Although early percutaneous coronary intervention has been demonstrated to reduce the risk of mortality in patients with non-ST-segment elevation acute coronary syndromes (NSTE-ACS), there are emerging conflicting data as to whether the catheterization needs to be done very early or whether it could be delayed while the patient receives medical therapy. The aim of the current study was to perform a meta-analysis of randomized controlled trials (RCTs) comparing early vs. delayed invasive strategies for NSTE-ACS patients. Medline/CENTRAL and the Web were searched for RCTs comparing early vs. delayed invasive strategies for NSTE-ACS patients. The primary endpoint was all cause mortality, whereas myocardial infarction (MI), coronary revascularizations and 30-day major bleeding complications were secondary end points. Fixed or random effects models were used based on statistical heterogeneity. As a sensitivity analysis, Bayesian random effects meta-analysis was performed in addition to the classical random effects meta-analysis. A total of 5 RCTs were finally included, enrolling 4155 patients. As compared with a delayed strategy, an early invasive approach did not significantly reduce the rates of death [odds ratio (OR) 95% confidence interval (95% CI) = 0.81 (95% CI 0.60–1.09), P = 0.17], MI [OR = 1.18 (95% CI 0.68–2.05), P = 0.55] or revascularizations [OR = 0.97 (0.77–1.24), P = 0.82]. There was a not significant trend toward fewer major bleeding complications for the early invasive approach [OR (95% CI) = 0.76 (0.55–1.04), P = 0.08]. The present meta-analysis shows that for NSTE-ACS patients a routine early invasive strategy does not significantly improve survival nor reduce MI and revascularization rates as compared with a delayed approach.
The optimal management of patients presenting with non-ST-segment elevation acute coronary syndromes (NSTE-ACS) has been extensively investigated. Several data and meta-analyses of contemporary randomized controlled trials (RCTs) in patients with NSTE-ACS show long-term mortality and morbidity benefits with a routine invasive strategy compared with an initially conservative one.1–3
Although early percutaneous coronary intervention (PCI) has been shown to reduce mortality in patients with NSTE-ACS,4–6 there are conflicting data as to whether catheterization should be done very early or whether it can be delayed beyond initial medical therapy. The aim of this work, therefore, was to assess the optimal timing of catheterization by performing a meta-analysis of all available RCTs comparing early vs. delayed invasive strategies in patients with NSTE-ACS.
The present meta-analysis was performed according to established methods7 and in compliance with the PRISMA statement for reporting systematic reviews and meta-analyses in health care interventions.8
A PRISMA flow chart providing detailed description of publication screening and reasons for exclusion is shown in Figure 1.
Flow diagram of the reviewing process according to the PRISMA statement.
A systematic search for eligible studies involving MEDLINE, CENTRAL, Google Scholar databases and conference proceedings from the American College of Cardiology, American Heart Association, European Society of Cardiology, Transcatheter Cardiovascular Therapeutics and EuroPCR scientific sessions was conducted without language restriction by two independent investigators using a combination of the following keywords: ‘non-ST-elevation myocardial infarction’, ‘early-delayed invasive’, ‘coronary angioplasty’, ’randomized trial’.
Study end points
The primary endpoint was overall mortality. Secondary endpoints were the rates of new myocardial infarction (MI), 30-day major bleeding complications and coronary revascularizations after the primary procedure. For mortality, new MI and revascularizations, the longest available follow-up was chosen.
Citations were screened at title/abstract level and retrieved as full reports. Inclusion criteria were:
(1) RCTs; (2) studies comparing an early vs. a delayed invasive approach; (3) studies including patients with NSTE-ACS.
Prespecified data were extracted concerning: time of catheterization, age, female gender, diabetes, ST-segment changes, 600 mg clopidogrel loading dose, use of glycoprotein IIb/IIIa inhibitors, rate of stent use and longest available follow-up time.
Exclusion criteria were: (i) studies comparing an invasive strategy vs. conservative medical treatment; (ii) studies comparing a routine early vs. a selective delayed invasive strategy and (iii) observational non-randomized studies.
