Association of Coronary Anatomical Complexity With Clinical Outcomes After Percutaneous or Surgical Revascularization in the Veterans Affairs Clinical Assessment Reporting and Tracking Program | Cardiology | JAMA Cardiology | JAMA Network
[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Figure 1.  Correlation of the Automatically and Manually Computed VA SYNTAX Score of Anatomical Complexity and Bland-Altman Plot of Differences
Correlation of the Automatically and Manually Computed VA SYNTAX Score of Anatomical Complexity and Bland-Altman Plot of Differences

The SYNTAX (Synergy Between Percutaneous Coronary Intervention With Taxus and Cardiac Surgery) score was adapted from data in the Veterans Affairs (VA) Healthcare System to derive the VA SYNTAX score. The VA SYNTAX score was calculated automatically and compared to a manual calculation of the SYNTAX score.12 A, A positive correlation was found between the automatically calculated and manually computed score for a random selection of angiograms (R = 0.683). Solid line represents the linear association between manually computed and automatically calculated score; dotted lines, 95% CIs for this association. B, No systematic bias was found in differences between manually calculated SYNTAX scores and automatically calculated VA SYNTAX scores across levels of anatomical complexity. Solid lines represent the mean difference between manually computed and automatically calculated scores; dotted lines, ±1.96 SDs of this mean difference. ICC indicates intraclass correlation.

Figure 2.  Adjusted Hazard Ratios for Composite Major Adverse Cardiac and Cerebrovascular Events (MACCEs) and Individual Event Outcomes
Adjusted Hazard Ratios for Composite Major Adverse Cardiac and Cerebrovascular Events (MACCEs) and Individual Event Outcomes

Outcomes were measured after percutaneous coronary intervention (PCI) and coronary artery bypass grafting (CABG) as a function of the Veterans Affairs (VA) SYNTAX (Synergy Between Percutaneous Coronary Intervention [PCI] With Taxus and Cardiac Surgery) score. As demonstrated, there is a strong association between increased SYNTAX score (adapted from data in the Veterans Affairs [VA] Healthcare System to derive the VA SYNTAX score) and MACCEs among patients undergoing PCI, whereas the hazard for MACCEs remains relatively unchanged with increasing scores for patients undergoing CABG.

Figure 3.  Adjusted Hazard Ratios for Individual Adverse Cardiac and Cerebrovascular Events as a Function of SYNTAX Score, Stratified by Revascularization Type
Adjusted Hazard Ratios for Individual Adverse Cardiac and Cerebrovascular Events as a Function of SYNTAX Score, Stratified by Revascularization Type

The SYNTAX (Synergy Between Percutaneous Coronary Intervention [PCI] With Taxus and Cardiac Surgery) score was adapted from data in the Veterans Affairs (VA) Healthcare System to derive the VA SYNTAX score. As shown, the hazard ratios increase with rising anatomical scores among those undergoing PCI. In contrast, the risk remains relatively constant among those undergoing coronary artery bypass grafting (CABG).

Table 1.  Demographic and Clinical Characteristics by Tertile of VA SYNTAX Score of Anatomical Complexity and Revascularization Modalitya
Demographic and Clinical Characteristics by Tertile of VA SYNTAX Score of Anatomical Complexity and Revascularization Modalitya
Table 2.  Major Adverse Cardiovascular and Cerebrovascular Events by Tertile of VA SYNTAX Score of Anatomical Complexity, Stratified by Revascularization Methoda
Major Adverse Cardiovascular and Cerebrovascular Events by Tertile of VA SYNTAX Score of Anatomical Complexity, Stratified by Revascularization Methoda
Supplement.

eMethods. Sources, Measures, Outcomes, and Analyses

eResults. Sensitivity Analyses

eTable 1. VA SYNTAX Scoring System, Defining Coefficients for Scoring by Anatomical Complexity

eTable 2. Administrative Codes Used to Identify Procedures and Readmission

eTable 3. Procedural Characteristics for Patients Undergoing PCI, by Tertile of VA SYNTAX Anatomical Complexity

eTable 4. Examination of Death as a Semicompeting Risk for Revascularization Through Comparison of Results From the Secondary End Points and Sensitivity Analyses

eTable 5. Examination of Death as a Semicompeting Risk for Rehospitalization for Stroke Through Comparison of Results From the Secondary End Points and Sensitivity Analyses

eTable 6. Examination of Death as a Semicompeting Risk for Rehospitalization for MI Through Comparison of Results From the Secondary End Points and Sensitivity Analyses

eFigure 1. Flow Diagram Depicting Inclusion and Exclusion Criteria for the Analytic Cohort

eFigure 2. Temporal Trends in the VA SYNTAX Anatomical Score

eFigure 3. Major Adverse Cardiovascular and Cerebrovascular Events (Composite) Among Patients Undergoing Percutaneous (A) or Surgical (B) Revascularization, Stratified by Tertile of VA SYNTAX Score

eFigure 4. Mortality Events Among Patients Undergoing Percutaneous (A) or Surgical (B) Revascularization, Stratified by Tertile of VA SYNTAX Score

eFigure 5. Readmission for Myocardial Infarction Among Patients Undergoing Percutaneous (A) or Surgical (B) Revascularization, Stratified by Tertile of VA SYNTAX Score

eFigure 6. Readmission for Stroke Among Patients Undergoing Percutaneous (A) or Surgical (B) Revascularization, Stratified by Tertile of VA SYNTAX Score

eFigure 7. Subsequent Revascularization Events Among Patients Undergoing Percutaneous (A) or Surgical (B) Revascularization, Stratified by Tertile of VA SYNTAX Score

eFigure 8. Major Adverse Cardiovascular and Cerebrovascular Events Among Patients From the EXCEL Trial Undergoing Percutaneous (A) or Surgical (B) Revascularization, Stratified by Tertile of VA SYNTAX Score

eReferences.

