Article Text

Download PDFPDF

Contemporary and evolving risk scoring algorithms for percutaneous coronary intervention
  1. Vasim Farooq,
  2. Salvatore Brugaletta,
  3. Patrick W Serruys
  1. Department of Interventional Cardiology, Erasmus University Medical Centre, Thoraxcenter, Rotterdam, The Netherlands
  1. Correspondence to Professor Patrick W Serruys, Professor of Medicine, Head of the Interventional Cardiology Department, Erasmus MC,'s-Gravendijkwal 230, 3015 CE Rotterdam, The Netherlands; p.w.j.c.serruys{at}


Risk stratification is an essential part of appropriately informing patients electing to undergo percutaneous coronary intervention (PCI). This process is also an integral part of the SYNTAX (Synergy between PCI with Taxus and Cardiac Surgery)-pioneered heart team approach in determining the most appropriate revascularisation modality for patients with complex coronary artery disease. The SYNTAX score was pioneered as an anatomical-based risk score to aid in this decision-making process; the lack of clinical variables in this score has, however, been its main limitation. This review examines the important established and evolving contemporary risk models used to aid this risk-stratification process. Risk scores based on clinical and anatomical variables alone and in combination—the latter of which is the subject of continuing research—are all explored. Other areas of discussion include risk scores based on the completeness of revascularisation and emerging concepts such as functional anatomical risk scores and the patient-empowered risk-benefit trade-off between PCI and coronary artery bypass grafting, to help personalise the choice of revascularisation modality.

  • PCI risk scores
  • SYNTAX score
  • EuroSCORE
  • CathPCI
  • NERS
  • ACEF
  • global risk
  • coronary artery disease
  • coronary angioplasty
  • stents
  • stent interventions
  • invasive cardiology

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Risk stratification is an essential component of appropriately informing patients electing to have percutaneous coronary intervention (PCI). This process is also an integral part of the Synergy between PCI with Taxus and Cardiac Surgery (SYNTAX) Trial-pioneered heart team approach in selecting the most appropriate revascularisation modality (coronary artery bypass grafting (CABG) or PCI) in patients with complex (three-vessel disease (3VD) or left main stem (LMS) disease) coronary artery disease. Despite the landmark SYNTAX Trial having established that surgery is the standard of care for patients with LMS disease or 3VD, an important finding from this study was that patients with less complex disease had equivalent outcomes to surgical and percutaneous revascularisation at up to 3 years follow-up.1–4 The heart team approach in managing patients with complex coronary disease has therefore recently been incorporated as a class I recommendation in recent European myocardial revascularisation guidelines.5

In cardiothoracic surgical practice, the use of risk models to appropriately risk-stratify patients is well established. These risk scores are predominantly related to clinical variables alone, with scores such as the EuroSCORE6 7 or Society of Thoracic Surgery Score8 being in widespread contemporary use. The use of anatomical variables for cardiothoracic risk models, eg, the SYNTAX Score, has been shown not to provide any additional predictive benefit over clinical variables. This is likely to be related to bypass grafts being anastomosed distal to the coronary disease, regardless of the complexity of the proximal segments, provided that there are suitable graftable targets.4 9

The SXscore was pioneered as an anatomical-based score to aid the heart team decision-making process—its main application being identification of less complex coronary artery disease, which would be equally amenable to surgery or PCI in terms of efficacy and safety. However, criticism emerged that clinical factors were not being taken into account in the risk stratification of the PCI patient, and potentially important morbidity and prognostic information may be missing.4 10–13 Several risk scores have consequently attempted to merge the SXscore with clinically based risk scores, such as the Clinical SYNTAX Score (a combination of the SXscore and the modified ACEF (age, creatinine clearance, ejection fraction) Score) and the Global Risk (a combination of the SXscore and the EuroSCORE). These are discussed later in this review.

The purpose of this review article is to give the clinician a concise overview of the important established and evolving contemporary risk models for risk-stratifying patients electing to undergo PCI. Risk scores based on clinical and anatomical variables alone and in combination, and on the completeness of revascularisation are all discussed. Furthermore, attention is drawn to which clinical outcomes—and over what time period (ie, in-hospital, short or long term)—the risk model is stratifying risk for, and also if the risk model is validated in a population other than that from which it was derived (table 1). Emerging concepts such as functional anatomical risk scores, based on invasive and non-invasive assessments, and the novel concept of patient-empowered risk–benefit trade-off between CABG and PCI to help personalise the choice of revascularisation modality are also explored.

