Author + information
- Received November 24, 2010
- Revision received February 28, 2011
- Accepted March 1, 2011
- Published online August 9, 2011.
- Pieter J. Vlaar, MD, PhD⁎,⁎ (, )
- Karim D. Mahmoud, BS⁎,
- David R. Holmes Jr, MD, PhD†,
- Gert van Valkenhoef, MS‡,
- Hans L. Hillege, MD, PhD⁎,‡,
- Iwan C.C. van der Horst, MD, PhD⁎,
- Felix Zijlstra, MD, PhD§ and
- Bart J.G.L. de Smet, MD, PhD⁎
- ↵⁎Reprint requests and correspondence:
Dr. Pieter J. Vlaar, Department of Cardiology, Thorax Center, University Medical Center Groningen, Groningen, the Netherlands
Objectives The purposes of this study were to investigate whether, in patients with ST-segment elevation myocardial infarction (STEMI) and multivessel disease (MVD), percutaneous coronary intervention (PCI) should be confined to the culprit or also nonculprit vessels and, when performing PCI for nonculprit vessels, whether it should take place during primary PCI or staged procedures.
Background A significant percentage of STEMI patients have MVD. However, the best PCI strategy for nonculprit vessel lesions is unknown.
Methods Pairwise and network meta-analyses were performed on 3 PCI strategies for MVD in STEMI patients: 1) culprit vessel only PCI strategy (culprit PCI), defined as PCI confined to culprit vessel lesions only; 2) multivessel PCI strategy (MV-PCI), defined as PCI of culprit vessel as well as ≥1 nonculprit vessel lesions; and 3) staged PCI strategy (staged PCI), defined as PCI confined to culprit vessel, after which ≥1 nonculprit vessel lesions are treated during staged procedures. Prospective and retrospective studies were included when research subjects were patients with STEMI and MVD undergoing PCI. The primary endpoint was short-term mortality.
Results Four prospective and 14 retrospective studies involving 40,280 patients were included. Pairwise meta-analyses demonstrated that staged PCI was associated with lower short- and long-term mortality as compared with culprit PCI and MV-PCI and that MV-PCI was associated with highest mortality rates at both short- and long-term follow-up. In network analyses, staged PCI was also consistently associated with lower mortality.
Conclusions This meta-analysis supports current guidelines discouraging performance of multivessel primary PCI for STEMI. When significant nonculprit vessel lesions are suitable for PCI, they should only be treated during staged procedures.
The primary objective of percutaneous coronary intervention (PCI) in patients with ST-segment elevation myocardial infarction (STEMI) is to restore epicardial flow and myocardial perfusion in the culprit vessel. However, the pathophysiological process of myocardial infarction is not limited to the culprit vessel (1). It is estimated that 40% to 65% of the patients presenting with STEMI have multivessel disease (MVD), which has been associated with worse clinical outcome as compared with single-vessel disease (2). Patients with MVD have in addition to the culprit lesion, 1 or more significant lesions in nonculprit vessels. When nonculprit vessel lesions are suitable for PCI and coronary artery bypass graft surgery (CABG) is not preferred, they can be treated according to 3 different strategies. After having treated the culprit vessel, the operator can choose to treat nonculprit vessel lesions conservatively, directly by means of a multivessel PCI, or during staged PCI procedures.
Although current international guidelines do not recommend performance of PCI for nonculprit vessels in patients unless there is hemodynamic instability (Class III, Level of Evidence: C) (3,4), no large randomized controlled trials have been performed or planned comparing these 3 strategies. Therefore, it remains uncertain whether treatment of nonculprit vessels is required and when it should be performed in patients presenting with STEMI.
Recently several small prospective and large retrospective studies have been published comparing these strategies. To evaluate the composite data, we performed a systematic review of all published data to summarize current evidence for these 3 current PCI strategies for MVD in STEMI patients. Pairwise meta-analyses were performed to compare these PCI strategies, and an additional network analysis was carried out to investigate the robustness of the pairwise meta-analyses, to combine both direct and indirect evidence, and to rank these 3 PCI strategies.
Definitions of the 3 PCI strategies
The 3 PCI strategies for STEMI patients with MVD were defined as follows: The culprit vessel only PCI (culprit PCI) strategy was defined as PCI confined to culprit vessel lesions only. The multivessel PCI (MV-PCI) strategy was defined as PCI in which lesions in the culprit vessel as well as ≥1 nonculprit vessel lesion were treated. All interventions should have had taken place within the same procedure. The staged PCI strategy was defined as PCI confined to culprit vessel lesions only, after which ≥1 lesions in nonculprit vessel were treated during planned secondary procedures. The timing of staged PCI procedures was defined as reported in each study (Table 1) (5–22).
