vs. 2.1%).

In the REACT trial, nearly 40% of the patients had fibrinspecific lytic agents, and although the trial definitions differ, fewer major bleeds were seen with no significant differences between the arms. However, minor bleeds were far more common in the rPCI arm, most relating to vascular access site complications. It is notable that in the recent ASSENT-4 PCI trial, in addition to the higher 30-day mortality rate, the in-hospital total stroke rates (1.8% vs. 0; p < 0.0001) and bleeding complications were higher in the tenecteplase and PCI arm compared to the PCI-alone arm.²⁴ Thus, even when using a fibrin-specific lytic agent like tenecteplase, PCI in combination with a lytic therapy is associated with an increased incidence of stroke and bleeding.

Clinical outcomes. The results in the RESCUE I trial appear different from the results in the anterior infarctioncohort in the MERLIN trial, with lower 30-day mortality (7.3% vs. 17.6%) and heart failure (4% vs. 34%) in spite of no stenting, no GP IIb/IIIa inhibitors, and without the use of an intra-aortic balloon pump. Although the results in heart failure may be due to definitional issues, the selection criteria used in the study and different timing suggests that the RESCUE I cohort was a significantly lower-risk group of patients.

When comparing the MERLIN and REACT trials, a number of differences in outcomes are worthy of comment. In the rPCI arm, reinfarction rates were higher in MERLIN (7.8% vs. 2.1%), possibly due to lower rates of stenting (50.3% vs. 68.5%) and less use of GP IIb/IIIa inhibitors (3.3% vs. 43.4%). Although coronary stenting is associated with less recurrent ischemia, restenosis and a reduction in the need for in-hospital and long-term target vessel revascularization, it has not been shown to have a major impact on mortality in this context. In the STOP-AMI 4 trial, stenting after failed fibrinolysis was associated with a greater salvage index compared with balloon angioplasty, but clinical benefits were not demonstrated.²⁵

Abciximab has been shown to be of benefit during primary angioplasty.^26,27 In one randomized study, treatment with abciximab during rPCI was associated with better left ventricular recovery, a lower incidence of adverse cardiac events at 30 days and 6 months and a low incidence of bleeding events, but no mortality benefit.²⁸ However, this trial was small, rPCI was performed mainly after the use of rPA and the procedure was performed relatively late (8.5 ± 5.4 hours following lytic therapy). Given the results of the TIMI 14 and ASSENT-4 PCI trials, we continue to recommend caution when using abciximab early after full-dose thrombolysis, especially when streptokinase is used. There may be a case for abandoning streptokinase as a lytic agent if a protocol of rPCI is to be followed.

It has been observed that mortality in the rescue arm of the REACT trial is surprisingly low.^10,29 Although the relative risk reduction in mortality seen in REACT was comparable to the estimated reduction used in the power calculation in MERLIN, given the delay to rPCI in REACT, a major mortality reduction would not have been expected by intervening 5–7 hours following fibrinolysis. As all of these trials have either been underpowered to show a reduction in mortality, or have not been powered for this endpoint, each of the trial results individually may have been due to chance.

In the conservative arms of the MERLIN and REACT trials, the mortality rate at 6 months was no different, and was lower than anticipated in both (expected vs. actual mortality: REACT trial [17% vs.12.8%], MERLIN trial [18% vs.12%]). We have proposed a number of reasons for a lower than anticipated mortality rate in the conservative arm of the MERLIN trial⁸ including the use of the diagnostic ECG at 60 minutes, a relatively high use of angiography and revascularization in the conservative arm following risk stratification and the inclusion of lower-risk patients with inferior infarction. In the conservative arm of both trials, the rate of subsequent revascularization was higher than in the rPCI arms, was similar in both studies (REACT 20.6%, MERLIN 25%) and was the major driver of the composite endpoint in MERLIN.

Failed rescue. The results of both the MERLIN and REACT trials add to the previous evidence outlining the poor outcomes of those with a failed rescue angioplasty.^30,31 In REACT, the incidence of the primary endpoint was higher in those with an unsuccessful procedure (55.6% vs. 11.3% with a successful procedure), and the poor outcome of the 6% with failed rPCI in MERLIN has been previously described.³² The morbidity and mortality rates in this group may exceed those in patients with failed fibrinolysis managed conservatively. It is likely then that the overall results for a population of patients undergoing rPCI will consist of a combination of good outcomes in those with a successful procedure countered by poor outcomes in the minority with an unsuccessful procedure. Whether results of rescue PCI can be improved by other adjuncts such as low-profile thrombectomy or embolic protection devices is unknown. Whereas their use has not translated into clinical benefits in primary angioplasty,³³ they may have a role in a patient group presenting later after infarction, and in the context of failed fibrinolysis where thrombus burden may be particularly high.