Table 1

Summary of a selection of established and contemporary risk models categorised by anatomical, clinical or combined types, for the assessment of risk in patients proposing to undergo percutaneous coronary intervention

Anatomy-based risk scores

In 1981, Leaman et al44 developed a scoring system that assessed the severity and extent of the underlying coronary artery disease. This system was based on the severity of luminal diameter narrowing and weighted according to the usual flow to the left ventricle in each coronary vessel. Consequently, the most weight was given to the LMS, followed by the left anterior descending, circumflex, and right coronary arteries. This early pioneering work, with further information derived from adverse characteristics of coronary lesions from the American College of Cardiology/American Heart Association (ACC/AHA) lesion classification system,45 46 and the modified Duke/Institut Cardiovasculaire Paris Sud (ICPS) System classification for bifurcation lesions,14 47 ultimately formed the basis of the SXscore.4 48 49

ACC/AHA lesion classification system

The ACC/AHA lesion classification system was one of the first angiography scoring systems developed, comprising 11 angiographic variables with all lesions categorised into types (A, B1, B2 and C).45 46 This system predicts the angiographic success of PCI with a subsequent prognostic effect on the early and late clinical outcomes in the pre-drug-eluting stent era (box 1).50 Registry data from the drug-eluting stent era have, however, had conflicting results. The German Cypher Registry (n=6755) failed to show any significant association with clinical outcomes at 6 months;15 conversely, data from a small registry (n=255) was potentially predictive of mortality in unprotected LMS PCI at 1-year follow-up.16

Box 1

Characteristics of American College of Cardiology/American Heart Association (ACC/AHA) type A, B and C lesions50

Type A lesions (high success, >85%; low risk)

Discrete (<10 mm length)


Readily accessible

Non-angulated segment <45°

Smooth contour

Little or no calcification

Less than totally occlusive

Not ostial in location

No major branch involvement

Absence of thrombus

Type B lesions (moderate success, 60–85%; moderate risk)

Tubular (10–20 mm length)


Moderate tortuosity of proximal segment

Moderately angulated segment, 45–90°

Irregular contour

Moderate to heavy calcification

Ostial in location

Bifurcation lesions requiring double guidewires

Some thrombus present

Total occlusion <3 months old

Subdivided into type B1 (one type B characteristic) and B2 (two type B characteristics) types

Type C lesions (low success, <60%; high risk)

Diffuse (>2 cm length)

Excessive tortuosity of proximal segment

Extremely angulated segments, >90°

Inability to protect major side branches

Degenerated vein grafts with friable lesions

Total occlusion >3 months old


The SXscore is an anatomical-based risk score that takes into account features such as bifurcations, total occlusions, thrombus, calcification and small vessels (figure 1). Each coronary lesion with a >50% luminal obstruction in vessels ≥1.5 mm is scored separately and the scores summated to provide the overall SXscore. This is calculated using dedicated software that integrates the number of lesions with their specific weighting factors, based on the amount of myocardium distal to the lesion and the morphological features of each lesion.4 44 48 49

Figure 1

SYNTAX Score (SXscore) algorithm. The algorithm is applied to each individual coronary lesion that has a diameter stenosis >50% and located in a vessel >1.5 mm in diameter. The individual lesion scores are added together to give the final SXscore.3 4 48 49 LAD, left anterior descending artery; LCx, left circumflex artery; LM, left main stem; RCA, right coronary artery. Reproduced, with permission, from Serruys et al.4

In the SYNTAX Trial,1 the distribution of the SXscore was found to be Gaussian in the randomised CABG and PCI populations, with the curves almost superimposable on each other (figure 2). When the scores of the randomised SYNTAX population were divided into tertiles, the upper boundary of the lowest tertile was 22 (low risk), the second tertile ranged from 23 to 32 (intermediate risk), and the lower boundary for the highest tertile was equal to or greater than 33 (high risk).