In studies investigating solely culprit PCI versus MV-PCI, the primary focus was often only on the strategy during the initial procedure and no details were given about whether planned staged procedures were allowed in patients treated according to the culprit PCI strategy. In these cases, studies were included but the applied definitions were extracted and used as a quality indicator. Authors were contacted in case of unclear definitions.
MEDLINE and Cochrane Controlled Trials Register searches were performed to identify relevant articles published between 1985 and August 2010. The following key words and medical subject headings (MeSH) terms were used: “percutaneous coronary intervention,” “angioplasty,” “stent,” “balloon,” “dilatat*,” “multivessel,” “multi-vessel,” “staged,” “culprit,” “infarct-related,” “myocard* infarct*,” “myocardial infarction[MeSH],” “angioplasty, transluminal, percutaneous coronary[MeSH],” “stents[MeSH],” and “balloon dilatation[MeSH].” Reference lists of selected articles were reviewed for other potentially relevant articles. Two independent reviewers (P.V. and K.M.) performed the study selection.
Both prospective and retrospective studies were considered for inclusion. Studies were selected if the study (sub)group consisted of STEMI patients with MVD who underwent acute PCI. At least survival data had to be available and stratified to at least 2 of the 3 PCI strategies for MVD. Studies investigating the impact of completeness of revascularization (so comparing complete versus incomplete revascularization) or surgical revascularization for MVD were excluded. In addition, studies investigating PCI in elective patients and acute coronary syndromes with MVD were also excluded. No studies were excluded based on baseline or angiographic criteria.
Data extraction was performed independently by 2 researchers (P.V. and K.M.). Information was collected with regard to study design, quality indicators, baseline clinical characteristics, procedural details, clinical outcomes, and safety outcomes. Authors were contacted in case of incomplete or unclear data.
The primary endpoint was short-term (in-hospital/30 days) mortality. Secondary endpoints were long-term mortality, reinfarction, any revascularization, major bleeding, and stroke. Unless otherwise specified, mortality included both cardiac and noncardiac death. Stroke included both ischemic and hemorrhagic stroke. Major bleeding was defined as the need for blood transfusion during hospitalization. Reinfarction as well as MVD and cardiogenic shock were defined as reported in the studies.
Absolute numbers and percentages of the endpoints were calculated for each study separately and all studies combined. For the direct pairwise meta-analyses, pooled estimates and 95% confidence intervals (CIs) were calculated assuming a random-effects model with inverse-variance weighting using the DerSimonian and Laird method to account for heterogeneity. The following pairs were analyzed: culprit PCI versus MV-PCI, culprit PCI versus staged PCI, and MV-PCI versus staged PCI. Heterogeneity across studies was tested by the Cochran's Q statistic and the I2 statistic. Funnel plots were used to assess potential publication bias. Subgroups were made based on design (prospective and retrospective studies) and shown for each comparison and endpoint. A subanalysis was performed on culprit PCI versus MV-PCI in cardiogenic shock patients. Pairwise analyses were performed using Review Manager (version 5.0.24).
A network analysis (23–25) was carried out to investigate the robustness of our findings and to combine both direct and indirect evidence about the 3 PCI strategies. The analysis was carried out using 3 types of random effects models: a consistency model and an inconsistency model (25), and a node-splitting model (26). Vague priors were specified in all of the models: N(0, 1,000) for effect parameters and U(0, 4) for variance parameters. As the evidence structure is a triangle, there is only 1 inconsistency factor w in the inconsistency model. In the node-splitting model, we split the node dm,s (MV-PCI vs. staged PCI) into direct evidence:
and indirect evidence:
Both hypothesis tests were performed using the Bayesian p value (26). All models were computed using Markov chain Monte Carlo simulation in JAGS (27) and R (28) using 3 chains with over-dispersed initial values. The models were run for 300,000 iterations, after which convergence was assessed using the Brooks-Gelman-Rubin diagnostic (29). After this, all inference was based on a further 100,000 iterations.
All p values were 2-tailed, with statistical significance set at <0.05. This meta-analysis was performed in compliance with published recommendations for meta-analyses (30). Two researchers (P.V. and G.V.) had full access to and take full responsibility for the integrity of the data.
