Author + information
- Received December 29, 2009
- Revision received April 9, 2010
- Accepted April 15, 2010
- Published online June 1, 2010.
- Shao-Liang Chen, MD⁎,⁎ (, )
- Jack P. Chen, MD#,
- Gary Mintz, MD⁎⁎,
- Bo Xu, MBBS†,
- Jing Kan, MD⁎,
- Fei Ye, MD⁎,
- Junjie Zhang, MD⁎,
- Xuewen Sun, MD‡,
- Yawei Xu, MD∥,
- Qing Jiang, MD¶,
- Aiping Zhang, MD§ and
- Gregg W. Stone, MD⁎⁎
- ↵⁎Reprint requests and correspondence:
Dr. Shao-Liang Chen, Nanjing First Hospital, Nanjing Medical University, Nanjing 210006, China
Objectives This study aimed to compare the NERS (New Risk Stratification) and SYNTAX (Synergy between Percutaneous Coronary Intervention with Taxus and Cardiac Surgery) scores for prognostication after stenting of unprotected left main stenosis in a ”real-world” setting.
Background In contrast to existing systems, the NERS score encompasses clinical, procedural, and angiographic characteristics.
Methods The NERS score was derived from 260 patients with unprotected left main stenosis who underwent percutaneous coronary intervention and tested in 337 patients in a consecutive left main registry (66.55 ± 10.49 years, 78.9% men) undergoing percutaneous coronary intervention in a prospective, multicenter trial. Six-month clinical and angiographic follow-up was obtained in 100% and 88.9% of patients, respectively. The primary end point was major adverse cardiac events (MACE), encompassing myocardial infarction, all-cause death, and target vessel revascularization. Receiver-operator characteristic (ROC) curve was generated for the comparison of NERS versus SYNTAX scores.
Results The NERS score consisted of 54 variables (17 clinical, 4 procedural, and 33 angiographic). A NERS score ≥25 demonstrated a sensitivity and specificity of 92.0% and 74.1% (MACE as state variable), respectively, significantly higher than SYNTAX intermediate risk (20.5% and 25.4%) or SYNTAX higher risk (70.5% and 35.2%, p for all <0.001). At follow-up, myocardial infarction, cardiac death, and target vessel revascularization occurred in 3.0%, 5.6%, and 13.1% of patients, respectively, for a composite MACE of 26.0%. A NERS score ≥25 (hazard ratio: 1.13; 95% confidence interval [CI]: 1.11 to 1.16; p < 0.001) was the only independent predictor of cumulative MACE and stent thrombosis at follow-up (odds ratio: 31.04; 95% CI: 19.36 to 67.07; p < 0.001).
Conclusions The NERS score was more predictive of MACE than the SYNTAX score was. Further study is needed to address their relative roles in assessment for appropriateness of coronary artery bypass graft versus percutaneous coronary intervention for unprotected left main coronary artery stenosis.
Percutaneous coronary intervention (PCI) of unprotected left main coronary artery (ULMCA) stenosis remains controversial during the drug-eluting stent (DES) era (1,2), with coronary artery bypass graft (CABG) surgery remaining the gold standard in many parts of the world. Although the presence of left main (LM) stenosis influences the complexity of PCI (3–7), clinical comorbidities likely also contribute to post-CABG clinical outcomes (8–13). Although pre-treatment assessments with the European system for cardiac operative risk evaluation (EuroSCORE) and Parsonnet score (12,13) are widely advocated, their complex calculation and exclusion of procedural or lesion characteristics potentially limit their applicability in everyday practice.
Furthermore, the SYNTAX (Synergy between Percutaneous Coronary Intervention with Taxus and Cardiac Surgery) score (14) for prediction of clinical events after CABG or PCI in UPLMS does not comprehensively incorporate clinical variables. Thus, the potential contributions of clinical, procedural, and angiographic indices in this patient cohort have not been fully elucidated. We hereby compare a novel scoring system, the NERS (New Risk Stratification) score, with the SYNTAX score in patients undergoing PCI for UPLMS.