Data were abstracted on prespecified forms by two unblinded reviewers not involved in any of the studies retrieved; divergences were resolved by discussion with a third investigator. The quality of included studies was appraised by two unblinded reviewers.
Odds ratios (ORs) and 95% confidence intervals (95% CIs) were computed from individual study data according to an intention-to-treat principle. Between-study heterogeneity was evaluated with the two test-based Q statistics and considered statistically significant at a level <0.10.9 The pooled OR was calculated using a Fixed-Effects Model with the Mantel–Haenszel method. The DerSimonian and Laird random effects model was used in case of significant heterogeneity and/or moderate or significant inconsistency (>50%) across studies. Potential publication bias was examined by constructing a ‘funnel plot’, in which sample size was plotted against ORs.10,11 A mathematical estimate of the asymmetry of this plot was provided by a linear regression approach.12 Asymmetry was considered to be present if the intercept of the regression line deviated significantly from zero.
As a sensitivity analysis, in case of statistical heterogeneity, in addition to the classical random effects model to pool the data, we performed a Bayesian hierarchical random effects meta-analysis.13 The Bayesian model describes the structural relationship between data and unknown parameters, whereby both data and parameters are considered random variables with uncertainty. This approach generates a posterior distribution of the effect size and a 95% credibility interval, as opposed to CIs generated by traditional meta-analyses. Because we incorporated all relevant past studies, we expected the posterior distribution not to be influenced by the selection of the initial (prior) distribution. Therefore, low information prior distributions were used throughout and sensitivity analyses with different choices of low information prior distributions showed robustness to the analysis. We estimated the posterior of the effect size using 20 000 iterations after a 10 000 iteration initial burn-in.
As a sensitivity analysis, we addressed the influence of each study, testing whether deleting each in turn would change significantly the pooled results of the meta-analysis. We considered a study influential if its exclusion changed the effect estimate by at least 20%. Statistical analyses were conducted using Review Manager Version 5.1 (Cochrane Collaboration, Software Update, Oxford, UK) and WinBUGS software (version 1.4, MRC Biostatistics Unit, Cambridge, UK). P-values are two-tailed.
Characteristics of the included studies
Among a total of 8471 potentially relevant articles screened, a total of 17 trials were initially identified (Figure 1). Five trials14–18 were finally included in the meta-analysis, enrolling a total of 4155 patients. The main characteristics of the included studies are summarized in Table 1. The median time from randomization to catheterization was 10 h for the patients assigned to an early intervention and 50 h for those who underwent a delayed strategy. The patients from the early and delayed group were comparable in term of several clinical characteristics. In the early and delayed groups, female gender was present in 35.7% vs. 35.6%, diabetes in 25.2% vs. 27.2% and ST-segment deviations in 77% vs. 76.9 %, respectively. The use of glycoprotein IIb/IIIa inhibitors did not differ significantly between early (36%) and delayed intervention (35.2%), as well.
N(%) = number of enrolled patients with relative percentages.
NA = not available.
Primary study endpoint
Follow-up data were available at 6 months in the OPTIMA and TIMACS trials, whereas the remaining three trials reported only 30-day follow-up data.
A total of 180 patients out of 4155 died at follow-up (4.3%). As shown in Figure 2, the lower mortality rate observed with the early strategy as compared with the delayed approach did not reach statistical significance (3.9% in the early vs. 4.7% in the delayed group): OR (95% CI) = 0.81 (0.60–1.09), P = 0.17 (fixed effects model), P for heterogeneity = 0.41.
Funnel plot of the randomized studies included in the meta-analysis. Sample size of each study was plotted against the odds ratio for overall mortality. No skewed distribution was observed, suggesting no publication bias.
Sensitivity analysis deleting in turn each study did not show significant deviations from the original overall estimate.
Inspection of the funnel plot revealed no publication bias (Figure 3), and Egger’s test was not significant.
Individual and summary ORs for overall mortality in patients treated with early vs. delayed PCI.