Eighteen studies, involving 40,280 patients, met our inclusion criteria (Fig. 1) (5–22). All included STEMI patients with MVD who underwent PCI. Of the 18 included studies, 4 studies were prospective studies and 14 were retrospective studies (Table 1). Two retrospective studies compared PCI strategies between matched populations (13,19). Five studies compared all 3 PCI strategies (7,10,13,15,22), 10 studies compared culprit PCI versus MV-PCI (5,8,9,11,14,16,17,19–21), 2 studies compared culprit PCI versus staged PCI (12,18), and 1 study compared MV-PCI versus staged PCI (6).
In the majority of the studies MVD was defined as a significant stenosis in ≥1 major epicardial vessel or side branch, but in 2 studies a left main stenosis was also defined as 2-vessel disease (20,21). Significant was defined as ≥70% stenosis, except for 2 studies that used ≥50% (19,20).
The timing of staged procedures was in the majority of the studies during hospitalization or within 1 month after index PCI (Table 1). Details about the quality of included prospective and retrospective studies are given in Tables 2 and 3.⇓⇓ Six of 14 retrospective studies were subanalyses of prospective registries. In addition, details are given regarding the studies in which planned staged procedures were allowed in patients treated according to the culprit PCI strategy. Assessment of funnel plots suggested no publication bias.
Analyses were performed on short- and long-term mortality. Available evidence for the primary endpoint of short-term mortality in pairwise and network comparisons is shown in Figure 2. A proper analysis on the secondary endpoints of reinfarction, any revascularization, major bleeding, and stroke was not possible because data were only available for a minority of the studies. Several authors provided additional information and outcome data (8,9,11,13,14,20–21).
Rate of the 3 PCI strategies across retrospective studies
In retrospective cohort studies (total 37,436 patients) (9–11,22) providing rates of all 3 PCI strategies for MVD in their populations, culprit PCI was always the most often performed PCI strategy (30,260 of 37,436 patients, 80.8%) as compared with MV-PCI (3,887 of 37,436 patients, 10.4%) and staged PCI (3,289 of 37,436 patients, 8.8%).
Baseline differences between the 3 PCI strategies
A summary of baseline variables for each included study group are detailed in Table 4. In addition, summaries of baseline mean/percentages are given for studies included in each pairwise analysis. In studies comparing culprit PCI versus MV-PCI, patients treated according to culprit PCI were older (62.3 years vs. 60.7 years) and had higher rates of 3-vessel disease (32.5% vs. 28.8%). No differences were observed regarding sex (male 72.7% vs. female 73.0%), diabetes mellitus (23.0% vs. 22.4%), and cardiogenic shock (9.4% vs. 10.2%).
The pairwise analyses
Pooled short-term outcome data are detailed in Figures 3A to 3C. The staged PCI strategy was superior in both the comparison with culprit PCI (odds ratio [OR]: 3.03, 95% CI: 1.41 to 6.51, p = 0.005), and MV-PCI (OR: 5.31, 95% CI: 2.31 to 12.21, p < 0.0001). In addition, mortality was lower in patients treated according to the culprit PCI strategy as compared with MV-PCI (OR: 0.66, 95% CI: 0.48 to 0.89, p = 0.007). Only in the pairwise analysis of culprit PCI versus MV-PCI, signs of heterogeneity were found across the trials (I2 = 47%).
Two studies investigated culprit PCI versus MV-PCI in patients presenting in cardiogenic shock (20,26). A total of 3,248 patients was included, of which 470 (14.4%) were treated according to the MV-PCI strategy. Short-term mortality was in both studies lower among patients treated according to the culprit PCI strategy (total effect OR: 0.68, 95% CI: 0.56 to 0.84, p = 0.0003).
The network analysis
All models had adequate convergence. There was no significant inconsistency in either short-term mortality (p = 0.94 in the inconsistency model, p = 0.75 in the node-splitting model, and similar deviance information criteria for all 3 models) or long-term mortality (p = 0.90 in the inconsistency model, p = 0.78 in the node-splitting model, and similar deviance information criteria for all 3 models).
Posterior means and 95% credibility intervals for the relative effects (OR) are shown for each comparison and for both short-term and long-term mortality in Figures 3 and 4.