1.
Levine  GN, Bates  ER, Blankenship  JC,  et al.  2011 ACCF/AHA/SCAI Guideline for Percutaneous Coronary Intervention: executive summary: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines and the Society for Cardiovascular Angiography and Interventions.  Circulation. 2011;124(23):2574-2609. doi:10.1161/CIR.0b013e31823a5596PubMedGoogle ScholarCrossref
2.
Windecker  S, Kolh  P, Alfonso  F,  et al; Authors/Task Force members.  2014 ESC/EACTS Guidelines on myocardial revascularization: The Task Force on Myocardial Revascularization of the European Society of Cardiology (ESC) and the European Association for Cardio-Thoracic Surgery (EACTS)Developed with the special contribution of the European Association of Percutaneous Cardiovascular Interventions (EAPCI).  Eur Heart J. 2014;35(37):2541-2619. doi:10.1093/eurheartj/ehu278PubMedGoogle ScholarCrossref
3.
Stone  GW, Sabik  JF, Serruys  PW,  et al; EXCEL Trial Investigators.  Everolimus-eluting stents or bypass surgery for left main coronary artery disease.  N Engl J Med. 2016;375(23):2223-2235. doi:10.1056/NEJMoa1610227PubMedGoogle ScholarCrossref
4.
Farkouh  ME, Domanski  M, Sleeper  LA,  et al; FREEDOM Trial Investigators.  Strategies for multivessel revascularization in patients with diabetes.  N Engl J Med. 2012;367(25):2375-2384. doi:10.1056/NEJMoa1211585PubMedGoogle ScholarCrossref
5.
Deb  S, Wijeysundera  HC, Ko  DT, Tsubota  H, Hill  S, Fremes  SE.  Coronary artery bypass graft surgery vs percutaneous interventions in coronary revascularization: a systematic review.  JAMA. 2013;310(19):2086-2095. doi:10.1001/jama.2013.281718PubMedGoogle ScholarCrossref
6.
Serruys  PW, Morice  MC, Kappetein  AP,  et al; SYNTAX Investigators.  Percutaneous coronary intervention versus coronary-artery bypass grafting for severe coronary artery disease.  N Engl J Med. 2009;360(10):961-972. doi:10.1056/NEJMoa0804626PubMedGoogle ScholarCrossref
7.
Farooq  V, van Klaveren  D, Steyerberg  EW,  et al.  Anatomical and clinical characteristics to guide decision making between coronary artery bypass surgery and percutaneous coronary intervention for individual patients: development and validation of SYNTAX score II.  Lancet. 2013;381(9867):639-650. doi:10.1016/S0140-6736(13)60108-7PubMedGoogle ScholarCrossref
8.
Iqbal  J, Vergouwe  Y, Bourantas  CV,  et al.  Predicting 3-year mortality after percutaneous coronary intervention: updated logistic clinical SYNTAX score based on patient-level data from 7 contemporary stent trials [published correction in JACC Cardiovasc Interv. 2014 Aug;7(8):946].  JACC Cardiovasc Interv. 2014;7(5):464-470. doi:10.1016/j.jcin.2014.02.007PubMedGoogle ScholarCrossref
9.
Weintraub  WS, Grau-Sepulveda  MV, Weiss  JM,  et al.  Comparative effectiveness of revascularization strategies.  N Engl J Med. 2012;366(16):1467-1476. doi:10.1056/NEJMoa1110717PubMedGoogle ScholarCrossref
10.
Sianos  G, Morel  MA, Kappetein  AP,  et al.  The SYNTAX Score: an angiographic tool grading the complexity of coronary artery disease.  EuroIntervention. 2005;1(2):219-227.PubMedGoogle Scholar
11.
SYNTAX score II. http://www.syntaxscore.com/calculator/start.htm. Published February 23, 2013. Accessed August 1, 2018.
12.
Waldo  SW, Secemsky  EA, O’Brien  C,  et al.  Surgical ineligibility and mortality among patients with unprotected left main or multivessel coronary artery disease undergoing percutaneous coronary intervention.  Circulation. 2014;130(25):2295-2301. doi:10.1161/CIRCULATIONAHA.114.011541PubMedGoogle ScholarCrossref
13.
Bhatt  DL.  CABG the clear choice for patients with diabetes and multivessel disease.  Lancet. 2018;391(10124):913-914. doi:10.1016/S0140-6736(18)30424-0PubMedGoogle ScholarCrossref
14.
Farooq  V, Serruys  PW, Bourantas  CV,  et al.  Quantification of incomplete revascularization and its association with five-year mortality in the synergy between percutaneous coronary intervention with taxus and cardiac surgery (SYNTAX) trial validation of the residual SYNTAX score.  Circulation. 2013;128(2):141-151. doi:10.1161/CIRCULATIONAHA.113.001803PubMedGoogle ScholarCrossref
Original Investigation
June 26, 2019

