Author + information
- Received January 10, 2003
- Revision received March 6, 2003
- Accepted March 12, 2003
- Published online October 1, 2003.
- Christophe L Dubois, MD*,
- Ann Belmans, MSc*,†,
- Christopher B Granger, MD, FACC‡,
- Paul W Armstrong, MD, FACC§,
- Lars Wallentin, MD, FACC∥,
- Paolo M Fioretti, MD, FACC¶,
- José L López-Sendón, MD, FACC#,
- Freek W Verheugt, MD, FACC**,
- Jürgen Meyer, MD, FACC††,
- Frans Van de Werf, MD, FACC*,* (, )
- ASSENT-3 Investigators
- ↵*Reprint requests and correspondence:
Prof. Frans Van de Werf, Department of Cardiology, University Hospital Gasthuisberg Leuven, Herestraat 49, B 3000 Leuven, Belgium.
Objectives The aim of this study was to evaluate percutaneous coronary intervention (PCI) in the Assessment of the Safety and Efficacy of New Thrombolytic Regimens (ASSENT-3) trial.
Background In the ASSENT-3 trial, co-therapy with abciximab (ABC) or enoxaparin (ENOX) reduced ischemic complications after ST-elevation acute myocardial infarction treated with tenecteplase when compared with unfractionated heparin (UFH). The effect of these new co-therapies on the results of PCI is unknown.
Methods Clinical outcomes in patients who received co-therapy with ABC, ENOX, or UFH and subsequently underwent an elective (n = 1,064) or urgent (n = 716) PCI in the ASSENT-3 trial were compared.
Results No significant differences in clinical end points were observed in patients who underwent an elective PCI. A non-significant trend toward fewer in-hospital myocardial re-infarctions was seen with ABC and ENOX when compared with UFH (0.5% vs. 0.6% vs. 1.5%, respectively). The incidence of bleeding complications was similar in the three treatment arms. Significantly fewer ABC- and ENOX-treated patients needed urgent PCI compared with UFH (9.1% vs. 11.9% vs. 14.3%; p < 0.0001), but outcomes in these patients were in general less favorable (30-day mortality: 8.2% vs. 5.4% vs. 4.5%; 1-year mortality: 11.0% vs. 8.5% vs. 5.6%; in-hospital re-infarction: 3.9% vs. 2.5% vs. 2.7%; major bleeding complications: 8.8% vs. 7.0% vs. 3.4%). In pairwise comparisons with UFH, the higher one-year mortality and major bleeding rates after ABC were statistically significant (p = 0.045 and p = 0.012, respectively).
Conclusions Clinical outcomes after elective PCI were similar with the three antithrombotic co-therapies studied in ASSENT-3. Although fewer patients needed urgent PCI with ABC and ENOX, clinical outcomes were less favorable in this selected population, especially with ABC.
Reperfusion therapy has been one of the major advances in the treatment of acute myocardial infarction (AMI) since the introduction of the coronary care unit. The combined administration of a fibrinolytic agent, unfractionated heparin (UFH), and aspirin became the standard of care in the late 1980s (1,2). Although significant improvements in fibrinolytic therapy were made, substantial challenges remain (3–5). Optimal reperfusion of the infarct-related epicardial coronary artery is achieved in only 50% to 60% of the patients 60 to 90 min after initiation of treatment (6–10). Other concerns are reperfusion at the tissue level, recurrent ischemia or re-infarction, and bleeding complications (11,12). Newer co-therapies, including low-molecular-weight heparin and platelet glycoprotein IIb/IIIa inhibitors were tested in recent trials in conjunction with fibrinolytics (13–15). Although no mortality benefit was seen, these newer treatment regimens led to a consistent reduction in ischemic complications, including re-infarction and refractory ischemia. This benefit was partly counterbalanced by an increased rate of non-intracranial bleeding complications.
