Percutaneous Coronary Intervention Versus Medical Therapy in the Treatment of Stable Coronary Artery Disease: An Updated Meta-Analysis of Contemporary Randomized Controlled Trials

Background. The net clinical benefit of percutaneous coronary intervention (PCI) compared with medical therapy (MT) alone for the treatment of stable coronary artery disease (CAD) remains uncertain. We conducted an updated meta-analysis of randomized controlled trials (RCTs) to compare PCI with MT for the treatment of patients with stable CAD. Methods. RCTs of PCI vs MT in patients with stable CAD were identified from MEDLINE, the Cochrane Library, and manual search of bibliographies to March 2020. Study-specific risk ratios (RRs) with 95% confidence intervals (CIs) were pooled for the composite of all-cause mortality, myocardial infarction (MI), revascularizations, rehospitalizations, or stroke and its individual components. Results. Eleven unique RCTs comprising 9629 patients were included. PCI reduced the overall risk of the composite outcome of all-cause mortality, MI, revascularizations, rehospitalizations, or stroke (RR, 0.63; 95% CI, 0.46-0.87); unplanned revascularization (RR, 0.58; 95% CI, 0.44-0.77); and fatal MI (RR, 0.69; 95% CI, 0.52-0.92). There were no significant differences in overall risk of all-cause mortality and other cardiovascular events comparing PCI with MT. The composite outcome of all-cause mortality, MI, revascularizations, rehospitalizations, or stroke was reduced with PCI at 2-5 years. Conclusions. In patients with stable CAD, overall, short-term and intermediate-term risks of all-cause mortality are not significantly different between PCI and MT. However, PCI may reduce the overall and intermediate-term risk of the combined outcome of all-cause mortality, MI, revascularizations, rehospitalizations, or stroke.

J INVASIVE CARDIOL 2021;33(8):E647-E657.

Key words: coronary artery disease, medical therapy, percutaneous coronary intervention

Coronary artery disease (CAD) is the major manifestation of cardiovascular disease (CVD), which is recognized as the leading cause of death and disability globally. An estimated 17.7 million people died from CVD globally (representing 31% of global deaths) in 2015; of this number, 7.4 million were due to CAD.¹ Evidence suggests that patients with stable CAD do experience relatively high annual CVD events, such as CVD death, myocardial infarction (MI), and stroke.² Reduction of death or MI and symptom relief are considered the main treatment goals in the management of stable CAD according to revascularization guidelines.^3-5 However, the optimal management strategy and implementation of invasive intervention of stable CAD is still debatable. Given that myocardial ischemia in patients with stable CAD confers an increased risk of death or MI, this factor plays a pivotal role in the selection of patients for revascularization.⁶ The potential benefit of revascularization has been reported to depend on the extent and severity of ischemia.^7,8

It is well known that percutaneous coronary intervention (PCI) increases survival and reduces the incidence of MI in patients with acute coronary syndromes (ACS).^9-11 Although PCI has been used as a common treatment option for patients with stable CAD over the last decades,¹² its effectiveness in increasing survival is not yet well established in patients with stable CAD.^13,14 Medical therapy (MT), which includes lifestyle intervention and aggressive modification of risk factors using medications including 3-hydroxy-3-methylglutarylcoenzyme A reductase inhibitors (statins), has been shown to improve clinical outcomes in stable CAD.^13,15 Evidence on the clinical benefit of PCI compared with MT alone for the treatment of stable CAD is, however, divergent and uncertain. Although it appears the incidence of CVD events is greater in MT alone compared with PCI in the treatment of stable CAD; PCI, on the other hand, seems to confer a greater risk of periprocedural complications and additional interventions compared with MT.¹⁶ In contrast, MT has also been shown to reduce the incidence of short-term CVD events and the need for additional revascularization.¹³

To date, no study has demonstrated a mortality benefit of PCI over MT in patients with stable CAD. Single randomized controlled trials (RCTs) on stable and optimally treated CAD patients may have been typical of relatively low risk for fatal CVD events. Some earlier meta-analyses of RCTs comparing PCI with MT have been conducted in an effort to address the existing uncertainties. Results of these previous reviews have suggested only minor differences in the risk of adverse outcome events, such as death or MI, between the two therapeutic approaches.^17-19 A limitation of these previous reviews is that they included some trials that utilized interventions that were not reflective of current guideline practices (eg, active use of drug-eluting stents in PCI and modern MT). In a recent review, Stergiopoulos and colleagues pooled the results of 5 contemporary trials comparing PCI + MT vs MT alone and showed no significant differences in mortality, nonfatal MI, unplanned revascularization, or angina.²⁰