Association of Coronary Anatomical Complexity With Clinical Outcomes After Percutaneous or Surgical Revascularization in the Veterans Affairs Clinical Assessment Reporting and Tracking Program

Author Affiliations
  • 1Department of Medicine, Division of Cardiology, Rocky Mountain Regional Veterans Affairs Medical Center, Aurora, Colorado
  • 2Department of Medicine, Division of Cardiology, University of Colorado School of Medicine, Aurora
  • 3Department of Medicine, Division of Cardiology, Veterans Affairs Boston Healthcare System, Boston, Massachusetts
  • 4Brigham and Women’s Hospital Heart & Vascular Center, Harvard Medical School, Boston, Massachusetts
  • 5Department of Medicine, Division of Cardiology, Durham Veterans Affairs Medical Center, Durham, North Carolina
  • 6National Heart and Lung Institute, Imperial College of Science, Technology and Medicine, London, United Kingdom
  • 7Thoraxcenter, Erasmus MC, Rotterdam, the Netherlands
  • 8Department of Surgery, University Hospitals of Cleveland Medical Center, Cleveland, Ohio
  • 9Clinical Trials Center, Cardiovascular Research Foundation, New York, New York
  • 10Center for Interventional Vascular Therapy, New York Presbyterian Hospital, Columbia University Medical Center, New York
JAMA Cardiol. 2019;4(8):727-735. doi:10.1001/jamacardio.2019.1923
Key Points

Question  Can a simplified anatomical scoring system adapted for registry data reproduce the established associations between increasing anatomical complexity and adverse outcomes after revascularization?

Findings  In this cross-sectional multiple-center study of 50 226 patients undergoing percutaneous or surgical revascularization, adaptation of the Veterans Affairs SYNTAX simplified anatomical scoring system for coronary artery disease complexity found an association between increasing anatomical complexity and adverse events among patients undergoing percutaneous revascularization but not among patients undergoing surgical revascularization.

Meaning  The Veterans Affairs SYNTAX score may be able to assess longitudinal risk of revascularization using registry data based on coronary anatomical complexity and may represent a powerful tool in understanding longitudinal risk in large data sets.

Abstract

Importance  Anatomical scoring systems for coronary artery disease, such as the SYNTAX (Synergy Between Percutaneous Coronary Intervention [PCI] With Taxus and Cardiac Surgery) score, are well established tools for understanding patient risk. However, they are cumbersome to compute manually for large data sets, limiting their use across broad and varied cohorts.

Objective  To adapt an anatomical scoring system for use with registry data, allowing facile and automatic calculation of scores and association with clinical outcomes among patients undergoing percutaneous or surgical revascularization.

Design, Setting, and Participants  This cross-sectional observational cohort study involved procedures performed in all cardiac catheterization laboratories in the largest integrated health care system in the United States, the Veterans Affairs (VA) Healthcare System. Patients undergoing coronary angiography in the VA Healthcare System followed by percutaneous or surgical revascularization within 90 days were observed and data were analyzed from January 1, 2010, through September 30, 2017.

Main Outcomes and Measures  An anatomical scoring system for coronary artery disease complexity before revascularization was simplified and adapted to data from the VA Clinical Assessment, Reporting, and Tracking Program. The adjusted association between quantified anatomical complexity and major adverse cardiovascular and cerebrovascular events (MACCEs), including death, myocardial infarction, stroke, and repeat revascularization, was assessed for patients undergoing percutaneous or surgical revascularization.

Results  A total of 50 226 patients (49 359 men [98.3%]; mean [SD] age, 66 [9] years) underwent revascularization during the study period, with 34 322 undergoing PCI and 15 904 undergoing coronary artery bypass grafting (CABG). After adjustment, the highest tertile of anatomical complexity was associated with increased hazard of MACCEs (adjusted hazard ratio [HR], 2.12; 95% CI, 2.01-2.23). In contrast, the highest tertile of anatomical complexity among patients undergoing CABG was not independently associated with overall MACCEs (adjusted HR, 1.04; 95% CI, 0.92-1.17), and only repeat revascularization was associated with increasing complexity (adjusted HR, 1.34; 95% CI, 1.06-1.70) in this subgroup.

Conclusions and Relevance  These findings suggest that an automatically computed score assessing anatomical complexity can be used to assess longitudinal risk for patients undergoing revascularization. This simplified scoring system appears to be an alternative tool for understanding longitudinal risk across large data sets.

Introduction

Anatomical scoring systems for coronary artery disease are established tools for physicians and researchers alike to estimate clinical outcomes for patients and offer a standardized method for risk assessment. Current professional guidelines endorse the use of scoring systems for clinical decision making,1,2 and they have been instrumental in clinical trials and comparative effectiveness research.3-5 The most established and widely used of these, the SYNTAX (Synergy Between Percutaneous Coronary Intervention [PCI] With Taxus and Cardiac Surgery) score,6 integrates the burden and anatomical complexity of atherosclerotic coronary artery disease and can be used to estimate patient risk through the association between the manually calculated score and clinical outcomes. Further research has expanded this score to include clinical characteristics that provide further discrimination in estimating longitudinal outcomes on a patient-by-patient basis.7,8

However, these scores can be challenging and cumbersome to compute, with significant limitations in their application to large existing data sets like those represented in cardiovascular registries.9 Adaptation of an anatomical risk scoring system to existing registry angiographic data may be able to create a powerful tool, enabling physicians to compute scores and estimate longitudinal outcomes in real time, thereby enhancing physician and patient understanding of risk. Adaptation of such a system would also allow for facile computation of individual patient scores and for broader application to large cohorts with similar angiographic information. In addition, such a score may have significant implications in understanding the risk of percutaneous or surgical revascularization at patient, operator, and institutional levels, allowing for broader assessment of cardiovascular care quality and a more thorough understanding of the validity of anatomical scoring systems across varied patient cohorts.