Study design and patient population
The NERS prospective registry involved UPLMS PCI from 4 centers in China from January 21, 1999, through March 1, 2009. Prior to February 2004, only bare-metal stents (BMS) were available, and implantation of DES commenced subsequently. The study protocol was approved by the Ethics Committee of each participating center by November 18, 1998.
The decision on revascularization strategy for each patient with UPLMS was jointly formulated by cardiologists and cardiac surgeons, based upon clinical and angiographic features. The UPLMS patients undergoing stent implantation were consecutively enrolled, and all data were entered into a dedicated database.
Variables for the SYNTAX and NERS scoring systems
The SYNTAX score was generated and classified as lower (0 to 22), intermediate (23 to 32), higher (≥33), and high-risk (≥23) scores. By September 29, 2006, 260 UPLMS patients were analyzed and reported in the DISTAL (Drug-Eluting Stent for the Treatment of Left Main Disease) study (4), from which 126 variables were selected and analyzed by logistic regression. Probability score for each variable was then calculated. Variables with p value ≤0.05 were considered significant predictors and prospectively evaluated in the present study. The NERS scoring system, derived from these 260 UPLMS patients, consisted of 54 variables that are defined and listed in Tables 1 and 2.⇓ Of these, double-stent techniques (4) were subdivided into classical crush, culotte, double kissing (DK) crush, T, and V/simultaneous kissing stent (SKS). Predicted scores for overall and each double-stent technique were then calculated.
Patients (n = 260) in the DISTAL study were excluded from the present study. All clinical variables were recorded by a follow-up team and reviewed/adjudicated by an independent committee. Scoring of angiographic and procedural variables was performed by 2 technicians from China Cardiovascular Research Foundation (Beijing, China), who were blinded to the present study.
In the NERS scoring system, scores for the coronary artery tree were numerically recorded as 10. Angiographic variables included chronic total occlusion (CTO), true bifurcation, long lesion, proximal vessel <2.0 mm, severe calcification, tortuous, and thrombotic. Left main lesions were classified by ostial, body, distal, and whole stem. According to the criterion of Leaman et al. (15), the right coronary artery supplies 16%, and the LM supplies 84% (66% by left anterior descending artery and 34% by left circumflex artery) of the myocardium in right dominance. Conversely, for left dominance, the right coronary artery is not scored. For ostial and shaft or distal LM stenosis, the highest score for distal LM lesion was considered the final score for the LM. Similarly, tandem stenoses within 1 vessel, such as CTO with bifurcation, were scored as the highest score of any lesion (i.e., CTO).
The primary end point was the rate of major adverse cardiac events (MACE), defined as all-cause mortality, myocardial infarction (MI), or target vessel revascularization (TVR) at follow-up. Stent thrombosis (ST) was assessed as a secondary event.
Procedural and periprocedural medications
After written informed consent, all interventional procedures were performed according to current standard guidelines. Decisions regarding glycoprotein IIb/IIIa inhibitors, low molecular-weight heparin, stent type, pre-dilation, and intravascular ultrasound were at operator's discretion. All patients were advised to take aspirin (100 mg daily) lifelong. Prior to or subsequent to January 2004, respective thienopyridine therapies consisted of 3 months of ticlopidine (250 mg, twice daily for 1 month, then 250 mg/day for 2 months) or 12 months of clopidogrel (75 mg/day).
UPLMS was defined as flow-limiting LM lesion(s) in the absence of a patent bypass conduit to the left system. Lesion variables, including isolated UPLMS disease, UPLMS with 1- to 3-vessel disease (VD), bifurcation or trifurcation lesions, CTO, severe calcification, severe tortuosity, thrombus, dominance, lesion number, and diseased vessel number, were classified according to SYNTAX (14) definitions. Lesion classifications in the LM, downstream lesion, restenosis in the LM, complete revascularization, stent number, stent length, and stent diameter were calculated based on the DISTAL criteria (4).
