The benefit of primary percutaneous coronary intervention (PCI) over hospital-based thrombolysis for ST-elevation myocardial infarction (STEMI) has been established beyond reasonable doubt.1 However, thrombolysis remains an effective and widely used therapy for a significant proportion of STEMI patients and is particularly effective within two hours of the onset of symptoms.2 There may also be logistical reasons for thrombolysis being used widely that need to be considered, as, without doubt, thrombolysis can be administered easily in most hospitals with an emergency room. Consequently, it is appropriate that guidelines from both Europe and North America have a place for thrombolytic therapy alongside primary PCI in the management of STEMI.
Based on the open artery hypothesis, restoration of epicardial blood flow has been shown to improve the long-term outcomes and reduce adverse clinical outcomes following a STEMI.2 In addition, the degree of reperfusion is key, with thrombolysis in myocardial infarction (TIMI) 3 flow (normal) after a therapeutic intervention (be it pharmacological or mechanical) being correlated with an improved outcome.3,4 It is thus unfortunate, particularly in early trial data, that in up to 40% of patients thrombolysed for a STEMI, artery patency was not achieved.5 More recently, adjunctive pharmacotherapy has been shown to improve these rates to the region of 68% for TIMI 3 flow.6 Nonetheless, a significant number of patients fail to restore coronary patency following thrombolysis.7,8
Reperfusion strategies for treating STEMI can involve both mechanical and pharmacological methods. However, combinations of these strategies are not quite black or white, and subtle differences influence outcome. On the one hand there is thrombolysis, and on the other is primary PCI; between these are rescue PCI, systematic PCI and facilitated PCI.
Considering the Treatment Strategies of Facilitated and Systematic Percutaneous Coronary Intervention
Unsurprisingly, these treatment strategies have often been considered together in early trials. It is worthwhile setting the stage by reviewing these data, as there are important lessons to be learned. Considering historical data, the 3,262-patient TIMI II trial randomised an invasive strategy at 18–48 hours versus a conservative strategy following recombinant tissue-type plasminogen activator (rtPA). The primary end-point of reinfarction or death was not significantly different between the invasive and conservative groups (10.9 versus 9.7%). At six weeks there was no difference in ejection fraction.9 SWIFT randomised 993 patients similarly after thrombolysis with anistreplase.
Again, there was no significant difference in death or reinfarction rates between invasive and conservative arms at 12 months (death: 5.8 versus 5%; p=0.6; reinfarction: 15.1 versus 12.9%; p=0.4).10 The Streptokinase Angioplasty in Myocardial Infarction (SAMI) trial attempted to compare primary PCI with systematic PCI after streptokinase. The trial demonstrated no difference in ventricular function or arterial patency rates at six months. However, in the context of this review the relevance of the data is questionable, as 40% of patients received thrombolysis after PCI.11
In the Assessment of the Safety and Efficacy of a new Thrombolytic regimens (ASSENT-4) study, the role of primary PCI and facilitated PCI were compared. The study had aimed to recruit 4,000 patients but was prematurely stopped at 1,667 patients. This was due to an excess of major adverse effects in the facilitated PCI group, including in-hospital mortality (6 versus 3%; p=0.0105); the primary end-point of death, heart failure or shock (19 versus 13%; p=0.0045); ischaemic cardiac complications (6 versus 4%; p=0.279); and stroke (1.8 versus 0%; p<0.0001).12 This trial did not support the use of facilitated PCI.