A substantial proportion of patients with ST-elevation AMI who receive thrombolytic therapy subsequently undergo a percutaneous coronary intervention (PCI). Unfractionated heparin has been the anticoagulant of choice during PCI. As in acute coronary syndromes, newer anticoagulants (low-molecular-weight heparin or direct antithrombins) and antiplatelet agents (platelet glycoprotein IIb/IIIa inhibitors) emerged and have been proven to be safe and perhaps even superior in elective and urgent PCI (16–24). However, the effect of these newer treatment regimens on the outcome of PCI performed after lytic therapy is largely unknown.
We analyzed clinical outcomes in patients who underwent PCI after lytic therapy in the Assessment of the Safety and Efficacy of New Thrombolytic Regimens (ASSENT-3) trial. In this trial, 6,095 patients with ST-elevation AMI of <6 h were randomly assigned to one of three regimens: half-dose tenecteplase (TNK-tPA) with weight-adjusted low-dose UFH and a 12-h infusion of abciximab (ABC), full-dose TNK-tPA and enoxaparin (ENOX) for a maximum of seven days, or full-dose TNK-tPA with weight-adjusted UFH for 48 h. The main results of this trial have been published (15).
Using the ASSENT-3 database, we selected all patients who underwent urgent or elective PCI during the initial hospitalization and evaluated them separately. Elective PCI referred to a planned procedure and was performed in 1,064 patients (17.5% of the total ASSENT-3 study population). A total of 716 patients (11.7%) underwent an urgent PCI, defined as an ischemia-driven revascularization before hospital discharge. Patients who underwent both an urgent and an elective PCI (n = 15) were included in the analysis of urgent PCI. Outcomes of early PCI (urgent or non-urgent), performed on the same day or the day after thrombolytic treatment, were also assessed separately. This group comprised 667 patients (11.0%). These patients are of special interest because the treatment effects of ABC and UFH have been thought to be most pronounced during the first 24 to 48 h. Furthermore, most efficacy end points of interest (death, re-infarction, refractory ischemia) occur early and could be influenced by early intervention (25).
Analyzed data included the allocated trial medication, time of randomization, death at 30 days, one-year mortality, in-hospital myocardial re-infarction, in-hospital refractory ischemia, in-hospital major and minor bleeds, puncture-site-related bleeds as well as day of PCI, re-infarction, or refractory ischemia. In the case of urgent PCI, reason for urgency was also assessed (per protocol defined as myocardial re-infarction, myocardial ischemia, or “others”). Data were checked for completeness and for internal consistency with source documents.
Definitions of efficacy and safety end points
We analyzed the composite end point of 30-day mortality, in-hospital re-infarction, or in-hospital refractory ischemia (efficacy end point) as well as the individual components of the efficacy end point. Because one-year follow-up was recently completed, one-year mortality data were also included in our analysis. The incidence of in-hospital intracranial hemorrhage (ICH), major non-cerebral bleeds, and major and minor bleeds at the arterial puncture site (safety end points) was also assessed. The same definitions for re-infarction, refractory ischemia, and major versus minor bleeding, as described in the main report of ASSENT-3, were also used in our analysis (15).
Baseline characteristics were tabulated for each treatment group. For binary and categorical variables, differences between treatment groups were assessed using the chi-square test. For continuous variables, an overall Ftest was performed.