With this in mind, we sought to adapt a simplified version of the SYNTAX anatomical risk score to readily available angiographic data from the largest integrated health care system in the United States, the Veterans Affairs (VA) Healthcare System. We subsequently sought to associate this anatomical risk score (VA SYNTAX) with clinical outcomes, stratified by percutaneous or surgical revascularization. Our findings suggest that physicians will be able to use the VA SYNTAX score to estimate the longitudinal risk of adverse events for patients undergoing revascularization at an individual level and across health care systems.

Methods
Population

Using data from the VA Clinical Assessment, Reporting, and Tracking Program, we evaluated all patients older than 18 years undergoing coronary angiography at VA catheterization laboratories from January 1, 2010, through September 30, 2017. Patients were included if they were found to have greater than 50% stenosis in at least 1 epicardial coronary artery at the time of angiography and underwent subsequent revascularization (PCI or coronary artery bypass grafting [CABG]) within 90 days inside or outside the VA Healthcare System. Patients with prior CABG or normal coronary arteries were excluded to ensure that any validated scoring system focused on individuals with unrevascularized coronary artery disease in native epicardial coronary arteries. Patients were also excluded for missing data on coronary anatomy, severity of coronary lesions, coronary dominance, or other key demographic or clinical variables. If a patient had multiple coronary angiograms within the time frame of the study, only the initial record was used. All procedures were reviewed and approved by the Colorado Multiple Institutions Review Board. Informed consent is not required for retrospective analyses primarily involving chart review and was therefore waived by the review board. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cross-sectional studies.

VA SYNTAX Score

An anatomical risk score (VA SYNTAX) was adapted from the original method used by Sianos et al10 for the SYNTAX trial. Scores and weights were derived from the original SYNTAX score and applied to existing registry data, including coronary anatomical dominance, location of stenosis and coronary segment, ostial location, noted calcification or thrombus associated with the lesion, and bifurcation status (eTable 1 in the Supplement). Anatomical characteristics not readily available in registry data, such as Medina classification and bifurcation angle, were excluded from this simplified score. The traditional SYNTAX score was manually calculated for 100 randomly selected patients via individual angiographic review and the use of a web-based calculator.11 Associations between the VA SYNTAX score and the manually calculated SYNTAX score were summarized using a Pearson correlation coefficient, intraclass correlation, and Bland-Altman analysis.

VA SYNTAX scores were calculated for each patient and categorized by tertiles of score across all patients who underwent revascularization. Subsequent analyses stratified patients by the first revascularization method (CABG or PCI) received within 90 days, and separate analyses assessed the association between the VA SYNTAX score and outcomes. Significant unmeasured confounding in the revascularization method chosen in observational cohorts is likely, so we purposefully did not compare revascularization methods. Finally, the score was applied to data from the EXCEL (Evaluation of XIENCE vs Coronary Artery Bypass Surgery for Effectiveness of Left Main Revascularization) trial3 to assess the feasibility of use in automatically calculating anatomical complexity and to validate the score’s ability to discriminate between outcomes across tertiles of anatomical complexity in external data sets. The score was applied in its entirety to the trial cohort (with the lone exception that mixed coronary dominance was not an available covariate from the trial data), with the primary outcome assessed across tertiles of anatomical complexity by revascularization method.

Outcomes

The primary outcome was a composite of death, revascularization, rehospitalization for myocardial infarction (MI), and rehospitalization for stroke after initial revascularization (major adverse cardiovascular and cerebrovascular events [MACCEs]). Individual clinical events were assessed as secondary end points. Outcomes were ascertained through a review of administrative billing codes for readmissions (eTable 2 in the Supplement) within and outside the VA system, using fee-basis data and Medicare claims as a sensitivity analysis.

Statistical Analysis

Data were analyzed from January 1, 2010, through September 30, 2017. Standard inferential statistical methods (χ2 test and analysis of variance) were used to assess differences in demographics, comorbidities, and clinical and procedural characteristics of patients across tertiles of anatomical score stratified by the first revascularization method received within 90 days, as well as across populations included and excluded from the analysis. Simplified anatomical scores were then assessed as a function of time. Cox proportional hazards regression models were then created for each revascularization group (eMethods in the Supplement). Two-sided P < .05 was considered statistically significant.

Results
Population

During the study period, 269 551 unique patients underwent coronary angiography and 122 882 unique patients underwent PCI or CABG within 90 days. Patients were excluded for having normal coronary arteries (n = 116 169), not undergoing revascularization (n = 226 779), and for having prior CABG (n = 82 673) or prior coronary angiography (n = 80 110); reasons were not mutually exclusive. Additional patients were excluded for missing procedural or other data (n = 19 164). The final analytic cohort included 50 226 patients across 83 facilities (49 359 men [98.3%] and 867 women [1.7%]; mean [SD] age, 66 [9] years). These patients were then stratified into those receiving PCI (n = 34 322) or CABG (n = 15 904) within 90 days (eFigure 1 in the Supplement).