The GRupo de Análisis de la Cardiopatía Isquémica Aguda (GRACIA-1) trial randomised 500 patients to angiography with or without PCI within 24 hours of fibrinolysis with rtPA to an ischaemia-guided post-fibrinolytic strategy.13 The primary end-point was combined death, reinfarction or revascularisation at 12 months. The facilitated angioplasty group had a significantly lower primary end-point (9 versus 21%; p=0.0008) and a trend towards lower death and reinfarction (7 versus 12%; p=0.07). The incidence of spontaneous recurrent in-hospital ischaemia was higher in the conservative group (2 versus 12%; p<0.0001). Thus, in contrast to ASSENT-4, GRACIA-1 favoured facilitated PCI. The differences in outcome between GRACIA-1 and ASSENT-4 have been attributed to the more contemporary nature of the later study. There was therefore a much higher use of stents (80%) and abciximab (32%) in GRACIA-1. The study benefited from a lower bleeding complication rate, likely to be due to heparin being discontinued after PCI. In addition, the medium time from thrombolysis to balloon inflation was 104 minutes in ASSENT- 4, which is too short a time to allow thrombolysis to be effective. The administration of clopidogrel was not mandated in this trial and was left to the discretion of the operators. With our current understanding of the importance of clopidogrel, this may have unwittingly introduced an uncontrolled variable.
In 2006, Keeley et al. published a meta-analysis of primary and facilitated PCI trial data.14 This meta-analysis included more contemporaneous data, with 2,237 patients assigned to facilitated PCI and 2,267 patients treated by primary PCI. Significantly more patients with facilitated PCI than primary PCI died (5 versus 3%, odds ratio [OR] 1.38), reinfarcted (3 versus 2%, OR 1.71) and had urgent target vessel revascularisation (4 versus 1%, OR 2.39). There were more adverse effects in the facilitated group versus primary PCI (7 versus 5%, OR 1.1 for major bleeding; 0.7 versus 0.1%; p=0.0014 for haemorrhagic and non-haemorrhagic stroke). One can conclude from this meta-analysis that there is no benefit of facilitated PCI over primary PCI.
More recently, the Facilitated InterventioN with Enhanced reperfusion Speed to Stop Events (FINESSE) trial has addressed the question once again in a multicentre randomised controlled trial of 2,452 patients.15 The facilitation of PCI with either abciximab alone or in combination with a reduced dose of reteplase was compared with primary PCI, where the abciximab was administered just before the PCI. The primary end-point was all-cause mortality, ventricular fibrillation (at >48 hours), cardiogenic shock and congestive cardiac failure at 90 days. No significant differences were demonstrated in the primary end-point (9.8 versus 10.5 versus 10.7% in the combination facilitated, facilitated and primary groups, respectively) nor in complications of MI (7.4 versus 7.5 versus 9%). Of note, however, is that significantly more patients had TIMI 3 flow in the combination facilitated group (32.8 versus 14.1 versus 12%; p<0.001). This was paralleled by a significantly higher rate of ST resolution (43.9 versus 33.1 versus 31%; p=0.003). However, there was a graded higher incidence of non-intracranial major and minor bleeding between the combination facilitated/facilitated/primary PCI groups (14.5 versus 10.1 versus 6.9%, p<0.001). It should be noted that the study was stopped prematurely due to difficulty in recruiting patients. The initial target had been to reach 3,000 patients. It is notable that in this study the median door-to-balloon time was 2.2 hours for all patients.
The data outlined above do not provide any convincing evidence that there is objective benefit in following a facilitated or systematic PCI strategy, and therefore these strategies cannot be recommended. However, a more recent study, Ongoing Tirofiban In Myocardial infarction Evaluation 2 (On-TIME-2), compared primary PCI with early tirofiban to that without tirofiban in 984 patients.
The mean residual ST elevation before PCI was found to be significantly lower in the tirofiban-treated group (10.9 versus 12.1; p=0.028). A similar significant reduction in residual ST segment one hour after PCI was seen in the tirofiban group (3.6 versus 4.8; p=0.003). There was no difference between the groups for major bleeding (p=0.36). Once again, this raises the possibility that facilitated PCI may be a reasonable strategy.
Rescue Percutaneous Coronary Intervention Following Failed Thrombolysis
This leaves us with the strategy of rescue PCI, i.e. PCI following the failure of thrombolysis to re-establish epicardial blood flow. Intuitively, this would appear to be a beneficial strategy, but, as shown above, intuition may not be enough, and robust investigation can disprove intuition. The at best equivocal results of studies of facilitated and systematic thrombolysis may be the reason why rescue PCI is often considered a conscience appeaser by cardiologists; however, is this truly the case?
