Introduction

In a lesion subset once routinely excluded from interventional cardiology trials, recent successes principally in nonrandomized comparative trials of drug-eluting stents (DES) and coronary artery bypass grafting (CABG) in unprotected left main coronary disease (ULM) revascularization have supported the rationale to revisit established conventions for treatment and broaden therapeutic options. Although varied in trial design, methods, and study population size, these trials suggest clinical equipoise for percutaneous coronary intervention (PCI) and CABG with consistently similar rates of combined safety outcomes of death, myocardial infarction, and stroke [1••]. In selected instances, new insights from recent studies have identified patient characteristics for which PCI may represent an acceptable alternative or possibly even preferred strategy. Accordingly, societal guidelines for ULM PCI have recently been revised. For example, the 2009 United States focused guidelines for PCI support consideration of ULM stenting in patients with anatomic conditions that are associated with a low risk of PCI-related procedural complications and clinical conditions that predict an increased risk of adverse surgical outcomes (class IIb) [2••]. Similarly, the 2010 European societal guidelines have advanced the indication for ULM PCI to either a IIa or IIb recommendation depending upon the extent and complexity of both left main and non-left main coronary disease [3••]. Considering the increasing focus on ULM PCI in clinical practice, the purpose of this document is to review the strengths and deficiencies of existing evidence that may support (or dismiss) ULM PCI, provide guidance regarding clinical decision making and address outstanding concerns that must be satisfied before further redefining standards for ULM revascularization.

Contemporary Trials in ULM Percutaneous Revascularization

Until recently, only modest evidence was available to support ULM PCI in patients ineligible for CABG with even less evidence to endorse PCI as a routine therapy for a broader patient population. Early discouraging reports with balloon angioplasty or bare metal stents (BMS) were confounded by poor patient selection and nascent procedural technique, contributing to at best inconsistent intermediate-term clinical outcomes and at worst unacceptably high rates of restenosis-related complications manifest as repeat revascularization, myocardial infarction, or even sudden cardiac death. However, encouraged by small, randomized experience and comparisons from observational trials or subgroup analysis, more contemporary studies evaluating DES for ULM disease and with longer-term follow-up have demonstrated remarkably consistent and favorable “hard” end points of death, myocardial infarction, and repeat revascularization [1••, 4, 5].

Outcomes of DES and BMS in ULM lesions parallel those in less complex lesion subsets, with significant reductions in restenosis and repeat revascularization and, at least similar safety outcomes of death, myocardial infarction, and stent thrombosis. In a recent meta-analysis of trials comparing BMS and DES in ULM revascularization (N = 5,081), treatment with DES was associated with significant reduction in the 3-year rates of death (odds ratio [OR], 0.70; 95% CI, 0.53–0.92; P = 0.01), myocardial infarction (OR, 0.49; 95% CI, 0.26–0.92; P = 0.03), and repeat revascularization (OR, 0.46; 95% CI, 0.30–0.69; P < 0.01) [6]. Although treatment with DES is preferred, outcomes with conventional BMS are important because their use (or CABG) may be favored over DES if there is increased bleeding risk, impending noncardiac surgery that requires discontinuation of dual antiplatelet therapy, and treatment of large diameter vessels (≥5 mm) that exceed DES diameter availability. Although outcomes may have been influenced by selection and treatment bias, recent nonrandomized comparison of BMS and CABG for ULM revascularization in the ASAN-MAIN registry reported lower unadjusted long-term rates of death and the composite of death, Q-wave myocardial infarction, and stroke with BMS compared with CABG (CABG; 15.9% vs 24.1%; P = 0.02 and 25.2% vs 32.1%; P = 0.04, respectively); however, rate of target vessel revascularization was significantly higher in the BMS cohort (36.7% vs 4.9%; P < 0.001) [7].

Recent trials and pooled analysis have provided important perspective to the safety and efficacy of ULM revascularization with DES compared with CABG. The nonrandomized MAIN COMPARE trial involving 2,240 patients with ULM disease compared outcomes with PCI (DES 71%/BMS 29%) or CABG [8•]. Notably there was less diabetes or multivessel coronary disease in the PCI cohort. At 5-year follow-up, according to propensity score adjustment including 542 matched patient pairs, ULM PCI was associated with similar mortality (hazard ratio [HR], 1.02; 95% CI, 0.74–1.39) and the composite outcome of death, Q-wave myocardial infarction, and stroke (HR, 1.10; 95% CI, 0.74–1.38) [9]. However, repeat revascularization was significantly more common with PCI compared with CABG (HR, 4.55; 95% CI, 2.88–7.20). Analysis following risk adjustment but limited to patients treated with CABG or DES (396 patient pairs) resulted in similar outcomes. These data represent an important contribution to our understanding of ULM PCI but must be cautiously interpreted given that propensity adjustment in nonrandomized trials may not fully correct for differences in patient populations.

