Culprit Vessel Only vs Immediate Complete Revascularization in Patients With Acute ST‐Segment Elevation Myocardial Infarction: Systematic Review and Meta‐Analysis

Abstract

Although multivessel coronary artery disease has been associated with poor health outcomes in patients with acute ST‐segment elevation myocardial infarction (STEMI), the optimal approach to revascularization remains uncertain. The objective of this review was to determine the benefits and harms of culprit vessel only vs immediate complete percutaneous coronary intervention (PCI) in patients with acute STEMI. We searched MEDLINE, EMBASE, the Cochrane Register of Controlled Trials, and the Cumulative Index to Nursing and Allied Health Literature (CINAHL) for randomized controlled trials (RCTs). Teams of 2 reviewers, independently and in duplicate, screened titles and abstracts, completed full‐text reviews, and abstracted data. We calculated pooled risk ratios (RRs) and associated 95% confidence intervals (CIs) using random‐effect models for nonfatal myocardial infarction (MI), revascularization, cardiovascular mortality, all‐cause mortality, and adverse events, and used the GRADE approach to rate confidence in estimates of effect. Of 341 patients randomized to complete revascularization and followed to study conclusion, 31 experienced revascularization, as did 80 of 324 randomized to culprit vessel only revascularization (RR: 0.35, 95% CI: 0.24‐0.53). Ten patients in the complete revascularization group and 28 patients in the culprit vessel only revascularization group experienced nonfatal MI (RR: 0.35, 95% CI: 0.17‐0.72). All‐cause mortality and cardiac deaths did not differ between groups (RR: 0.69, 95% CI: 0.40‐1.21 for all‐cause mortality; RR: 0.48, 95% CI: 0.22‐1.04 for cardiac deaths). Pooled data from 3 RCTs suggest that immediate complete revascularization probably reduces revascularization in patients with acute STEMI; although results suggest possible benefits on MI and death, confidence in estimates is low.

Introduction

ST‐segment elevation myocardial infarction (STEMI) may occur in up to 25% to 40% of patients with acute coronary syndromes, though the incidence of STEMI has declined after 1999.1, 2 Although randomized controlled trials (RCTs) have demonstrated that primary percutaneous coronary intervention (PCI) of the occluded artery responsible for the infarct (the culprit artery) decreases mortality and recurrent myocardial infarction (MI) in comparison with thrombolysis,3, 4, 5 which itself is superior to no thrombolysis,6, 7, 8, 9 other aspects of the management of STEMI remain uncertain.10, 11, 12, 13 In particular, how to address disease in nonculprit vessels remains a matter of controversy.

The prevalence of such multivessel coronary artery disease (CAD) ranges between 40% and 65% in patients with acute MI.13, 14, 15, 16 Although the extent and severity of atherosclerosis in coronary arteries have been correlated with health outcomes—for instance, multivessel vs single‐vessel disease is associated with a higher risk for urgent coronary artery bypass grafting (CABG)—there exists limited evidence to support optimal management in the acute phase.10, 14, 15, 16

A number of studies have addressed preventive PCI in nonculprit arteries in the context of STEMI. In some studies, complete revascularization in STEMI patients has been associated with lower cardiac events,17, 18, 19, 20 whereas other reports failed to show benefit.15, 21, 22, 23 The most recent and comprehensive systematic review suggested that complete revascularization in acute STEMI patients was associated with longer survival; however, primary PCI for complete revascularization during the index catheterization (immediate PCI) was associated with higher in‐hospital mortality rates.14 Further, immediate revascularization may be associated with an increased risk for periprocedural MI and is associated with increased radiation exposure and contrast.10, 24 A staged procedure can be employed but is associated with the risks and resource use associated with a second procedure.

The most recent systematic review did not include a large RCT conducted in the latter part of 2013.14 We therefore conducted an updated systematic review and meta‐analysis of RCTs comparing health outcomes of immediate complete revascularization vs culprit vessel only revascularization in patients with acute STEMI.

Methods

Eligibility Criteria

We included studies meeting the following criteria: (1) patients with acute STEMI and multivessel CAD; (2) patients randomized to PCI for culprit and nonculprit coronary arteries vs culprit‐only PCI; (3) report of ≥1 of the following outcomes: nonfatal MI, revascularization (PCI or CABG), cardiovascular mortality, all‐cause mortality; and (4) minimum follow‐up of 1 year. We excluded studies in which ≥20% of patients suffered from cardiogenic shock, significant stenosis in a prior coronary bypass graft, or significant (>50%) left‐main coronary artery stenosis in which the outcomes of patients without these conditions were not reported separately. We also excluded studies in which PCI in nonculprit arteries occurred days or weeks after the initial PCI (staged PCI).