The end point data were analyzed using a proportional hazards regression. To allow for the timing of the outcome (before or after PCI) to be taken into account, PCI was included as a time-dependent covariate (coded as 0 before the performance of PCI, coded as 1 afterwards). In addition, an interaction term between this time-dependent covariate and treatment was also included in the model to allow for different treatment effects before and after the performance of a PCI. Furthermore, propensity scores (probability of receiving treatment based on baseline characteristics) were calculated for all patients and included in the model (26). By including propensity scores in the model, it is hoped that any imbalances in risk factors between the treatment groups after the performance of a PCI will be reduced or removed. For each of the three subgroups that were analyzed, a propensity score analysis was performed, in which the probability of receiving one of the three treatments was modeled on the basis of patients’ baseline characteristics, using a logistic regression. The models used to determine the propensity scores included patients’ baseline characteristics and their second-order interactions: gender, age, weight, infarct location, previous myocardial infarction, systolic and diastolic blood pressure, Killip class at entry, time from symptom onset to treatment initiation, heart rate, diabetes, previous coronary artery bypass surgery, previous PCI, country (U.S./Europe/other), and randomization date. Visual inspection of exploratory plots revealed that no transformation of the continuous variables was necessary for inclusion into the propensity model. For the urgent and early PCI groups, the concordance obtained with the propensity model was good with C-statistics of 0.82 (UFH vs. ENOX) and 0.83 (UFH vs. ABC) for the group of patients with urgent PCI and C-statistics of 0.82 (UFH vs. ENOX) and 0.85 (UFH vs. ABC) for the patients who underwent an early PCI. For the subgroup of patients who underwent an elective PCI, the C-statistics were 0.74 (UFH vs. ENOX) and 0.73 (UFH vs. ABC). Then, based on the obtained propensity scores, all patients were divided into quintiles. For all subgroups studied, patient numbers and baseline characteristics were tabulated by treatment group within each of the obtained quintiles. The results indicated that baseline characteristics were balanced within quintile and that none of the covariates that were included in the propensity model induced an inappropriate separation of the treatments.
Based on the proportional hazards regression, pair wise treatment comparisons between ENOX and UFH and between ABC and UFH were calculated, together with their corresponding 95% confidence interval. A chi-square test was performed to assess the statistical significance of the pairwise comparisons.
The distribution of patients who underwent an early, elective, or urgent PCI in the three treatment arms is shown in Table 1. Fewer patients in the ABC arm than in the UFH arm underwent an urgent or early PCI. There was a similar trend for urgent PCI with ENOX.
The baseline characteristics of the patients who underwent an elective PCI were similar in the three treatment arms (Table 2). Median time to revascularization was five days in each group. The composite end point of 30-day mortality, in-hospital myocardial re-infarction, or in-hospital refractory ischemia, its individual components, one-year mortality and non-intracranial bleeding complications in patients who underwent elective PCI are shown in Figure 1.
There was one ICH in the UFH arm (0.30%) and one in the ENOX arm (0.29%). No ICH was seen in the ABC arm. No significant differences were observed between the different treatment regimens. A trend towards fewer in-hospital myocardial re-infarctions was seen in the ABC and ENOX arms compared with the UFH arm, but statistical significance was not reached. Low 30-day and one-year mortality rates were observed after elective PCI, the lowest in the ENOX arm. The incidences of in-hospital major non-cerebral bleeds and arterial puncture-related major bleeds were also similar in the three treatment arms. The proportion of patients who received additional UFH during PCI was similar in the three groups (77% for ABC, 79% for ENOX, and 75% for UFH). Pair wise treatment comparisons with hazard ratios and corresponding 95% confidence intervals for outcome after elective PCI are also shown in Figure 1. Overall, there was a slightly more favorable composite efficacy outcome with ENOX treatment, whereas safety outcome was slightly better with ABC.
