Identifying the most appropriate revascularisation strategy for patients affected by unprotected left main coronary artery (ULMCA) disease has been considered a topic of great interest and discussion in the past few years. Although coronary artery bypass grafting (CABG) is still conventionally regarded as the standard of care for patients with significant ULMCA disease eligible for surgery, percutaneous coronary intervention (PCI) has emerged as a reasonable alternative mode of revascularisation in carefully selected patients with anatomical conditions that are associated with a low risk of procedural complications and clinical co-morbidities that predict an increased risk of adverse surgical outcomes.1,2 When considering PCI for ULMCA disease, physicians are invited to focus on lesion location, possible complex or extensive coronary artery disease and existing co-morbidities, and, a groundbreaking notion included in the current guidelines for ULMCA revascularisation, it is suggested the use of risk-scoring systems to better select the most suitable treatment.1–4
Risk Stratification – The Risk/Benefit Trade-Off
Myocardial revascularisation is appropriate when the expected benefits, in terms of survival and/or health outcomes, outweigh the expected harmful effects of the procedure. Consequently, as an essential aspect of contemporary clinical practice, a proper risk stratification is needed in order to assess the most suitable procedure to perform in case of ULMCA disease.5 In the short term, procedural risk estimation (for both PCI and CABG) offers clinicians complementary information that can help guide treatment strategy, while over the long term, it allows quality control and the assessment of health economics, while providing individual operators, institutions and regulatory bodies with the means to evaluate and compare performance. Additionally, procedural risk stratification allows patients to be more adequately informed about the risk/benefit trade-off of the available revascularisation strategy, enabling them an interactive discussion with the Heart Team, leading to the making of an informed decision.
Risk Assessment for Unprotected Left Main Coronary Artery Disease
Several scoring systems have been developed for risk assessment in the case of myocardial revascularisation. Until recently, estimation of procedural and clinical outcome following percutaneous ULMCA revascularisation has been limited by extrapolation of risk models from bypass surgery for left main and multi-vessel coronary disease, such as EuroSCORE or Parsonnet score, and/or retrospective assessment within individual trials.6–9
In many cases, these risk-scoring systems have been applied for descriptive aims as a reference to surgical patients with complex coronary artery disease rather than to prospectively determine clinical treatment or predict outcome. Moreover, risk models with a good predictive ability for CABG, essentially based on clinical variables, are often not equally valid in the case of PCI. Anatomical variables, in fact, although less important in determining surgical outcomes, play an essential role in the setting of percutaneous revascularisation. In addition to the anatomical location of the disease within the left main segment, which has just been identified as an important predictor of clinical outcome following PCI, the extent, severity and complexity of the non-left main disease may be an even greater determinant of risk. However, there is recognition that both anatomical and clinical variables are required to appropriately risk-stratify patients contemplating percutaneous revascularisation.
Available Methods of Risk Stratification
Essentially, scoring systems in current use can be divided into those based on clinical variables, those using angiographic data, and those combining both clinical and angiographic elements (see Table 1).
These scoring models, only incorporating clinical data, are relatively quick to perform, often at the bedside, and are essentially objective, including variables that are not liable to user interpretation, thus offering excellent reproducibility. However, the use of too many individual variables has been suggested to reduce the overall accuracy of data, besides making the risk model less practical to apply.10
The EuroSCORE is an established risk model, using 17 clinical variables within cardiothoracic practice to predict operative mortality.6 The additive EuroSCORE has been identified as an independent predictor of major adverse cardiovascular and cerebrovascular events (MACCE) both in patients with ULMCA disease undergoing percutaneous or surgical revascularisation.8,11–15 At one-year follow-up, irrespective of the treatment strategy (CABG or PCI), the additive EuroSCORE emerged as an independent predictor of MACCE also in the left main stem subgroup of the Synergy between percutaneous coronary intervention with TAXUS and cardiac surgery (SYNTAX) study.13
Although the additive EuroSCORE in isolation is most likely not particularly useful in selecting the revascularisation strategy, it is certainly an effective tool in identifying which patients have a high risk of adverse events following either PCI or CABG. The role of the logistic EuroSCORE has not been well clarified.
