Abstract
Background
A recent large-scale clinical trial found that an initial invasive strategy does not improve cardiac outcomes beyond optimized medical therapy in patients with stable coronary artery disease (CAD). Novel methods to stratify at-risk patients may refine therapeutic decisions to improve outcomes.
Methods and Results
In a cohort of 815 consecutive patients referred for evaluation of myocardial ischemia, we determined the net reclassification improvement of the risk of cardiac death or nonfatal MI (MACE) incremental to clinical risk models, using guideline–based low (<1%), moderate (1–3%), and high (>3%) annual risk categories. In the whole cohort, inducible ischemia demonstrated strong association with MACE (hazard ratio 14.66, P<0.0001) with low negative event rates of MACE and cardiac death (0.6% and 0.4%). This prognostic robustness maintained in patients with prior CAD (hazard ratio 8.17, P<0.0001, and 1.3% and 0.6%, respectively). Adding inducible ischemia to the multivariable clinical risk model (age and prior CAD adjusted) improved discrimination of MACE (C-statistic 0.81 to 0.86, P=0.04; Adjusted hazard ratio 7.37, P<0.0001) and reclassified 91.5% of patients at moderate pre-test risk (65.7% to low risk; 25.8% to high risk) with corresponding changes in the observed event rates (0.3%/year and 4.9%/year, for low and high risk post-test, respectively). Categorical net reclassification index was 0.229 (95% CI 0.063–0.391). Continuous NRI was 1.11 (95% CI 0.81–1.39).
Conclusions
Stress CMR effectively reclassifies patient risk beyond standard clinical variables, specifically in patients at moderate to high pre-test clinical risk and in patients with prior CAD.
Clinical Trial Registration Information
http://clinicaltrials.gov/. Identifier: NCT01821924.
Keywords: chronic ischemia, cardiac magnetic resonance imaging
Introduction
Clinical risk assessment is limited in patients with established stable coronary artery disease (CAD). The Clinical Outcomes Utilizing Revascularization and Aggressive Drug Evaluation (COURAGE) trial demonstrated that patients with stable CAD may be safely managed with an initial strategy of optimal medical therapy.1 However, patients with extensive ischemia may benefit from mechanical coronary revascularization.2, 3 American College of Cardiology Foundation (ACC)/American Heart Association (AHA) practice guidelines also recommend that CAD patients deemed high risk for adverse events be considered for revascularization.4 Therefore, an effective risk assessment method may improve management decisions particularly towards utilization of invasive investigations.
Stress cardiac magnetic resonance imaging (CMR) offers a comprehensive assessment of the presence and extent of myocardial ischemia and viability. Although numerous studies have demonstrated an excellent diagnostic accuracy for CAD detection,5, 6 the ability of stress CMR to influence clinical decision-making by reclassification of patient risk has not been investigated. In this study, we sought to specifically test the hypothesis that stress CMR would effectively reclassify patients across AHA/ACC-recommended cardiac risk categories, the basis for management decisions in these patients.
Methods
Study population
We performed vasodilator stress CMR in 815 patients referred for assessment of myocardial ischemia. Patients were consecutively enrolled between 2001–2011 from the Brigham and Women’s Hospital inpatient and outpatient cardiology and general medical service. Indication for referral was the assessment of suspected myocardial ischemia. Inclusion criteria included age > 18 years and a clinical suspicion of myocardial ischemia at the discretion of the referring clinician. Exclusion criteria consisted of absolute contraindications to CMR (e.g., metallic hazards, pregnancy, severe renal dysfunction) or contraindications to vasodilator stress testing. A physician obtained detailed medical history from the patient prior to CMR study. A history of prior CAD was defined by evidence of MI, prior PTCA or CABG, or angiographically significant coronary stenosis (>70% stenosis in any epicardial or > 50% of the left main coronary artery). Prior MI was confirmed by definitive clinical evidence in the medical record or presence of pathologic Q-wave(s) by published criteria.7 Institutional review board approved patient follow-up using standardized questionnaire and telephone script. For the purpose of evaluating a clinical question different from the current study, 424 patients from our current study were used in a prior report.8 Patients with images adequate for CMR analysis and any clinical follow-up were included in the analysis (see Results below). The study is registered at clinicaltrials.gov (NCT01821924).
