JACC: Cardiovascular Interventions
A Global Risk Approach to Identify Patients With Left Main or 3-Vessel Disease Who Could Safely and Efficaciously Be Treated With Percutaneous Coronary InterventionThe SYNTAX Trial at 3 Years
Author + information
- Received February 7, 2012
- Revision received March 16, 2012
- Accepted March 22, 2012
- Published online June 1, 2012.
Author Information
- Patrick W. Serruys, MD, PhD⁎,⁎ (p.w.j.c.serruys{at}erasmusmc.nl),
- Vasim Farooq, MBchB⁎,
- Pascal Vranckx, MD⁎,
- Chrysafios Girasis, MD⁎,
- Salvatore Brugaletta, MD⁎,
- Hector M. Garcia-Garcia, MD⁎,
- David R. Holmes Jr, MD†,
- Arie-Pieter Kappetein, MD, PhD‡,
- Michael J. Mack, MD§,
- Ted Feldman, MD∥,
- Marie-Claude Morice, MD¶,
- Elisabeth Ståhle, MD#,
- Stefan James, MD#,
- Antonio Colombo, MD⁎⁎,
- Peggy Pereda, MS††,
- Jian Huang, MD, MS††,
- Marie-Angèle Morel, BSc‡‡,
- Gerrit-Anne Van Es, PhD‡‡,
- Keith D. Dawkins, MD††,
- Friedrich W. Mohr, MD§§ and
- Ewout W. Steyerberg, PhD∥∥
- ↵⁎Reprint requests and correspondence:
Dr. Patrick W. Serruys, Department of Interventional Cardiology, Erasmus Medical Center, ‘s-Gravendijkwal 230, 3015 CE, Rotterdam, the Netherlands
Abstract
Objectives The aim of this study was to assess the additional value of the Global Risk—a combination of the SYNTAX Score (SXscore) and additive EuroSCORE—in the identification of a low-risk population, who could safely and efficaciously be treated with coronary artery bypass graft surgery (CABG) or percutaneous coronary intervention (PCI).
Background PCI is increasingly acceptable in appropriately selected patients with left main stem or 3-vessel coronary artery disease.
Methods Within the SYNTAX Trial (Synergy between PCI with TAXUS and Cardiac Surgery Trial), all-cause death and major adverse cardiac and cerebrovascular events (MACCE) were analyzed at 36 months in low (GRCLOW) to high Global Risk groups, with Kaplan-Meier, log-rank, and Cox regression analyses.
Results Within the randomized left main stem population (n = 701), comparisons between GRCLOW groups demonstrated a significantly lower mortality with PCI compared with CABG (CABG: 7.5%, PCI: 1.2%, hazard ratio [HR]: 0.16, 95% confidence interval [CI]: 0.03 to 0.70, p = 0.0054) and a trend toward reduced MACCE (CABG: 23.1%, PCI: 15.8%, HR: 0.64, 95% CI: 0.39 to 1.07, p = 0.088). Similar analyses within the randomized 3-vessel disease population (n = 1,088) demonstrated no statistically significant differences in mortality (CABG: 5.2%, PCI: 5.8%, HR: 1.14, 95% CI: 0.57 to 2.30, p = 0.71) or MACCE (CABG: 19.0%, PCI: 24.7%, HR: 1.35, 95% CI: 0.95 to 1.92, p = 0.10). Risk-model performance and reclassification analyses demonstrated that the EuroSCORE—with the added incremental benefit of the SXscore to form the Global Risk—enhanced the risk stratification of all PCI patients.
Conclusions In comparison with the SXscore, the Global Risk, with a simple treatment algorithm, substantially enhances the identification of low-risk patients who could safely and efficaciously be treated with CABG or PCI.
The SYNTAX score (SXscore) (1–4) has established itself as an important tool in the SYNTAX trial (The Synergy between PCI with TAXUS and Cardiac Surgery Trial) pioneered Heart Team approach, in which the cardiac surgeon and interventional cardiologist determined the optimal revascularization modality for patients with untreated left main stem (LMS) or 3-vessel (3VD) coronary artery disease (1,5–7). The SXscore has since been validated in the LMS percutaneous coronary intervention (PCI) population at short- and long-term follow-up (8–11); the 3VD PCI population at short-term (1-year) follow-up (12,13); and “All-Comers” patients undergoing PCI in contemporary stent trials at 1-year follow-up (14,15). In addition, both the current U.S. and European Guidelines on myocardial revascularization (16–18) advocate the use of the SXscore to determine the optimal revascularization modality in patients with unprotected left main or complex coronary disease, without the explicit use of clinical variables. Furthermore, the U.S. guidelines now gives surgical revascularization for unprotected left main coronary disease a Class 1B recommendation (16,17), compared with a Class 1A recommendation in previous guidelines (19).