The Randomized Evaluation of Salvage angioplasty with Combined Utilization of End-points (RESCUE 1) trial was the first significant trial to look at rescue PCI in isolation.16 In the trial, 151 patients with an anterior STEMI with failed thrombolysis defined at angiography with TIMI 0 or 1 flow were randomised to rescue PCI or conservative treatment. Notably, in this trial stents were not deployed and glycoprotein IIb/IIIa inhibitors were not used, reflecting the period during which the trial was conducted. Important exclusion criteria included patients presenting at less than three hours, left main stem stenosis ≥50% and patients with a prior MI. Despite these shortcomings, there was a significant difference in the primary end-point of left ventricular ejection fraction at 30 days, measured by radionuclide imaging with exercise. In the PCI group the 30-day exercise ejection fraction was 43%, versus 38% in the conservative group (p=0.04). There was no significant difference in resting ejection fraction. Non-pre-defined analysis of the composite secondary end-points showed a significant difference between the two groups, favouring PCI when looking at death or severe heart failure in isolation (6 versus 17%; p=0.05). There was no significant difference in the secondary end-point of 30-day mortality, heart failure and ventricular tachycardia (>48 hours) (18 versus 28%; p=0.31).
The Middlesbrough Early Revascularisation to Limit Infarction (MERLIN) trial randomised 307 patients with ≥50% ST-segment persistent elevation at 60 minutes following thrombolysis.17 The primary end-point of all-cause mortality at 30 days was not significantly different between the two groups (9.8 versus 11%; p=0.7). The secondary composite end-point of mortality, reinfarction, stroke, revascularisation and heart failure at one month favoured the rescue PCI group (27.3 versus 50%; p=0.02). This was mainly driven by the reduced need for revascularisation. These benefits came at the expense of higher rates of stroke and transfusion in the PCI group (4.6 versus 0.6%; p=0.03). Essentially, these findings were maintained at three-year follow-up. There was no late survival benefit in the rescue PCI group at three years (16.9 versus 17.6%; p=0.9), and the incidence of the composite secondary end-point was higher in the conservative group (49 versus 64.3%; p=0.01), driven by the relative lack of early unplanned revascularisation in the rescue PCI group.8
The MERLIN trial protocols have been criticised. First, the trial was underpowered to meet the primary end-point. However, demonstrating a difference in all-cause mortality may have been ambitious, as this has proved difficult to demonstrate in trials involving primary PCI. There were additional concerns regarding the study design. Randomisation at 60 minutes is clearly too early for thrombolysis to have definitely failed. This early randomisation has led to some criticism that MERLIN was effectively a study of early systematic PCI. This may go some way to explaining the relatively high mortality rates observed. Although better than RESCUE 1 in reflecting current practice, the use of stents and glycoprotein IIb/IIIa inhibitors was woefully low at 50.3 and 3.3%, respectively, in MERLIN, and this attracted some criticism.
The REscue Angioplasty vs Conservative treatment of repeat Thrombolysis (REACT) study was somewhat better designed and reflects current practice as regards stent and glycoprotein IIb/IIIa use.7 This study randomised 427 patients to rescue PCI, repeat thrombolysis and conservative management. The composite primary endpoint of all-cause mortality, recurrent infarction, stroke and heart failure at six months was 15.3, 31 and 29.8%, respectively (p<0.01). This difference was accounted for by a lower rate of recurrent infarction (2.1, 10.6 and 8.5%, respectively; p<0.01). Again, the study was not powered to detect a difference in mortality, but a trend was observed (6.2, 12.7 and 12.8%, respectively; p=0.12).