In addition, the rank-probability of the 3 PCI strategies was investigated. This analysis demonstrated that for the primary endpoint short-term mortality, the staged PCI strategy had a 0.9998 probability of being the best treatment as compared with culprit PCI (second rank probability of 0.94) and MV-PCI (third rank probability of 0.94). For long-term mortality, the rank-probability analysis also demonstrated that staged PCI had 0.995 probability of being the best treatment as compared with culprit PCI (second rank probability of 0.990) and MV-PCI (third rank probability of 0.996).
This meta-analysis supports current guidelines advising the performance of primary PCI for STEMI confined to the culprit vessel only. Multivessel PCI should be discouraged, and significant nonculprit vessel lesions should only be treated during planned staged procedures. Although considered safe, PCI remains associated with potential serious procedural complications, such as restenosis, stent thrombosis, and contrast-induced nephropathy (3,4). International guidelines, therefore, recommend using PCI selectively in cases in which the benefit of a revascularization outweighs the risk of complications. For elective PCI, extensive research has resulted in the consensus that it should be performed selectively in significant coronary lesions that cause myocardial ischemia (3,4). For lesions that do not induce ischemia, the benefit of revascularization is less clear. For these patients, an initial conservative medical strategy is likely to be as effective (3,4,31). However, in the context of STEMI, it is less clear how significant lesions in nonculprit vessels should be treated.
Current guidelines indicate that MV-PCI should not be performed in hemodynamic stable STEMI patients (3,4). Only for patients in cardiogenic shock, PCI maybe recommended for all critically stenosed large epicardial coronary arteries. However, no randomized data have been published indicating that MV-PCI is beneficial for cardiogenic shock patients. Because of limited evidence on this subject, different opinions exist on the use of MV-PCI for STEMI across centers and operators. That is illustrated by a recent analysis of the National Cardiovascular Data Registry database, which found incidences of MV-PCI ranging between 0% and 38% in some participating centers (9). One potential source of this variability may be the result of the operator considering that there were multiple infarct-related artery lesions/arteries.
It has been hypothesized that for selected STEMI patients (e.g., cardiogenic shock), PCI of the nonculprit vessel in the acute phase is able to reduce (border zone) ischemia and improve survival (32,33). In addition, when >1 culprit lesion is suspected, multivessel PCI may also be beneficial (1). Multivessel PCI may also be more convenient for the patient, as no secondary procedures are necessary. Further, there are logistic and economic reasons to perform MV-PCI as it may limit staged procedures and reduces length of hospital stays, and thereby medical costs. However, the present meta-analysis found that MV-PCI during the acute phase of STEMI was associated with higher mortality rates as compared with culprit PCI or staged PCI. A small subanalysis in cardiogenic shock patients also did not show any mortality benefit of a MV-PCI strategy in these patients. A recently published smaller pairwise meta-analysis on culprit PCI versus MV-PCI also found no benefit of MV-PCI over culprit PCI (34). However, it did not find MV-PCI to be associated with increased long-term mortality. These inconsistent long-term results are likely explained by methodological and statistical differences between the 2 meta-analyses. Nevertheless, these data indicate that the possible benefits of multivessel PCI do not outweigh the adverse effects associated with this aggressive strategy. These adverse effects are likely to be explained by the following factors.
First, the enhanced thrombotic and inflammatory environment of STEMI contributes to a higher risk of procedural complications as compared with elective procedures (35–39). Factors that increase risk are related to the complexity and duration of the procedure, which is the case with multivessel PCI for STEMI. Although a secondary staged PCI may also relate to an increased risk of complications, our results indicate that the risk associated with a secondary staged PCI is lower than that with an acute PCI.
Second, when performing multivessel PCI of significant nonculprit vessel, the PCI will be performed without objective evidence for the presence of myocardial ischemia. As the actual significance of a stenosis may be difficult to determine due to several factors in the acute phase of STEMI (39,40), routine multivessel PCI of nonculprit vessel lesions can result in PCI of clinical irrelevant lesions. The benefit of not treating nonculprit vessel lesions during the acute phase is that coronary angiograms can be discussed within a joint heart team to determine the best strategy for each individual patient (3,4). In the case of intermediate lesions, additional noninvasive ischemia tests and fractional flow reserve measurements can be performed before deciding to perform additional revascularizations (3,4,41).
