Left main coronary artery disease (LMD), defined as >50% diameter stenosis, is found during 8–10% of diagnostic cardiac catheterisation procedures.1 LMD is associated with the highest mortality among patients with stable coronary artery disease (CAD) and preserved left ventricular systolic function.2
Analyses of randomised controlled clinical trials (RCTs) in stable CAD patients (n=2,234) from the late 1970s and early 1980s reported lower mortality with a strategy of routine coronary artery bypass graft (CABG) surgery versus initial medical therapy.2 However, these analyses were based on a small, post hoc subset of only 150 patients with stable LMD, comprising only 6.7% of the combined study population (Table 1).2
Nonetheless, over the past 40+ years, clinical practice guidelines have assigned revascularisation with CABG a class I indication for all patients with LMD with ≥50% obstruction, regardless of left ventricular function, symptom severity and the magnitude/extent of inducible ischaemia (Figure 1).3
For several decades, patients with LMD have been excluded from RCTs comparing routine revascularisation to medical therapy with selective revascularisation. In particular, there are no RCT data comparing a strategy of initial revascularisation plus optimal medical therapy (OMT) to OMT alone, so we lack contemporary evidence from carefully designed, prospective RCTs comparing an invasive versus a non-invasive approach to the management of this LMD subset.
While there are numerous reports from registries and small, observational studies suggesting lower mortality in patients undergoing revascularisation versus OMT alone, their results are inherently confounded by unequal distribution of comorbidities, despite attempted adjustments.
In clinical practice, revascularisation may be withheld in patients with extensive vascular disease, advanced kidney failure, malignancy, frailty, cerebrovascular disease, cognitive impairment and other conditions, creating a considerable bias against OMT alone for outcome analyses.
Given their inherent bias, observational studies are insufficient to inform clinical practice. Recommendations for altering our approach in patients with LMD requires evidence from an adequately powered RCT.
Ever since the early RCTs of CABG versus rudimentary medical therapy in the 1970s and the subsequent evolution of percutaneous coronary intervention (PCI) in the 1980s, the focus has shifted to the question of whether PCI may provide results equivalent to CABG in patients with LMD. With improvements in stent design, several RCTs have suggested similar outcomes with PCI compared to CABG after 5 years, with 10-year follow-up data confirming these results.4–6 However, none of these trials included an arm for OMT only.
Evolution of Medical Therapy
During the same time period, medical therapy has seen an ever-expanding armamentarium of therapeutic interventions that, in addition to anti-ischaemic approaches (e.g. nitrates, β-blockers, calcium antagonists and ranolazine), now include a robust portfolio of agents, including antiplatelet/anti-thrombotic, lipid-lowering and intensive anti-hypertensive drugs, as well as glucose-modulating drugs (sodium–glucose cotransporter 2 inhibitors and glucagon-like-peptide-1 agonists) with beneficial systemic effects and potentially anti-inflammatory therapy.7
The interventions in this portfolio reduce mortality and major adverse cardiovascular events (MACE) and, particularly in the case of lipid-lowering drugs, slow or even reverse the progression of atherosclerosis.8 In conjunction with agents aimed at lowering myocardial oxygen consumption, they comprise a powerful arsenal directed at halting atherosclerosis disease progression, mitigating its adverse impact and better controlling symptoms.7
In trials using contemporary OMT, the overall mortality benefit of revascularisation, previously observed with CABG in the early studies, has not been associated with revascularisation (most commonly with PCI) in more recent RCTs of stable CAD patients, e.g. COURAGE, FAME-2 and ISCHEMIA.7
Even in the setting of multi-vessel CAD and type 2 diabetes (BARI-2D), a subset universally regarded as high risk, no overall mortality benefit was evident with routine revascularisation when used with the best available medical therapy provided at the time these trials were conducted.9
A meta-analysis suggested lower cardiac mortality with a strategy of routine revascularisation without a difference in overall death rates, but this research included many studies from decades ago, before multi-faceted OMT became available.10,11
Nevertheless, ISCHEMIA-EXTEND confirmed lower cardiovascular mortality with routine revascularisation, albeit with a greater risk of non-cardiovascular mortality, resulting in a net zero benefit in terms of mortality.12
In essence, the benefit of routine revascularisation over OMT alone in patients with stable CAD has become so small that it does not outweigh its inherent procedural risks.
