05 Jul 2023
Percutaneous coronary intervention vs coronary artery bypass graft surgery for left main disease in patients with and without acute coronary syndromes: a pooled analysis of 4 randomized clinical trials
Selected in JAMA Cardiology by G. Occhipinti , D. Giacoppo
This patient-level analysis of randomized trials comparing PCI vs CABG for the treatment of low-to-intermediate CAD involving the LM confirms the absence of long-term differences in mortality between revascularization techniques, regardless of the clinical presentation.
References
Authors
Prakriti Gaba; Evald H. Christiansen; Per H. Nielsen; et al
Reference
JAMA Cardiol. Published online May 31, 2023. doi:10.1001/jamacardio.2023.1177
Published
31 May 2023
Link
Read the abstract
Reviewers
Our Comment
Why this study – the rationale/objective?
The optimal revascularization strategy for patients with low-to-complex coronary artery disease involving the left main remains a matter of debate.
Recent long-term data from randomized trials comparing percutaneous coronary intervention (PCI) and coronary artery bypass grafting (CABG) support the conclusion that there are no significant differences in mortality and stroke between revascularization techniques, but CABG is associated with lower incidences of spontaneous myocardial infarction and repeat revascularization compared with PCI.
Differences between PCI and CABG for the treatment of left main disease may be influenced by clinical presentation. In patients with acute coronary syndrome (ACS), the benefits of achieving a quicker and less invasive revascularization by PCI should be weighed against the higher rates of complete revascularization provided by CABG.
How was it executed? - the methodology
The present study is a pooled analysis of the four major randomized clinical trials (i.e., SYNTAX, PRECOMBAT, NOBLE, EXCEL trials) comparing PCI vs CABG for the treatment of low-to-complex coronary artery disease involving the left main. Patients were grouped according to their clinical presentation, ACS or stable coronary artery disease.
The inclusion and exclusion criteria related to ACS significantly differed across trials. SYNTAX allowed the inclusion only of patients with unstable angina, PRECOMBAT, EXCEL, and NOBLE allowed the inclusion of patients with unstable angina and myocardial infarction. In the PRECOMBAT trial, patients with myocardial infarctions could be included only 1 week after the onset, in the EXCEL trial patients with recent myocardial infarctions with elevated creatine kinase myocardial band (CK-MB) levels were excluded, and in the NOBLE trial patients with ST-segment elevation myocardial infarction with onset within the preceding 24 hours were excluded.
All the trials employed PCI with drug-eluting stents and CABG with almost complete use of a left internal mammary among grafts. The primary endpoint was all-cause death at 5 years. Secondary endpoints included cardiovascular death, noncardiovascular death, spontaneous myocardial infarction, procedural myocardial infarction, stroke, and coronary revascularization.
What is the main result?
Diagram showing the design of the study and its main results.
Courtesy of Giovanni Occhipinti and Daniele Giacoppo. Source: PCRonline
Among 4,394 patients randomly assigned to PCI or CABG, 1,466 (33 %) presented with ACS, and 2,927 (67 %) had stable coronary artery disease. Baseline clinical, angiographic, and procedural characteristics were balanced between treatments within ACS and stable coronary artery disease. Prescribed medications significantly differed between treatment groups in both ACS and stable coronary artery disease.
At 5 years, the rates of all-cause death were not significantly different between PCI and CABG in patients presenting with ACS (10.9 % vs 11.5 %; HR 0.93, 9 5% CI 0.68-1.27) and stable coronary artery disease (11.3 % vs 9.6 %; HR 1.19, 95 % CI 0.95-1.50), without a significant treatment-by-subgroup interaction (Pinteraction = 0.22).
In a sensitivity analysis based on the SYNTAX and PRECOMBAT trials, at 10 years, the results between PCI and CABG remained unchanged (ACS: HR 0.97; 95 % CI 0.74-1.27; stable coronary artery disease: HR 1.18; 95 % CI 0.96-1.44; Pinteraction = 0.27). A landmark analysis at 30 days including all the trials revealed consistent results between PCI and CABG (ACS: 1.7 % vs 2.2 %; HR 0.75, 95 % CI 0.37-1.59; stable coronary artery disease: 0.7 % vs 0.3 %; HR 2.18, 95 % CI 0.76-6.27; Pinteraction = 0.11).
Cardiovascular death, non-cardiovascular death, and stroke did not significantly differ between PCI and CABG in both ACS and stable coronary artery disease, without significant treatment-by-subgroup interaction.
At 5 years, the incidences of spontaneous myocardial infarction were approximately 2- and 3-fold higher in patients undergoing PCI compared with those undergoing CABG, in the settings of ACS (HR 1.74; 95 % CI 1.09-2.77) and stable coronary artery disease (HR 3.03; 95 % CI 1.94-4.72), respectively. The interaction did not reach the threshold of statistical significance (Pinteraction = 0.09). Rates of procedural myocardial infarction according to the pre-specified protocol definition in patients with ACS undergoing PCI were significantly lower than those undergoing CABG (3.1 % vs 5.7 %; HR 0.52, 95 % CI 0.30-0.90), while the difference was not significant in patients with stable coronary artery disease (3.2 % vs 4.0 %; HR 0.78, 95% CI 0.51-1.21). However, there was no significant treatment-by-subgroup interaction (Pinteraction = 0.26).