In the randomized SYNTAX trial comparing CABG with PCI for left main/multivessel disease, patient treatment assignment was stratified according to presence of significant ULM disease. Of the ULM cohort (N = 705), there were approximately 60% with bifurcation disease and 13% with isolated left main disease [10•]. In the ULM subgroup, despite significantly higher repeat revascularization for the PCI group at 3 years (20.0% PCI vs 11.7% CABG; P = 0.004), and lower incidence of stroke in the PCI group (1.2% PCI vs 4.0% CABG; P = 0.02), outcomes of death (7.3% PCI vs 8.4% CABG; P = 0.64), myocardial infarction (6.9% PCI vs 4.1% CABG; P = 0.14), and the composite end point of death/myocardial infarction/stroke (13.0% PCI vs 14.3% CABG; P = 0.60) remained similar between treatment groups [11•, 12]. These results are consistent with those from a recent systematic overview of comparative trials in ULM revascularization (N = 3,773) reporting similar safety outcomes between percutaneous and surgical revascularization strategies but a higher rate of repeat revascularization with PCI [13].

Regarding selection of DES, few comparative studies have evaluated outcomes relative to DES type. In the randomized ISAR-LEFT MAIN trial [14•], PCI with either sirolimus-eluting stents (SES) or paclitaxel-eluting stents (PES) was associated with similar 2-year clinical events in both stent groups (target lesion revascularization, 9.2% PES vs 10.7% SES; P = 0.47). Although to date no studies have compared newer-generation DES to PES or SES in ULM PCI, ongoing trials are comparing newer-generation DES to CABG. The Nordic-Baltic-British Left Main Revascularization trial is randomizing 1,200 patients with ULM and SYNTAX score less than 22 to CABG or DES with end points of major adverse cardiac events (MACE) at 2 years and death at 5 years. The EXCEL trial is a large, international trial in which approximately 3,000 patients with ULM disease and a SYNTAX score ≤ 32 will be randomized to revascularization with everolimus-eluting stents or CABG.

Patient Selection and Predictive Models for Outcomes Assessment Following ULM Revascularization

Although not all uncertainties must be satisfied to extend ULM PCI as an alternative to surgery in broad patient populations, presently available data do permit advancing ULM PCI in more narrowly defined subgroups of patients. However, translation of clinical data into results meaningful to an individual patient is challenging, and the presentation of clinical data to patients may be biased by preferential selection of clinical studies and differential emphasis on end points (eg, death vs repeat revascularization vs recovery and quality of life). Specific to ULM revascularization, for which a surgical standard of care has been historically established, and in which both PCI and surgical alternatives may result in divergent clinical outcomes favoring either therapy depending upon risk, a predictive model based on clinical and angiographic characteristics seems essential to clinical decision making and informing patients for consent.

The overall extent and complexity of both left main and non-left main coronary disease are important for deciding revascularization strategy. Although anatomic location of disease within the left main segment has been identified as an important predictor of clinical outcome following PCI (detailed below), the extent, severity, and complexity of non-left main disease may be an even greater determinant of risk. Specifically, non-left main lesion complexity (eg, chronic total occlusion, bifurcation disease, calcification) may jeopardize the likelihood of procedural success and limit completeness of revascularization, which was significantly less common among PCI than CABG patients in the ULM subgroup of the SYNTAX trial [10•]. Among patients undergoing ULM PCI, the extent of coronary disease also predicts likelihood of late-term major adverse events, principally driven by increasing risk of repeat revascularization [11•, 12].

Risk scores are useful in determining the early and late outcomes after PCI and CABG for ULM disease, and discriminating between these two modalities for the individual patient [15•]. The EuroSCORE and Parsonnet score, which are typically used to risk stratify CABG candidates, have been applied both prospectively and retrospectively to patients undergoing ULM PCI. An analysis from the MAIN-COMPARE trial demonstrated that the EuroSCORE ≥ 6 was an independent predictor of mortality in ULM patients who undergo both percutaneous and surgical revascularization [16]. Similarly, increasing Parsonnet score was also identified as a significant predictor of major adverse cardiac and cerebrovascular events [16].