Given the existing controversy, the high level of clinical interest in the topic, and the publication of newer trials, there is a need to conduct a more detailed and updated synthesis of existing evidence with comprehensive assessment of outcomes. In this context, we conducted an updated meta-analysis of contemporary RCTs to evaluate the overall risk of all-cause mortality, MI, unplanned revascularizations, stroke, angina during follow-up, and other CVD events when PCI was compared with MT for the treatment of stable CAD. Subsidiary analyses evaluated outcomes by trial duration to investigate short- and long-term effects.

Data sources and search strategy. This review was conducted using a predefined protocol and in accordance with PRISMA guidelines (Appendix 1, Part 1; Part 2).¹² We searched MEDLINE and the Cochrane Library for published studies from November 2012 (date of search for the last relevant review) to March 21, 2020. Studies were limited to humans and no language restrictions were applied. The computer-based searches combined terms related to the interventions (eg, PCI and medical therapy), population (eg, CAD), and a filter for RCTs. Full details on the search strategy are provided in Appendix 2. The titles and abstracts of citations retrieved were initially screened to assess their suitability for inclusion, after which full texts of potentially relevant articles were acquired for detailed evaluation. To identify potential articles missed by search strategy, the reference lists of eligible studies and relevant review articles were scanned manually. Finally, the “cited reference search” function in Web of Science was used to check for additional eligible studies.

Data extraction and assessment of risk of bias. A predesigned data extraction form was used to extract information on study design, patient characteristics (eg, average age, sex), study location, numbers enrolled and randomized, allocation concealment, blinding, and outcomes at specific time points and their risk ratios. Primary outcomes evaluated were all-cause mortality and the composite of all-cause mortality, MI, revascularizations, rehospitalizations, or stroke. Secondary outcomes were the individual components of the composite primary outcome, angina during follow-up, and other CVD endpoints. Endpoint definitions were those used in the individual trials. If information was unavailable from a published report, we collected relevant data by extracting from previously published reviews or by contacting investigators of these published studies. Risk of bias was assessed using the Cochrane Collaboration’s Risk of Bias tool.¹³ This tool evaluates seven possible sources of bias: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other bias. For each individual domain, studies were classified into low, unclear, and high risk of bias.

Statistical analyses. Summary measures were presented as relative risks (RRs) with 95% confidence intervals (CIs). Where appropriate, hazard ratios and odds ratios approximated the same measure of RR based on the assumptions of rare outcomes and relatively short follow-up times.^21,22 For each outcome, risk estimates for the longest follow-up of each study were used for the primary analysis; subsidiary analyses used risk estimates for specific time points categorized as short-term (≤1 year), intermediate (>1 to 5 years), and long-term (>5 years). The inverse-variance weighted method was used to combine summary measures using random-effects models to minimize the effect of between-study heterogeneity.²⁵ Statistical heterogeneity across studies was quantified using the Cochrane Chi-squared statistic and the I² statistic.¹⁴ Study-level characteristics, including year of enrollment (before 2000 vs after 2000), type of population (stable CAD vs stable after recent MI), design characteristics (eg, allocation concealment, outcome assessment blinding), PCI type (fractional flow reserve [FFR] guided vs non-FFR guided) and average duration of follow-up, were prespecified as characteristics for assessment of heterogeneity, which was conducted using stratified analysis and random effects meta-regression.²³ STATA, version MP 16 (StataCorp LP) was used for all statistical analyses.

Study identification and selection. Our initial search of relevant databases and manual scanning of reference lists identified 298 potentially relevant citations. After screening based on titles and abstracts, 30 articles remained for further evaluation. Following detailed assessments, 16 articles were excluded for the following reasons: (1) reviews (n = 9); (2) population not relevant (n = 3); (3) comparator not relevant (n = 2); and (4) duplicates of an eligible study (n = 2). The remaining 14 articles based on 11 unique RCTs met our inclusion criteria and were included in the meta-analysis (Figure 1).^13,20,24-36 Some of the trials were based on multiple reports because outcomes were reported at different follow-ups.