Figure 2

Distribution of the SYNTAX Score (SXscore) in the randomised controlled trial (RCT) and nested registry percutaneous coronary intervention (PCI) and coronary artery bypass grafting (CABG) populations from the SYNTAX Trial. Note how the distributions for both the RCT, CABG and PCI populations are almost superimposable on each other, whereas the nested registries—consisting of patients with more complex coronary disease—are shifted to the right.1 4 Adapted and reprinted from EuroIntervention, Serruys PW et al, Assessment of the SYNTAX score in the Syntax study. Copyright 2009, with permission from Europa Edition.4

The SXscore has since consistently been shown to identify poorer outcomes and to be an independent predictor of major adverse cardiovascular and cerebrovascular events (MACCE) in the high-tertile group of risk for PCI at 1 year,4 19 51 and in the Arterial Revascularization Therapies Study II (ARTS II) population (a population with two (46%) or three (54%) vessel disease) at up to 5 years of follow-up (figure 3).52 53 Furthermore, the 3-year SYNTAX Trial showed that a low SXscore (ie, <23) in the 3VD cohort and low–intermediate SXscore (ie, <33) in the LMS disease cohort were able to identify a subset of patients who could safely and efficaciously be treated with CABG or PCI and have comparable clinical outcomes in terms of death and MACCE.2

Figure 3

Kaplan–Meier curves of freedom protocol-defined major adverse cardiac and cerebrovascular event (MACCE) rate according to SYNTAX Score tertiles, from the 5-year clinical outcomes of the ARTS II (Arterial Revascularization Therapies Study II) Trial. Reproduced, with permission, from Serruys et al.53 MACCE, major adverse cardiovascular and cerebrovascular events; interm., intermediate; SES, sirolimus eluting stent.

Early-submitted results from the 3-year SYNTAX Trial have, however, indicated that the SXscore appears to be poorly predictive of clinical events in the 3VD cohort, unless combined with clinical variables; the high SXscore was, however, predictive of clinical events in the LMS disease population at 3 years.36 37 It was hypothesised that a high SXscore in patients with LMS disease and an intermediate–high SXscore in patients with 3VD were markers of a more adverse risk profile. This is supported by the 10-year predicted Framingham risk scores, which were recently shown to have a significant and direct relationship to the prevalence and magnitude of coronary artery calcium scores.54 Furthermore, the ankle–brachial index and common carotid intima–media thickness have both previously been reported to be associated with the extent and severity of coronary artery disease.55–57 Consequently, not only is the presence of a higher SXscore associated with anatomical complexities, such as multiple bifurcations and the presence of total occlusions—which would potentially make PCI more technically challenging with consequent increased procedural risk—but the apparent association with clinical comorbidity would also place these patients at greater long-term risk. As was clearly demonstrated in the SYNTAX Trial, patients with a higher SXscore would be better managed by CABG provided that an acceptable threshold of risk for the patient and surgeon was achievable.36 37 The reason for this is that CABG protects the entire coronary vessel compared with PCI which treats the individual.

With the next generation drug-eluting stent in the All-Comers LEADERS (Limus Eluted from A Durable vs ERodable Stent coating)58 and All-Comers Resolute59 randomised populations undergoing PCI for a broad spectrum of indications, it has been shown that the highest tertile of the SXscore was associated with a significantly higher incidence of mortality and major adverse cardiac events (MACE). Moreover, in the MULTISTRATEGY and STRATEGY Registries, composed of patients presenting with a ST-elevation myocardial infarction (MI), it was reported that the SXscore was an independent predictor of mortality, MACE, and stent thrombosis at 1-year follow-up; when however the SXscore was combined with clinical variables, the risk model proved significantly more predictive of adverse clinical events at 1 year.60

Functional anatomy-based risk scores

Functional SXscore

Fractional flow reserve (FFR) is a technique that uses a pressure wire to assess physiological variables that reflect both the severity of epicardial stenosis and the amount of myocardium supplied.61 This was investigated in the The FFR versus Angiography for Guiding PCI in Patients with Multivessel Evaluation (FAME) Study, which determined the potential prognostic impact of PCI guided by FFR measurements to determine the functional significance of an individual coronary lesion before intervention.62 63

Consequently, by incorporating FFR measurements into the SXscore to form the recently dubbed ‘Functional SXscore’, it was shown in a retrospective sub-analysis of almost 500 patients with multivessel disease from the FFR-guided arm of the FAME Study that this improved the risk stratification of patients compared with the conventional angiography-based approach to the calculation of the SXscore.21 The primary benefit appeared to be in reclassifying a significant proportion of the higher-risk groups into lower-risk categories while still maintaining a significantly higher event rate (death/MI and MACE at 1 year) in the high-risk groups. It should, however, be emphasised that no patients with LMS disease were involved in this study, and prospective validation of the Functional SXscore in LMS disease and multivessel disease is required.