Predictive models, such as the SYNTAX score, may also help discussions of risk and benefit during the consent process for ULM revascularization [11•, 17]. When the SYNTAX score is in the highest tercile (≥ 33) indicating extensive and/or complex coronary artery disease, surgical revascularization may be favored over PCI. Alternatively, patients with low (0–22) and intermediate (23–32) SYNTAX scores may have comparable composite safety and efficacy outcomes with either treatment strategy [11•]. At 3 years, in patients with ULM disease and SYNTAX scores ≥ 33, PCI patients had higher rates of repeat revascularization than CABG (27.7% vs 9.2%; P < 0.001) and death (13.4% vs 7.6%; P = 0.10) [12]. Conversely, patients with ULM with low (0–22) and intermediate (23–32) SYNTAX scores had similar rates of repeat revascularization, myocardial infarction, death, or stroke or any combination of these if treated by PCI or CABG. Although these analyses from the left main cohort of the SYNTAX trial are post hoc and statistically underpowered, they currently represent the best level of evidence to guide revascularization decisions, and have thus recently been incorporated into societal guideline recommendations. As an example, the 2010 European Society of Cardiology/European Association for Cardio-Thoracic Surgery guidelines for myocardial revascularization have revised the indication for ULM PCI involving the ostium/shaft with or without single vessel coronary disease to a class IIa recommendation (weight of evidence favors its efficacy) [3••]. ULM PCI for ostial/shaft disease with two- or three-vessel disease, or distal bifurcation disease of the left main with two- or three-vessel disease and a SYNTAX score less than 33 is provided a class IIb recommendation (usefulness is less well established). Alternatively, a class III recommendation for PCI is applied for ULM patients with a SYNTAX score ≥ 33 (evidence that treatment is not useful and may be harmful).

For any risk model, prospective validation in an independent patient population with ULM disease is essential. Observational studies have externally validated the utility of the SYNTAX score to predict mortality and major adverse cardiac events in usual practice [18, 19]. In addition to the SYNTAX score, the incorporation of clinical risk factors besides angiographic characteristics may improve the predictive utility. The New Risk Stratification (NERS) model includes clinical, procedural, and angiographic characteristics [20]. For MACE, the sensitivity and specificity of an NERS score ≥ 25 were 92.0% and 74.1%, respectively, representing significantly higher predictive measures than SYNTAX intermediate risk (20.5% and 25.4%) or SYNTAX higher risk scores (70.5% and 35.2%; P < 0.001 for all comparisons). Furthermore, an NERS score ≥ 25 was the only independent predictor of MACE and stent thrombosis.

Currently, no uniform consensus exists regarding the most practical and accurate risk model for evaluating patients with ULM disease. One additional challenge is that patients considered for ULM PCI have been systematically excluded from many clinical trials, further restricting the generalizability of data to any individual patient. Acknowledging the limitation that risk scores cannot predict outcomes for individual patients with characteristics not included in the model, construction of a risk model must include appropriate and necessary preprocedural clinical and angiographic characteristics while avoiding overfitting the model [15•].

Finally, excepting emergency indications, ad hoc PCI should not be performed. Instead, the process for patient selection and informed consent ideally should involve a collaborative, multidisciplinary approach with a “Heart Team” represented by both a cardiac surgeon and interventional cardiologist, with both having the opportunity to discuss with the patient the relative merits and risk of each strategy [3••]. Consultation with a non-interventional cardiologist may also provide the patient more insight into the treatment options. Recognizing the informed consent process as an opportunity to optimize patient understanding and objective decision making, the Heart Team represents a balanced and integrated approach for multidisciplinary decision making and consensus building.

Special Considerations for ULM Percutaneous Revascularization

Left Main Lesion Complexity

Left main coronary lesion complexity has clear procedural and clinical implications underscoring the need for proper evaluation of both the distribution and severity of disease. To this purpose, hemodynamic and intravascular ultrasound (IVUS) assessment of left main coronary atherosclerotic disease has associated the functional and/or anatomic relevance of stenosis with the need for treatment and clinical outcome [21, 22]. Specifically, an IVUS-derived minimal luminal area of less than 6.0 mm2 has been found to be a useful cutoff value for clinically significant left main coronary artery disease [23], and has been correlated with hemodynamic significance by fractional flow reserve assessment [22]. Recent IVUS studies have demonstrated that left main atherosclerotic plaque burden is frequently more extensive than predicted by angiography alone [24, 25]; in particular, whereas atherosclerotic plaque is rarely present in the carina (flow divider), extension of disease from the ostia of the left anterior descending artery or left circumflex arteries into the left main segment is very common.