Secondary study endpoints
New myocardial infarction
A total of 299 out of 4155 patients experienced a new MI (7.2%). The ABOARD, ELISA and ISAR-COOL trials reported only 30-day MI data. As shown in Figure 4, patients receiving early coronary angiography showed the same incidence of new MIs in comparison to those treated with delayed catheterization (7.2% vs. 7.2%): OR (95% CI) = 1.18 (0.68–2.05), P = 0.55 (random effects model), P for heterogeneity = 0.01.
Individual and summary ORs for new MIs in patients treated with early vs. delayed PCI.
In addition, Bayesian analysis performed to strengthen the results of the classical meta-analysis, showed a posterior distribution with final estimates comparable with the previous ones: OR (95% credibility interval): 1.16 (0.60–2.43).
Visual inspection of the funnel plot did not reveal asymmetry and Egger’s regression test was not statistically significant.
Major bleeding complications
Data on major bleeding complications at 30 days were reported in all the included studies. As shown in Figure 5, there was a not significant trend toward fewer major bleeding complications in the early as compared with the delayed group [3.4% vs. 4.5%: OR (95% CI) = 0.76 (0.55–1.04), P = 0.08 (Fixed Effects Model), P for heterogeneity = 0.78].
Individual and summary ORs for major bleedings in patients treated with early vs. delayed PCI.
Sensitivity analysis, performed by removing each of the studies one at time, did not demonstrate any single study to influence the overall results. No skewed distribution of the included studies was noted in the funnel plot.
Three studies reported data on repeated revascularizations that were performed in 292 out of 3525 patients (8.3%). As shown in Figure 6, the rate of new PCIs was not different among the two groups: 8.2% vs. 8.4%; OR (95% CI) = 0.97 (0.77–1.24), P = 0.82 (fixed effects model), P for heterogeneity = 0.50.
Individual and summary ORs for repeated revascularization in patients previously treated with early vs. delayed PCI.
The direction and magnitude of the effect were quite similar in the three studies after deleting each study in turn (sensitivity analysis).
The main finding of the present meta-analysis is that an early invasive strategy does not significantly improve clinical outcomes as compared with a late invasive strategy. Several RCTs and meta-analyses1–3 have shown that an invasive strategy is superior to an initially conservative strategy whereby selective angiography is mandated only by clinical or non-invasive evidence of recurrent ischemia. However, whether an early invasive approach can provide further benefits as compared with a delayed one is still debated. The former strategy would lead to a rapid diagnosis and, likely, to an earlier discharge; on the other hand, there might be a potential for early hazard because of intervention on unstable plaques with fresh thrombus.19–21 Conversely, a delayed strategy might offer benefits through plaque passivation by optimal medical treatment followed by intervention on more stable plaques.22–24 This potential advantage, however, might be counterbalanced by a higher risk of events while waiting for angiography.
The specific question of optimal timing was addressed for the first time in the ISAR-COOL trial,16 in which 410 patients with ischemic symptoms and ST-segment depression or positive troponin were randomized to a very early (median = 2.4 h) versus a delayed (median = 86 h) invasive strategy. All patients were aggressively treated with antithrombotic therapies, including aspirin, heparin, 600 mg loading dose of clopidogrel and upstream tirofiban. The primary end point of death or large, non-fatal MI at 30 days was reduced in patients undergoing an early invasive strategy (5.9% vs. 11.6%; RR 1.96; P = 0.04). The Angioplasty to Blunt the Rise of Troponin in Acute Coronary Syndromes Randomized for an Immediate or Delayed Intervention (ABOARD) trial15 evaluated the potential benefits of a very early (‘primary PCI like’) strategy in high-risk patients (TIMI risk score ≥3) randomly assigned to immediate (median 70 min) or delayed (median 21 h) catheterization. The primary outcome, peak troponin during hospitalization, was not found to differ between the two groups as well as the secondary outcome of death, MI or urgent revascularization. The findings of ABOARD are consistent with those of the Timing of Intervention in Acute Coronary Syndromes (TIMACS) trial,14 that investigated 3031 patients with unstable angina and NSTE-ACS. In TIMACS, the primary endpoint (death, new myocardial infarction or stroke) did not differ significantly between groups. The study did, however, demonstrate a significant reduction in death and MI at 6-month follow-up (14.1% vs. 21.6%, hazard ratio 0.65, P = 0.005) among the subgroup with high GRACE risk score undergoing early intervention.