In order to optimise the EuroSCORE application in contemporary practice and improve its discriminatory ability, a new scoring system has been developed, the EuroSCORE II. It has been validated by the EuroSCORE Project Group and, presented at the annual European Association for Cardio-Thoracic Surgery meeting (Lisbon, 2011), awaits validation by users worldwide.
Mayo Clinic Risk Score
The Mayo Clinic risk score is based on seven clinical variables incorporated in a computational model to predict in-hospital mortality after revascularisation with either PCI or CABG.16,17
No specific studies have been performed applying this model to patients with ULMCA disease and the use of this tool for either defining procedural risk and/or deciding on revascularisation treatment in this complex subset remains unclear.
The ACEF score is a relatively simple risk model consisting of only three clinical variables (age, pre-operative serum creatinine value and left ventricular ejection fraction) and assessing operative risk of mortality in case of elective cardiac operations.18
The ACEF model was applied to the all-comers limus eluted from a durable versus erodable stent (LEADERS) coating population undergoing PCI and, at one-year follow-up, it was shown to be superior to the SYNTAX score alone as a predictor of cardiac death and myocardial infarction. However, it was found to be inferior to the SYNTAX score at predicting overall major adverse cardiovascular events (MACE) and the risk of repeat revascularisation.19 At two-year follow-up, considering a quite wide cohort of patients with ULMCA disease undergoing revascularisation, the ACEF score, if compared with other stand-alone or combined scores, nevertheless displayed good calibration characteristics in PCI patients, exhibiting the best prognostic accuracy in CABG patients.20
These scoring methods are only based on angiographic data. Nevertheless, paying great attention to location, extension and burden of coronary artery disease, these scores are often complex to calculate, time-consuming and, introducing a subjective element to the assessment of risk, they are often not well reproducible without appropriate training.21,22
American College of Cardiology/American Heart Association Lesion Classification
The ACC/AHA lesion classification system, developed in the pre drug-eluting stent era and based on 11 angiographic variables, was predictive of the angiographic success of PCI with a subsequent prognostic effect on the early and late clinical outcomes.23 Nevertheless, conflicting results have been reported during the drug-eluting stent era.24–26
Specifically analysing a cohort of patients with ULMCA disease undergoing percutaneous revascularisation, Capodanno et al. showed that the ACC/AHA risk model was significantly predictive of both cardiac death and MACE at one-year follow-up.26
The SYNTAX score is an anatomical scoring system based on 11 angiographic variables, enabling the complexity of coronary anatomy and disease.27,28
The value of SYNTAX score in case of ULMCA disease has been specifically assessed in over 3,000 patients at up to four years follow-up in four different studies: the SYNTAX trial; the appraise a Customized strategy for left main revascularisation (CUSTOMIZE) registry; the Revascularisation for unprotected left main coronary artery stenosis: the comparison of percutaneous coronary angioplasty versus surgical revascularisation from multicenter registry (MAIN COMPARE); and the Rotterdam left main (Rotterdam LM) registry.13,15,26,29–31 Despite the SYNTAX scoring model being identified as an independent predictor of MACE for patients undergoing PCI, the same has not been demonstrated for those undergoing CABG, with only a few exceptions.26,30–37 This is probably due to the fact that the bypass is anastomosed distal to the severe coronary disease, regardless of the anatomical complexity, provided there are suitable graftable targets.
According to the extension and degree of concomitant coronary artery disease assessed by the SYNTAX score, differential treatment effects on long-term mortality have been shown in patients with ULMCA disease undergoing drug-eluting stent (DES) or CABG, with CABG comparing favourably with PCI in the high atherosclerotic burden of disease.38 The good MACCE and cardiac mortality predictive ability for PCI, along with the poor predictive ability for the same outcomes in CABG, make the SYNTAX score the preferable decision-making tool in ULMCA disease.13,15,20,39
Functional SYNTAX Score
This scoring system, developed incorporating fractional flow reserve (FFR) measurements into the SYNTAX score model and applied in a retrospective sub-analysis involving 497 patients with multi-vessel disease enrolled in the FFR-guided arm of the Fractional flow reserve versus angiography for multi-vessel evaluation (FAME) study, was shown to improve risk stratification of patients, if compared to the conventional SYNTAX score model.40 A relevant proportion of the higher-risk patients were favourably reclassified into lower-risk categories while a significantly higher event rate was maintained in the high-risk cohort.40 Nevertheless, the score was not validated in ULMCA disease patients.