Left main lesions not involving the distal bifurcation (representing <40% of patients undergoing revascularization) are associated with high procedural success rates and favorable late-term outcomes of death, myocardial infarction, and repeat revascularization. In a multicenter study of 147 patients undergoing ostial or shaft ULM PCI with DES, rates of cardiac death rate and repeat target lesion revascularization were 2.7% and 0.7%, respectively, during an average follow-up period of approximately 2.5 years [26]. In comparison, the presence of distal bifurcation disease has repeatedly been identified as one of the most significant predictors of repeat revascularization and overall MACE after ULM PCI. A meta-analysis of 17 trials involving ULM PCI identified the presence of bifurcation disease as the most significant predictor of repeat revascularization and overall MACE [27].

Procedural Technique

Percutaneous coronary intervention for ULM disease can be technically challenging, requiring optimal strategies for treatment of either complex distal bifurcation disease or ostial/shaft stenoses that jeopardize a large myocardial territory. Somewhat surprisingly, however, procedural strategy and technique are common practical considerations that are poorly addressed in clinical trials describing left main revascularization. In part related to this reason, optimal PCI strategies for ULM disease are yet to be clearly defined. Moreover, strategies may vary depending on different anatomic features of the entire coronary anatomy and lesion morphology.

Treatment of distal bifurcation lesions is technically more challenging and is associated with higher rates of restenosis than isolated ostial or shaft disease, particularly when two stents are used. When angiographic restenosis does occur, it is most common at the ostium of the left circumflex artery [28], a clinical observation that may be dependent upon bifurcation angle and carina shift rather than change in plaque geometry. In accord with published results involving bifurcation stenting in non-left main disease, uncontrolled studies of ULM bifurcation PCI favor a single-stent provisional approach compared with intentional two-stent techniques. Notably, one large study reported comparatively higher rates of cardiovascular death and target lesion revascularization with two-stent treatment of bifurcation disease, including instances of unsuccessful provisional approaches that required additional stent placement [28]. In addition, several recent observational studies have demonstrated nearly equivalent clinical outcomes with single stent ULM bifurcation revascularization and left main stenting for ostial or shaft disease [29–32]. Nevertheless, approximately 40% of ULM bifurcation treatment involves a two-stent method [14•, 28], yet the optimal two-stent technique (eg, crush, culotte, V- or T-stenting) has not been identified, and the procedure is instead determined more by operator and institutional preference.

Intended as a solution to limitations of existing stent designs in complex anatomy, novel dedicated bifurcation stent designs are in early clinical development [33], yet evidence to support their procedural and clinical superiority over existing standards has not been sufficiently demonstrated.

Aside from stent technique, additional procedural uncertainties relate to use of IVUS and hemodynamic support. In many circumstances, the application of IVUS may be invaluable to assess optimally plaque distribution, bifurcation involvement, and vessel calcification that are characteristics often poorly defined by angiography alone. IVUS may also provide important information regarding stent sizing, post deployment stent expansion, and stent-wall apposition. Nevertheless, the application of IVUS in ULM PCI trials has been inconsistent, with some studies reporting improved survival with IVUS-guided ULM PCI [34] and others describing favorable outcomes despite negligible use of IVUS [14•].

Similarly, patient-specific angiographic- and procedural-related factors that predict the unplanned requirement for adjunctive hemodynamic support during ULM PCI are poorly characterized. In most instances, however, pharmacologic or mechanical circulatory support is not required; in the ISAR-LEFT MAIN trial, for example, intra-aortic balloon pump counterpulsation was used in less than 1% of the 607 patients undergoing ULM PCI [14•]. Although procedural-related complications [35] or hemodynamic compromise [36] may be reduced with elective use of intra-aortic balloon pump counterpulsation or alternative methods of hemodynamic support [37], a reduction in adverse clinical outcomes (eg, myocardial infarction, death) compared with their provisional use has not been demonstrated.

Antiplatelet Therapy and Stent Thrombosis

DES placement in ULM disease represents a dilemma between the importance of avoiding restenosis and the risk of stent thrombosis associated with delayed vessel healing. Given that stent thrombosis is a devastating complication of PCI, associated with a near uniform rate of myocardial infarction and considerable mortality [38], its occurrence in the ULM territory may have catastrophic clinical consequences. Fortunately, ULM stent thrombosis is uncommon, and several recent multicenter registries evaluating the occurrence of late and very late stent thrombosis provide some reassuring and remarkably consistent evidence to support DES treatment in ULM disease. In the ISAR-LEFT MAIN trial, for example, among 607 patients receiving DES for ULM disease, the overall 2-year rate of Academic Research Consortium–defined definite stent thrombosis was 0.5%, with no instances beyond 30 days of the index procedure [14•].