Because of limited randomized data, this meta-analysis included both prospective and retrospective studies. As a consequence, the majority of the included studies were retrospective cohort analyses. The inclusion of studies with different designs and retrospective studies is likely to have induced heterogeneity in the results, as illustrated by the differences found between prospective and retrospective studies. Further, in most studies, the operator's intent to perform 1 of the 3 PCI strategies was not prospectively registered and may be influenced by important patient characteristics for which we were not able to adjust. The results and conclusions should be interpreted with these limitations in mind. However, we have carried out a network analyses to assess the robustness of our findings and combining direct and indirect evidence about the 3 strategies. In this analysis, staged PCI was also consistently associated with significant lower mortality rates at both short-term and long-term follow-up as compared with culprit PCI and MV-PCI. Only the comparison between culprit PCI and MV-PCI for short-term mortality lost significance. Additional analyses demonstrated that was not due to the indirect comparisons, but to the direct comparison in the network analysis being performed according to the Bayesian instead of the DerSimonian and Laird method.
Furthermore, the analyses between culprit PCI and MV-PCI in the included studies were primarily focused on the index PCI procedure. Therefore, in some studies, staged PCI was allowed in the culprit PCI group. That may have influenced the results, however, only in the culprit PCI versus MV-PCI and not the other pairwise comparisons. In addition, when excluding studies that allowed staged PCI procedures or did not provide information on this, a significant short-term mortality benefit of culprit PCI over MV-PCI was still observed. Finally, the role of timing of staged PCI procedures, CABG, and use of noninvasive ischemia testing in the management of MVD were neither investigated nor discussed in the majority of the studies.
This meta-analysis supports current guidelines discouraging performance of multivessel primary PCI for STEMI. When significant nonculprit vessel lesions are suitable for PCI, they should only be treated during staged procedures. More prospective research should be performed to investigate which strategy is superior, in both hemodynamic stable and unstable STEMI patients (CABG vs. culprit PCI vs. staged PCI vs. MV-PCI). We, therefore, propose a prospective international registry to investigate these strategies. To facilitate intention-to-treat comparisons between the different strategies for multivessel PCI, this registry should focus on registering the operator's intent to perform 1 of the strategies at the time of the initial PCI.
Ms. van Valkenhoef has received funding from the project Eschen (T6-202) of Top Institute Pharma (the Netherlands). All other authors have reported that they have no relationships to disclose.
- Abbreviations and Acronyms
- coronary artery bypass graft surgery
- multivessel disease
- odds ratio
- percutaneous coronary intervention
- ST-segment elevation myocardial infarction
- Received November 24, 2010.
- Revision received February 28, 2011.
- Accepted March 1, 2011.
- American College of Cardiology Foundation
- Sorajja P.,
- Gersh B.J.,
- Cox D.A.,
- et al.
- Kushner F.G.,
- Hand M.,
- Smith S.C. Jr..,
- et al.
- Wijns W.,
- Kolh P.,
- Danchin N.,
- et al.
- Politi L.,
- Sgura F.,
- Rossi R.,
- et al.
- Cavender M.A.,
- Milford-Beland S.,
- Roe M.T.,
- Peterson E.D.,
- Weintraub W.S.,
- Rao S.V.
- Dziewierz A.,
- Siudak Z.,
- Rakowski T.,
- Zasada W.,
- Dubiel J.S.,
- Dudek D.
- Han Y.L.,
- Wang B.,
- Wang X.Z.,
- et al.
- Hannan E.L.,
- Samadashvili Z.,
- Walford G.,
- et al.
- Kong J.A.,
- Chou E.T.,
- Minutello R.M.,
- Wong S.C.,
- Hong M.K.
- Poyen V.,
- Labrunie P.,
- Silvestri M.,
- Valeix B.
- van der Schaaf R.J.,
- Claessen B.E.,
- Vis M.M.,
- et al.
- Toma M.,
- Buller C.E.,
- Westerhout C.M.,
- et al.
- Caldwell D.M.,
- Ades A.E.,
- Higgins J.P.
- Plummer M.
- R Development Core Team
- Thiele H.,
- Allam B.,
- Chatellier G.,
- Schuler G.,
- Lafont A.
- Heusch G.,
- Kleinbongard P.,
- Böse D.,
- et al.
- Gonzalo N.,
- Barlis P.,
- Serruys P.W.,
- et al.
- Sianos G.,
- Papafaklis M.I.,
- Daemen J.,
- et al.
- Peterson E.D.,
- Dai D.,
- DeLong E.R.,
- et al.
- Hanratty C.G.,
- Koyama Y.,
- Rasmussen H.H.,
- Nelson G.I.,
- Hansen P.S.,
- Ward M.R.