The trial design for REACT reflected current practice, as 68.5% had stent implantation and 43.4% had glycoprotein IIb/IIIa administered during PCI. Randomisation at 90 minutes also made the study a valid investigation of rescue PCI, as opposed to systematic PCI following thrombolysis. Interestingly, no significant differences were found in major bleeding events. A clear mandate for rescue PCI was established.
This discussion would not be complete without including the recent Combined Abciximab REteplase Stent Study in Acute Myocardial Infarction (CARESS-in-AMI).18 This study included 600 high-risk patients with one or more high-risk features (extensive ST-segment elevation, new-onset left bundle branch block, previous MI, Killip class >2 or left ventricular ejection fraction ≤35%). Patients received half-dose reteplase, abciximab, unfractionated heparin (40μ/kg) and aspirin. Following this they were randomised to immediate PCI (with clopidogrel) or a rescue PCI strategy. The primary outcome was a composite of death, reinfarction or refractory ischaemia at 30 days. This occurred in 4.4% of the immediate PCI group and 10.7% of the rescue group (p=0.004). There was no difference in major bleeding complication rates (3.4 versus 2.3%; p=0.47) or stroke (0.7 versus 1.3%; p=0.50). This study clearly showed a benefit for this high-risk STEMI group, favouring a facilitated strategy. It is worth noting that half-dose reteplase was used in conjunction with abciximab. Nonetheless, there may be a role for facilitated PCI in high-risk STEMI patients.
Meta-analyses of Rescue Percutaneous Coronary Intervention
Several meta-analyses have been carried out that study the issue of rescue PCI.19–21 Broadly, these show that rescue PCI reduces mortality and heart failure at 30 days. However, this is countered by an increase in the incidence of stroke. The meta-analysis by Patel et al.19 included the significant trials mentioned above (RESCUE, MERLIN and REACT), as well as the early study by Belenkie et al.22 with 28 patients, and RESCUE 2, which recruited 29 patients before being terminated early.23 Nonetheless, this meta-analysis included 800 patients and demonstrated a 36% risk reduction (RR) for short-term mortality (p=0.048). There was a trend towards reduced heart failure (RR 28%; p=0.06) and thromboembolic stroke (RR 3.61; p=0.07). Collet et al.20 excluded RESCUE from their meta-analysis, as TIMI 2 flow patients were randomised. In total, five studies with 920 patients were included.
At 30 days a non-significant 37% reduction in mortality was demonstrated (p=0.055). The benefit of rescue PCI in reducing reintervention was underscored by the significant reduction in the combined end-point of death or reinfarction at 30 days (p=0.012). Wijeysundera et al.21 conducted a meta-analysis of eight trials including 1,177 patients. Again, no significant reduction in mortality was demonstrated in the rescue PCI group versus conservative treatment (RR 0.69; p=0.09). Heart failure (RR 0.73; p=0.05) and reintervention (RR 0.58; p=0.84) were significantly reduced. The risk of stroke increased with rescue PCI from 0.7 to 3.4% (p=0.02). Almost certainly, more relevant data will become available to aid us. We await the Trial of Routine ANgioplasty and Stenting after Fibrinolysis to Enhance Reperfusion in Acute Myocardial Infarction (TRANSFER-AMI), which will include 1,200 high-risk STEMI patients.24 This is a contemporary study that compares routine PCI within six hours after fibrinolysis, rescue PCI and delayed cardiac catheterisation.
Conclusions
On reviewing the wealth of data available, several common features arise. First, there is little doubt that timely primary PCI for STEMI is an effective treatment with a strong body of evidence behind it. However, thrombolysis is still widely used but fails to restore target epicardial blood flow in a significant proportion. With this in mind, how does the evidence guide us? It is clear that rescue PCI is beneficial compared with conservative management in the studies outlined above. Proving a mortality benefit is difficult, as significant differences appear to be driven by reduced rates of reintervention and heart failure. However, there is evidence of a trend towards reduced mortality with rescue PCI. Overall, we can conclude that there is an evidence-based mandate for rescue PCI and that with its use we would be acting in the best interests of our patients with a clear conscience.