Data Sources and Search Strategy

We searched MEDLINE, EMBASE, the Cochrane Register of Controlled Trials, and the Cumulative Index to Nursing and Allied Health Literature (CINAHL) from 1946 up to March 2014 without any restrictions. For every eligible study we identified, and for studies such as review articles that included citations to potentially eligible studies, 1 reviewer examined the reference list (Figure 1). (See Supporting Information, Appendix, in the online version of this article for the search strategy for the MEDLINE database).

Figure 1.

PRISMA 2009 Flow Diagram.

Study Selection

Teams of 2 investigators (NS, LCL) independently screened each title and abstract. If either reviewer identified a citation as potentially relevant, we obtained the full‐text article. Three reviewers working in pairs (NS, LCL, FAS) independently determined the eligibility of all studies that underwent full‐text evaluation. Disagreements were resolved through discussion between the reviewers.

Data Abstraction

Using a custom‐made data collection form, 2 reviewers (NS, FAS) abstracted the following information from each study: study characteristics, study population, follow‐up time (overall and per cohort, including minimum, maximum, and mean follow‐up), loss to follow‐up, intervention, and outcomes.

Risk of Bias and Confidence in Effect Assessment

A modified version of the Cochrane risk for bias tool (http://distillercer.com/resources/) was employed by 2 independent reviewers (NS, FAS). The assessment included the following components: adequacy of sequence generation, allocation sequence concealment, level of blinding, incomplete outcome data, selective outcome reporting, incomplete reporting for loss to follow‐up, and stopping early for benefit (Figure 2).25

Figure 2.

Risk of Bias Summary.

We used the Grades of Recommendation, Assessment, Development and Evaluation (GRADE) methodology to rate confidence in estimates of effect for each outcome as high, moderate, low, or very low.26 We used detailed GRADE guidance to assess overall risk of bias,27 imprecision,28 inconsistency,8 indirectness,29 and publication bias30 and summarized results in an evidence profile.

For decisions regarding eligibility, risk of bias assessment, and data abstraction, reviewers resolved disagreement through discussion. A simple (unweighted) κ value for full‐text selection showed substantial agreement (κ = 0.82).

Data Synthesis and Statistical Analysis

We report descriptive statistics as proportions for categorical variables and mean or medians for continuous variables. Our primary analysis for nonfatal MI, revascularization, cardiovascular mortality, and all‐cause mortality included only patients with complete follow‐up.

We conducted 1 sensitivity analysis, a worst‐case scenario in which we assumed that all patients with missing data in the complete‐revascularization arm suffered the adverse outcome and all patients lost to follow‐up in the culprit vessel only revascularization arm did not.

We calculated pooled risk ratios (RRs) and associated 95% confidence intervals (CIs) for each outcome using random‐effects models applying the Mantel‐Haenszel method. Absolute effects and 95% CI were calculated by multiplying pooled RRs and 95% CI by the pooled control rate of outcomes. Statistical heterogeneity was assessed by the χ2 test and I² statistic. When event rates were very low (nephropathy, stroke, and bleeding), we calculated risk difference (RD) and 95% CI directly by pooling across trials.31 We planned to assess publication bias by visual inspection of funnel plots, but too few studies were available. Analyses were performed using RevMan version 5.2 (Cochrane Collaboration, Nordic Cochrane Centre, Copenhagen, Denmark).

Results

Trial Identification

Our search yielded 2837 abstracts, of which 83 were duplicates. Sixteen proved potentially eligible and underwent full review. Of these studies, 13 were excluded—5 had an ineligible population, 3 did not have nonculprit artery revascularization as a comparator, 1 was a prospective cohort study, 3 had follow‐up <1 year, and 1 was an editorial letter (Figure 1). We included 3 randomized trials enrolling 683 patients.

Trial and Total Population Characteristics

Table 1 presents the characteristics of the 3 eligible studies. In 1 of the 3 studies, PCI was performed using heparin‐coated stents; in the other 2, clinicians used a mix of drug‐eluting and bare‐metal stents or angioplasty alone. All studies excluded patients with cardiogenic shock. All patients enrolled had a STEMI with <12 hours of symptoms.

Table 1.