VA SYNTAX Anatomical Risk Score

The VA SYNTAX simplified anatomical risk score was automatically computed based on the data provided by physicians interpreting each angiogram. A mild increase in the anatomical risk score occurred for patients undergoing PCI (9.5 in 2010 to 10.5 in 2017; P < .001 for trend) and CABG (10.5 in 2010 to 11.0 in 2017; P = .02 for trend) (eFigure 2 in the Supplement) over time. The Pearson correlation between the VA SYNTAX anatomical score and the manually computed SYNTAX score was moderate (0.683) (Figure 1A), as was the intraclass correlation (0.541). Figure 1B shows a Bland-Altman analysis of manually and automatically calculated anatomical scores, demonstrating lack of significant bias in differences across magnitudes of scores.

When we divided VA SYNTAX anatomical complexity scores into tertiles, tertile 1 had scores less than or equal to 7, and tertile 3 had scores greater than 15. These tertiles were then applied to each revascularization subgroup. Among patients undergoing PCI, the largest proportion (46.2%) were in tertile 1, with a minority falling into tertile 3 (18.2%). Most patients undergoing CABG had the most anatomically complex disease (60.6%), with a small proportion (10.0%) falling in tertile 1.

Clinical Characteristics

The clinical characteristics of patients were stratified by VA SYNTAX score and revascularization modality (Table 1). Patients undergoing PCI in increasing tertiles of anatomical complexity were older (mean [SD] age, 68.0 [9.7] vs 65.9 [9.2] vs 64.6 [8.8] years) and had more comorbidities and more frequent presentation for angiography in the setting of acute coronary syndrome than in lower tertiles (Table 1). Procedural characteristics of PCI by VA SYNTAX tertile are demonstrated in eTable 3 in the Supplement. Among those undergoing CABG, patients were of similar age (mean [SD] age, 66.9 [7.9] vs 66.2 [7.9] vs 67.5 [8.3] years) and with less consistent increases in comorbidities across tertiles. However, patients with higher scores more frequently presented for angiography in the setting of acute coronary syndrome (Table 1), similar to the trend seen in the PCI cohort.

Clinical Outcomes

Rates of clinical outcomes as a function of VA SYNTAX score were stratified by revascularization method. Before adjustment, proportions of composite MACCEs and individual major events of death, MI, repeat revascularization, and stroke increased among increasing tertiles of VA SYNTAX scores of patients undergoing PCI (eFigures 3A-7A in the Supplement). Among patients undergoing CABG, the number of deaths was not different between tertiles of VA SYNTAX score, whereas MI, revascularization, and stroke increased with increasing anatomical complexity (eFigures 3B-7B in the Supplement). After adjustment, among patients undergoing PCI, the adjusted hazard ratio (aHR) for MACCEs was significantly higher in tertiles 2 (1.44; 95% CI, 1.38-1.51) and 3 (2.12; 95% CI, 2.01-2.23) when compared with tertile 1, as seen in Table 2. In Figure 2A, aHRs for MACCEs after PCI increased as a function of VA SYNTAX score. The aHRs for individual major adverse events of death, MI, repeat revascularization, and stroke were also associated with increasing anatomical complexity among patients undergoing PCI (Figure 3A-D). Among patients undergoing CABG, the aHR of repeat revascularization was significantly higher in the highest tertile of anatomical complexity (1.34; 95% CI, 1.06-1.70), but no such association was noted for MACCEs (1.04; 95% CI, 0.92-1.17) or for individual major adverse events of death, MI, or stroke (Table 2). Figure 2B demonstrates the relative stability of aHRs with increasing VA SYNTAX score after CABG. The aHRs for individual clinical outcomes among patients undergoing surgical revascularization as a function of VA SYNTAX score are demonstrated in Figure 3E and H. Sensitivity analyses excluding facilities with low rates of CABG and evaluating mortality as a semicompeting risk are reported in eResults and eTables 4 to 6 in the Supplement.

External Validation

The VA SYNTAX score was applied to patients in the EXCEL trial3 data set, stratified by revascularization method. Proportions of composite MACCEs increased among increasing tertiles of VA SYNTAX scores of patients undergoing PCI (tertile 1: 51 [17.6%]; tertile 2: 114 [18.3%]; tertile 3: 117 [21.9%]; log-rank P = .29) (eFigure 8A in the Supplement); however, these differences did not reach statistical significance. Among patients undergoing CABG, composite MACCEs had similar proportions across tertiles (tertile 1, 48 [16.2%]; tertile 2, 56 [16.4%]; tertile 3, 49 [16.7%]; log rank P = .98) (eFigure 8B in the Supplement).

Discussion

This analysis adapted an anatomical risk score to existing registry data for patients undergoing coronary revascularization, with results suggesting that this novel VA SYNTAX score may be easily and automatically implemented across large data sets. After adjustment for demographic, clinical, and procedural comorbidities, increasing anatomical complexity was associated with an increased hazard of adverse events among patients undergoing PCI but not among patients undergoing CABG. Anatomical complexity was associated with each of the individual clinical outcomes among those undergoing PCI but only with repeat revascularization among those undergoing CABG. These findings suggest that a simplified anatomical risk model can be automatically calculated from registry data and may be useful in predicting longitudinal outcomes for patients undergoing revascularization, independent of clinical characteristics.