Meta-analysis. The mortality results from the meta-analysis suggest that a definitive trial to demonstrate a 4% absolute mortality rate with rescue PCI (power of 80%) compared with a conservative approach would require 803 patients in each arm with a two-sided alpha value of 0.05 (1074 in each arm for 90% power). A multicenter trial of this size would be feasible and perhaps easier to recruit for than was the case with the REACT trial, which incorporated a third arm (of repeat fibrinolysis). However, given the contribution of the REACT trial results to the meta-analysis, it is possible that such a trial would remain underpowered, and it might be more appropriate to design a larger trial with prespecified interim analyses after 1,500 and 2,000 patients. With the increasing use of prehospital fibrinolysis, early decisions on the use of rPCI will be an important issue. An international trial of prehospital fibrinolysis with intended rPCI versus a strategy of primary PCI is being planned.

Although there are clearly differences in the design and results of these trials, the basic strategy of comparing a protocol of rescue PCI with an alternative strategy is similar, and the meta-analysis shows that the results are actually also reasonably similar. Heterogeneity testing is, in effect, a means of assessing the degree of variation between the trials. The test in this context demonstrates that the results are not so varied that it is inappropriate to combine the results in a metaanalysis. However, when there are only a few studies in such a meta-analysis, the heterogeneity test is not very sensitive in detecting excess variation.

Conclusion

Although some outcomes were similar, differences in patient recruitment and protocols may have accounted for thenumerically different mortality rates at 6 months seen in the MERLIN and the REACT trials. The randomized trials do not clearly demonstrate a mortality benefit in favor of rescue angioplasty in the setting of failed fibrinolysis, but the recent meta-analyses suggest that this is probably the case. Other benefits of rescue PCI (lower rates of unplanned revascularization, reinfarction and heart failure) are achieved at the expense of higher rates of bleeding and a higher risk of stroke.

For centers that are unable to deliver a primary PCI program, the results of these trials are supportive of the use of rescue angioplasty. If rescue angioplasty is to be considered as part of routine clinical practice, standardization of protocols to minimize bleeding (possibly by avoiding streptokinase as a lytic agent) is recommended, with a greater use of stenting and perhaps a bolder use of GP IIb/IIIa inhibitors, especially if rescue angioplasty is performed several hours after thrombolysis. However, it is probably still important to minimize the delay in getting patients to the catheterization laboratory once a diagnosis of failed thrombolysis is made.

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 randomized trials of more than 1000 patients. Lancet 1994; 343:311–322.

2. Eagle KA, Goodman SG, Avezum A, et al. Practice variation and missed opportunities for reperfusion in ST-segment-elevation myocardial infarction: Findings from the Global Registry of Acute Coronary Events (GRACE). Lancet 2002;359:373–377.

3. Keeley EC, Boura JA, Grines CL. Primary angioplasty versus intravenous thrombolytic therapy for acute myocardial infarction: A quantitative review of 23 randomized trials. Lancet 2003;361:13–20.

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 Engl J Med 1993; 329:1615–1622.

5. Herrmann HC. Triple therapy for acute myocardial infarction: Combining fibrinolysis, platelet IIb/IIIa inhibition, and percutaneous coronary intervention. Am J Cardiol 2000;85:10C–6C.

6. Anderson JL, Karagounis LA, Califf RM. Metaanalysis of five reported studies on the relation of early coronary patency grades with mortality and outcomes after acute myocardial infarction. Am J Cardiol 1996;78:1–8.

7. Ross AM, Coyne KS, Reiner JS, et al. A randomized trial comparing primary angioplasty with a strategy of short-acting thrombolysis and immediate planned rescue angioplasty in acute myocardial infarction: The PACT trial. PACT investigators. Plasminogen-activator Angioplasty Compatibility Trial. J Am Coll Cardiol 1999;34:1954–1962.

8. Sutton AG, Campbell PG, Graham R, et al. A randomized trial of rescue angioplasty versus a conservative approach for failed fibrinolysis in ST-segment elevation myocardial infarction: the Middlesbrough Early Revascularization to Limit INfarction (MERLIN) trial. J Am Coll Cardiol 2004;44:287–296.

9. Ellis SG, Da Silva ER, Heyndrickx G, et al. Randomized comparison of rescue angioplasty with conservative management of patients with early failure of thrombolysis for acute anterior myocardial infarction. Circulation 1994;90:2280–2284.

10. Gershlick AH, Stephens-Lloyd A, Hughes S, et al. Rescue angioplasty after failed thrombolytic therapy for acute myocardial infarction. N Engl J Med 2005;353:2758–2768.

11. Patel TN, Bavry AA, Kumbhani DJ, Ellis SG. A meta-analysis of randomized trials of rescue percutaneous coronary intervention after failed fibrinolysis. Am J Cardiol 2006; 97:1685–1690.

12. Collet JP, Montalescot G, Le May M, et al. Percutaneous coronary intervention after fibrinolysis: A multiple meta-analyses approach according to the type of strategy. J Am Coll Cardiol 2006;48:1326–1335.

13. Wijeysundera HC, Vijayaraghavan R, Nallamothu BK, et al. Rescue angioplasty or repeat fibrinolysis after failed fibrinolytic therapy for ST-segment myocardial infarction: A meta-analysis of randomized trials. J Am Coll Cardiol 2007;49:422–430.