Because the SXscore relies on the scoring of the coronary anatomy in isolation to objectively select the appropriate revascularization strategy for the individual patient, criticism has emerged, due to potentially important prognostic information being missing secondary to the absence of clinical factors (20,21). The EuroSCORE (22,23), a cardiac surgery-based clinical risk score, has been shown to have a reliable impact on prognosis and to be an independent predictor of major adverse cardiac events and mortality in surgical and percutaneously treated patients in the SYNTAX trial, thus confirming findings from previous studies (24–28).
Attempts have previously been made to combine the SXscore and additive EuroSCORE to form the Global Risk classification (29,30) and have been shown to potentially improve the risk stratification of patients undergoing surgical or percutaneous LMS intervention, compared with the SXscore alone. The Global Risk categorical approach reported in the present study, adapted from the previously reported Global Risk classification (29,30), uses the historically accepted cutoff levels for tertiles of the additive EuroSCORE (22,23)—of which a high EuroSCORE tertile has previously been shown to be an independent predictor of adverse outcomes after PCI (24,26–28)—and the now broadly accepted tertiles of the SXscore (Fig. 1) (1–4). Therefore the goal of the Global Risk is to improve in the identification of low-risk groups with LMS or 3VD, compared with the SXscore, who would achieve comparable surgical and percutaneous outcomes in terms of efficacy and safety at 3 years.
The Proposed Global Risk
The Global Risk demonstrating recategorization of established tertiles of risk for the SYNTAX Score and additive EuroSCORE to form low (GRCLOW), intermediate (GRCINT), and high (GRCHIGH) Global Risk groups. Highlighted (dashed) boxes indicate patients moved (reclassified) into or out of the low SYNTAX Score group to form the low Global Risk group (GRCLOW).
Adapted from Capodanno et al. (29,30).
Methods
The SYNTAX trial is a randomized, prospective, multicenter trial that incorporated an “All-Comers” design and consisted of pre-specified LMS (isolated or associated with 1-, 2-, or 3-vessel disease) and 3VD cohorts (1,6,7). Patients were randomized on a 1:1 basis to undergo either coronary artery bypass graft surgery (CABG) or PCI or placed in nested registries when considered unsuitable for randomization by the Heart Team (CABG nested registry for PCI-ineligible patients and PCI nested registry for CABG-ineligible patients). Exclusions were only limited to patients with prior coronary revascularization, the requirement of concomitant cardiac surgery, or ongoing acute myocardial infarction (MI). Recent MI with resolution of cardiac enzymes (<2× upper limit of normal) and unstable angina (no elevation in biomarkers) were not exclusion criteria.
The PCI techniques between the randomized and nested registry populations were similar except that TAXUS Express (Boston Scientific Corporation, Natick, Massachusetts) paclitaxel-eluting stents were only permitted in the randomized population. Within the PCI nested registry the implantation of any type of drug-eluting stent or bare-metal stent (BMS) was permitted, although the use of TAXUS paclitaxel-eluting stents was encouraged. Of the 589 stents implanted within the PCI nested registry, 57% were TAXUS Express or Liberté (Boston Scientific Corporation), 19% were another drug-eluting stent, and 24% were BMS; 4 PCI nested registry patients did not receive a stent. Conversely CABG techniques between the randomized and nested registry were broadly similar, except that double left and right internal mammary artery grafts were more frequently performed in the randomized (27.6%) CABG population, compared with the CABG nested registry (16.1%).
The Global Risk was calculated by combining the SXscore (2–4)—calculated by an independent core laboratory (Cardialysis BV, Rotterdam, the Netherlands)—and the additive EuroSCORE, as assessed by the Heart Team before randomization (22,23). The additive EuroSCORE was used to form the Global Risk, because this was shown to be more predictive of clinical outcomes in the PCI population compared with the logistic EuroSCORE, findings consistent with a previous study (29). Patients were subsequently categorized into 3 classes of Global Risk, on the basis of predefined anatomical and clinical-risk categories (low, intermediate, and high) for the SXscore (2–4) and additive EuroSCORE, respectively (22,23) (Fig. 1).