There was a significantly lower need for urgent PCI in patients given ABC or ENOX when compared with UFH (Table 1). Fewer early PCI were performed in the ABC arm (Table 1). Of the urgent PCI performed, 75% were early PCI. The baseline characteristics in patients who needed urgent PCI or who underwent an early PCI are given in Tables 3 and 4. ⇓Except for age in the urgent PCI analysis and for time from onset of symptoms to TNK-tPA in the early PCI analysis, ⇓baseline characteristics were similar. Efficacy end points are summarized in Figures 2 and 3. ⇓There was a trend towards reduced efficacy in the ABC arm, with even a significantly higher one-year mortality when compared with UFH. ⇓
The safety end points in the urgent and early PCI patient population as well as the pairwise comparisons are shown in Figures 2 and 3. One ICH was seen in each group (0.55% for ABC, 0.41% for ENOX, and 0.34% for UFH) in urgent PCI. Significantly more major, non-cerebral bleeding complications were seen in the ABC group compared with the UFH group, whereas major bleeds at the arterial puncture site were only slightly more frequent. There was a modest increase in major, non-cerebral bleeds with ENOX compared with UFH, mainly as a result of more puncture-related bleeding complications. The proportion of patients who received UFH during PCI was similar in the three groups (71% for ABC, 68% for ENOX, and 71% for UFH). Major bleeding rates were more frequent in patients who needed urgent PCI compared with those in whom an elective PCI was performed (Table 5).
A significant proportion of patients with ST-elevation AMI who receive thrombolytic therapy subsequently undergo urgent or elective PCI.
The aim of our study was to evaluate clinical outcomes in patients who underwent PCI during initial hospitalization in the ASSENT-3 trial. In light of the pharmacologic properties attributable to ENOX, ease of administration and proven safety and efficacy in non–ST-elevation acute coronary syndromes and PCI, evaluation of the possible benefit or harm of ENOX in PCI early after thrombolysis of an ST-elevation AMI is important (16–18,27,28). Although there is clear evidence that ABC reduces thrombotic cardiac complications in patients undergoing PCI, it is not known whether ABC is beneficial after PCI in patients with ST-elevation AMI previously treated with reduced dose thrombolytics (19–22).
No significant difference in the primary composite efficacy end point or its components between the three treatment regimens was seen in the population who underwent elective PCI in ASSENT-3. Nevertheless, a trend towards fewer in-hospital myocardial re-infarctions was seen in the ENOX and ABC arms. Median time to elective revascularization was rather long (five days after randomization). Because per protocol ENOX was to be administered until revascularization or discharge with a maximum of seven days and because prolonged low-level platelet receptor blockade may still be present up to 10 days after the cessation of an ABC infusion, the antithrombotic effects of ABC and ENOX may have been responsible for these favorable outcomes, supporting the results of trials in elective PCI (16,17,19,20,22).
In patients who underwent an urgent PCI during the initial hospital stay in the ASSENT-3 trial, the overall low event rates after the procedure make the interpretation of differences in post-procedural outcome very hazardous. Our analysis shows a somewhat higher incidence of the efficacy end points after PCI in the ABC arm. More specifically, there was a significantly higher one-year mortality and also a slight excess in re-infarction. These worse outcomes are difficult to understand because one would have expected benefit in this regard from ABC (19,21,24). Although baseline characteristics were similar in the three post-randomization subgroups, differences in risk-profile (selection of a higher-risk population) could have played a role despite our attempts to adjust for these imbalances.
An alternative explanation could be the observed differences as to the urgency of PCI. In ASSENT-3, total PCI rates were similar in the three treatment arms with a significantly lower need for urgent intervention in the ABC arm and to a lesser extent in the ENOX arm and a borderline significant excess of elective PCI in the ABC arm. Patient characteristics (and hence probably lesion characteristics and extent of disease) were similar in the three treatment groups in ASSENT-3 (15). Because differences in urgency for revascularization are likely to be a direct treatment effect, patients with similar lesion complexity could potentially have been shifted from urgent to elective PCI or even medical therapy, as a consequence of the plaque-stabilizing characteristics of ABC. This could have resulted in a concentration of very complex “ABC-resistant” lesions in a smaller group of patients undergoing urgent revascularization and a selection of “ABC-stabilized” but complex lesions in the group of patients undergoing an elective intervention. Percutaneous coronary intervention in these patients with complex lesions might have a worse prognosis, and this could explain in part the higher mortality rates. On the other hand, some “ABC-stabilized” patients who were treated medically could have benefited from a PCI, preventing late re-infarction and death.