Double-stent technique for distal UPLMS was defined as either intention-to-treat or intraprocedural crossover from a single-stent technique. All deaths were considered cardiac unless a noncardiac cause was established clinically or at autopsy. Myocardial infarction was diagnosed as plasma level of creatine kinase-myocardial band ≥3 times the upper normal limit. In-stent restenosis was defined as ≥50% diameter stenosis at follow-up; TVR was considered as repeat PCI or CABG involving the index vessel. Stent thrombosis was classified according to Academic Research Consortium definitions as definite, probable, or possible; and as early (0 to 30 days), late (31 to 360 days), or very late (>360 days) (16). Angiographic success was defined as residual stenosis <10% with TIMI (Thrombolysis In Myocardial Infarction) flow grade 3 in both branches. Procedural success referred to angiographic success in the absence of in-hospital MACE.
Phone interviews were conducted monthly for 12 months and at 6-month intervals thereafter; clinical assessment was performed for suspected ischemic symptomatology. Clinical events were adjudicated by an independent committee from 3 nonparticipating institutions; all interventionalists and committee members had free access to the database.
Quantitative coronary angiographic analyses
Repeat angiography at 6 months after index procedure was scheduled unless clinically indicated earlier. Quantitative coronary angiographic (QCA) analysis at baseline, post-stenting, and at follow-up were performed by China Cardiovascular Research Foundation using edge detection techniques (CAAS II, version 5.0, Pie Medical, Maastricht, the Netherlands). Binary restenosis was defined as luminal diameter stenosis ≥50%; angiographic measurements included the stented segment as well as margins 5 mm proximal and distal to stent edge.
Continuous variables were expressed as mean ± SD and were compared using Student t test. Categorical variables were presented as counts and percentages and were compared using chi-square or Fisher exact test. Survival curves were generated by the Kaplan-Meier method, and survival among groups was compared using the log-rank test. Cox proportional hazards models were used to assess risk factors for adverse events. Multivariable analysis, with all significant variables (defined as p <0.05 for the association), was performed to adjust for possible confounders and to identify independent predictors of adverse events. Sensitivity, specificity, false-positive and false-negative, and positive and negative predictive values were calculated and analyzed by crosstable after weighted cases. Receiver-operator characteristic (ROC) curves for NERS and SYNTAX scores were generated; area under the curve (AUC) for each scoring method and end point was compared to the null hypothesis true area (= 0.5), with sorting out of 1-specificity, sensitivity, 95% confidence interval (CI), and significance. The value corresponding to the highest accuracy (i.e., minimal false-negative and false-positive rates) was chosen as the optimal cutoff value. To testify the difference of AUC and standard error from ROC curve between 2 scoring systems by Delong, Delong method, all variables, including cumulative MACE, stent thrombosis, TVR for left main, MI, and cardiac death, were entered into Analyse-it (Analyse-it Software, Ltd., Hearne Scientific Software Pty Ltd., Melbourne, Australia). A p value <0.05 was considered statistically significant. All statistical tests were 2-tailed. Statistical analysis was performed using SPSS version 16.0 (SPSS Inc., Chicago, Illinois).
Baseline clinical characteristics
From January 21, 1999, to March 1, 2009, of 854 patients with UPLMS, 517 (260 formed the basis of DISTAL study, 257 underwent CABG) were excluded. The remaining 337 patients (age 66.58 ± 10.69 years, 78.9% men) treated by stent implantation were studied (Table 3). Sixty-four patients (19.0%) had decreased left ventricular ejection fraction (<40%), and emergency LM stenting was needed in 22 patients (6.5%).