Stress CMR protocol
We performed stress CMR with a 1.5-Tesla scanner before 2006 (N=381, 47%, Signa CV/i, General Electric, Milwaukee; 8-element coil) or a 3.0-Tesla scanner (N=434, 53%, TIM TRIO/VERIO, Siemens, Erlangen, Germany; 16-element coil) thereafter. CMR protocol consisted of stress and rest myocardial perfusion, ventricular function, and late gadolinium enhancement (LGE). An electrocardiogram was obtained before and after CMR. Adenosine (N=396), regadenoson (N=389) (both from Astellas Pharma US, Deerfield, IL), or dipyridamole (N=30) (Boehringer Ingelheim, Germany); were prescribed as intravenous stress agents.8 Myocardial perfusion images were acquired during bolus injection of 0.1 mmol/kg intravenous gadolinium-DPTA (Magnevist, Bayer, Wayne, NJ). Stress CMR perfusion, when performed at 3T, used a saturation-recovery prepared turbo FLASH sequence with the following specifications: TR/TE/flip = 2.4, 1.0, 18 degree; 10 msec delay after saturation before read-out; linear phase-encoding order; acceleration (factor of 2) using Generalized Autocalibrating Partially Parallel Acquisition (GRAPPA); in-plane resolution 2.2×2.7 mm; slice thickness 10 mm; receiver bandwidth 800–900 Hz/pixel. At 1.5T, it was performed with a notched saturation-prepared, fast gradient echo sequence with echo-planar readout (TR/TE 6, 2.3 msec; echo-train length 4; SENSE 2; slice thickness 8 mm) in 3 parallel short-axis views of the left ventricle (LV) sampled every R-R interval. For perfusion images acquired on 3T, a 4-chamber long-axis view was also obtained. Cine and LGE imaging (15–20 minutes after contrast) were acquired via similar protocols on 1.5T and 3T as previously published.8, 9 All images were acquired at end-expiration using ECG or pulse oximetry gating.
The presence and segmental extent of perfusion defects were confirmed by the consensus of two independent reviewers blinded to clinical and follow-up information. Perfusion defects were defined as hypoenhanced regions that persisted for at least 3 phases after peak contrast enhancement, were >1 pixel in thickness, and followed a coronary distribution. Following the AHA/ACC 16-segmental model, segments were graded as “1” if endocardial (<50% wall thickness) and “2” if transmural (≥50% of wall thickness). Inducible ischemia was defined by the presence of a stress perfusion defect in any segment without matching LGE. Global extent of myocardial ischemia (ischemia score) was calculated as the sum of stress perfusion grades from all segments with inducible ischemia. In addition, we defined a binary variable stratified by the presence of > 10% ischemia, estimated by the presence of inducible ischemia in ≥ 3 of 32 subsegments (endocardial and epicardial sectors for each of the 16 segments). This criterion of >10% ischemia has been observed as high-risk in the nuclear substudy of the COURAGE trial.2 Infarct volume was defined as the myocardial volume with signal intensity ≥2 standard deviations above remote normal myocardium. All CMR analyses used validated commercial software (Mass®, Medis, Leiden, the Netherlands).
