Lancet 1990;336:65-71.
Background Large trials up to this point had established the role of thrombolytic therapy and aspirin in patients with acute MI. The next question centered on the different types of thrombolytic agents as well as the merits of adding high dose heparin to aspirin following revascularization.
Data from smaller studies showed that alteplase (tPA) had a higher recanalization rate at 90 minutes compared to streptokinase (SK). “Recanalization” is a surrogate endpoint – a person doesn’t necessarily care whether their artery is open or not at 90 minutes, but instead, whether they live or die, and how they live in the aftermath of a heart attack.
The GISSI-2 trial sought to test the hypothesis that tPA would reduce the composite hard endpoint of mortality and extensive LV dysfunction compared to streptokinase (SK) and that heparin plus aspirin compared to aspirin alone would do the same.
Cardiology Trial’s Substack is a reader-supported publication. To receive new posts and support our work, consider becoming a free or paid subscriber.
Patients Patients were eligible who presented with chest pain and ST segment elevations of ≥1 mm in any limb lead of the ECG and/or ≥2 mm in any precordial leads and were admitted to the CCU within 6 hr from the onset of symptoms.
Absolute contraindications included: recent or current bleeding, stroke within the previous 6 months, a surgical or invasive procedure or trauma within the previous 2 weeks, uncontrolled hypertension defined as SBP ≥200 mmHg or DBP ≥110 mmHg, or previous treatment with SK within 6 months.
Baseline characteristics The number of patients who were admitted to CCUs with acute myocardial infarctions (MI) over the study period was 38,086 and 12,490 (33%) were ultimately enrolled. Reasons for exclusion included: more than 6 hr from onset of symptoms (61%), contraindications to fibrinolysis (15%), unlikely to have acute MI (13%), administrative reasons (5%), ST depressions (4%) and 1% were not reported.
Similar to GISSI-1, which also provided information on patients enrolled versus those who were not, women were more likely to be excluded (30% of excluded vs 20% of enrolled) as were patients >70 years of age (41% of excluded vs 22% of enrolled). The death rate of those excluded was higher (11% of excluded vs 9% of enrolled).
Approximately 80% of patients enrolled were men under the age of 70. Patients with inferior (34%) and anterior STEMI’s (31%) composed more than half of the cohort. 72% of patients presented within 3 hours of symptom onset and over 95% had Killip scores of II or below with the vast majority being Killip I (78%).
Procedures Immediately following randomization half of all patients received either 1.5 MU of SK infused over 1 hour or 100 mg of tPA infused over 3 hours. Since this was a 2x2 factorial design, half of all patients also received 12,500 U of subcutaneous heparin twice daily starting 12 hours after the beginning of the tPA or SK infusion and to be continued until hospital discharge.
All patients without specific contraindications were recommended to receive oral aspirin 325 mg/day and 5-10 mg of IV atenolol, even before randomization, as soon as evolving MI was diagnosed.
Endpoints The combined primary endpoint consisted of all-cause mortality plus the number of patients who had late (day 4 or later) clinical congestive heart failure, or extensive LV damage (LVEF <35%) in the absence of clinical heart failure.
Results 12,490 patients were randomized from 223 CCUs in Italy. Alteplase did not significantly reduced the composite primary endpoint compared to SK overall (RR 1.04; 23.1% vs 22.5%; 95% CI 0.95-1.13) or in any important subgroups based on hours from symptom onset or site and type of infarct. Alteplase did not significantly reduce any of the individual components of the combined endpoint including death (9.0% vs 8.6%). There were no differences in adverse events or post-infarction in-hospital complications between patients receiving alteplase or streptokinase.
Heparin also did not significantly reduce the composite primary endpoint (RR 0.99; 22.7% vs 22.9%; 95% CI 0.91-1.08). Data on subgroups are not presented for the comparison of heparin vs controls.
Regarding individual components of the primary endpoint, heparin was associated with a reduction in death (8.3% vs 9.3%); however, the authors do not present results of statistical tests for it. It should be noted that the trial was specifically designed to detect a 15% reduction in the primary endpoint, which was estimated to occur in about 20% of patients. Over 25,000 patients would need to be enrolled to detect a 1% difference in death at the event rates noted for this comparison based on normal confidence levels. The authors were right to not emphasize this difference.
Unlike the comparison of alteplase versus streptokinase, heparin did increase total bleeding events (RR 1.87; 10.6% vs 5.9%; 95% CI 1.65-2.13), which translates to a number needed to harm of only 21 patients. This difference was driven primarily by minor bleeding events from an absolute standpoint; however, the difference in major bleeding events was also statistically significant. There were no significant differences between groups in post-infarct in-hospital complications.
Conclusions Alteplase did not reduce the primary composite endpoint of death, clinical heart failure or extensive LV dysfunction in patients with AMI treated within 6 hours of symptom onset nor did it cause significant harm compared to the old standard – streptokinase. Furthermore, the use of therapeutic heparin did not reduce events compared to control but did significantly increase harm, mainly via minor bleeding events, with a NNH of approximately 20 patients.
These results are notable for several reasons. First, it shows the importance of testing interventions in large scale trials that are based on hard outcomes and not surrogate endpoints. Existing evidence supported the superiority of alteplase over streptokinase for vessel patency at 90 minutes, and while it was presumed to be important, it did not translate to any meaningful improvement in outcomes that actually matter to patients, such as staying alive and free of heart failure.
The true danger of adopting practices based on surrogate endpoints will be highlighted in our next trial but it is worth remembering the lessons from GISSI-2 in this regard.
The second takeaway involves patient selection. Note how patient selection was narrowed in GISSI-II compared to GISSI-I. This was done for the purpose of detecting a smaller benefit. Patients were excluded if they presented 6 hours of more following symptom onset. As a result, the overall mortality rate was lower in this trial for all patients compared to those who received thrombolysis in GISSI-1.
As a general rule, detecting smaller benefits requires more careful patient selection. On the other hand, it limits external validity (generalizability) of results. Had this trial been positive, it would only apply to patients presenting less than 6 hours from symptom onset and the majority of patients present outside this window (see baseline characteristics).
Finally, a point about beta blocker use in acute MI is worth pointing out here. Despite the fact that immediate intravenous atenolol was recommended for all patients, unless contraindicated, only 45% of patients in the trial received it. This is far less than the overall percent of patients that would be predicted based on their Killip scale at randomization (78% were Killip I).
This reflects the fact that the evidence for beta blockade in AMI was and still is limited to highly selected patients as we discussed in prior reviews (BHAT and ISIS-1 trials). The present-day, indiscriminate use of beta blockers in AMI patients is not justified based on evidence from clinical trials.
Thank you for reading Cardiology Trial’s Substack. This post is public so feel free to share it.
Cardiology Trial’s Substack is a reader-supported publication. To receive new posts and support our work, consider becoming a free or paid subscriber.