Characteristics of the Eligible Studies

First author	Di Mario et al17		Politi et al18		Wald et al19
Study type	RCT		RCT		RCT
Number randomized in complete PCI	52		65		234
Number randomized in culprit‐only PCI	17		84		231
Funding	NR		NR		Private, not for profit
Inclusion criteria	Acute STEMI with multivessel coronary artery disease		Acute STEMI with multivessel CAD		Acute STEMI with multivessel CAD
Exclusion criteria	Significant lesions in vein grafts, previous stent implantation in the same segments, cardiogenic shock, recent thrombolysis, and left‐main CAD		Cardiogenic shock, left‐main CAD, previous CABG, and severe valvular heart disease		Cardiogenic shock, left‐main CAD, and previous CABG
Description of multivessel disease	1 to 3 lesions in nonculprit arteries		>70% stenosis in ≥2 nonculprit arteries		≥50% stenosis in ≥1 nonculprit arteries
Maximum follow‐up	1 year		4 years		3 years
	Complete PCI	Culprit‐Only PCI	Complete PCI	Culprit‐Only PCI	Complete PCI	Culprit‐Only PCI
Age, mean ± SD	64 ± 12	65 ± 7	64.5 ± 12	66 ± 13	62 (33–92)a	62 (33–90)a
Female sex, %	12	15	23	24	24	19
DM, %	12	41	13	23	15	21
Hypertension, %	37	59	49	59	40	40
3‐Vessel disease, %	31	47	25	29	39	33
DES, n	52b	17b	5	10	147c	135c

Abbreviations: CABG, coronary artery bypass grafting; CAD, coronary artery disease; DES, drug‐eluting stent; DM, diabetes mellitus; MI, myocardial infarction; N, total number of participants in both study arms; n, number of cases; NR, not reported; PCI, percutaneous coronary intervention; RCT, randomized controlled trial; SD, standard deviation; STEMI, ST‐segment elevation myocardial infarction.

^aRange.

^bHeparin‐coated stents.

^cNumber of drug‐eluting stents in the culprit artery.

One study explicitly reported loss to follow‐up and reported 18 participants lost. Loss to follow‐up was slightly higher in the complete‐revascularization arm (n = 10, 4.3%) than the culprit vessel only revascularization arm (n = 8, 3.5%).18 One study failed to report follow‐up completeness,17 and the other reported that follow‐up was complete.19 The 3 studies enrolled similar patients in terms of age, with some differences in medical history (Table 1).

Assessment of Risk of Bias

Overall risk of bias was deemed moderate for all 3 studies; in all 3 due to lack of blinding (including, in 2 studies, failure to blind outcome assessors); in 2 due to questionable allocation concealment; and in 1 due to stopping early for apparent benefit (Figure 2). We further dealt with loss to follow‐up as a risk‐of‐bias issue in our sensitivity analysis.

Outcomes Assessment

Nonfatal Myocardial Infarction

Among 341 patients randomized to complete revascularization, 10 suffered a nonfatal MI, as did 28 of 324 patients randomized to culprit vessel only revascularization (RR: 0.35, 95% CI: 0.17‐0.72, heterogeneity: P = 1.00, I² = 0%; Figure 3). Using our best estimate of baseline risk of nonfatal MI from the largest randomized trial18 (20 nonfatal MIs in 223 patients treated with culprit vessel only revascularization), complete revascularization may be associated with 58 fewer cases per 1000 treated (95% CI: 75 fewer to 25 fewer; Table 2). Our worst‐case analysis changed the results substantially, with loss of statistical significance of the effect (RR: 0.72, 95% CI: 0.41‐1.27, heterogeneity: P = 0.53, I² = 0%). We therefore rated down confidence in estimates for both risk of bias and imprecision (small number of events) to low confidence (Table 2).

Figure 3.

Pooled risk of non‐fatal myocardial infarction with complete PCI versus culprit only PCI.

Table 2.