Statistical methods
Comparisons of all-cause death and major adverse cardiac and cerebrovascular events (MACCE) (a composite of all-cause death, MI, stroke, and all-cause revascularization) (1,7) were performed (Kaplan-Meier curves) with the log-rank test and Cox proportional hazard ratios between the low Global Risk (GRCLOW) groups for CABG and PCI. This prognostic model-based approach to identify a low subgroup of patients defined by prognostic factors is in contrast to the traditional approach to subgroup analysis (31–33). A logistic regression model that incorporated the SXscore and EuroSCORE as covariates was used to give each score proper relative weighting to each other within the Global Risk. This analysis allowed for the assessment of the predictive ability of the Global Risk: 1) the Hosmer–Lemeshow test for calibration—the assessment of the correctness of the prediction by the risk model, with poor fit indicated by a significant p value (<0.05); 2) receiver operator curves for discrimination (C-statistic)—the ability of the risk model to appropriately assign the correct risk prediction in patients who have the outcome, ranging from 0.50 (no discrimination) to 1.0 (perfect discrimination); 3) the Brier score—an overall risk model performance measure capturing both discrimination and calibration aspects of the risk model, ranging from 0 to 1, with a lower value (closer to 0) suggestive of a more predictive risk model (34–36). Comparisons were made with other risk models, namely, the SXscore (2,4); age, creatinine, and ejection fraction/modified age, creatinine, and ejection fraction scores (37,38); the Clinical SXscore (39,40); and the additive/logistic EuroSCOREs (22,23)—a brief description of which is enclosed in the Online Appendix.
Analyses on those groups that were reclassified according to the Global Risk compared with the SXscore alone were undertaken, following the principles of Net Reclassification Improvement (38,41) (Fig. 1). These reclassification analyses were to test whether the low Global Risk group (GRCLOW) appropriately risk-stratified patients, compared with a low SXscore. Higher anatomical risk patients (i.e., intermediate SXscores with low-intermediate EuroSCOREs) would be appropriately reclassified as lower-risk (GRCLOW) if they had comparable (or more favorable) PCI outcomes, compared with CABG. Conversely lower anatomical risk patients with a high clinical comorbidity (i.e., low SXscore with a high EuroSCORE) would only be appropriately reclassified to a higher risk group (GRCINT) if they had more favorable surgical outcomes, compared with PCI. Further detailed methodology, including illustrative figures describing the reclassification concepts, is included in the Online Appendix. A 2-sided p value <0.05 was considered significant for all tests. All analyses were conducted with SAS System Software (version 8.0 or higher, SAS Institute, Cary, North Carolina) and SPSS (version 17.0, SPSS, Inc., Chicago, Illinois).
Results
All randomized patients underwent planned follow-up. Within the nested registries, all PCI patients underwent planned follow-up, and 649 of the 1,077 CABG patients were randomly allocated for follow-up, on the basis of the original study protocol (1,7). Complete data, including clinical outcomes relating to the Global Risk, were available in 1,789 of 1,800 randomized patients (PCI, n = 899; CABG, n = 890), and 2,610 of 3,075 “All-Comers” patients (PCI, n = 1,088; CABG, n = 1,522) at 3-year follow-up.
Within the randomized SYNTAX population baseline demographic data and clinical characteristics for the treatment arms have previously been described and were well-balanced (1). Within the “All-Comers” population more complex coronary anatomy was present in the CABG population (mean SXscore ± 1 SD: CABG: 32.7 ± 12.9, PCI: 28.9 ± 11.7, p < 0.001) (Table 1). Conversely, significantly more comorbidity was present in the “All-Comers” PCI population, as evidenced by significantly greater Parsonnet (42) and EuroSCOREs (Table 1).
Comparison of Anatomical Factors (SXscore) and Clinical Risk Scores for the CABG and PCI Populations
Clinical outcomes with PCI
Within the randomized and “All-Comers” LMS (Fig. 2) and 3VD (Fig. 3) PCI populations, a low Global Risk group (GRCLOW) could be differentiated from the higher risk groups (GRCINT-HIGH) for All-Cause death and MACCE. Furthermore, within the LMS PCI population, the Global Risk demonstrated a clear incremental increase in predictive ability (C-statistics and overall risk model performance measures), compared with either the SXscore or the EuroSCORE in isolation (Fig. 4). Within the randomized and “All-Comers” 3VD PCI population, the additive EuroSCORE had a superior predictive ability for all-cause death and MACCE compared with the SXscore alone, with little or no additional improvements in the predictive ability of the Global Risk compared with the additive EuroSCORE (Fig. 4).
Comparisons of Clinical Outcomes by Categories of Global Risk in the SYNTAX LMS Population
Kaplan-Meier curves for cumulative rate of death (A) and MACCE (B) at 3-year follow-up stratified according to the Global Risk group; event rate ± 1.5 SE. CABG = coronary arty bypass grafting; GRCLOW = Low Global Risk; GRCINT = Intermediate Global Risk; GRCHIGH = High Global Risk; LMS = left main stem; MACCE = major adverse cardiac and cerebrovascular event(s); PCI = percutaneous coronary intervention.