One further caveat to the Functional SXscore approach is that two-dimensional (2D) and three-dimensional (3D) quantitative coronary angiography (QCA) have been shown to be far more reliable than visual estimation of vessel size, the latter being associated with poor reproducibility and often overestimation of lesion significance.17 64–67 Yong et al65 recently showed that 3D QCA assessment of the minimum lumen area was superior to the 2D QCA-derived minimum lumen diameter in determining the functional significance of a coronary lesion as assessed by FFR. The 3D QCA measurements of the minimum lumen area were derived from two orthogonal angiographic views of the target lesion, enabling better assessment of vessel tortuosity and coronary lesion asymmetry compared with 2D QCA.

If 2D QCA and, in particular, 3D QCA had been used to assess vessel size to calculate the SXscore, this may have reduced the benefits of the Functional SXscore compared with the SXscore derived from visual estimation. Visual assessment of vessel/lesion size is, however, representative of real-life practice and was also the basis for calculation of the SXscores in the SYNTAX Trial. It should also be emphasised that, although FFR may be regarded as the ‘gold standard’ method for detecting reversible ischaemia, the ischaemic potential of the small grey zone of FFR (between 0.75 and 0.80) remains unclear, and an abnormal FFR (ie, >0.80) excludes ischaemia in 90% of cases.68–70 Further study is required to validate the Functional SXscore compared with a visually or QCA derived SXscore.

Non-invasive Functional SXscore

Novel techniques in development to potentially simplify the generation of the newly developed Functional SXscore include the use of non-invasive coronary CT angiography (CTA), which allows the simultaneous assessment of anatomy and measurement of the haemodynamic significance of lesions—to permit non-invasive computation of FFR—using computational fluid dynamic techniques applied to the coronary CTA (Heartflow, Redwood City, California, USA) (figure 4). Preliminary validation of this technique was recently reported in the DISCOVER FLOW Trial,71 where it was found that the non-invasive FFR technique could dramatically improve the diagnostic accuracy of CT imaging without any immediate need for invasive FFR imaging. The larger-scale multicentre DeFACTO (Determination of Fractional Flow Reserve by Anatomic Computed Tomographic Angiography) Trial (NCT01233518) is ongoing.

Figure 4

The principle of non-invasive CT angiography that allows the simultaneous assessment of anatomy and measurement of the haemodynamic significance of lesions by fractional flow reserve (FFR) guidance. This principle can allow for the non-invasive assessment of the Functional Syntax Score. In this clinical case example, tandem lesions in the left anterior descending artery (LAD) are evident, the haemodynamic significance of which was unclear from the CT coronary angiogram: non-invasive FFR in the distal LAD was 0.60 and therefore haemodynamically significant, but FFR just proximal to the distal lesion was 0.80, indicating that the proximal LAD lesion was not haemodynamically significant; FFR at the ostial LAD was normal (0.98, not illustrated).

Limitations of anatomy-based and functional anatomy-based risk scores

The progressive development of anatomical-based risk scores culminating in the Functional SXscore has undoubtedly improved the performance of these risk models in terms of risk stratification for the individual patient. However, a limiting factor is the inevitable intervariability in coronary angiogram assessment if visual assessment is used to assess the vessel/lesion size.4 With the potential use of the QCA-derived SXscore or Functional SXscore, this problem may be circumvented.

Another limiting factor is that no clinical variables are used, which are less subjective than angiographic variables; this is of paramount importance given that the lack of clinical variables is likely to reduce the predictive ability of the risk model if anatomical variables alone are relied on.