Challenge of Using Biomarker Criteria for MI in Interpreting Results of Comparative Effectiveness RCTs
Given the absence of mortality benefit with routine revascularisation, the focus has shifted increasingly to other study endpoints and to discussion of their value.
MI is an important study endpoint but its definition and associated impact have become controversial.
Both CABG and PCI are associated with risk of periprocedural MI (types 5 and 4A, respectively), which is now commonly distinguished from a spontaneous MI event (type 1).
The prognostic significance of periprocedural infarcts depends on their size, typically measured by the extent of the associated troponin or creatine-kinase release. The Universal Definition of Myocardial Infarction (UDMI) requires additional evidence of myocardial ischaemia, while some studies merely use the biomarker release for defining MI.13
Furthermore, the Society of Coronary Angiography and Intervention uses much more restrictive criteria for periprocedural MIs than the UDMI, often with troponin values in excess of 70–100 × the upper limit of normal (ULN) for the 99th percentile, thereby complicating planning and interpretation of RCTs.14
As such, the criteria used may substantially affect a composite study endpoint consisting typically of death, MI or stroke, with MI being the predominant contributor.
For example, in the original FAME study, periprocedural MIs were defined by an elevation of 3 × ULN of creatine kinase-MB, which led to a large number of MIs in the early part of the trial, favouring the experimental arm guided by fractional flow reserve (FFR), which entailed fewer procedures.15 On the other hand, the subsequent FAME-2 study defined periprocedural MIs as 10 × ULN, leading to significantly fewer MIs overall and, again, favouring the FFR-guided PCI arm.16
Recent analyses support the notion that revascularisation is associated with a greater risk of periprocedural MI but a lower risk of spontaneous MI compared with an initial strategy of OMT.17,18 This benefit, however, is confined to revascularisation using CABG, not PCI.19
The mechanism underlying a reduction of MI risk with revascularisation has not been fully elucidated. The cardioprotective superiority of CABG is postulated to result from bypass grafts to the mid-coronary vessels that not only treat obstructive lesions (even anatomically complex ones) but also afford prophylaxis against new, proximal disease. On the other hand, stents treat only suitable stenotic segments with no benefit against native coronary disease progression.20
Antiplatelet Therapy and MI after PCI
The benefit of covering advanced atheroma with a stent is offset by the implantation of a highly thrombogenic material that requires intensified antiplatelet therapy.
The need for dual antiplatelet therapy (DAPT) introduces a critical bias in favour of PCI for the endpoint of MI. DAPT has been shown to result in a 47% relative risk reduction for MI versus aspirin alone when treatment is extended beyond 12 months.21
Thus, the small effect of lower spontaneous MI risk after PCI observed in RCTs may be at least partly, if not fully, explained by the larger proportions of patients taking DAPT versus those assigned to medical therapy without DAPT.12
Similarly, stroke is another endpoint also likely to be impacted by the use of DAPT, again favouring PCI. Stroke risk is lower after PCI versus CABG.22 Despite the effect of DAPT, however, meta-analyses have revealed a strong trend of higher stroke rates after PCI versus OMT.18
While the absolute stroke risk is rather small, its clinical impact is substantial given the potential for disabling neurologic impairment, and such risk should be discussed openly and transparently with patients before PCI.
While the benefit of DAPT in reducing vascular events is unequivocal, allowing for equal allocation with OMT alone may be challenging given the increased bleeding risk. The use of DAPT is supported in patients with a history of MI and study trial designs may incorporate appropriate adjustments for the effect of DAPT.