Thienopyridine discontinuation within 6 months of DES implantation is a predictor of stent thrombosis [38]. In a multicenter observational study, clopidogrel discontinuation within the first 31 to 180 days following ULM DES implantation was associated with a more than fourfold risk-adjusted increase in cardiovascular mortality and myocardial infarction compared with discontinuation beyond 180 days [39]. Alternatively, with dual antiplatelet therapy extended beyond 1 year, combined data from two randomized trials with DES recently reported no significant reduction in cardiovascular death, myocardial infarction, or stent thrombosis [40] but only 3% of patients included in this analysis underwent ULM PCI.

In ULM PCI clinical trials, the duration of antiplatelet therapy has been variable, with no standardized recommendations for aspirin and thienopyridine dosing or duration. In addition, limited studies have focused on the role of genomic and/or platelet reactivity testing to identify patients with high on-treatment residual platelet activity who may be at increased risk for subsequent ischemic events [41]. Accordingly, the optimal duration of dual antiplatelet therapy in patients stented for ULM disease has not been established. Thus, expectations for patient compliance and a reasonable assurance of no foreseeable circumstances that might necessitate premature discontinuation of antiplatelet therapy are important considerations. CABG or even BMS should be considered if concerns are present regarding bleeding risk, noncompliance, foreseeable need for interruption, or contraindications to prolonged dual antiplatelet therapy.

Angiographic and Clinical Surveillance

Historically, routine surveillance angiography was common, but considering the low rates of DES restenosis and stent thrombosis following ULM revascularization, the clinical utility of surveillance of angiography has been challenged. In the LE MANS substudy (N = 145) of the SYNTAX trial, for example, angiographic restenosis (>50% stenosis) at 15-month follow-up was identified in only 2% (1/48) of patients with ostial/shaft disease and 10% (10/97) of patients with bifurcation disease [42]. In consideration of these data and the recognition that scheduled nonsymptom-driven angiography could result in unnecessary procedures, the most recent guidelines no longer endorse its performance [2••]. However, noninvasive assessment of ischemia is reasonable at 6 months and annually thereafter [1••].

When clinical restenosis following ULM stenting is identified, the most appropriate treatment is also uncertain. In cases of intermediate significance, assessment of hemodynamic significance with fractional flow reserve should be performed. If lesion significance is confirmed, IVUS may also inform the mechanism of restenosis and should be performed routinely for restenosis if repeat PCI is considered.

For patients with ULM restenosis, repeat revascularization seems imperative, and PCI may be associated with favorable outcomes. In a multicenter observational study of patients with ULM restenosis treated with repeat PCI (N = 70), the risk of cardiovascular death was 1.7% over a mean follow-up period of 35 months, with no occurrences of stent thrombosis [43]. During the follow-up period, the risk of MACE was lowest with CABG or repeat PCI compared with medical therapy alone (MACE at 35 months, 14% CABG, 25% PCI, 50% medical therapy). Unlike after DES treatment of de novo ULM disease, early follow-up angiography at 4 to 6 months should be considered if repeat PCI is performed for ULM restenosis given higher risk of disease recurrence.

Conclusions

Over the past 5 years, an evidence base has emerged supporting the consideration of ULM PCI as an alternative revascularization strategy to CABG in selected patients. In particular, trials reporting late-term safety outcomes of cardiovascular death, myocardial infarction, and stroke comparable to CABG have fostered renewed enthusiasm for ULM PCI. In parallel, application of DES and advances in technique and strategy have enhanced the rates of early procedural and long-term clinical success of ULM PCI.

Ultimately, demonstration of at least clinical equivalence in a randomized trial comparing ULM PCI with surgery is necessary before percutaneous revascularization can be routinely accepted as an alternative to bypass surgery; to this purpose, forthcoming trials designed with careful attention to patient selection, timing of end-point ascertainment, and relevance of safety and efficacy end points should inform clinical decision making and treatment guidelines. However, consistent with many large comparative trials, it is more likely that such studies may clarify which patients with ULM are suitable for both or either therapies rather than demonstrate clinical equivalence for a broadly defined patient population with individual characteristics that pose variable risk.

Presently available data are substantive enough to support ULM PCI in the absence of coexisting complex coronary disease (eg, SYNTAX score ≥ 33). Thus, an important focus should be the responsibility of appropriate patient selection, with consensus that represents full consultation ideally with both an interventionalist and cardiac surgeon. Integration of clinical and angiographic variables into risk models may further refine assessment of patient risk for either revascularization method and responsibly inform the patient for consent. Aside from an emphasis on comparative clinical outcomes, ongoing clinical trials intended to address the practical and technique-related issues of ULM PCI and promote consensus building between cardiology and surgical societies are essential to advance treatment recommendations and optimize the outcomes for patients with ULM coronary artery disease.