Myocardial jeopardy scores

Myocardial jeopardy scores are used to estimate the amount of myocardium at risk on the basis of assessment of both the severity of the coronary artery lesion and the volume of myocardium it supplies. Examples of such scores include the Duke Jeopardy Score, the Myocardial Jeopardy Index from the Bypass Angioplasty Revascularization Investigation (BARI) Score, and the Alberta Provincial Project for Outcome Assessment in Coronary Heart Disease (APPROACH) Score (figure 5; table 1). The Duke and BARI Scores were developed and validated in relatively small populations. All three models have since been validated in one population-based cohort consisting of >20 000 patients and were found to be predictive of 1-year mortality in patients treated with PCI (or medically); in this population, all three scores also had similar risk model performance measures with only minor differences in c-statistics evident.22–24

Figure 5

Example of a Myocardial Jeopardy Score. The APPROACH Lesion Score illustrating the weighting factors for myocardial regions is illustrated. LAD, left anterior descending artery; RV, right ventricle; PDA, posterior decending artery; PL, posterolateral. Reproduced, with permission, from Graham et al.23

The Jeopardy Score has since been shown to be an independent predictor of adverse clinical outcomes (ie, death/MI) in medically treated patients with acute coronary syndromes at up to 1 year, in the post hoc Acute Catheterization and Urgent Intervention Triage Strategy (ACUITY) Trial.72

The recently described BCIS-1 Myocardial Jeopardy Score, a variant of the Duke Jeopardy Score which has been reported to be simpler to use, has been shown to have a strong correlation with the myocardial ischaemia burden as assessed by cardiac magnetic resonance perfusion imaging.73 74 A BCIS-1 Jeopardy Score of 10–12 and a Revascularisation Index (pre- minus post-procedural Jeopardy Scores divided by pre-procedural Jeopardy Score, with 1 indicating complete revascularisation) of 0–0.33 were both shown to be highly predictive of mortality after contemporary PCI in a single UK centre experience involving over 600 patients.75 Larger-scale, multicentre trials are, however, required. A correlation with the underlying complexity of coronary anatomy (SXscore), which appears to have different prognostic outcomes as previously described, and the correlation between QCA- and FFR-derived jeopardy scores should in our opinion be incorporated into future trials.

Clinically based risk scores

The main advantages of these scores are that they are potentially easier to perform and less subjective than purely anatomical-based scores, which require interpretation of the coronary angiogram. They can also be performed relatively quickly and often at the bedside if necessary.

New Mayo Clinic Risk Score

The New Mayo Clinic Risk Score was designed to replace the Mayo Clinic Risk Score by predominantly excluding angiographic variables, namely the presence of LMS or multivessel disease, and a few of the interaction effects of specific clinical variables.25 26 43

The New Mayo Clinic Risk Score is based solely on baseline clinical and non-invasive assessments and incorporates seven pre-procedural variables (age, serum creatinine, left ventricular ejection fraction, MI ≤24 h, pre-procedural shock, congestive heart failure and peripheral vascular disease) for the prediction of procedural Death or MACE. The risk model had c-statistics of 0.74 and 0.89 for MACE and procedural death, respectively, in the population from which the risk model was derived.25 26 The risk model has since been validated for in-hospital mortality in the National Cardiovascular Data Registry (NCDR);26 it has, however, not been validated for MACE. The New Mayo Clinic Risk Score has also been shown to be predictive of in-hospital mortality after CABG surgery.27

CathPCI Risk Score System

The NCDR CathPCI Risk Score System was developed from 181 775 procedures performed in Medicare patients over a 2-year period (table 2); the model was independently validated in two separate validation cohorts.31 The risk model was based on eight key pre-procedural factors and was found to have excellent discriminative ability in predicting in-hospital mortality (c-statistic 0.89), with the model still able to have good predictive ability for 30-day mortality after PCI (c-statistic 0.83). A full risk model, from which this model was derived (incorporating 35 pre-procedural and angiographic features), resulted in a marginally better risk model (c-statistic 0.90 and 0.86 for in-hospital and 30-day mortality, respectively).

Table 2

NCDR CathPCI Risk Score System

European system for cardiac operative risk evaluation (EuroSCORE)

The EuroSCORE is an established risk model, using 17 clinical variables, in cardiothoracic surgical practice for predicting operative mortality and has been validated in many populations around the world.6 7 76 In use since 1999, the model was derived from almost 20 000 consecutive patients from 128 hospitals in eight European countries. The additive EuroSCORE assigns an individual score to 17 clinical variables (table 3), with a low EuroSCORE risk tertile ranging from 1 to 2, intermediate risk tertile from 3 to 5, and a high risk tertile of 6+.