Owing to the significant advances in OMT and implementation of effective lifestyle changes, adverse event rates in patients with stable CAD have decreased substantially over past decades. The lower MACE rates compared to earlier studies have led to the need for much larger sample sizes for sufficiently powered clinical trials.
To circumvent this problem, contemporary study designs have included additional components in their composite endpoints, which are commonly described as ‘soft’, given their less critical impact and often somewhat ambiguous definitions.
Among such soft endpoints, downstream revascularisation is particularly problematic because it is expected that some patients (perhaps 20–30%) will crossover from OMT to revascularisation during follow-up owing to intractable symptoms or acute events.
Therefore, downstream, symptom-driven revascularisation may be considered not an adverse event but an anticipated component of a strategy of starting with a focus on OMT alone rather than routinely proceeding with revascularisation; revascularisation might be regarded more accurately as a clinical decision rather than as an endpoint.
Furthermore, treatment assignment in most clinical studies is unblinded to both patients and providers, which may inherently result in a strong bias against OMT, as seen in FAME-2.
If included in a composite endpoint, downstream revascularisation should, at a minimum, require objective criteria for its indication, including ECG and imaging evidence for new or worsening myocardial ischaemia in conjunction with documentation of adequate OMT.
The ISCHEMIA study demonstrated conclusively that reduction of inducible myocardial ischaemia using revascularisation is not associated with a lower risk of MACE, disproving a long-held belief in cardiovascular medicine.23,24 ISCHEMIA also did not identify a patient subgroup that derived a mortality benefit from routine revascularisation; however, patients with LMD were excluded from randomisation.
Implications for Practice and Research
Thus, the question of whether LMD still requires routine revascularisation in all patients with stable angina, as universally recommended by clinical practice guidelines, remains unanswered.19
In most instances, LMD is associated with multi-vessel CAD and it is less commonly isolated.11 Mean survival in patients with LMD was 79.8 months with medical therapy (versus 99.6 months with CABG) in clinical trials in the 1970s and 1980s.2
Given the advancements in OMT in the past several decades, it seems likely that the difference of 19.3 months after a 10-year follow-up will be substantially smaller with contemporary OMT.20–22
Revascularisation is associated with risks often not considered in RCTs, e.g. bleeding, kidney failure, hospitalisation, procedural complications, vascular injury and cognitive impairment. In the setting of near or full equipoise for MACE, such major secondary endpoints should be considered in the decision to offer patients revascularisation and thus should be included in RCTs.
Apart from MACE, symptom control and quality of life indices are important considerations in the management of patients with CAD. Both COURAGE and ISCHEMIA revealed a small benefit of symptom control and quality of life metrics with routine revascularisation versus initial OMT alone; this benefit was most marked in those with severe or frequent symptoms.25,26
In ISCHEMIA, the symptom burden was remarkably small despite the highly selected study population with mostly severe, inducible ischaemia. Furthermore, treatment assignment was not blinded and a placebo effect is likely, as suggested by the negative results from ORBITA, which may well account for the small difference seen in ISCHEMIA.27
In times of significant resource restrictions, costs should also be considered. Revascularisation is expensive, accounts for billions of dollars each year in the US alone, and has not been shown to be cost-effective in recent trials.28–30
In contrast, risk factor control for both primary and secondary prevention has been shown to be cost-effective.31 Nonetheless, risk factor management and medical therapy are less well incentivised and, even in RCTs, only 20% or fewer patients receive full OMT.32
The above suggests that there is need for a definitive, well-powered prospective RCT to test the hypothesis that in patients with stable LMD, routine revascularisation is truly superior to an initial strategy of OMT.
Ideally, such an RCT would use a composite endpoint of death, MI or stroke. All-cause death is preferable over cardiovascular death in this context since the risk of non-cardiac related events (e.g. infection, bleeding and adverse outcomes owing to comorbidities) should be considered during a decision to revascularise routinely.
For practical considerations, it may be necessary to add further endpoint components, such as worsening or refractory angina as well as ischaemia-driven revascularisation, with objective findings including ischaemic ECG changes and/or regional wall motion abnormalities on non-invasive testing, which should be well defined and established.