In contrast, the rates of procedural myocardial infarction between PCI and CABG according to the universal definition in the SYNTAX and EXCEL trials were not significantly different in the setting of ACS (3.6 % vs 3.6 %; HR 1.00, 95 % CI 0.50-1.98), while PCI was associated with a significant excess of events compared with CABG in the setting of stable coronary artery disease (2.9 % vs 1.5 %; HR 2.02, 95 % CI 1.00-4.07). However, even in this case, there was no significant interaction (Pinteraction = 0.17).
At 5-year, the rates of repeat revascularization were approximately 1.5- to 2.0-fold higher after PCI compared with CABG both in patients with and without ACS (ACS: HR 1.57, 95 % CI 1.19-2.09; stable coronary artery disease: HR 1.90, 95 % CI 1.54-2.33; Pinteraction = 0.31).
The analysis comparing ACS vs stable coronary artery disease, regardless of the revascularization technique employed, showed substantial differences in baseline clinical, angiographic, and procedural characteristics between settings. Results across different time points indicated an increased risk of cardiovascular death at 30 days and 1 year associated with ACS. Spontaneous myocardial infarction was significantly more frequent at 30 days, 1 year, and 5 years in patients admitted for ACS.
Critical reading and the relevance for clinical practice
This patient-level analysis of randomized trials comparing PCI vs CABG for the treatment of low-to-intermediate coronary artery disease involving the left main confirms the absence of long-term differences in mortality between revascularization techniques, regardless of the clinical presentation. Results remained consistent at different time points and there was no inconsistency in the components of cardiovascular and non-cardiovascular death.
This study also shows that in the ACS setting, PCI was associated with an excess of spontaneous myocardial infarction compared with CABG. However, the interaction testing was not significant and, although this may be a result of insufficient statistical power, formally it is not possible to declare an association between clinical presentation and treatment effect in terms of spontaneous myocardial infarction. Any conclusion about this endpoint should refer to the overall population analysis published elsewhere, in which a significant excess of events was associated with PCI compared with CABG, regardless of the clinical presentation.
Interestingly, the risk of procedural myocardial infarction was lower in patients admitted for ACS assigned to PCI if the protocol definition was applied, while the risk of procedural myocardial infarction was higher in patients admitted for stable coronary artery disease assigned to PCI if the universal definition was applied. Although these differences likely reflect the influence of myocardial infarction definitions, any speculation needs to be avoided since no significant treatment-by-subgroup interaction was observed.
Long-term differences in cardiovascular mortality and spontaneous myocardial infarction between ACS and stable coronary artery disease in patients with low-to-moderate complexity coronary artery disease involving the left main, regardless of the revascularization technique employed, are consistent with previous results in other coronary artery disease patterns. However, in the context of unadjusted analyses is difficult to define the influence of possible confounders and the inspection of possible associations may delineate whether specific mechanisms leading to worse outcomes have played a role.
Some relevant considerations when interpreting the results of this study are required. Firstly, the authors of the study do not seem to have employed meta-analytic methods and individual patient data were merged to produce an individual, larger randomized trial. However, by disregarding the correlation between individual trials and treatment effect the results do not account for between-trial heterogeneity (i.e., meta-analysis). In addition, in consideration of some relevant differences in clinical and procedural characteristics across trials, a sensitivity analysis with multivariable-adjusted outcomes might have improved the strength of the study.
Secondly, higher-risk, hemodynamically unstable ACS and recent myocardial infarctions were generally excluded from the four randomized trials. Moreover, differences in eligibility criteria across trials may have further enhanced the impact on summary pooled analysis.
Thirdly, differences in the definitions used to adjudicate the events as well as the spectrum of data collected across trials may have significantly influenced the results of the present study.
Finally, the SYNTAX and PRECOMBAT trials were conducted several years before the EXCEL and NOBLE trials. Variations in devices (e.g., drug-eluting stent generation), revascularization techniques (e.g., intravascular imaging guidance during PCI, off-pump CABG, number of arterial grafts), and medical therapy (e.g., antithrombotic medications) may not reflect the contemporary era.
In conclusion, based on the findings of the present study, PCI and CABG are associated with similar mortality in patients with unstable angina and low-risk, stabilised myocardial infarction. Differences in spontaneous myocardial infarction and repeat revascularization between PCI and CABG does not seem to be significantly influenced by the clinical presentation.
2 comments
What about QoL? Evidently, with less reinterventions, its better with CABG.
Congratulations to the study Authors and to the comments by Giovanni and Daniele! Unfortunately, we are still dealing with old-fashioned studies, the last patient in these trials being enrolled 10 years ago. The need for a modern study for comparison with CABG, including mandatory IV imaging (pre and post), avoidance of complex 2-stenting procedures and implementation of DCB in the protocol is eagerly needed! So far, PCI is a little bit back to CABG for LM management in terms of scientific data.