The present meta-analysis shows no significant advantage from a routine early strategy in reducing overall mortality, new MI or coronary revascularizations as compared with a delayed approach. There was a suggestion toward a lower rate of deaths (3.9% vs. 4.7%, P for the overall OR = 0.17) and a stronger non-significant trend toward fewer major bleeding complications (3.4% vs. 4.5%, P for the overall OR = 0.08) in the early group. The patients’ characteristics were comparable between the two arms as well as the use of glycoprotein IIb/IIIa inhibitors. Bayesian computations applied as sensitivity analysis in case of statistical heterogeneity confirmed the findings of the classical random effects model for new MI, with comparable pooled estimates. Thus, the timing of the invasive approach in NSTE-ACS remains a challenging question. Currently, we do not have a gold standard ‘threshold-time’ to perform interventional procedures.
Based on our meta-analysis and RCTs available to date, an early invasive procedure in NSTE-ACS is not as effective as it is in the ST-elevation myocardial infarction population. A routine early invasive strategy in patients with NSTE-ACS therefore does not appear to be warranted.
Future large RCTs that include cost-effectiveness analyses, patient ischemic and bleeding risk profiles and the use of new antiplatelet agents are needed to identify the appropriate timing and the subgroups of patients who could derive the greatest benefit from an early invasive approach.
Significant statistical heterogeneity was found for the results on new MIs. In order to mitigate this, a random effects model was used to compute estimates. This model relies on the assumption that the studies are a random sample from a hypothetical population of studies and that the heterogeneity among studies can be represented by a single variance. The model tends to give a more conservative estimate. In addition, we performed, as a sensitivity analysis, a random Bayesian analysis which gave a posterior distribution and a related credibility interval comparable to the estimates obtained with the classical random effects model. Time of early and delayed catheterization was quite variable in the included studies and thus far no consensus exists on a univocal definition of early and delayed timing. A high loading dose of clopidogrel was used in a minority of trials. Second, there were limited data to perform subgroup analyses to identify whether any specific populations (e.g. those with high or low GRACE risk score) may have greater benefits or harms with either of the two invasive strategies.
Finally, the availability of individual patients’ data would have further improved the results of the current study.
This meta-analysis shows that among NSTE-ACS patients an early invasive strategy as compared with a delayed approach does not confer significant benefits in terms of mortality, new MIs and coronary revascularizations.
. Outcome at 1 year after an invasive compared with a non-invasive strategy in unstable coronary-artery disease: the FRISC II invasive randomised trial. FRISC II Investigators. Fast revascularisation during instability in coronary artery disease. Lancet 2000;356:9-16.
. Interventional versus conservative treatment for patients with unstable angina or non-ST-segment elevation myocardial infarction: the British Heart Foundation RITA 3 randomised trial. Randomized Intervention Trial of Unstable Angina. Lancet 2002;360:743-51.
FRagmin and Fast Revascularisation during instability in coronary artery disease Investigators. Invasive compared with non-invasive treatment in unstable coronary-artery disease: FRISC II prospective randomised multicentre study. Lancet 1999;354:708-15.
. A comparison of two invasive strategies in patients with non-ST elevation acute coronary syndromes: results of the Early or Late Intervention in unStable Angina (ELISA) pilot study. 2b/3a upstream therapy and acute coronary syndromes. Eur Heart J 2003;24:1401-5.
. Intracoronary thrombus and platelet glycoprotein IIb/IIIa receptor blockade with tirofiban in unstable angina or non-Q-wave myocardial infarction: angiographic results from the PRISM-PLUS trial. Circulation 1999;100:1609-15.
. Assessment of coronary angiograms prior to and after treatment with abciximab, and the outcome of angioplasty in refractory unstable angina patients: angiographic results from the CAPTURE trial. Eur Heart J 1999;20:1572-8.