Cite this asAbudukeremu N, Yu ZX, Yang YN, Chan SP, Ma YT, et al. (2018) An Updated Systematic Review and Meta-analysis of the Short- and Long-term Outcomes of Percutaneous Coronary Intervention for Patients with Severe Left Ventricular Systolic Dysfunction. J Cardiovasc Med Cardiol 5(4): 073-080. DOI: 10.17352/2455-2976.000076
Background: Coronary artery disease (CAD) is the most common cause of left ventricular dysfunction (LVD). Conflicting evidence exists with regards to available treatments and patient prognosis. Revascularization may improve ventricular function while coronary artery bypass surgery (CABG) has significantly improved survival. The effectiveness of percutaneous coronary intervention (PCI) has also not been thoroughly investigated to date.
Objectives: To ascertain the in-hospital and long-term (≥1 year) outcomes of CAD patients with LV systolic dysfunction (ejection fraction ≤40%) after PCI according to a meta-analysis.
Methods: A systematic literature search and a series of random-effect meta-analyses were conducted to evaluate the short- and long-term outcomes of PCI of the selected studies. Single-center studies and those that did not report evidence on long-term mortality were excluded in the analysis. All statistical tests were performed with 95% confidence intervals. A p-value of less than 0.05 was considered statistically significant.
Results: A total of 25 studies involving 5,471 patients (78% males, average age 65.1 years) were identified. The average follow-up duration was approximately 27 months. The majority of patients had multi-vessel disease (68%), hypertension (66%), hypercholesterolemia (59%), and prior myocardial infarction (MI) (58%). The meta-analysis showed that the in-hospital occurrence of major adverse cardiac events (MACE), deaths, MI, and repeat revascularization (RR) after PCI were controlled at 4%, 2%, 2%, and 1%, respectively. The pooled estimates for long-term outcome were 40% MACE, 20% deaths, 4% MI, and 21% RR. There was no significant difference in mortality risk when PCI was compared with CABG (p=0.71).
Conclusion: PCI carries acceptable short- and long-term outcomes for CAD patients with LV systolic dysfunction.
Coronary artery disease (CAD) is the most common cause for left ventricular dysfunction (LVD). Patients with CAD and reduced LV systolic function present a challenge to current treatment modalities, although coronary artery bypass grafting (CABG) has traditionally been the preferred revascularization strategy for patients with reduced LV systolic function [1,2].
However, controversy persists in light of the published results that compare clinical outcomes after PCI or CABG for revascularization of coronary lesions in patients with LV systolic dysfunction [1,3,4]. In general, PCI for patients with decreased left ventricular ejection fraction may result in higher mortality rates in the short- and long-term [5-8], increased risk for nonfatal myocardial infarction (MI) , and a greater need for repeat revascularization (RR) . On the other hand, the high perioperative mortality rate after CABG among patients with CAD and LV systolic dysfunction [10,11], suggests that PCI may be a better option. It is also noteworthy that revascularization with PCI for patients suffering from left main stem disease has shown similar safety and efficacy outcomes when compared with CABG at 1 year . A number of studies also demonstrated equivalent mid- and long-term outcomes between CABG and PCI for patients with CAD and LV systolic dysfunction [1,3,13-15].
While there are several studies confirming the efficacy of PCI for CAD patients with severe LV dysfunction, it is timely and relevant to conduct a comprehensive systematic review and meta-analysis to clarify this issue.
To identify relevant published studies concerning CAD patients with LV systolic dysfunction (LVEF ≤40%) who had undergone PCI (e.g., drug eluting stents, bare metal stents), a search of full manuscripts and abstracts with Medical Subject Headings (MeSH) terms was conducted on the electronic database PubMed, with no language or other methodological restrictions (authors, study design, location, and sample size). The search period was between January 1990 and September 2014. The following terms were searched: “poor left ventricle function,” “percutaneous coronary intervention,” “revascularization,” “LV dysfunction” and “heart failure.”