Study characteristics and quality. Table 1 (Part 1; Part 2; Part 3) summarizes the key characteristics of the RCTs included in the review. In aggregate, the included trials, which were published between 2002 and 2018, comprised 9629 patients (4673 assigned to PCI and 4956 assigned to MT) with stable CAD. All RCTs were prospective, open-label RCTs. Six trials were single-country studies conducted in Brazil, Germany, Denmark, the United Kingdom, France, and Japan, and the other 5 recruited patients from multiple countries in Europe, Asia, and North and South America. The baseline average age of participants ranged from 57-64 years. Using the Cochrane Collaboration tool, all 11 trials demonstrated a high risk of bias for blinding of participants and personnel and a low risk of bias for random sequence allocation, incomplete outcome data, and selective reporting. Two trials had a high risk of bias for blinding of outcome assessment (Appendix 3).

Outcomes for overall follow-up. Figure 2 presents the pooled RRs for primary outcomes based on the longest follow-up of all included studies. In pooled analysis of 6 trials, PCI reduced the risk of the composite outcome of mortality, MI, revascularizations, rehospitalizations, or stroke compared with MT (0.63; 95% CI, 0.46-0.87). There was evidence of heterogeneity between the contributing trials (I²=79%; 53%-90%; P<.001). Comparing PCI with MT, there was no statistically significant difference in risk of all-cause mortality in 11 trials (RR, 0.93; 95% CI, 0.80-1.07) with no evidence of heterogeneity between contributing trials (I²=0%; 0%-60%; P=.94). Secondary outcomes are presented in Figure 3. PCI reduced the risk of unplanned revascularizations (11 trials) and fatal MI (2 trials) (RR, 0.58; 95% CI, 0.44-0.77 and RR, 0.69; 95% CI, 0.52-0.92, respectively). There was evidence of substantial heterogeneity between the contributing trials of unplanned revascularization (I²=79%; 53%-90%; P<.001), which seemed to be partly explained by year of enrollment, study design characteristics such as whether there was allocation concealment and blinding of outcome assessment, and whether PCI was FFR guided or not (Appendix 4).

Comparing PCI with MT, there were no statistically significant differences in the following: risk of non-fatal MI (9 trials; (RR, 0.95; 95% CI, 0.74-1.23); stroke (9 trials; RR, 0.98; 95% CI, 0.58-1.65); angina during follow-up (6 trials; RR, 0.83; 95% CI, 0.61-1.11); composite of death and non-fatal MI (2 trials; RR, 1.11; 95% CI, 0.95-1.29); CVD death (3 trials; RR, 0.94; 95% CI, 0.48-1.85); composite of CVD death or MI (2 trials; RR, 0.73; 95% CI, 0.53-1.01); fatal and non-fatal MI (4 trials; RR, 1.28; 95% CI, 0.89-1.85); heart failure (3 trials; RR, 0.91; 95% CI, 0.62-1.34); and CVD death (2 trials; RR, 1.08; 95% CI, 0.76-1.53).

Outcomes for specific time points. Figure 4 presents the pooled RRs for all outcomes at time points up to 1 year for PCI compared with MT. PCI reduced the risk of stroke (5 trials; RR, 0.43; 95% CI, 0.29-0.63). Comparing PCI with MT, there were no statistically significant differences in the risk of the composite outcome of mortality, MI, revascularizations, rehospitalizations, or stroke (3 trials); all-cause mortality (5 trials), non-fatal MI (4 trials), angina during follow-up (3 trials), and unplanned revascularization (5 trials). Results from single reports showed no significant differences in the risk of fatal MI, CVD death, MI, and heart failure (Figure 4).

At follow-up time 2-5 years, PCI reduced the risk of the composite outcome of mortality, MI, revascularizations, rehospitalizations, or stroke (4 trials; RR, 0.63; 95% CI, 0.46-0.86) and unplanned revascularization (8 trials; RR, 0.58; 95% CI, 0.43-0.7)  (Appendix 5 and Appendix 6). Comparing PCI with MT, there were no statistically significant differences in the risk of all-cause mortality (8 trials), non-fatal MI (7 trials), stroke (6 trials); fatal MI (2 trials), angina during follow-up (6 trials), composite of death and non-fatal MI (2 trials), CVD death (3 trials), composite of CVD death or MI (2 trials), fatal and non-fatal MI (3 trials), heart failure (2 trials), and CVD death (2 trials).