Comparisons of Clinical Outcomes by Categories of Global Risk in the SYNTAX 3VD Population
Kaplan-Meier curves for cumulative rate of death (A) and MACCE (B), at 3-year follow-up stratified according to the Global Risk group; event rate ± 1.5 SE. 3VD = 3-vessel disease; other abbreviations as in Figure 2.
Comparison of Different Risk Models for the Randomized LMS and 3VD Populations for Death and MACCE at 3 Years
Comparison of different risk models for the randomized LMS and 3VD populations for death (A) and MACCE (B) at 3 years, with calibration (Hosmer–Lemeshow test), discrimination (C-statistic, y axes), and overall model performance measures (Brier score, x axes) (36). Yellow arrows in the LMS cohort demonstrate the incremental benefit, in terms of predictive ability, of the Global Risk compared with the additive EuroSCORE and SYNTAX Score as evidenced by a greater C-statistic and lower Brier score. Yellow arrows in the 3VD cohort demonstrate the incremental benefit of the additive EuroSCORE compared with the SYNTAX Score with little/no improvement of the predictive ability of the Global Risk compared with the additive EuroSCORE, as evidenced by comparable C-statistics and Brier scores. Note how even minor differences in the Brier score reflect overall improvements in the model performance and the different scales for the Brier scores for death and MACCE reflecting the findings that the risk models are superior in predicting death. ACEF = age, creatinine, and ejection fraction (score); CSS = clinical SYNTAX score; other abbreviations as in Figures 2 and 3.
Clinical outcomes with CABG
At 36 months the Global Risk could differentiate between the GRCINT-HIGH groups only for all-cause death and MACCE in the randomized and “All-Comers” CABG populations (Figs. 2 and 3). The Global Risk added little or no improvement to the predictive ability, compared with the additive EuroSCORE used in isolation (Fig. 4).
Comparison of CABG and PCI: low-risk LMS population
Within the GRCLOW group of the randomized LMS population (n = 701), CABG resulted in significantly higher 3-year mortality compared with PCI (CABG: 7.5%, PCI: 1.2%, HR: 0.16, 95% CI: 0.03 to 0.70, p = 0.0054), with a trend toward a lower incidence of MACCE (CABG: 23.1%, PCI: 15.8%, HR: 0.64, 95% CI: 0.39 to 1.07, p = 0.088) and stroke (CABG: 3.5%, PCI: 0.6%, HR: 0.17, 95% CI: 0.02 to 1.46, p = 0.067). No statistically significant differences in MI or all-cause revascularization were found.
Within the GRCLOW group of the “All-Comers” LMS population (n = 1,079), no statistically significant differences in mortality (CABG: 5.3%, PCI: 2.7%, HR: 0.51, 95% CI: 0.18 to 1.44, p = 0.19) or MACCE (CABG: 18.0%, PCI: 18.5%, HR: 1.02, 95% CI: 0.65 to 1.60, p = 0.94) were observed. A significantly greater incidence of stroke was evident with CABG (CABG: 4.0%, PCI: 0.6%, HR: 0.13, 95% CI: 0.02 to 1.05, p = 0.025), and a significantly greater frequency of MI was evident with PCI (CABG: 0.9%, PCI: 3.9%, HR: 4.30, 95% CI: 0.89 to 20.70, p = 0.047). No statistically significant differences in all-cause revascularization were seen (CABG: 10.7%, PCI: 14.8%, HR: 1.40, 95% CI: 0.81 to 2.42, p = 0.23).
Comparison of CABG and PCI: low-risk 3VD population
Within the GRCLOW group of the randomized 3VD population (n = 1,088), no statistically significant differences in 3-year mortality (CABG: 5.2%, PCI: 5.8%, HR: 1.14, 95% CI: 0.57 to 2.30, p = 0.71) or MACCE (CABG: 19.0%, PCI: 24.7%, HR: 1.35, 95% CI: 0.95 to 1.92, p = 0.10) were observed. Percutaneous coronary intervention was associated with a significantly increased risk of all-cause revascularization (CABG: 10.5%, PCI: 18.5%, HR: 1.88, 95% CI: 1.19 to 2.96, p = 0.0055). No statistically significant differences in the risk of stroke were seen (CABG: 2.9%, PCI: 1.3%, HR: 0.45, 95% CI: 0.13 to 1.49, p = 0.18).