Combined Risk Scores
In order to achieve a more complete assessment of risk, new scoring models have been proposed, combining clinical and angiographic variables. The stand-alone clinical and angiographic-based scores assess totally diverse but equally important variables and are suitable in predicting different outcomes. Particularly in the setting of PCI, one common concern of using stand-alone clinical risk scores is that they do not include any information about the anatomy and extent of coronary artery disease, essential elements in influencing the outcome following the procedure. In consideration of this, some combined clinical and angiographic risk scores have been developed, although their validation is still at an early stage.
Society of Thoracic Surgery Score
This risk model, incorporating two angiographic and 40 clinical variables, predicts the risk of operative mortality and morbidity after adult cardiac surgery.41,42
No data exists to our knowledge regarding the utility of the STS score in patients undergoing PCI and this scoring model seems not to be particularly relevant in the assessment of patients with ULMCA disease prior to the selection of a strategy of surgical revascularisation.43
Clinical SYNTAX Score
This risk-scoring model was obtained by multiplying the SYNTAX score with a variant of the ACEF score (the modified ACEF score, more accurately assessing the underlying renal function).18,44,45
Applied to the Arterial revascularization therapies study II (ARTS II) population undergoing PCI with sirolimus-eluting stents for multi-vessel coronary artery disease, the Clinical SYNTAX score was shown to better predict MACCE and death at five-year follow-up when compared to the SYNTAX score or modified ACEF score alone.45,46 An improved predictive accuracy for five-year all-cause death mortality using the Clinical SYNTAX score was also reported in the SIRTAX trial.47 Also confirmed in a registry including patients with ULMCA disease treated percutaneously, despite accurately predicting events in the high-risk tertile, the Clinical SYNTAX score was unable to discriminate well between the end points for the low and intermediate tertile.20,46
Global Risk Classification
The Global Risk Classification intends to combine the EuroSCORE established clinical variables with the SYNTAX score anatomical variables to provide a global and as complete as possible risk assessment for patients proposing to undergo PCI revascularisation.48 This combination appears to be particularly attractive considering the EuroSCORE’s ability to select patients at high risk of adverse events irrespective of treatment strategy and the SYNTAX score’s ability in identifying the optimal revascularisation strategy.
Applying this risk model, considering a cohort of 255 patients with ULMCA disease undergoing PCI, a significant improvement in the prediction of cardiac mortality was found at two-year follow-up, in addition to a better ability in discriminating between patients at intermediate risk from those at low and high risk, respectively.48 Compared to other stand-alone or combined risk scores, the Global Risk Classification seems to offer the best balance in terms of discrimination and calibration for cardiac mortality for ULMCA disease treated with PCI.20
New Risk Classification Score
The NERS score, consisting of 54 variables (17 clinical, four procedural and 33 angiographic), was developed to predict outcomes for ULMCA disease patients undergoing PCI.49
The NERS score was found to be more predictive of MACE than the SYNTAX score alone and, at six-month follow-up, a NERS score ≥25 was shown as the only independent predictor of cumulative MACE and stent thrombosis.49 Nevertheless, the NERS score was derived from a relatively small number of patients undergoing PCI in four different centres in one country and validation of this model in a larger and randomised population is required.
Is there a Specific Risk Stratification Model to Recommend?
Comparative analyses of the numerous different risk assessment models are limited due to the fact that each has been applied to different study populations with different outcome measures reported at various times. These limits restrict the ability to recommend one specific risk model.
To be effective and routinely applicable, an ideal risk model should combine a reliable ability to predict clinical outcomes with a scoring system that is as simple as possible – a really difficult mix to obtain. The number of incorporated variables should be sufficient for the model to properly predict risk, but it should not be excessive so as to avoid restraining user uptake. In addition, in a good risk model, the selected variables should account for inter-patient variations in co-morbidity. Assessment of prospective data collected from a wide population undergoing long-term follow-up is needed to develop a comprehensive stratification scoring system capable of optimally discriminating between PCI and CABG in patients with ULMCA disease requiring revascularisation.
Finally, risk stratification models should only be considered as guides, while clinical judgement and multidisciplinary dialogue should remain the cornerstones of patient management.