Only 1 trial reported outcomes at 10-year follow-up; except for a reduced risk of non-fatal MI for PCI, there were no significant differences in the risk of all other outcomes when PCI was compared with MT (Appendix 7).²⁴

Subgroup analyses and publication bias. PCI reduced the risk of unplanned revascularizations in trials with: (1) patients enrolled after year 2000 vs those enrolled before 2000 (meta-regression P=.01); (2) adequate allocation concealment vs those with unclear allocation concealment (meta-regression P=.01); and (3) adequate outcome blinding assessment vs those with no or unclear outcome blinding assessment (meta-regression P=.05). Furthermore, FFR-guided PCI substantially reduced the risk of unplanned revascularization vs non-FFR-guided PCI (meta-regression P<.001) (Appendix 4).

Under visual examination, funnel plots for those analyses that involved 10 or more studies were mostly symmetrical and Egger’s regression tests showed no statistical evidence of publication bias for all analyses (Appendix 8).

Whether PCI improves survival and clinical outcomes compared with MT in patients with stable CAD is controversial. Results from previous published trials have been divergent and hence previous attempts have been made to synthesize the existing evidence. In the first-ever meta-analysis on the topic, which considered trials including patients with stable CAD, Stergiopoulos and colleagues in pooled analysis of 5 trials showed that PCI in combination with MT provided no significant reduction in mortality, non-fatal MI, unplanned revascularization, or angina compared with MT alone.²⁰ In an individual participant data meta-analysis of 3 trials, Zimmerman and colleagues showed that FFR-guided PCI compared with MT reduced the risk of the composite outcome of CVD death or MI, which appeared to be driven by a decreased risk of MI.⁶ Although some trials have indicated potential benefits of FFR,^29,30 this may not be reflective of real-life FFR use between centers, countries, and invasive cardiologists; FFR-guided PCI is only used in a small percentage of stable CAD patients.

In the current study, which is the largest meta-analysis of contemporary trials (based on their longest available follow-up) conducted to date on the topic and evaluated several CVD endpoints following treatment of stable CAD patients with PCI or MT, PCI when compared with MT reduced the overall risk of the composite outcome of mortality, MI, revascularizations, rehospitalizations, or stroke; unplanned revascularization; and fatal MI. There was no significant reduction in all-cause mortality, non-fatal MI, stroke, angina during follow-up, CVD death, or the composite of CVD death or MI overall. An unexpected finding was that PCI reduced the risk of stroke at 1 year. This was likely due to the use of effective antithrombotic drug treatment, using a combination of aspirin with adenosine diphosphate receptor blockers, such as clopidogrel, prasugrel, or ticagrelor (dual-antiplatelet therapy) in the PCI arm following the procedure. However, we have no specific data to investigate whether PCI could also prevent larger-size MIs or atrial fibrillation, which could subsequently increase the risk of cerebrovascular events. Larger MI size may lead to left ventricular dysfunction, which may increase the risk of embolic stroke of cardiac origin. This finding deserves further investigation. Furthermore, PCI reduced the risk of unplanned revascularizations and the composite outcome of mortality, MI, revascularizations, rehospitalizations, or stroke at 2-5 years. In stratified analyses, there were suggestions of differences in the effect of PCI by year of enrollment, study design characteristics, such as allocation concealment and blinding of outcome assessment, and whether PCI was FFR guided or not, on the risk of unplanned revascularizations. However, given the multiple tests for interaction, these results should be interpreted with caution.

PCI-based treatment options usually effectively relieve angina symptoms and improve quality of life compared with MT alone. However, published single studies or previous meta-analyses have not been able to provide comprehensive data for application in current guidelines due to limited single trial size, short-term observation time until crossover, changes in treatments over the years, and insufficient power; therefore, this updated meta-analysis from RCTs on stable CAD management was necessary. Although the ischemia-guided approach to revascularization is recommended, there are challenges with the use of invasive interventions, including the inability to make an asymptomatic patient feel better or PCI to prevent MI in stable CAD patients with angina, and the risk of periprocedural complications, including death, stroke, MI, and hemorrhage at the access site.^37,38 Patients with more extensive and severe ischemia or left main disease experience increased long-term mortality or reduced event-free survival compared with patients without ischemia, necessitating the use of FFR-guided PCI in these patients. The association between ischemia and mortality is mediated by obstructive coronary stenosis justifying revascularization.^39-42 However, the systematic onsite use of FFR is still relatively low in all cases of stable coronary stenosis due to various reasons based on preangiographic tests, experience, and local systems.⁴³ Patients with more extensive ischemia have poorer prognoses compared with CAD patients without ischemia. Therefore, FFR has been recommended as an additional tool for optimizing the indication for invasive interventions in CAD by reducing potentially unbeneficial PCI.