Within the “All-Comers” 3VD population (n = 1,531) no statistically significant differences in mortality were observed (CABG: 5.1%, PCI: 5.9%, HR: 1.16, 95% CI: 0.62 to 2.17, p = 0.65). A significantly higher incidence of MACCE was evident with PCI (CABG: 17.9%, PCI: 24.4%, HR: 1.42, 95% CI: 1.03 to 1.96, p = 0.031) secondary to predominantly greater all-cause revascularization (CABG: 9.1%, PCI: 18.5%, HR: 2.20, 95% CI: 1.44 to 3.35, p = 0.0002). No statistically significant differences in the incidences of stroke were seen (CABG: 3.2%, PCI: 1.5%, HR: 0.46, 95% CI: 0.16 to 1.30, p = 0.13).
Analyses of reclassified patients
Within the LMS population, patients (i.e., patients with an intermediate SXscore and low-moderate EuroSCOREs) were appropriately reclassified to the GRCLOW group (Fig. 1). More favorable outcomes were seen with PCI, compared with CABG (randomized population: 3-year all-cause death: CABG 10.8%, PCI 1.3%; 3-year MACCE: CABG 24.4%, PCI 15.6%). Conversely, patients reclassified to a higher-risk (GRCINT) group (i.e., patients with a low SXscore and a high EuroSCORE) had more favorable surgical outcomes in the larger “All-Comers” population (Fig. 1), predominantly secondary to reduced MACCE with CABG (3-year MACCE: CABG 20.4%, PCI 27.0%).
Within the 3VD population more favorable surgical outcomes were evident in patients reclassified to the GRCLOW group. Further analyses indicated that a GRCLOW with an intermediate SXscore would remain better-managed by CABG, and a GRCLOW with a low SXscore would have comparable surgical and PCI outcomes. More favorable surgical outcomes were evident in patients (i.e., with a low SXscore and a high EuroSCORE) reclassified to the higher-risk (GRCINT) group (randomized population: 3-year death: CABG 9.6%, PCI 21.3%; 3-year MACCE: CABG 22.5%, PCI 39.1%). Further detailed results of the reclassification analyses are included in the Online Appendix.
Discussion
The main findings of this study are: 1) clinical variables (EuroSCORE) per se are more predictive of clinical outcomes (all-cause death and MACCE), compared with anatomical variables (SXscore) in the PCI population; 2) within the LMS PCI population the Global Risk demonstrated a clear incremental benefit in its predictive ability, compared with the SXscore or EuroSCORE used in isolation; 3) within the 3VD PCI population, the Global Risk improved the risk stratification of patients, compared with the SXscore alone, by demonstrating that low SXscore patients with a high EuroSCORE to attain a mortality benefit in undergoing CABG compared to PCI; and 4) that the Global Risk substantially enhanced the identification of a low-risk population who could safely and efficaciously be treated with CABG or PCI at 3 years.
The application of the Global Risk to the SYNTAX population was complicated by the differing prognostic and morbidity outcomes between the LMS and 3VD populations. However, a low-risk population was identified with outcomes comparable to CABG and PCI at 3 years in terms of efficacy and safety, namely a GRCLOW group in the LMS population and a GRCLOW group with a low SXscore in the 3VD population. Ultimately reclassification analyses proved vital in ensuring the optimal revascularization modality in specific groups of patients. For example, high-EuroSCORE patients with a low SXscore were shown to confer a clear mortality benefit from undergoing CABG in the 3VD population, and intermediate-SXscore patients with low-moderate EuroSCOREs were shown to confer a potential survival advantage in undergoing PCI in the LMS population. On the basis of these findings a treatment algorithm is proposed to simplify these concepts, which admittedly will require further validation in other unselected registries (Fig. 5).
Proposed Treatment Algorithm for the Management of LMS and 3VD Incorporating Clinical and Anatomical Variables
Proposed treatment algorithm for the management of LMS and 3VD incorporating clinical (additive EuroSCORE) and anatomical (SYNTAX Score) variables. Although not explicitly stated PCI might be the preferred revascularization modality in low-risk patients subject to the Heart Team discussion. *If an acceptable threshold of operative risk is exceeded for the patient and cardiac surgeon, consideration of PCI should be considered—appropriate discussion concerning risk stratification should be undertaken. Abbreviations as in Figures 2 and 3.
The main strengths of the Global Risk are that the additive EuroSCORE is a simple bedside calculation and that the Global Risk can be applied across the entire spectrum of surgical and percutaneously treated patients. Within the “All-Comers” SYNTAX population the adoption of this treatment algorithm (Fig. 5) would potentially identify a smaller population of patients, compared with using the SXscore alone, who would have similar outcomes to CABG and PCI at 3 years in terms of efficacy and safety—namely, 39% of the LMS population (compared with 51% with low-moderate SXscores) and 21% of the 3VD population (compared with 26% with a low SXscore).