Multi-center studies were eligible for inclusion if the long-term outcomes and all-cause mortality were reported for patients whose LVEF was ≤40. The full-text article of any identified study that appeared initially to meet the above-mentioned criteria was retrieved for closer examination by two reviewers. The final selection was based on consensus. In the event of a disagreement, a third reviewer was summoned to independently determine the article’s inclusion in this study.The search strategy is described below:
﹟1Search poor left ventricle function
﹟2Search left ventricular dysfunction
﹟3Search low left ventricular ejection fraction
﹟4Search heart failure
﹟5 ﹟1 OR﹟2 OR﹟3 OR﹟4
﹟6Search percutaneous coronary intervention
﹟8Search stent implantation
﹟9Search coronary angioplasty
﹟10﹟6 OR﹟7 OR﹟8 OR﹟9
The endpoints of concern included in-hospital and long-term (≥1 year) major adverse cardiac events (MACE), mortality, MI, and RR among CAD patients with LV systolic dysfunction (LVEF ≤40%) after PCI. MACE was a composite of all-cause death, MI, and RR. The baseline data, including age, gender, comorbid condition (i.e., hypertension, diabetes, hypercholesterolaemia, stable and unstable angina, prior myocardial infarction [Q wave, Non-Q wave], prior PCI, prior CABG, multi-vessel disease, smoking status, LVEF and follow-up duration) were tabulated by the reviewers. Each identified study was rated with a quality score based the Newcastle-Ottawa Scale .
The meta-analysis were performed with the Review Manager (RevMan) software (version 5.3, Cochrane collaboration, http\:ims.cochrane.org/revman/download) and Stata MP Version 14 (Stata Corp., College Station, Texas). Pooled values of the event rates (MACE, mortality, MI, and RR) and 95% confidence intervals were calculated using the random-effects model in anticipation of study heterogeneity. Nevertheless, heterogeneity among the outcomes of enrolled studies was evaluated with Q, based on the chi-squared test. In addition, I2 statistics (ranging from 0% to 100%) were generated to quantify the total variation, consistent with inter-study heterogeneity. A study was deemed to deviate from acceptable homogeneity if the I2 statistic exceeded 50% and when the p-value of the Q-test was below 0.05. Lastly, funnel plots were generated to assess publication bias.
A total of 717 records were identified in the search process and 41 were thoroughly screened. Thirteen studies were excluded and 28 were assessed for eligibility. The search eventually ended with 25 observational studies (5,471 patients with PCI) satisfying selection criteria. There was no disagreement between reviewers in the selection process, as summarized in figure 1. The features of the included studies are presented in table 1 and patient characteristics in table 2. The quality of the selected studies was remarkable, with 22 (88%) assigned a maximum score of 9. All 25 studies reported long-term mortality (Table 3), of which 16 (64%) provided data on in-hospital mortality (Table 4) and 6 (24%) compared long-term mortality between PCI and CABG.
The pooled estimates of PCI patients’ baseline characteristics are shown in table 2. With an average follow-up period of about 27 months, the mean age of the 5,471 patients was 65 years. The majority (78%) were male subjects and 42% were smokers. The mean EF was 30% prior to PCI and diabetes was reported in 34% of patients. While 66% had hypertension, 59% had hypercholesterolemia and 58% had a history of prior MI. PCI for multi-vessel disease was performed in 68% of patients. Lastly, 24% had undergone PCI and 27% had undergone CABG.
The individual and pooled event rates of the long-term and in-hospital outcomes based on the meta-analysis is summarized in tables 3 and 4 and figures 2 ,3. All but one meta-analysis showed that there was a substantial amount of study heterogeneity according to the Q and I2 statistics.