Baseline angiographic and procedural characteristics
LM-CTO lesions were rare (2.1%); however, 23.1% had non–LM-CTO lesions (Table 4). Most UPLMS (83.7%) were distal, with requirement of double-stent techniques in 122 (36.2%), including 41 DK crush (33.6%), 32 classical crush (26.2%), 21 culotte (17.2%), 20 T (16.4%), and 8 V/SKS (6.6%). Downstream lesions were seen in most patients (75.4%), with 1-VD in 23.4%, 2-VD in 23.4%, and 3-VD in 28.5%, respectively. Staged procedures were performed in 78.6% of patients for CTO (39.4%) or multivessel disease (39.2%). Finally, complete revascularization was achieved in 202 (59.9%) patients.
In total, 441 LM stents were implanted (56 BMS and 385 DES). Overall LM stent diameter and length were 3.44 ± 0.50 mm and 22.79 ± 14.14 mm, respectively. Angiographic LM success rate was 99.7%, significantly higher than that of non-LM vessels (88.4%, p = 0.022).
Variables in NERS scoring system
The NERS score system consisted of 17 clinical, 4 procedural, and 33 angiographic variables (Tables 1 and 2). Of clinical variables, acute MI had the highest predicted score (7.3), followed by cardiogenic shock (5.6), previous MI (5.5), and intra-aortic balloon pump requirement (4.9). Of angiographic variables, complex LM lesions, including distal stenosis, CTO, restenosis, severe calcification, and thrombotic lesions, had higher predicted scores (5.9 to 10). Among procedural variables, failure of non–LM-CTO recanalization was associated with the highest predicted score (10.2), independent of dominance. Conversely, predicted score of incomplete revascularization for right coronary or left circumflex arteries was dependent on dominance. Importantly, logistic regression analysis demonstrated that classical crush stenting conferred higher predicted score (3.3), followed by SKS (2.6), and culotte (2.1), significantly different from that of DK crush and T-stenting (0.02 and 0.6, respectively) (p for all < 0.001).
Comparison of NERS versus SYNTAX score
The AUC (Figs. 1 and 2)⇓ was 0.89 for NERS score (95% CI: 0.86 to 0.93, p < 0.001) and 0.69 for SYNTAX score (95% CI: 0.63 to 0.76, p < 0.001), when cumulative MACE was considered as “state variable.” The absence of overlap of 95% CI between the 2 scoring systems, as well as larger AUC (p < 0.0001), suggested that NERS score may be more predictive of cumulative MACE. Similarly, there were significant differences in AUC for cardiogenic death (0.86 vs. 0.74, 95% CI: 0.79 to 0.94 vs. 0.54 to 0.75, p = 0.0209), MI (0.78 vs. 0.55, 95% CI: 0.69 to 0.86 vs. 0.31 to 0.58, p = 0.0316), TVR (0.84 vs. 0.68, 95% CI: 0.79 to 0.89 vs. 0.59 to 0.76, p < 0.0001) and ST (0.82 vs. 0.58, 95% CI: 0.74 to 0.89 vs. 0.45 to 0.71, p = 0.0007) between NERS and SYNTAX scores, respectively (Table 5).
Comparisons of subgroups stratified by NERS and SYNTAX scores
Per ROC curve, a value of 25 stratified scores as NERS-lower (0 to 24) and NERS-higher (≥25). The NERS-higher scores demonstrated enhanced MACE sensitivity and specificity of 92.0% and 74.1%, respectively (Table 6), which is significantly higher than those for SYNTAX-intermediate (20.5% and 25.4%, respectively, p for all < 0.001), SYNTAX-higher (70.5% and 35.2%, respectively, p for all < 0.001), and SYNTAX high-risk (90.9% and 32.4%, respectively, p for all <0.001) subgroups. Furthermore, a NERS-higher score surpassed SYNTAX score in prediction of ST, MI, LM-TVR, cardiac death, and cumulative MACE at follow-up; no differences were observed for in-hospital MACE and all-cause mortality at follow-up, however. Interestingly, the sensitivity and specificity for NERS-lower scores (8.0% and 3.7%) were similar to those of SYNTAX-lower (9.1% and 8.9%, p for all > 0.05) scores.