Prior data have shown that increasing coronary anatomical complexity is associated with worse clinical outcomes,6 with subsequent analyses demonstrating similar associations after adjusting for demographic and clinical characteristics.7,8 However, these anatomical scoring systems have been challenging and time consuming to compute, making them difficult to implement for large data sets.9 In the present study, a simplified anatomical score derived from the well-validated SYNTAX score10 and applied to existing registry data appeared to demonstrate similar associations of increasing anatomical complexity with adverse clinical events across 50 226 patients undergoing revascularization. Furthermore, successful automated calculation of the VA SYNTAX score from the EXCEL3 trial data suggests the ability to extend this score to other data sets, independent of the derivation cohort. Although the difference in outcomes did not reach statistical significance in the EXCEL data set, the point estimates were similar to those in the derivation cohort. We suspect this finding was likely influenced by a greater than 25-fold difference in cohort sizes but acknowledge that the finding is speculative and may require future validation in a larger population-based cohort. However, the scores derived in this analysis serve as a proof of concept that anatomical risk scores may be developed and automatically calculated for a population-based evaluation of risk for major adverse events after revascularization, across varied patient cohorts.

These findings appear to validate prior analyses, suggesting that increasing anatomical complexity should be considered when providing prognostic information to patients undergoing PCI independent of the clinical risk from medical comorbidities. These findings also underscore the importance of these associations by demonstrating temporal increases in anatomical complexity of patients undergoing revascularization. This model was easily implemented in a large data set with extensive demographic and clinical data including more than 30 000 patients undergoing PCI, allowing for the implementation of robust methods to account for comorbidities and features shown to be associated with increased anatomical complexity. With this implementation, the VA SYNTAX anatomical risk score continued to demonstrate consistent associations between anatomical complexity and major adverse events for percutaneous revascularization after adjustment. These findings are similar in trend and magnitude to previously reported hazards of mortality after PCI for increasing anatomical complexity, after adjustment for clinical features.7,8

The association between anatomical complexity and clinical outcomes, however, was less pronounced for patients undergoing CABG. Previous data have demonstrated only a mild increase in adverse events with increasing coronary complexity among patients undergoing CABG, and adjustment for clinical comorbidities nullifies this association.6,7 Similarly, among 15 904 patients in the largest integrated health care system in the United States, increasing VA SYNTAX scores were not associated with increases in adjusted rates of most adverse events after CABG. These findings suggest that clinical comorbidities may be better associated with adverse outcomes for patients undergoing surgical revascularization as opposed to anatomical complexity. In addition, although anatomical complexity was associated with an increased burden of medical and clinical comorbidity among patients undergoing PCI, this association was not noted for patients undergoing CABG. These findings seem to intimate previously suspected bias in selection of revascularization modality, wherein patients with more clinically complex disease may undergo PCI rather than CABG despite similar anatomical complexity.13

Together, these findings for both revascularization methods suggest that a simplified anatomical scoring system was capable of generating useful prognostic information to establish patient risk after revascularization and may be adaptable to registry data. With such a scoring system, scores could be automatically calculated in real time to offer physicians valuable insights into the longitudinal outcomes for patients and assist in physician-patient discussions and shared decision making. In addition, such a scoring system may provide more granular and precise risk adjustment in assessment of institutional and operator outcomes, for internal or public reporting of quality and performance. This system might add significant value for the public reporting of interventional outcomes, moving beyond administrative or claims data to an integrated clinical and anatomical model for refinement of adjustment methods. Simplified and adapted from existing standardized registry data, this VA SYNTAX score might be applicable not only in the largest integrated health care system in the United States but also in community facilities that participate in systematized data repositories. More specifically, sites that participate in the National Cardiovascular Data Registries might use this simplified and validated model to allow automatic risk adjustment for clinical outcomes based on the anatomical complexity of patients treated.

Limitations

This analysis should be interpreted in the context of its limitations. First, residual confounding may exist beyond the VA SYNTAX score. Using rigorous adjustment methods, we attempted to limit residual confounding as much as possible. Furthermore, by stratifying the cohort by revascularization methods, we minimized effects of selection bias. Second, we were unable to determine the completeness of revascularization, a factor known to be associated with longitudinal outcomes. Patients with incomplete revascularization could represent a mediator in the demonstrated associations. However, because baseline anatomical complexity is known to be associated with the degree of incomplete revascularization,14 much of this association may remain captured in our models. Furthermore, our models were stratified by revascularization method and did not attempt to compare strategies between CABG and PCI, limiting any influence of differential rates of incomplete revascularization by strategy.12

Third, we did not differentiate among subsets of revascularization such as unprotected left main stenoses or multivessel coronary disease. Because this analysis was designed as an assessment of overall anatomical complexity irrespective of indication to better characterize longitudinal outcomes after revascularization, we believed that a more inclusive approach would be more generalizable for physicians. Fourth, many patients were excluded from the development of the anatomical risk model and assessment of its association with outcomes. However, most (>270 000) of these exclusions were for not having significant coronary disease by angiography or for not undergoing revascularization. Because the present study focused on the association between anatomical complexity and outcomes after revascularization, these patients were necessarily removed from the analytic cohort.