In terms of long-term MACE, 12 studies were gathered for meta-analysis and the pooled estimate was found to be 40% (95% CI 25%–55%; table 3 and figure 2A). The individual rates of MACE ranged from 8.9% to 81.8%. There was substantial publication bias as evidenced in the funnel plot (Figure 2A). Similarly, the pooled long-term mortality rate based on all 25 selected studies was 20% (95% CI 4%–36%; table 3 and figure 2B) and publication bias could be a major concern (Figure 2B). Long-term rate of MI, based on 14 studies was substantially lower at 4% (95% CI 2%–6%; table 3 and figure 2C). This is not surprising as the majority of the studies reported a mortality rate below 5%. The pooled estimate for the rate of RR based on 16 studies was 21% (95% CI 15%–27%; table 3 and figure 2D). Note that with the exception of MI, all meta-analyses generated wide 95% CIs owing to the large variations in MACE, mortality, and RR rates reported by the individual studies (Table 3).
As shown in table 4, the in-hospital pooled estimate for MACE, mortality, MI, and RR rates were substantially lower compared with the long-term rates. This was expected given the shorter time frames. The pooled estimate in-hospital MACE based on 10 studies was 4% (95% CI 2%–6%; table 4 and Figure 3A). Remarkably, there was substantially less publication bias (Figure 2A) when compared with that of the long-term meta-analysis of MACE described above. The pooled in-hospital mortality rate based on 16 selected studies turned out to be a mere 2% (95% CI 1%–3%; table 4 and figure 3B) and publication bias was not severe (Figure 3B). In-hospital MI rate based on 12 studies was also 2% (95% CI 1%–3%; table 4 and figure 3C). The pooled estimate for the rate of RR based on 10 studies was 1% (95% CI 0%–2%; table 4 and figure 3D), given that half the studies reported no occurrence of RR. The random-effects model was applied throughout for pooling in-hospital outcomes. The results were more credible than that of the long-term outcomes, as the 95% CIs were narrow and publication bias was not a major concern.
Six studies were gathered to compare the long-term mortality of CABG and PCI. As shown in table 5 and figure 4, the pooled odds ratio was 1.05 (95% CI 0.81–1.38), calculated using the fixed-effects model as there was no significant evidence against study homogeneity (Q5 2.55; I2 0%). As such, the results suggest that the long-term mortality rate was not significantly higher for patients who had undergone PCI compared with those who underwent CABG.
The presented meta-analyses based on 25 selected observational studies have provided evidential support for PCI as an acceptable treatment for patients with severe LVD. Patients who underwent PCI benefitted from favorable short-term outcomes (2% mortality) and acceptable long-term outcomes (20% mortality) among CAD patients with severe LV systolic dysfunction (EF ≤40%), while the perioperative mortality rate after CABG among patients with CAD and left ventricular systolic dysfunction was 4%–5% [36,37].
The results were consistent with those of the individual selected studies, which reported near 0% in-hospital mortality rates after PCI. For example, Biondi-Zoccai  reported similar results (in-hospital MACE 5%, in-hospital mortality 2%, long-term MACE 33%, long-term mortality 11%) for 1,284 patients with an EF ≤50%. In the Heart Failure Revascularization Trial , none of the 15 patients who underwent PCI died in the hospital but the long-term mortality rate was 27%. In the SYNTAX trial , the 3-year MI rate and RR for three-vessel disease in the PCI group were 7.1% and 19.1%, respectively. In general, the results were also in agreement with a recently-published random-effects meta-analysis involving 4,766 patients with an LVEF ≤40% . It reported a 1.8% in-hospital mortality rate and a 15.6% long-term mortality rate. However, our meta-analysis involved more patients who were followed up in a longer time frame, and with a wider range of high-risk patients characterized by a higher proportion of multi-vessel disease.
This study demonstrated no survival benefit for CABG compared with PCI in patients with severe LV systolic dysfunction (EF ≤40%). Several previous observational studies comparing CABG with PCI also reported similar results [1,3,13–15]. Resent large-scaled trials, which proceeded PCI with more advanced techniques and latest generation of stents,also showed similar survival rates in both groups [38,39]. However, after a long period of follow-up, Jeevan et al. found that CABG was associated with lower rates of RR and improved survival over PCI. In their trial,Lin Jiang and his team demonstrated that Compared with PCI, CABG has a lower risk of cardiac death, repeat revascularization, and MACCE . Another meta-analysis comparing PCI (with drug-eluting stent) with CABG reported a higher mortality rate for the PCI group .