Safety of the new antithrombotic co-therapies was not an issue after elective PCI. The rates of in-hospital non-cerebral major bleeds as well as puncture-related bleeds were low and similar in the three treatment arms. These results are in accordance with other recent trials (16,17,20,22). The time to elective PCI in ASSENT-3 was long (±5 days), which may partly explain the low rates of major bleedings at the puncture site.
An excess of major bleeds was observed in patients who needed urgent PCI compared with those who underwent an elective PCI (Table 5). This excess was small in the UFH group but larger in the ABC and ENOX groups. When compared with UFH, the incidence of major, non-cerebral bleeds in patients undergoing urgent PCI was significantly higher with ABC, whereas there was a modest increase in patients treated with ENOX. Overall, puncture-related bleeds (major and minor) were similar in the three arms. Nevertheless, the incidence of major puncture-related bleeds was higher in the ENOX arm. It is unlikely that this difference could be attributed to additional administration of UFH during the procedure, because the proportion of patients who received additional UFH during PCI was similar in the three groups.
Bleeding sites differed among the three treatment arms: one-fourth of the major bleeds occurred at the arterial puncture site in the ABC arm, in contrast with one-half and almost two-thirds in the UFH and ENOX arms, respectively (Fig. 2). These observations are in line with those of the Global Use of Strategies To open Occluded coronary arteries (GUSTO V) trial (14). No single bleeding site could explain the relatively higher proportion of non-puncture-related major bleeding episodes in the ABC arm. New trials of facilitated PCI will evaluate the safety of different pre-PCI pharmacologic reperfusion regimens.
Our study has several limitations. Because this was a post hoc analysis, caution is needed when interpreting differences between treatment groups, because the performance of PCI could have been influenced by the study treatments. In addition, interactions between study treatment and PCI can further obscure the study treatment effect. We tried to address some of these issues by analyzing data using a proportional hazards regression that included the interaction between occurrence of PCI and treatment, in which PCI was included as a time-dependent covariate, and by calculating propensity scores in order to avoid or to reduce imbalances between treatment groups after the performance of a PCI. It should be kept in mind that even when taking the above measures, it is impossible to ascertain the true effect of the study treatment in the group of patients who underwent a PCI. No adjustments for multiplicity were made, and false positive results can therefore not be excluded. On the other hand, event rates were low, and true differences might have been missed because of the small sample size.
Clinical outcomes in patients with an ST-elevation AMI undergoing an elective PCI after treatment with TNK-tPA were similar in the three treatment arms studied in ASSENT-3, and no excess of bleeding complications was observed.
Although fewer patients needed urgent PCI after co-therapy with ABC and ENOX, clinical outcomes were less favorable in this selected population, especially with ABC.
- acute myocardial infarction
- ASsessment of the Safety and Efficacy of New Thrombolytic Regimens
- intracranial hemorrhage
- percutaneous coronary intervention
- unfractionated heparin
- Received January 10, 2003.
- Revision received March 6, 2003.
- Accepted March 12, 2003.
- American College of Cardiology Foundation
- Smalling R.W.,
- Bode C.,
- Kalbfleisch J.,
- et al.
- Bode C.,
- Smalling R.W.,
- Berg G.,
- et al.
- Cannon C.P.,
- Gibson C.M.,
- McCabe C.H.,
- et al.
- den Heijer P.,
- Vermeer F.,
- Ambrosioni E.,
- et al.
- White H.,
- Van de Werf F.
- Armstrong P.W.,
- Collen D.
- Ross A.M.,
- Molhoek P.,
- Lundergan C.,
- et al.
- Collet J.P.,
- Montalescot G.,
- Lison L.,
- et al.
- Hudson M.P.,
- Granger C.B.,
- Topol E.J.,
- et al.
- Rosenbaum P.R.,
- Rubin D.B.
- Antman E.M.,
- McCabe C.H.,
- Gurfinkel E.P.,
- et al.