Table 3

Additive EuroSCORE, calculated by summating the individual scores from 17 different variables6

The subsequently developed logistic EuroSCORE has been suggested to allow a more accurate risk prediction in the CABG cohort, in particular for the high-risk population, where the additive model was found to lead to a potential underestimation of risk.6 7 76 77 Conversely, the logistic EuroSCORE has been shown to potentially overestimate observed mortality, with its accuracy for predicting risk varying in different surgical subgroups.77 78

Kim et al first demonstrated that the high-risk tertile of the additive EuroSCORE was an independent predictor of Death/MI after unprotected LMS intervention with sirolimus-eluting stents.11 Subsequently, Romangoli et al applied the additive EuroSCORE to predict in-hospital mortality in 1173 consecutive patients undergoing PCI in a single high-volume centre and correlated the higher-risk tertiles of the EuroSCORE with in-hospital mortality; the study population also included patients who had undergone unprotected LMS PCI.10 In addition, several studies have since all identified the additive EuroSCORE as an independent predictor of MACCE in patients with unprotected LMS PCI at up to 4 years follow-up.11 12 30 Only one study has examined the logistic EuroSCORE in PCI patients, with few differences being demonstrated in stratifying risk when compared with the additive EuroSCORE.10 Our group has recently applied the additive and logistic EuroSCOREs to the SYNTAX PCI population; risk model performance measures79 80 suggested that the additive EuroSCORE was superior to the logistic EuroSCORE in risk-stratifying PCI patients.36 37

ACEF Score

Ranucci et al demonstrated a relatively simple risk model (consisting of only three clinical variables, namely age, preoperative serum creatinine value and left ventricular ejection fraction) for assessing operative mortality risk in elective cardiac operations. It is noteworthy that, despite the simplicity of the model, its clinical performance appeared to be comparable to either the additive or the logistic EuroSCORE.34 35

The ACEF Score is calculated using the formula:ACEF=[Age/Ejection fraction(%)]+[1(if creatinine>2mg/dl)]

From this score, a mortality risk can be calculated from a graphic relationship of the score with an operative risk or an equation.34 35

The ACEF model was recently applied to PCI patients from the All-Comers LEADERS population at 1-year follow-up.32 Despite the ACEF Score being demonstrated to be superior to the SXscore alone as a predictor of cardiac death and MI after PCI, it was found to be inferior to the SXscore at predicting overall MACE and the risk of repeat revascularisation, reflecting the observation that anatomical and clinical variables appear to be necessary requirements for a comprehensive risk model in predicting clinical outcomes with PCI. It should however be emphasised that the ACEF model has not been validated in the PCI population.

Limitations of clinically based risk scores

The main limitations of the clinically based risk scores, apart from not incorporating anatomical-based variables, are that they rely on predominantly registry data—proposed to be more representative of contemporary ‘real-life’ practice. The potential for selection bias in patients receiving coronary angiography—for example, in limiting the number of octogenarians—may lead to the risk model potentially underestimating risk in these patient subsets.

Combined (anatomical and clinical based) risk scores

SXscore and Parsonnet Score

Combining the Parsonnet Score, an operative risk score published in 1989 consisting of 14 clinical variables,29 with the SXscore has been shown to potentially improve the performance of the SXscore alone.

In 2005, Valgimigli et al showed that the Parsonnet Score was an independent long-term (∼3 years) predictor of MACE after LMS intervention from the Rotterdam RESEARCH and T-SEARCH Registries.28 More recently, Chakravarty et al showed that adding the Parsonnet Score as a covariate to the SXscore improved the long-term (∼4 years) predictive ability of the score in predicting MACCE after LMS PCI.19

Clinical SYNTAX Score

The underlying rationale for the Clinical SXscore was to combine the SXscore and a variant of the ACEF Score (modified ACEF Score).34 35 The modified ACEF Score was used instead of the ACEF Score in this model because it had previously been shown to potentially allow a more accurate assessment of the underlying renal function, and had subsequently improved the accuracy of cardiac prediction models such as the EuroSCORE in patients receiving CABG.40 81 82 The modified ACEF Score is calculated using the formula: age/ejection fraction +1 point for every 10 ml/min reduction in creatinine clearance below 60 ml/min/1.73 m2 (up to a maximum of 6 points).