The completeness of revascularization was found to be an important predictor of long-term survival and functional status after CABG . However, a randomized study demonstrated that incomplete revascularization by PCI in patients with multi-vessel disease and preserved LVEF did not affect in-hospital mortality and long-term survival, although freedom from angina and RR at 5 years were significantly lower than for patients randomized to CABG .
Another alternative treatment for patients with CAD and LV systolic dysfunction is the drug therapy alone. The HEART study performed in patients with myocardial viability and LVD demonstrated that a conservative strategy might not be inferior to an invasive strategy . Another study comparing three treatment strategies for patients with triple-vessel coronary disease and LV systolic dysfunction, revealed that both CABG and PCI were associated with a lower risk of mortality compared with medical therapy alone . A recent meta-analysis of observational data suggests that CABG may offer superior outcomes compared with PCI, with either modality being preferable to medical therapy alone .
Most of studies selected for this meta-analysis were conducted many years ago. Technical advances in the field of interventional cardiology with the introduction of new coronary stents and potent antiplatelet therapies have resulted in a marked increase in the complexity of procedures and expanded indications, even in patients considered to be high-risk such as those included in the present study. Newer-generation drug-eluting stents have been developed since the use of the pacilitaxel-eluting stent and have replaced the latter in current clinical practice owing to significant reduction in MI, stent thrombosis, and RR .
In actual practice, the reason for avoiding interventional treatments in diseased vessels might be explained by the occurrence of serious medical conditions (e.g., severe LV systolic dysfunction, cancer, old age, and unsuitable anatomic conditions such as chronic total occlusion or severe vessel complexity) . Therefore, a careful evaluation should include appropriate procedural considerations for optimal revascularization strategies.
The meta-analyses concerning the long-term outcomes presented in this study were affected by severe publication bias (see Figure 2 a–d). One possible explanation could be the huge variations in MACE, mortality, MI, and RR. To a large extent, this could be explained by the vast differences in follow-up periods. As such, the results for in-hospital outcomes were more reliable than those reported for the long-term analyses.
The present study remains subject to the inherent caveats of a meta-analysis including publication bias;however, in-depth statistical analysis was carried out to account for these limitations. Given the fact that the present meta-analysis included patients over a long study period and from different organizations around the world, patient-level data were not available and the results were obtained from published pooled data.During the time period of the studies included in the present meta-analysis, there have been significant developments in PCI in terms of improved pharmacotherapy and device technology.This may improve the overall outcomes of CAD patients with LV systolic dysfunction.
This study has several limitations. First, all the selected studies were observational by nature. As such, care must be taken in interpreting the strength of evidence generated from this meta-analysis. Second, the selected studies were carried out many years ago and not all studies used drug-eluting stents as the routine procedure. As a result, the reported outcomes might not accurately reflect current practice. With the significant improvement in PCI technology and pharmacotherapy, it is recommended that large-scale randomized studies be conducted. Third, it was impossible to perform multiple meta-regression analyses to adjust for confounding factors, as no individual-level data were available at the time of analysis. It is acknowledged that the considered studies differed considerably in their study populations, the usage of stents, pharmacotherapy, and study duration. Fourth, it was also impossible to compare the results of PCI with medical therapy, which could also be considered appropriate in the absence of anginal symptoms or viability, but may be more appropriate according to specific patient or anatomic features .
Our meta-analysis has shown that PCI has acceptable short- and long-term outcomes in CAD patients with LV systolic dysfunction. There was no significant difference in long-term mortality risk with PCI compared with CABG.
The authors have no conflicts of interest to declare.
Conceived and designed the experiments: NA. Performed the experiments: NA, SBF. Analyzed the data: NA, SBF. Contributed reagents, materials, or analysis tools: NA,SBF. Contributed to the writing of the manuscript: NA.
Supported by the Program for Innovative Research Team, University of Ministry of Education of China (grant No. IRT13094).
Subscribe to our articles alerts and stay tuned.