By Kaplan-Meier analysis, in-hospital MACE-free survival differed significantly between NERS-lower (100%) and NERS-higher (89.1%, p < 0.001) groups (Fig. 2A). This difference was sustained throughout the entire follow-up period (Fig. 2B). However, in-hospital (Fig. 2C) and follow-up (Fig. 2D) cumulative MACE were similar between SYNTAX-intermediate (90.0% and 72.0%) and SYNTAX-higher (94.8% and 65.3%, p for all > 0.05) subgroups. Moreover, SYNTAX-lower individuals enjoyed improved survival with freedom from either in-hospital (100%, p = 0.013) or cumulative MACE at follow-up (91.4%, p < 0.001), as compared to 2 other SYNTAX groups.
QCA and clinical outcomes
Six-month angiographic follow-up (194 ± 18 days post-procedure) was performed in 300 (89.0%) patients, with 100% 6-month clinical follow-up. Rates of LM in-stent restenosis, target lesion revascularization, and -TVR were 12.2% (n = 41), 9.8% (n = 33), and 13.1% (n = 44), respectively; with non-LM in-stent restenosis, target lesion revascularization, and TVR rates of 16.9% (n = 57), 15.5% (n = 52), and 20.2% (n = 68), respectively. Not unexpectedly, in distal LM stenting, significantly lower follow-up cumulative MACE rates were observed for single- versus double-stent techniques (20.4% vs. 32.5%, p < 0.001).
In-hospital follow-up was available for all patients, with cumulative MACE rate of 4.7% (MI, TVR, and all-cause death in 4 [1.2%], 3 [0.9%], and 7 [2.1%] patients, respectively), translating into 95.0% procedural success rate. Early ST occurred in 6 patients (1.8%).
At average follow-up of 1,764 days, rates of MI, TVR, and cardiac mortality were 3.0%, 13.1%, and 5.6%, respectively. Final cumulative MACE rate of 26.0% was driven mainly by subsequent TVR. The overall ST rate was 6.0% (definite 3.6%, probable 1.0%, and possible 1.4%). By regression analysis, NERS score ≥25 (hazard ratio: 1.13; 95% CI: 1.11 to 1.16; p < 0.001) was the only independent predictor of cumulative MACE and ST at follow-up (odds ratio: 31.04; 95% CI: 19.36 to 67.07; p < 0.001).
The present study yielded several notable findings: 1) when compared with SYNTAX score, NERS score demonstrated enhanced sensitivity and specificity for prediction of cumulative MACE, ST, cardiogenic death, TVR, and MI; 2) NERS-higher score had significantly greater predictive value over SYNTAX-intermediate, SYNTAX-higher, and SYNTAX-high-risk scores; 3) SYNTAX-intermediate and SYNTAX-higher scores demonstrated similar MACE predictability; and 4) SYNTAX-lower score had equivalent predictive value to NERS-lower score.
Elements of NERS score
Clinical, angiographic, and procedural characteristics likely all contribute to prognosis after UPLMS stenting. However, quantification of each component's relative contribution is problematic, as there is little supporting data from previous studies.
Findings indicative of decreased left ventricular ejection fraction and/or high-risk ischemic syndromes (including acute MI, cardiogenic shock, previous MI, intra-aortic balloon pump requirement), as well as complexity of lesion morphologies and stenting techniques, have been associated with worse outcomes after stenting (4,12–22). Of note, for double-stent techniques, in concurrence with previous reports (3,4,9,10,23), DK-crush and T-stenting demonstrated lower scores, as compared with classic crush, culotte, and SKS. This finding may be of interest for the practicing interventionalist.