GRADE Assessment of Confidence in Estimates of Effect

Outcome	No. of Participants (Studies)	Risk of Bias	Consistency	Directness	Precision	Publication Bias	Quality	Relative Effect (95% CI)	Risk Difference
Nonfatal MI	665 (3)	Serious limitationsa	No serious limitations	No serious limitations	Serious limitationsb	Not detected	Low	RR 0.35 (0.17‐0.72)	5.8% fewer (7.5% fewer to 2.5% fewer)
Revascularization	665 (3)	No serious limitations	No serious limitations	No serious limitations	Serious limitationsb	Not detected	Moderate	RR 0.35 (0.24‐0.53)	13.4% fewer (15.6% fewer to 9.7% fewer)
Cardiac death	665 (3)	Serious limitationsa	No serious limitations	No serious limitations	Serious limitationsb	Not detected	Low	RR 0.49 (0.23‐1.04)	2.3% fewer (3.5% fewer to 0.2% more)
All‐cause mortality	665 (3)	Serious limitationsa	No serious limitations	No serious limitations	Serious limitationsb	Not detected	Low	RR 0.69 (0.40‐1.21)	2.2% fewer (4.3% fewer to 1.5% more)
Nephropathy	596 (2)	Serious limitationsa	No serious limitations	No serious limitations	No serious limitations	Not detected	Moderate	RD −0.01 (−0.03 to 0.01)	1% fewer (3% fewer to 1% more)
Stroke	447 (2)	Serious limitationsa	No serious limitations	No serious limitations	No serious limitations	Not detected	Moderate	RD 0.01 (−0.01 to 0.02)	1% more (1% fewer to 2% more)
Bleeding requiring transfusion or surgery	447 (2)	Serious limitationsa	No serious limitations	No serious limitations	Serious limitationsb	Not detected	Low	RD 0.00 (−0.03 to 0.04)	0% fewer (3% fewer to 4% more)

Abbreviations: CI, confidence interval; GRADE, Grades of Recommendation, Assessment, Development and Evaluation; MI, myocardial infarction; RD, risk difference; RR, risk ratio.

^aSerious risk of bias due to the lack of blinding of participants and personnel in all trials and stopping early for benefit in largest study. See text for other potential sources of bias in individual studies.

^bSerious limitations due to uncertainty about magnitude of effect, small sample size, and number of events.

Revascularization

Among 341 patients randomized to complete revascularization, 31 underwent subsequent revascularization, as did 80 cases among 324 patients randomized to complete revascularization (RR: 0.35, 95% CI: 0.24‐0.53, heterogeneity: P = 0.63, I² = 0%). Using our best estimate of baseline rates of follow‐up revascularization from the available randomized trials of 46 revascularization in 223 patients treated with culprit vessel only revascularization, complete revascularization may be associated with 134 fewer events per 1000 treated (95% CI: 156 fewer to 97 fewer, moderate confidence in estimates; Table 2). Our worst‐case analysis changed results little (RR: 0.47, 95% CI: 0.32‐0.70, heterogeneity: P = 0.32, I² = 9%). We therefore rated down confidence in estimates only for imprecision (small number of events) to moderate confidence (Table 2).

Cardiovascular Mortality

Complete case data from the 3 studies showed 9 vs 20 deaths due to cardiac causes in the complete revascularization and culprit vessel only revascularization arms, respectively (RR: 0.48, 95% CI: 0.22‐1.04, heterogeneity: P = 0.84, I² = 0%). Cardiovascular mortality may be associated with 23 fewer cardiac deaths in 1000 patients treated with multivessel revascularization (95% CI: 35 fewer to 2 more, low confidence because of risk of bias, uncertainty associated with imprecision and inconsistency; Table 2). Our worst‐case analysis changed results substantially (RR: 0.99, 95% CI: 0.53‐1.87, heterogeneity: P = 0.37, I² = 0%), and we therefore rated down confidence in estimates for both risk of bias and imprecision to low confidence (Table 2).

All‐Cause Mortality

There were 19 deaths in the complete revascularization vs 29 deaths in the culprit vessel only revascularization arm of the 3 studies (RR: 0.69, 95% CI: 0.40‐1.21; heterogeneity: P = 0.9, I² = 0%). Complete PCI may have no effect on all‐cause mortality (95% CI: 43 fewer to 15 more, low confidence because of risk of bias and imprecision; Table 2).

Adverse Events

Two studies documented adverse events, which included contrast‐induced nephropathy,18, 19 stroke,18 and serious bleeding.18 Pooling data from these 2 studies, nephropathy occurred in 2 vs 6 in the complete‐revascularization vs culprit vessel only revascularization arms (RD: −0.01, 95% CI: −0.03 to 0.01, heterogeneity: P = 0.64, I² = 0%, moderate confidence because of risk of bias; Table 2). Only 1 study18 reported stroke and serious bleeding (RD: 0.01, 95% CI: −0.01 to 0.02, moderate confidence because of risk of bias for stroke, risk difference 0.00, 95% CI: −0.03 to 0.04, low confidence because of risk of bias and imprecision for serious bleeding; Table 2).