14. Kunadian B, Vijayalakshmi K, Dunning J, et al. Rescue angioplasty versus conservative management for failed thrombolysis in acute myocardial infarction: A metaanalysis. J Am Coll Cardiol 2005;47(suppl 4):929–935.

15. Belenkie I, Traboulsi M, Hall CA, et al. Rescue angioplasty during myocardial infarction has a beneficial effect on mortality: A tenable hypothesis. Can J Cardiol 1992;8:357–362.

16. Ellis SG, Da Silva ER, Spaulding CM, et al. Review of immediate angioplasty after fibrinolytic therapy for acute myocardial infarction: Insights from the RESCUE I, RESCUE II, and other contemporary clinical experiences. Am Heart J 2000;139:1046–1053.

17. Silber S, Albertsson P, Aviles FF, et al. Guidelines for percutaneous coronary interventions. The Task Force for Percutaneous Coronary Interventions of the European Society of Cardiology. Eur Heart J 2005;26:804–847.

18. Zeymer U, Schroder R, Tebbe U, et al. Non-invasive detection of early infarct vessel patency by resolution of ST-segment elevation in patients with thrombolysis for acute myocardial infarction; Results of the angiographic substudy of the Hirudin for Improvement of Thrombolysis (HIT)-4 trial. Eur Heart J 2001;22:769–775.

19. Zeymer U, Schroder K, Wegscheider K, et al. ST resolution in a single electrocardiographic lead: A simple and accurate predictor of cardiac mortality in patients with fibrinolytic therapy for acute ST-elevation myocardial infarction. Am Heart J 2005;149:91–97.

20. Kovac JD, Gershlick AH. How should we detect and manage failed thrombolysis? Eur Heart J 2002;22:405–407.

21. Barbash GI, Reiner J, White HD, et al. Evaluation of paradoxic 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. Global Utilization of Streptokinase and Tissue-Plasminogen Activator for Occluded Coronary Arteries. J Am Coll Cardiol 1995;26:1222–1229.

22. O'Neill WW, Weintraub R, Grines CL, et al. A prospective, placebo-controlled, randomized trial of intravenous streptokinase and angioplasty versus lone angioplasty therapy of acute myocardial infarction. Circulation 1992;86:1710–1717.

23. Antman EM, Giugliano RP, Gibson CM, et al. Abciximab facilitates the rate and extent of thrombolysis: results of the thrombolysis in myocardial infarction (TIMI) 14 trial. The TIMI 14 Investigators. Circulation 1999;99:2720–2732.

24. Assessment of the Safety and Efficacy of a New Treatment Strategy with Percutaneous Coronary Intervention (ASSENT-4 PCI) investigators. Primary versus tenecteplase-facilitated percutaneous coronary intervention in patients with STsegment elevation acute myocardial infarction (ASSENT-4 PCI): randomized trial. Lancet 2006;367:569–578.

25. Schomig A, Ndrepepa G, Mehilli J, et al. A randomized trial of coronary stenting versus balloon angioplasty as a rescue intervention after failed thrombolysis in patients with acute myocardial infarction. J Am Coll Cardiol 2004;44:2073–2079.

26. De Luca G, Suryapranata H, Stone GW, et al. Abciximab as adjunctive therapy to reperfusion in acute ST-segment elevation myocardial infarction: A meta-analysis of randomized trials. JAMA 2005;293:1759–1765.

27. De Luca G, Suryapranata H, Piscione F, et al. Stenting and abciximab in primary angioplasty: A review of current status. Ital Heart J 2005;6:874–880.

28. Petronio AS, Musumeci G, Limbruno U, et al. Abciximab improves 6-month clinical outcome after rescue coronary angioplasty. Am Heart J 2002; 143:334–341.

29. Aziz S, Ramsdale D. Rescue angioplasty after thrombolysis. N Engl J Med 2006;354:1639–1641.

30. Ross AM, Lundergan CF, Rohrbeck SC, et al. Rescue angioplasty after failed thrombolysis: Technical and clinical outcomes in a large thrombolysis trial. GUSTO-1 Angiographic Investigators. Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries. J Am Coll Cardiol 1998;31:1511–1517.

31. Abbottsmith CW, Topol EJ, George BS, et al. Fate of patients with acute myocardial infarction with patency of the infarct-related vessel achieved with successful thrombolysis versus rescue angioplasty. J Am Coll Cardiol 1990;16:770–778.

32. Sutton AG, Campbell PG, Graham R, et al. One year results of the Middlesbrough early revascularisation to limit infarction (MERLIN) trial. Heart 2005;10:1330–1337.

33. Kunadian B, Dunning J, Vijayalakshmi K, et al. Meta-analysis of randomized trials comparing anti-embolic devices with standard PCI for improving myocardial reperfusion in patients with acute myocardial infarction. Catheter Cardiovasc Interv 2007;69:488–496.