The bias of unequal DAPT allocation described should be addressed, e.g. by statistical adjustments or by altering study designs, including equal prescription of DAPT.
Secondary endpoints should include the risk of bleeding, kidney failure, procedural complications, all-cause hospitalisations, cognitive decline, symptom control, quality of life indices, healthcare expenditures and differential assessments of resource use and costs.
Exclusion criteria should include left ventricular systolic dysfunction (ejection fraction <50%), and class IV angina.
Diagnostic Considerations
Given the advances of non-invasive coronary angiography, CT angiography may emerge as the key modality for identifying patients with LMD, though clinical practice may still require confirmation by cardiac catheterisation in many cases.
A dedicated RCT on LMD may also lend itself to facilitate a transition to a routine, non-invasive diagnostic strategy, which may reduce the impetus to immediately proceed with revascularisation and lead instead to stronger consensus-building in the decision-making process.33
CT coronary angiography now provides detailed atherosclerotic plaque analysis, information on functional significance of stenosis and metrics on vascular inflammation.34 These additional features may allow more individualised approaches to patients with LMD, which should be tested prospectively.
While significant LMD has been largely and consistently defined in clinical trials as ≥50% diameter stenosis, management implications may differ for 50–69% versus 70–90% stenoses. Significant blood flow restriction is unlikely with stenoses <70% (by visual inspection). However, LMD with the 50% threshold has been associated with greater mortality than three-vessel CAD in the same cohorts, which may suggest that the atherosclerotic burden rather than haemodynamic significance is the driver of outcome.2
Regardless of this, mortality may be higher in patients with very severe LMD, and a first RCT comparing OMT to revascularisation alone may establish the safety with moderate LMD first before addressing more severe disease.
Trial Feasibility
Is a conclusive RCT in LMD feasible? Even when using a composite endpoint including soft endpoints (e.g. ischaemia-driven revascularisation), an adequately powered RCT likely requires approximately 1,200 participants.
We estimate that only 5–6% of patients referred for coronary angiography would be eligible for inclusion when considering all criteria, posing challenges to enrolling the required population within a reasonable time frame. Furthermore, referring providers may be hesitant to have their patients approached for study recruitment out of concern for their safety.
For too long, LMD has had a reflex referral for revascularisation, and deviating from this half-century-old paradigm may cause anxiety among both providers and patients. It may require considerable education of the referral base and establishment of community equipoise before a RCT is logistically feasible.
Providers may be open to randomising borderline cases, such as those with 40–60% stenosis, but may be reluctant to have their patients with severe disease (≥70% stenosis) participate.
Resistance may also come from interventional cardiologists and cardiothoracic surgeons, given the potential for losing another group of patients in case of a negative trial.
These obstacles may be overcome with involvement of all major stakeholders and open discussions of this topic on multiple platforms, including conferences, webinars, presentations, debates and publications.
Conclusion
Management of LMD may be the last frontier for contemporary OMT in patients with stable CAD. Available data suggest that a strategy of initial OMT may well have closed the gap, favouring routine revascularisation of all patients with stable LMD, particularly among those of low to intermediate risk.
Furthermore, advances in non-invasive coronary atherosclerosis imaging may allow more individualised management. The outcome gap among patients with stable CAD receiving revascularisation in addition to OMT versus OMT alone is rapidly closing and the impact of lifestyle interventions increasing. Current data and options allow a range of management options, which should be tailored to each patient’s preferences.35
At the same time, there is a massive void regarding prospective RCT data on this topic because patients with LMD have been excluded from relevant trials for decades, while clinical practice guidelines promulgating routine revascularisation rely on studies performed almost half a century ago.
A dedicated RCT in LMD testing the hypothesis that the practice of routine revascularisation is truly superior to an initial strategy of OMT in patients with stable CAD is urgently needed to adequately inform providers and patients of the most appropriate treatment in this clinical setting.