Lyse and Statistics: GRACIA-2 vs. ASSENT-4
The first Journal Club of the year got off to a thunderous start (there was lightning and flooding, too, and an explosion a little bit later). Those who braved the elements participated in a discussion about thrombolysis before angioplasty for STEMI, centered on two papers with two different conclusions about its benefit.
The primary paper we covered was from the GRACIA-2 non-inferiority RCT, called Primary angioplasty vs. early routine post-fibrinolysis angioplasty for acute myocardial infarction with ST-segment elevation (from the European Heart Journal 2007 Vol. 28, p949-960).
Our secondary paper was from the ASSENT-4 RCT, called Primary verses tenecteplase-facilitated percutaneous coronary intervention in patients with ST-segment elevation acute MI (from the Lancet 2006: Vol 367, p569-578).
The GRACIA authors had their hearts in the right place — they were thinking about the STEMI patients that experience significant delays before PCI, and wondered if thrombolytics beforehand were safe or helped. Thus, the timings for lysis and PCI is different than in the ASSENT trial — GRACIA patients got PCI within 3-12 hours of lytics (average: 5.9 hours) whereas ASSENT patients waited just 135 minutes on average (interval quartiles were 90 and 210 minutes).
In 2002-2003, the GRACIA authors enrolled (presumably) all patients over 18 from fifteen hospitals in Spain and Portugal (10 of which had interventional facilities). They were included if they had symptom onset within 12 hours, lasting longer than 30 minutes, with ST-segment elevation of at least 0.1 mV in at least two limb leads, or at least 0.2 mV in two or more contiguous precordial leads, or left bundle-branch block or paced rhythm. Patients were exlcuded if they were in shock, had a terminal illness, had evidence of cardiac rupture, liver dysfunction, creatinine over 2.5, or contraindication to lytics, aspirin, plavix, heparin (or more arcane reasons described in the text).
Patients who met criteria were randomized to receive PCI of the culprit artery within three hours of randomization, or fibrinolysis plus delayed PCI. All patients received some amount of aspirin (200-500 mg) in the ED. The fibrinolysis-first group, 104 patients in all, received a 5 second bolus of tenecteplase over a period of 5 s based on bodyweight (between 30-50 mg) and an enoxaparin bolus of 30 mg followed by a subcutaneous dose of 1 mg/kg. Then, between 3 and 12 hours later, PCI was performed on 87 of these patients. The culprit vessel was dilated if its TIMI flow was less than 3 or it was more than 50% stenosed. They’d stent more lesions if a lot of myocardium was threatened, and patients who remained in pain or maintained ST elevations 90 minutes after lytics went straight to PCI.
The 108 primary-PCI patients received unfractionated heparin (to achieve an activated clotting time of 350–450 s during the invasive procedure) and abciximab, given as bolus of 0.25 mg/kg, followed by a drip rate of 0.125 mg/kg/min. One patient died before catheterization. Of the PCI grouped patients, one died, 107 got angiography, and of those patients, 95 received Guidant stents. Of the 104 fibrinolysis patients, 84% continued to PCI and received stents. “Beta blockers, ACE inhibitors, statins, and post-interventional antithrombotic therapy were administered to all patients, as outlined in the international guidelines.” No patients got more heparin or lovenox if recanalization was successful.
Results made sense in terms of reperfusion — 67% of the early fibrinolysis group had TIMI 3 flow in the culprit artery by the time angioplasty was performed, compared to 14% in the early PCI group. Post-procedure numbers were similar, at around 85%, and all but one patient maintained TIMI-3 flow at 6-week followup. Infarct size was similar for both groups, and both groups had similar good LV function at 6-week followup.
In terms of bleeding events, “there was no significant difference in major bleeding rate among the groups; two patients (2%) in the post-fibrinolysis angioplasty group experienced major bleeding, with one episode of intracranial haemorrhage, and three patients (3%) in the primary angioplasty group.”
At six months, “the incidence of death, reinfarction, stroke (no cases of stroke were observed during follow-up), or revascularization was 10% in the early routine post-fibrinolysis angioplasty group compared with 12% in the primary angioplasty group, (relative risk, 0.80; 95% confidence interval, .37–1.74).”
In table 4, you see that the primary PCI group had 6 deaths (5.6%) and the early fibrinolysis group had three deaths (2.9%). Based on this we calculated a number-needed-to-treat of 37 (taking a 5.6% in the PCI-alone group, minus 2.9% in the lytics+PCI group = 2.7% absolute risk reduction) . That’s not a bad NNT, considering what’s at stake and the availability of lytics. But can we put faith in these results?
Our biggest problem with the GRACIA-2 data is in Table 1 — it seems the PCI-only patients were sicker. They were much more likely to have three-vessel disease (15 vs. 3 for the lytics+PCI group), degree of stenosis was greater, and number of non-stenotic vessels was less. The authors try to write this off as an effect of the lytics, but we were skeptical: I’d trust tenecteplase to relieve a vessel occluded by thrombus, but not to widen every stenotic coronary. Rather, we thought the degree of sick hearts in the PCI arm was a failure of randomization, either by fluke or by nonblinded decisionmaking from the caregivers. We found ourselves wishing patients had undergone CT-angio to better ensure equality.
Also, when you compare this data to the ASSENT trial, it makes less sense. The ASSENT trial took patients from 24 countries in an attempt to see if patients with large MIs did better with “facilitated PCI” than with delayed PCI. They wanted 4000 patients, but had to stop early because of higher in-hospital mortality in the group receiving PCI and tenecteplase (6% vs. 3% in PCI alone). The facilitated group had more strokes, more reinfarction, and more need for revascularization. This was puzzling, as the infarct patency was improved with fibrinolysis before angioplasty (similar to GRACIA data).
The ASSENT authors speculate the bad outcomes with lytics + PCI are due to suboptimal antithrombotic cotherapy, such as withholding heparin — but the authors were concerned about intracranial bleeds (which turned out to be a nonissue). They also point to treatment delays in the lytics+PCI group, compared to PCI alone.
So, with the disturbing ASSENT-4 data and the inapplicability of GRACIA to our patient population (for instance, the Iberian diet, while delicious, is quite different than ours) we cannot apply these findings to our practice. As much as we’d like to give lytics before PCI, the data at this time doesn’t support it.
Of course, ACEP provides guidelines on fibrinolytic therapy in acute MI, and this summer the AHA came out with new recommendations for unstable angina / NSTEMI. And, as always, EMcrit.org has useful references, including the 2004 STEMI treatment guidelines and a section on PCI vs. lytics (scroll down).