By identifying a low Global Risk (GRCLOW) group within the randomized LMS population, a significant mortality benefit and trend toward a reduction in MACCE was evident for PCI at 3 years. This was not apparent in the “All-Comers” population. Several factors might explain this disparity in results. First, the “All-Comers” PCI population had significantly greater comorbidity (Table 1)—factors well known to be associated with in-hospital and long-term adverse outcomes after PCI (24,26–28). Second, a low EuroSCORE could not exclude nonadjustable characteristics, such as the judgment of the treating clinician in declining a patient for CABG and thus undergoing PCI instead, such as patients with a short-term survival—as were recruited within the “All-Comers” SYNTAX trial. Third, the CABG nested registry might potentially have been diluted with a proportion of lower-anatomical-risk patients who might have been suitable for PCI, because the concept of the clinical outcomes on the basis of tertiles of the SXscore were unknown at the time of the SYNTAX trial. Fourth, is the use of BMS in the nested PCI registries: although specific data pertaining to the indications for use of BMS were not collected, it is probable that a sizeable proportion of these patients might have had comorbidities that precluded the use of prolonged dual antiplatelet therapy. The adoption of the “All-Comers” approach is nonetheless more likely to mirror contemporary clinical practice and is perceived by many as the recommended approach (18,43).
Adverse clinical comorbidity: CABG or PCI?
In both the LMS and 3VD populations the surgical benefit in the higher comorbidity patients (i.e., with a low SXscore and high EuroSCORE) was more pronounced, compared with PCI. In particular a substantial mortality benefit favoring CABG was evident in low-SXscore patients with a high EuroSCORE in the 3VD population. Although it has been previously demonstrated that a high EuroSCORE is a potential predictor of adverse outcomes after PCI (24,26–28), conventional clinical practice has suggested that high-comorbidity patients might be more suitable for PCI compared with CABG. In contrast to this accepted clinical practice, it seems that it is precisely these types of patients who would potentially stand to gain more from CABG on prognostic and morbidity grounds. These findings are probably related to PCI treating the individual lesion, whereas CABG would potentially protect the entire treated vessel from future cardiac events for the lifespan of the graft.
There is nevertheless a recognition that a certain threshold of operative risk would have to be acceptable for the cardiac surgeon and patient, and the latter of who may adamantly refuse a surgical approach due to the anticipated prohibitive risk of the proposed surgical intervention. It should, however, be recognized that PCI may potentially be more hazardous for the patient with regard to long-term outcomes, compared with CABG. A greater understanding of this phenomenon may potentially reduce the threshold value for which surgical revascularization is declined during the Heart Team discussion.
Anatomical and clinical variables in the LMS PCI population
Before the introduction of the SXscore, heterogeneity of the LMS anatomical description and its impact on clinical outcomes were previously recognized by the use of classical terminology describing isolated LMS or LMS + 1-, 2-, or 3-vessel disease. The SXscore was an attempt to eliminate the historical and arbitrarily defined subdivisions of the coronary tree into LMS and 3VD and to create a common anatomical denominator (2,3). It now seems that the LMS outcomes are largely a reflection of the presence of distal LMS bifurcation disease, clinical comorbidity, and importantly, the increasing prevalence of 3VD and its association with clinical comorbidities (as discussed in the following) and anatomical complexities, such as multiple bifurcations and the presence of total occlusions leading to higher SXscores.
Anatomical and clinical variables in the 3VD PCI population
Within the 3VD PCI population the SXscore added very little incremental benefit to the additive EuroSCORE in predicting death and MACCE (Fig. 4). It might be hypothesized that the severity of 3VD (as evidenced by a higher SXscore) might be representative of patients with a more adverse risk profile who have evidence of systemic atherosclerosis and therefore are at greater longer-term cardiovascular and cerebrovascular risk. Consequently these patients might potentially benefit from CABG on prognostic and morbidity grounds due to the bypass grafts protecting the coronary vessel as discussed. This might also be an explanation for the comparability in long-term stroke outcomes between CABG and PCI for the 3VD SYNTAX population (6).
This hypothesis is supported by the significant and direct relationship of the 10-year predicted Framingham risk scores with the prevalence and magnitude of coronary artery calcium scores (44). Furthermore, the ankle-brachial pressure index (45–49) and common carotid intima-media thickness (50–53), both markers of peripheral vascular disease, have been correlated with the severity of coronary artery disease and clinical events.
Implications of the Global Risk for future trials
Given the heterogeneity of the outcomes between LMS and 3VD, the strategy for the future is to construct separate trials for the LMS and 3VD populations that should incorporate anatomical and clinical variables. Left main stem disease is currently subject to the ongoing EXCEL (Evaluation of Xience Prime or Xience V versus Coronary Artery Bypass Surgery for Effectiveness of Left Main Revascularization) trial, recruiting patients on anatomical entry criteria only—namely low-moderate SXscores—and does not directly take into account clinical variables except within the Heart Team discussion.