This model was applied to the ARTS II population treated with sirolimus-eluting stents for multivessel (two or three) coronary artery disease.40 83 By dividing the calculated Clinical SXscores into tertiles of risk, it was demonstrated that the risk model for predicting outcomes for MACCE and death at 5 years was superior to the SXscore or modified ACEF Score alone. One of the limiting factors of the Clinical SXscore that has prevented its clinical use is that, despite being able to potentially predict events more accurately in the high-risk tertile, the risk model was unable to differentiate between the clinical events for the low- and intermediate-risk tertiles40; this was also demonstrated when applied to a different similar-sized registry by another group.33

New Risk Classification Score (NERS)

NERS39 is a risk model developed within four centres in China (n=260) to predict long-term outcomes after unprotected LMS PCI. Reflecting the long time period over which this registry was performed (∼10 years), the patients included either had bare metal or drug-eluting stent implantation. The model was subsequently tested (internally validated) in a different consecutive group of patients in the same registry all treated with drug-eluting stents (n=337).

This risk model consists of 54 variables (17 clinical, four procedural and 33 angiographic features). A substantially higher c-statistic was evident for NERS compared with the SXscore (0.89 vs 0.69, respectively), indicating that it had excellent discriminatory ability. When NERS was separated into two groups of risk (high and low) and clinical outcomes were assessed, the NERS model was able to identify a high-risk population for MACE, at 30 days and at over 5 years follow-up. Importantly, the high-risk NERS group was shown to be significantly more predictive of MACE compared with the intermediate or high SXscore tertiles.

Conversely, in the low-risk NERS group, outcomes were similar to the low SXscore group, suggesting, at least from this study, that anatomical variables alone may be sufficient to predict clinical outcomes in the low-risk group. One of the main limitations of this risk model is that comorbidity in the NERS patient population was significantly less prevalent compared with the All-Comers SYNTAX population;1 3 the latter was designed to overcome many of the limitations/selection bias inherent in small registries. Validation of the NERS risk model in a much larger All-Comers type population is therefore required to overcome many of these issues.

Global Risk

The SYNTAX Trial established a complex interaction between the EuroSCORE and SXscore in preliminary unpublished data.84 Given that the EuroSCORE has been shown to be an independent predictor of MACE for either CABG or PCI as previously described,30 85 the need to combine the angiography and clinical scores into a single approach has become evident.86

One of the main advantages of potentially combining the EuroSCORE and SXscore to give a ‘Global Risk’ assessment is that the same risk model can be used during the heart team approach in selecting the optimal revascularisation modality for the patient. To facilitate this process, it is our opinion that the historically accepted cut-offs for the level of risk (low, intermediate and high) for the additive EuroSCORE, each of which have previously been demonstrated to have a different prognostic value,4 6 7 48 49 and the SYNTAX Trial-defined anatomical SXscore ranges4 48 should be incorporated into the Global Risk model (table 4).

Table 4

The Global Risk approach, combining historically defined tertiles of risk for the EuroSCORE and the SYNTAX Trial-defined ranges for SXscore36 37

In variants of this model using differing and inconsistent cut-off levels of risk for the SXscore and/or additive EuroSCORE in a non-randomised population, it has recently been shown that this may potentially improve the ability to predict outcomes in patients undergoing surgical or percutaneous LMS revascularisation.33 87

The main goal of using the Global Risk model is therefore to combine anatomical and clinical variables to potentially further aid risk stratification of patients with 3VD or LMS disease considering revascularisation, and to identify a low-risk group of surgically or percutaneously treated patients who would have comparable outcomes in terms of safety and efficacy. This concept has recently been applied by our group to the randomised and All-Comers SYNTAX population, and proved to be potentially useful in the identification of the previously described low-risk population in the LMS disease cohort. In the SYNTAX 3VD cohort, the interaction between anatomical and clinical variables proved to be more complex; the Global Risk model, however, still proved to be clinically useful when interpreted through a treatment algorithm. The results of these studies are forthcoming.36 37

EuroHeart Score

On the basis of the Euro Heart Survey (a European PCI registry consisting of over 46 000 patients from 176 European centres who underwent PCI for different indications), a logistic regression model comprising 10 clinical variables and six anatomical variables was developed (figure 6).38 The risk model was shown to be highly predictive of in-hospital mortality (c-statistic 0.91). The strengths of the risk model are that it was internally validated in the registry population and it retained its discriminatory power (c-statistic 0.90).