Discussion

This systematic review suggests a possible benefit of immediate complete revascularization with low confidence in reductions in nonfatal MI (RR: 0.35, 95% CI: 0.17‐0.72) and moderate confidence in revascularization (RR: 0.35, 95% CI: 0.24‐0.53). We also found a possible reduction in cardiovascular mortality (RR: 0.48, 95% CI: 0.22‐1.04, low confidence) with no suggestion of important adverse effects. In particular, we found a lower incidence of contrast‐induced nephropathy in the complete‐revascularization arms, though there were very few events. If our estimates of effect on outcomes of MI and revascularization were accurate, the impact of immediate complete revascularization would represent an important benefit (point estimates of absolute effects: 4 fewer nonfatal MIs per 100 patients, 13.4 fewer revascularizations per 100 patients; Table 2).

Strengths of this review include explicit eligibility criteria with a restriction to RCTs; a comprehensive search; rigorous and duplicate assessment of eligibility, risk of bias, and data abstraction; conduct of a sensitivity analysis exploring the impact of loss to follow‐up; and use of the GRADE approach to rate confidence in effect estimates.

Limitations are primarily those of the available evidence. Only 3 studies have addressed the question, 2 of which were very small. The total sample size and number of events are modest. We therefore rated down confidence in estimates of effect for imprecision even when the conservative boundary of the confidence interval suggested a substantial benefit (Table 2). The 2 small studies have limitations of questionable concealment and failure to blind outcome assessors. The larger study likely overestimated effects by stopping early for apparent benefit. The results of the studies were, however, strikingly consistent for all major outcomes.

Prior systematic reviews of this question are limited by a focus on observational studies, a failure to systematically assess confidence in effect estimates, and the fact that they were conducted prior to the report of the recent, and by far the largest, RCT.11, 32, 33 Their results were, therefore, very different, suggesting harm rather than benefit for immediate culprit and nonculprit PCI. Prior reviews have also addressed the issue of delayed rather than immediate nonculprit artery revascularization and suggested benefit from delayed PCI. Inferences from these results are very limited, however, because evidence comes from observational studies rather than RCTs.

Guidelines published before the large recent study understandably recommend against immediate primary PCI in nonculprit arteries in hemodynamically stable patients:29, 34 Available evidence prior to that publication suggested, if anything, harm from nonculprit PCI. A guideline published since the recent publication continues to recommend against immediate nonculprit PCI.35 Reasons for this persistent recommendation include indirect evidence suggesting the RCT results are implausible.15, 21, 23 This indirect evidence includes biological rationale:24 It is difficult to assess stenosis in the context of STEMI, and increasing the duration of PCI could increase contrast‐induced nephropathy,24 stent thrombosis, and impaired recovery of injured myocardium.

Moreover, the results of RCTs of PCI vs medical management have shown no benefits in mortality or MI in stable coronary artery disease.36, 37 The limitations of the available trials, which we have described in detail, also fuel reluctance to recommend immediate nonculprit PCI.

Partial or complete occlusion of a nonculprit artery has, however, been associated with mortality in a larger cohort of STEMI patients in the 3‐year follow‐up period.16 In a large‐scale study, intravascular ultrasound was performed in nonculprit arteries after STEMI,38 and it was shown that the atheromatous plaques may become unstable over time.38 As a result, lesions in a nonculprit artery may be correlated with adverse health outcomes in those with STEMI, providing a biological rationale for immediate PCI of nonculprit lesions.

Most compelling, of course, are the pooled data suggesting a possible decrease in recurrent MI and a likely reduction in revascularization. The latter is particularly persuasive: In the presence of angina or other symptoms after MI, patients are less likely to undergo repeat catheterization if they have undergone complete revascularization, and recurrence of restenosis after a successful procedure requires time for newly developed lesions. Results from RCTs provide no support for concerns about at least 1 complication of nonculprit PCI, contrast‐induced nephropathy, which occurred less frequently in patients undergoing PCI of nonculprit lesions.

Conclusion

Multivessel CAD is frequent and has been associated with lower survival rates in STEMI patients. Recurrent MI and repeat revascularization are both relatively frequent, and reductions in risk would therefore represent substantial patient benefit. The large apparent benefits in these outcomes seen with immediate nonculprit PCI in the available RCTs are most likely overestimates. Nevertheless, dismissing the findings would be inappropriate. Definitely establishing the role of immediate nonculprit PCI will require larger, well‐designed RCTs. In the interval, clinicians should seriously consider immediate revascularization of nonculprit arteries in patients suffering STEMI.

Supporting information

Appendix S1. SUPPORTING APPENDIX

Figure S1.

FigureS2.