Fifth, we were unable to ascertain granular procedural details for CABG or PCI such as percentage of arterial grafts owing to limitations of the data set. However, this cohort consists of all patients undergoing revascularization across a national integrated health care system. Thus, these findings represent real-world practice, likely increasing their external validity. Sixth, these findings may not be generalizable to populations not well represented in the VA. In particular, the applicability of this risk score to female patients is uncertain, given their low representation in the analytic cohort. However, the successful use of the score in the EXCEL3 trial data set demonstrated feasibility and was suggestive of similar findings in a more diverse population. Although the separation in outcomes across tertiles of anatomical complexity did not reach statistical significance, the greater than 25-fold difference in cohort size from registry derivation to trial validation likely influenced the ability to reach thresholds for significance across tertiles but may limit the score’s generalizability. Further analysis in population-based registries not limited to left main coronary artery revascularization remains necessary to fully validate the score. Finally, because the VA SYNTAX score depends on registry data, missing data or entry errors could influence the score and the present findings.

Conclusions

These findings suggest that a simplified anatomical scoring system can be adapted to registry data to provide valuable prognostic information after coronary revascularization, independent of clinical characteristics. Implementation of this simplified scoring system may have implications for real-time assessment of procedural risk, with additional potential for refining risk adjustment when evaluating interventional quality in a variety of health care settings.

Back to top
Article Information

Accepted for Publication: April 30, 2019.

Corresponding Author: Stephen W. Waldo, MD, Rocky Mountain Regional Veterans Affairs Medical Center, 1700 N Wheeling St, Aurora, CO 80045 (stephen.waldo@va.gov).

Published Online: June 26, 2019. doi:10.1001/jamacardio.2019.1923

Author Contributions: Drs Valle and Waldo had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Valle, Bricker, Armstrong, Bhatt, Serruys, Kappetein, Sabik, Waldo.

Acquisition, analysis, or interpretation of data: Valle, Glorioso, Bricker, Baron, Rao, Plomondon, Kappetein, Dressler, Stone, Waldo.

Drafting of the manuscript: Valle, Glorioso, Bricker, Serruys, Waldo.

Critical revision of the manuscript for important intellectual content: Valle, Baron, Armstrong, Bhatt, Rao, Plomondon, Serruys, Kappetein, Sabik, Dressler, Stone, Waldo.

Statistical analysis: Valle, Glorioso, Serruys, Dressler.

Obtained funding: Kappetein.

Administrative, technical, or material support: Bricker, Armstrong, Rao, Waldo.

Supervision: Valle, Armstrong, Plomondon, Kappetein, Sabik, Waldo.

Conflict of Interest Disclosures: Dr Barón reported receiving grants from University of Colorado during the conduct of the study. Dr Armstrong reported receiving nonfinancial support from Abbott Vascular, Boston Scientific Corporation, Cardiovascular Systems, Medtronic plc, and Philips outside the submitted work. Dr Bhatt reported receiving research support from the Department of Veterans Affairs (VA) during the conduct of the study; grants from Amarin Corporation, AstraZeneca, Bristol-Myers Squibb, Eisai Co, Ltd, Ethicon Inc, Medtronic plc, Sanofi Aventis, The Medicines Company, Roche, Pfizer, Inc, Forest Laboratories/AstraZeneca, Ischemix, Amgen, Inc, Eli Lilly and Co, Chiesi Farmaceuticia, SpA, Ironwood Pharmaceuticals, Inc, Abbott Vascular, Regeneron Pharmaceuticals, Inc, PhaseBio Pharmaceuticals, Inc, Idorsia, and Synaptic Pharmaceutical Corporation outside the submitted work; unfunded research collaboration from FlowCo, PLx Pharma, Inc, and Takeda Pharmaceutical Company, Ltd, personal fees from Duke Clinical Research Institute, Mayo Clinic, Population Health Research Institute, Medscape Cardiology, Regado Biosciences, Inc, Boston VA Research Institute, Clinical Cardiology, Journal of the American College of Cardiology, the VA, St. Jude Medical (now Abbott), Biotronik, Cardax, Inc, Boston Scientific Corporation, Merck & Co, Svelte, Boehringer Ingelheim, and Novo Nordisk outside the submitted work; personal fees from Belvoir Publications, Slack Publications, WebMD, Elsevier, Society of Cardiovascular Patient Care, HMP Global, Harvard Clinical Research Institute (now Baim Institute for Clinical Research), Cleveland Clinic, Mount Sinai School of Medicine, TobeSoft, Boehringer Ingelheim, Bayer, and American College of Cardiology outside the submitted work; and nonfinancial support from the American Heart Association, the Society of Cardiovascular Patient Care, and the American College of Cardiology outside the submitted work. Dr Serruys reported receiving personal fees from Abbott Vascular, Biosensors International, Ltd, Cardialysis, Medtronic plc, Micell Technologies, Sino Medical-Device Technology Co, Ltd, Philips/Volcano, and Xeltis outside the submitted work. Dr Kappetein reported receiving personal fees from Erasmus MC during the conduct of the study and personal fees from Medtronic plc outside the submitted work. Dr Stone reported receiving personal fees from Medical Development Technologies, Claret, Backbeat, Sirtex, Matrizyme Pharma Corporation, Miracor Medical SA, Neovasc, Inc, Shockwave, V-wave, VALFIX, TherOx, Inc, Reva Pharma, Vascular Dynamics, Inc, Robocath, HeartFlow, Ablative Solutions, Inc, Gore, Terumo Medical Corporatoin, and Amaranth and equities and options from Ancora, Cagent, Qool Therapeutics, Aria, Caliber Therapeutics, MedFocus family of funds, Biostar family of funds, Applied Therapeutics, and SpectraWave outside the submitted work; and receipt of royalties by employer Columbia University for sale of the MitraClip from Abbott Vascular. Dr Waldo reported receiving grants from Abiomed, Merck & Co, and Cardiovascular Systems Incorporated outside the submitted work. No other disclosures were reported.