Study strengths. Compared with previous reviews,^6,20 the current study has several advantages that deserve mention. It is an updated assessment and the largest meta-analysis on the topic to date. Our study has more clinical relevance compared with earlier meta-analyses on this topic. We reported on a comprehensive list of outcomes by the longest available follow-up and at specific time points. The generalizability of our findings was enhanced by the involvement of data from all RCTs published so far. We quantified the extent of heterogeneity and systematically explored for possible sources of heterogeneity using stratified and meta-regression analyses where possible. Formal tests to detect small-study bias were also conducted.

Study limitations. There are also some limitations that deserve consideration. The definition of outcomes, such as the composite outcome of mortality, MI, revascularizations, rehospitalizations, or stroke, was not uniform across all included RCTs. Although comprehensive, our review was based on pooled analysis of 11 available RCTs, which precluded the ability to perform detailed and clinically relevant subgroup analyses. Second, only 1 trial reported long-term follow-up data (10 years), which limited interpretation of the analyses based on short-, intermediate-, and long-term risk. Third, we could not account for the effects of crossover between study groups, as this analysis would require individual participant data. However, some of the studies incorporated assumptions about crossover between study groups and loss to follow-up in their power calculations.²⁷ Fourth, we could not include results of the recently published ISCHEMIA (International Study of Comparative Health Effectiveness with Medical and Invasive Approaches) trial,⁴⁴ because there were no published data on the comparison between PCI and optimal MT. We contacted the investigators, and the response was that this analysis hadn’t been conducted yet. The ISCHEMIA trial, however, showed that any mode of revascularization, including PCI or coronary artery bypass grafting compared with MT, did not reduce the risk of ischemic CVD events or death from any cause in patients with stable CAD.⁴⁴ Fifth, MT varied across trials, hence, a potential source of bias. An inherent limitation is that not all patients in included trials could be blinded because of CAD lesions, which may not be treated invasively based on the study protocol (some of them would have returned for “unplanned” active treatment). In patients with CAD without significant left main stenosis and/or multivessel disease or low ejection fraction, optimal MT is a key therapy for reducing symptoms, preventing atherosclerosis and atherothrombotic CVD events. The gold-standard MT includes lifestyle interventions and disease-modifying secondary prevention therapies, such as 3-hydroxy-3-methylglutarylcoenzyme A reductase inhibitors (statins), proprotein convertase subtilisin/kexin type 9, and renin-angiotensin system inhibitors; antithrombotic agents, such as aspirin; as well as symptom control agents (eg, calcium-channel blockers and nitrates), which are known to improve clinical outcomes and prognosis in stable CAD. Data on beneficial lifestyle changes, such as increased physical activity levels (which improves physical fitness) and health dietary patterns, known to be associated with reduced risk of CVD,^45-48 were not available. Structured cardiac rehabilitation programs should be available for all patients undergoing CAD treatment procedures.^49,50 Myocardial revascularization by PCI is needed in the management of angina pectoris on top of MT when symptoms cannot be treated properly by the drugs and lifestyle changes only. PCI has proven benefits in worsening CAD symptoms, although it does not lower mortality risk. We were not able to report the dosage of medications administered on the proportion of patients with stent usage. We did not include RCTs that did not use stents or those studies with drug-eluting balloon only in stable CAD. Finally, the duration of antithrombotic drugs may have varied between the trials; knowledge of the effect of patient characteristics with the new-generation drug-eluting stents has led to modifications of drug therapy after PCI, such as the duration of antithrombotic therapy. The focus of this study was to investigate PCI vs MT in stable CAD, and thus we also included studies with CAD patients who were successfully stabilized after acute MI with other angiographically non-culprit significant stenosis, and therefore they were considered as stable patients (Table 1).

Updated aggregate analysis indicates that in patients with stable CAD, overall, short-term, and intermediate-term risks of all-cause mortality are not significantly different between the main treatment lines of PCI and MT. However, PCI may reduce the overall and intermediate-term risk of the combined outcome of mortality, MI, revascularizations, rehospitalizations, and stroke.