Within the 3VD population perhaps a more targeted identification of patients by other markers of atherosclerotic burden, such as ankle-brachial pressure index and common carotid intima-media thickness as discussed, in conjunction with the SXscore or Global Risk might prove beneficial.Another recently described approach is the “Functional SYNTAX Score”—utilizing the functional assessment of coronary lesions—and has potentially been shown to improve the identification of low and higher risk patients (54). In addition, the potential noninvasive calculation of the Functional SYNTAX Score, with computational fluid dynamics applied to coronary computed tomography angiography, has shown significant promise (55,56).
Summary of the potential clinical implications of the Global Risk
The practical clinical application of the Global Risk is summarized in a treatment algorithm in Figure 5. Although validation of the Global Risk concept is required, the practicalities are that not only high-anatomical-risk patients but also patients with significant comorbidity are best served by undergoing surgical revascularization, because they seem to be one of the patient groups that stand to potentially gain more from surgical revascularization on prognostic and morbidity grounds, particularly if they have 3VD. Clearly a threshold of operative risk for surgical revascularization should not be exceeded; therefore these issues are vital in the Heart Team discussion in selecting the most appropriate revascularization modality.
Study limitations
This study represents a post hoc analysis of the original SYNTAX Trial, and the predictive models were developed retrospectively at 3-year follow-up. The further analyses undertaken in the LMS and 3VD cohorts (1,7) should be considered as hypothesis-generating. Furthermore, the focus of this study was on low Global risk (GRCLOW) groups, because CABG is the standard of care in the management of patients in the higher-Global Risk groups (1,5,6). Comparisons between the higher-Global Risk groups are nevertheless provided in the Online Appendix. In addition there was limited statistical power for the comparison between CABG and PCI for events such as stroke and MI and analyses of reclassified patients. Consequently external validation of the proposed Global Risk is required in unselected registries, with numbers greater than the SYNTAX “All-Comers” population. The “All-Comers” concept of the SYNTAX trial, although more representative of contemporary clinical practice compared with the randomized approach (18,43), has been reported to potentially not result in the inclusion of consecutive patients, predominantly due to the inability to gain appropriate informed consent and refusal to participate (57).
It is not possible to judge and account for the decisions made by the Heart Team in selecting a patient for randomization. However, this approach is representative of contemporary practice. It should also be acknowledged that, although the SYNTAX Trial was based on contemporary revascularization practice at the time, improvements in technology in both CABG and PCI might yield differences in clinical outcomes in future trials. Noninvasive or invasive carotid imaging to screen for the presence of significant carotid disease was undertaken by the clinical consensus of the Heart Team to calculate the EuroSCORE. The possibility of a small minority of patients with clinically silent carotid disease cannot be excluded. The cardiac-related comorbidities within the EuroSCORE are more likely to reflect outcomes after PCI, whereas extracardiac factors (e.g., the presence of chronic obstructive pulmonary disease and poor neurological status) are unlikely to reflect outcomes (24–28)—this point should be borne in mind when interpreting the Global Risk. Furthermore, the use of the EuroSCORE with a continuous approach might have affected the results of the analysis. However, the categorical approach was adopted from the outset to allow application of the same risk model in CABG and PCI patients, given that a high SXscore tertile has consistently been shown to be an independent predictor of adverse outcomes after PCI (24,26–28). The newly developed EuroSCORE II (58) cannot be applied to the concept of the Global Risk, because this information was not collected during the original SYNTAX trial.
Conclusions
In comparison with the SXscore, the Global Risk—with a simplified treatment algorithm—substantially enhances the identification of low-risk patients who could safely and efficaciously be treated with CABG or PCI.
Appendix
For supplementary figures, tables, text, and references, please see the online version of this article.
Appendix
A Global Risk Approach to Identify Patients With 3-Vessel and/or Left Main Stem Disease Who Could Safely and Efficaciously Be Treated With Percutaneous Coronary Intervention
[S1936879812003664_mmc1.docx]Footnotes
The SYNTAX study was funded by Boston Scientific. Dr. Mack has served on the Speaker's Bureau of Boston Scientific, Cordis, and Medtronic. Dr. Feldman serves on the Speaker's Bureau of Boston Scientific; receives grant support from Abbott, Atritech, Boston Scientific, Edwards, and Evalve; and consults for Abbott, Coherex, Intervalve, Square One, and W.L. Gore. Dr. Morice reports that her institution received a research grant from Boston Scientific. Dr. Dawkins, Ms. Pereda, and Dr. Huang are all full-time employees of Boston Scientific. Dr. Dawkins holds stock in Boston Scientific. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. Stephen G. Ellis, MD, served as Guest Editor for this paper. Drs. Serruys and Farooq contributed equally to this paper.