Funding/Support: This study was supported by projects SDR 02-237 and 98-004Support from the VA, Veterans Health Administration, Office of Research and Development, Health Services Research and Development, VA Information Resource Center (VA and Centers for Medicare & Medicaid Services data).

Role of the Funder/Sponsor: The sponsor had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the VA or the US government.

References
1.
Levine  GN, Bates  ER, Blankenship  JC,  et al.  2011 ACCF/AHA/SCAI Guideline for Percutaneous Coronary Intervention: executive summary: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines and the Society for Cardiovascular Angiography and Interventions.  Circulation. 2011;124(23):2574-2609. doi:10.1161/CIR.0b013e31823a5596PubMedGoogle ScholarCrossref
2.
Windecker  S, Kolh  P, Alfonso  F,  et al; Authors/Task Force members.  2014 ESC/EACTS Guidelines on myocardial revascularization: The Task Force on Myocardial Revascularization of the European Society of Cardiology (ESC) and the European Association for Cardio-Thoracic Surgery (EACTS)Developed with the special contribution of the European Association of Percutaneous Cardiovascular Interventions (EAPCI).  Eur Heart J. 2014;35(37):2541-2619. doi:10.1093/eurheartj/ehu278PubMedGoogle ScholarCrossref
3.
Stone  GW, Sabik  JF, Serruys  PW,  et al; EXCEL Trial Investigators.  Everolimus-eluting stents or bypass surgery for left main coronary artery disease.  N Engl J Med. 2016;375(23):2223-2235. doi:10.1056/NEJMoa1610227PubMedGoogle ScholarCrossref
4.
Farkouh  ME, Domanski  M, Sleeper  LA,  et al; FREEDOM Trial Investigators.  Strategies for multivessel revascularization in patients with diabetes.  N Engl J Med. 2012;367(25):2375-2384. doi:10.1056/NEJMoa1211585PubMedGoogle ScholarCrossref
5.
Deb  S, Wijeysundera  HC, Ko  DT, Tsubota  H, Hill  S, Fremes  SE.  Coronary artery bypass graft surgery vs percutaneous interventions in coronary revascularization: a systematic review.  JAMA. 2013;310(19):2086-2095. doi:10.1001/jama.2013.281718PubMedGoogle ScholarCrossref
6.
Serruys  PW, Morice  MC, Kappetein  AP,  et al; SYNTAX Investigators.  Percutaneous coronary intervention versus coronary-artery bypass grafting for severe coronary artery disease.  N Engl J Med. 2009;360(10):961-972. doi:10.1056/NEJMoa0804626PubMedGoogle ScholarCrossref
7.
Farooq  V, van Klaveren  D, Steyerberg  EW,  et al.  Anatomical and clinical characteristics to guide decision making between coronary artery bypass surgery and percutaneous coronary intervention for individual patients: development and validation of SYNTAX score II.  Lancet. 2013;381(9867):639-650. doi:10.1016/S0140-6736(13)60108-7PubMedGoogle ScholarCrossref
8.
Iqbal  J, Vergouwe  Y, Bourantas  CV,  et al.  Predicting 3-year mortality after percutaneous coronary intervention: updated logistic clinical SYNTAX score based on patient-level data from 7 contemporary stent trials [published correction in JACC Cardiovasc Interv. 2014 Aug;7(8):946].  JACC Cardiovasc Interv. 2014;7(5):464-470. doi:10.1016/j.jcin.2014.02.007PubMedGoogle ScholarCrossref
9.
Weintraub  WS, Grau-Sepulveda  MV, Weiss  JM,  et al.  Comparative effectiveness of revascularization strategies.  N Engl J Med. 2012;366(16):1467-1476. doi:10.1056/NEJMoa1110717PubMedGoogle ScholarCrossref
10.
Sianos  G, Morel  MA, Kappetein  AP,  et al.  The SYNTAX Score: an angiographic tool grading the complexity of coronary artery disease.  EuroIntervention. 2005;1(2):219-227.PubMedGoogle Scholar
11.
SYNTAX score II. http://www.syntaxscore.com/calculator/start.htm. Published February 23, 2013. Accessed August 1, 2018.
12.
Waldo  SW, Secemsky  EA, O’Brien  C,  et al.  Surgical ineligibility and mortality among patients with unprotected left main or multivessel coronary artery disease undergoing percutaneous coronary intervention.  Circulation. 2014;130(25):2295-2301. doi:10.1161/CIRCULATIONAHA.114.011541PubMedGoogle ScholarCrossref
13.
Bhatt  DL.  CABG the clear choice for patients with diabetes and multivessel disease.  Lancet. 2018;391(10124):913-914. doi:10.1016/S0140-6736(18)30424-0PubMedGoogle ScholarCrossref
14.
Farooq  V, Serruys  PW, Bourantas  CV,  et al.  Quantification of incomplete revascularization and its association with five-year mortality in the synergy between percutaneous coronary intervention with taxus and cardiac surgery (SYNTAX) trial validation of the residual SYNTAX score.  Circulation. 2013;128(2):141-151. doi:10.1161/CIRCULATIONAHA.113.001803PubMedGoogle ScholarCrossref
×