Comparison of NERS versus SYNTAX scores
We believe that the augmented predictive accuracy of NERS over SYNTAX score is reflective of the comprehensive inclusion of clinical, angiographic, and procedural features. The apparent lower sensitivity and specificity of the SYNTAX score in the present study, as compared with the SYNTAX study, may be partially ascribed to the study design. First, the SYNTAX study did not provide the results of ROC for the SYNTAX score (20,24); moreover, repeat calculation might have included single lesions in the SYNTAX study. Importantly, exclusion of clinical variables may be a potential drawback of the SYNTAX score.
Cardiac risk factors and comorbid conditions were more prevalent in the PCI arm of SYNTAX (14). Whereas such clinical variables are known to influence the prognosis of patients with UPLMS (3,4,7,8,18–24), their relative contributions were omitted from the SYNTAX score. Our findings underscore the potential importance of the interplay of clinical, procedural, as well as angiographic variables.
Subgroup analysis stratified by NERS and SYNTAX scores
The SYNTAX-lower scores had equivalent predictive value to NERS-lower scores (less complex lesions, fewer comorbid conditions and cardiac risk factors, and simple stenting techniques), with fewer associated follow-up events. In contrast to the original SYNTAX trial, our data revealed similar sensitivity and specificity between SYNTAX-intermediate and SYNTAX-higher groups. Of note, the predictive value of NERS-higher in the present study has greater predictive value over SYNTAX high-risk groups. This observation likely again reflects the cumulative contribution of clinical and angiographic/procedural features.
First, its nonrandomized nature likely influences the power of the results. Second, both BMS and DES were included. Although no score was generated for BMS patients, their inclusion influences the relevance of our results to modern practice. Moreover, comparison of non-QCA baseline angiography (reflective of the “real world”) with follow-up QCA undoubtedly influenced luminal measurements. Without randomization, operator selection of the specific 2-stent technique likely reflects lesion complexity, thereby influencing outcome. Additionally, the NERS score was derived from a relatively small number of patients undergoing PCI in 4 centers in 1 country. Clearly, its applicability to other demographics requires further study.
The NERS score demonstrated improved predictive value for primary end points after UPLMS-PCI, as compared with SYNTAX score. Further, randomized studies are needed to elucidate its role for decision of PCI versus CABG for UPLMS and complex anatomies in a more diverse population.
Funded by the Nanjing and Jiangsu Provincial Health Bureau, China.
- Abbreviations and Acronyms
- area under the curve
- bare-metal stent(s)
- coronary artery bypass graft
- chronic total occlusion
- drug-eluting stent(s)
- double kissing
- left main
- major adverse cardiac events
- myocardial infarction
- percutaneous coronary intervention
- quantitative coronary angiography
- receiver-operator characteristic
- simultaneous kissing stent(s)
- stent thrombosis
- target vessel revascularization
- unprotected left main coronary artery
- vessel disease
- Received December 29, 2009.
- Revision received April 9, 2010.
- Accepted April 15, 2010.
- American College of Cardiology Foundation
- White A.J.,
- Kedia G.,
- Mirocha J.M.,
- et al.
- Valgimigli M.,
- Malagutti P.,
- Rodriguez-Granillo G.A.,
- et al.
- Chen S.L.,
- Ye F.,
- Zhang J.J.,
- et al.
- Chen S.L.,
- Zhang J.J.,
- Ye F.,
- et al.
- Vaguerizo B.,
- Lefevre T.,
- Darremont O.,
- et al.
- Palmerini T.,
- Sangiorgi D.,
- Marzocchi A.,
- et al.
- Nashef S.A.,
- Roques F.,
- Michel P.,
- et al.
- Leaman D.M.,
- Brower R.W.,
- Meester G.T.,
- Serruys P.,
- van den Brand M.
- Teirstein P.S.
- Romagnoli E.,
- Burzotta F.,
- Trani C.,
- et al.
- Ralf L.,
- Ioakim S.,
- Janine K.,
- Andreas Z.M.,
- Volker S.,
- Stephan F.
- Gyöngyösi M.,
- Christ G.,
- Lang I.,
- et al.,
- AUTAX Investigators