- Abbreviations and Acronyms
- 3VD
- 3-vessel disease
- BMS
- bare-metal stent(s)
- CABG
- coronary artery bypass graft surgery
- CI
- confidence interval
- GRCLOW
- Low Global Risk
- GRCINT
- Intermediate Global Risk
- GRCHIGH
- High Global Risk
- HR
- hazard ratio
- LMS
- left main stem
- MACCE
- major adverse cardiac and cerebrovascular event(s)
- MI
- myocardial infarction
- PCI
- percutaneous coronary intervention
- SXscore
- SYNTAX Score
- Received February 7, 2012.
- Revision received March 16, 2012.
- Accepted March 22, 2012.
- American College of Cardiology Foundation
References
- ↵
- ↵
- SYNTAX score calculator: SYNTAX working-group. http://www.syntaxscore.com. Accessed April 1, 2012.
- Morice M.C.,
- Serruys P.W.,
- Kappetein A.P.,
- et al.
- ↵
- Kappetein A.P.,
- Feldman T.E.,
- Mack M.J.,
- et al.
- ↵
- Park D.W.,
- Kim Y.H.,
- Yun S.C.,
- et al.
- Capodanno D.,
- Capranzano P.,
- Di Salvo M.E.,
- et al.
- ↵
- Valgimigli M.,
- Serruys P.W.,
- Tsuchida K.,
- et al.
- ↵
- Garg S.,
- Sarno G.,
- Girasis C.,
- et al.
- Palmerini T.,
- Genereux P.,
- Caixeta A.,
- et al.
- ↵
- Levine G.N.,
- Bates E.R.,
- Blankenship J.C.,
- et al.
- Hillis L.D.,
- Smith P.K.,
- Anderson J.L.,
- et al.
- ↵
- Wijns W.,
- Kolh P.,
- Danchin N.,
- et al.
- ↵
- Kushner F.G.,
- Hand M.,
- Smith S.C. Jr..,
- et al.
- ↵
- Applegate R.J.
- Serruys PW. Integrating lessons from recent clinical trials into practice: beyond the SYNTAX score. Presented at: Transcatheter Therapeutics. Asia Pacific, April 29, 2010; Seoul, South Korea.
- ↵
- Nashef S.A.,
- Roques F.,
- Michel P.,
- Gauducheau E.,
- Lemeshow S.,
- Salamon R.
- ↵
- Rodés-Cabau J.,
- Deblois J.,
- Bertrand O.F.,
- et al.
- Romagnoli E.,
- Burzotta F.,
- Trani C.,
- et al.
- Kim Y.-H.,
- Park D.-W.,
- Kim W.-J.,
- et al.
- ↵
- Capodanno D.,
- Caggegi A.,
- Miano M.,
- et al.
- ↵
- ↵
- Moons K.G.,
- Altman D.G.,
- Vergouwe Y.,
- Royston P.
- Cook N.R.
- ↵
- ↵
- Ranucci M.,
- Castelvecchio S.,
- Menicanti L.,
- Frigiola A.,
- Pelissero G.
- ↵
- ↵
- ↵
- ↵
- Okwuosa T.M.,
- Greenland P.,
- Ning H.,
- et al.
- ↵
- ↵
- Adams M.R.,
- Nakagomi A.,
- Keech A.,
- et al.
- Steinvil A.,
- Sadeh B.,
- Arbel Y.,
- et al.
- Ikeda N.,
- Kogame N.,
- Iijima R.,
- Nakamura M.,
- Sugi K.
- ↵
- Nam C.W.,
- Mangiacapra F.,
- Entjes R.,
- et al.
- ↵
- Koo B.K.,
- Erglis A.,
- Doh J.H.,
- et al.
- Farooq V.,
- Brugaletta S.,
- Serruys P.W.
- ↵
- de Boer S.P.,
- Lenzen M.J.,
- Oemrawsingh R.M.,
- et al.
- ↵EuroSCORE, II. http://www.euroscore.org/calc.html. Presented at the 25th EACTS Annual Meeting, 3rd October 2011, Lisbon, Portugal Accessed April 1, 2012.
Toolbox
Citation Manager Formats
Figures in Article
Tables
Supplementary Materials
- Online Appendix
A Global Risk Approach to Identify Patients With 3-Vessel and/or Left Main Stem Disease Who Could Safely and Efficaciously Be Treated With Percutaneous Coronary Intervention
[S1936879812003664_mmc1.docx]