Figure 6

EuroHeart PCI Score. STE-ACS, ST-elevation acute coronary syndrome; NSTE-ACS, non ST-elevation acute coronary syndrome; BMI, body mass index; CABG, coronary artery bypass grafting; LAD, left anterior descending artery; PCI, percutaneous coronary intervention; TIMI flow, thrombolysis in myocardial infarction flow. Reproduced, with permission, from de Mulder et al.38

Risk–benefit analysis

Undoubtedly, both CABG and PCI result in an improvement in the quality of life of the patient. One of the main drawbacks of using contemporary risk models for both CABG and PCI is that the role of the individual patient and their personal preferences and perception of risk may be underestimated. To address this issue, the novel concept of a clinical model that balances the risks and benefits of the proposed revascularisation procedure has recently emerged.88

Remaining active in their professional/personal lives may be vital for some people, and they would thus be more prepared to accept the longer-term risks of PCI (in particular an increased risk of repeat revascularisation) in order to remain at their present functional state, compared with the short-term morbidity effects associated with CABG, the latter being predominantly related to the intrinsically more invasive nature of the CABG procedure (eg, thoracotomy and vein harvesting and subsequent sternotomy and leg pain, etc).89 90

Individual patients may, however, value this risk–benefit trade-off differently. For some, exchanging the increased risk of repeat PCI or CABG to obtain short-term pain relief and a rapid return to full mobility will be acceptable, while others may prefer to endure short-term pain to obtain a higher probability of avoiding a subsequent revascularisation. Some patients may also prefer to risk undergoing multiple PCI procedures compared with a single CABG, or they may prefer to avoid the risk of requiring CABG subsequent to PCI and instead have CABG initially. Consequently, from the patient's perspective, the balance between these conflicting considerations plays a crucial role in selecting the preferred revascularisation strategy.

Federspiel et al recently applied this concept to the ARTS II population, by quantifying the trade-off between the risks and benefits of PCI versus CABG, such as freedom of chest pain and improvement in health-related quality of life measures, for patients with multivessel disease (figure 7).83 88 Although this study was performed on data that were over 10 years old, in a population who had implantation of bare metal stents, the results nevertheless supported this concept and have allowed, for the first time, quantification of a level of risk that a patient would be able to accept in order to maintain their present functional state. Data from the SYNTAX Trial on this concept, reflecting more contemporary practice with drug-eluting stents, are forthcoming.

Figure 7

Principle of the risk–benefit trade-off in determining the choice of revascularisation modality from the ARTS II multivessel study population. Left panel, between 0 and 6 months, patients who selected percutaneous coronary intervention (PCI) were more likely to be free from pain compared with those who selected coronary artery bypass grafting (CABG). After 6 months, no statistically significant differences in pain between CABG and PCI were evident. Right panel, risk–benefit acceptability curve for incremental risk of additional revascularisation in 3 years versus freedom from pain. A patient would be required to be willing to tolerate a risk of 1.06 additional revascularisation procedures at 3 years in exchange for being pain free at 1 month, in order to be 95% confident that the risk is worth the benefit in choosing PCI over CABG. Reprinted from EuroIntervention, Federspiel et al, Risk-benefit trade-offs in revascularisation choices. Copyright 2009, with permission from Europa Edition.88


It would appear that a combination of clinical and anatomical variables is required for an effective, clinically useful risk model for patients. The SXscore, while prognostically useful in risk-stratifying patients proposing to undergo PCI, in itself appears to carry important information on clinical comorbidity and outcomes for the individual patient. However, this clearly is not the whole picture, and it is our view that the incremental value of adding clinical variables to the SXscore, as demonstrated with risk models such as Global Risk, will ultimately prove to be more clinically useful compared with the SXscore alone. Ongoing and future risk models may answer these questions. Novel concepts such as the Functional SXscore, performed invasively or non-invasively as discussed, and the patient-empowered risk–benefit trade-off are all further areas in current development where additional clinically relevant information may become available.


VF thanks the Dickinson Trust Travelling Scholarship, Manchester Royal Infirmary, Manchester, England, UK.



  • Competing interests None.

  • Provenance and peer review Commissioned; internally peer reviewed.