Follow-up of Cardiac Events
We first reviewed available electronic medical records (EMR) across teaching hospitals of Partners HealthCare, for notes from any physicians caring for the patient to assess for interval MI or mortality. Cardiac mortality was defined by any death preceded by an acute MI, decompensated heart failure, or ventricular arrhythmia. Non-fatal MI was defined by the presentation of an acute coronary syndrome and elevation of cardiac biomarkers (>99th percentile of the upper limit of normal) temporally consistent with an acute injury. In cases where records were not found in the EMR, we contacted treating physicians using a standardized institute-approved medical questionnaire to inquire about events and follow-up. Regardless of the results of the follow-up procedures above, annually we checked the mortality status from the Social Security Death Index of all patients and mailed a standardized IRB-approved medical questionnaire to all patients to inquire in detail about MACE. We confirmed cases of mortality by review of both medical records and the Social Security Death Index. If there remains inadequacy of follow-up information, we then telephoned to speak with the patient or a family member using the standardized medical questionnaire. Details of any outside hospitalization, including suspected acute MI, were checked by retrieving medical records from outside hospitals. Two cardiologists blinded to CMR results performed all standardized follow-up procedures. Our primary outcome included a composite of cardiac mortality or acute, non-fatal MI (MACE). All-cause mortality was a secondary outcome.
Statistics
Univariable associations with MACE were determined by Cox proportional hazards regression, and event-free survival stratified by inducible ischemia was estimated with Kaplan-Meier survival methods. A multivariable clinical risk model for MACE was constructed using a backward elimination Cox regression strategy (P<0.10 for model retention). Prior CAD and age were forced into this clinical risk model, given their prognostic importance in patients with suspected ischemia. To form this clinical risk model, all clinical, electrocardiographic, and CMR covariates, including LV/RV volumes, mass, and ejection fraction, were considered in the model selection process. We also built another clinical risk model without CMR covariates considered. Time zero was defined as the day of the CMR study for all Cox regression analyses.
To address the incremental prognostic value of inducible ischemia on MACE, we added inducible ischemia to the multivariable clinical risk model to obtain an adjusted hazard ratio for ischemia. In addition, we assessed for significant incremental changes in model C-statistic before and after addition of inducible ischemia. To assess if early revascularization performed in response to the extent of inducible ischemia modified the association between ischemia and outcome, we included an interaction term between early coronary revascularization (<90 days after CMR) and ischemia score as a covariate in Cox models for prediction of MACE. For all multivariable models, proportional hazards assumptions were validated and calibrated as previously described.10
In order to evaluate the ability of inducible ischemia to reclassify patients, we performed net reclassification improvement (NRI) analyses11 using categories of <1%, 1–3%, and >3%/year event rates per ACC/AHA guidelines) to define low, moderate, and high-risk categories, respectively.12 NRI calculations were performed using three-year event data because this time point is between the median and 75th percentile for follow-up and optimizes the trade-off between loss of power for earlier time points and potential bias for later time points. Three-year event rates were then annualized to facilitate comparison with prior studies. Pre-test risk was defined by the annualized probability of MACE estimated by the multivariable clinical risk model. Post-test risk was defined by the annualized probability of MACE estimated by a model combining the multivariable clinical model and inducible ischemia. NRI was computed by pooling all upward reclassified subjects, calculating their Kaplan-Meier event rate, and performing the same calculation for all downward reclassified subjects.
All statistical analysis was performed using SAS (SAS Institutes, version 9.2, Cary, North Carolina). A two-tailed P-value < 0.05 was considered significant.
Results
Patient Characteristics
Indications for stress CMR referral included chest pain or dyspnea (N=700), palpitations or arrhythmias (N=57), syncope (N=20), and abnormality on ECG (N=38). Out of the 815 patients, 13 (2%) could not undergo or complete stress CMR due to severe claustrophobia (N=8), adverse or intolerable reaction to adenosine (N=4), or inadequate gating (N=1). Of the remaining 802 patients, 8 patients (1%) had non-diagnostic perfusion quality. Clinical follow-up was successful in all but 2 patients (99%). The remaining 792 patients formed the cohort for analysis (273 with prior CAD, 34%). Baseline patient and stress CMR characteristics stratified by presence or absence of prior CAD are shown in Table 1.
Table 1.
Baseline characteristics of the cohort and patients with prior CAD.
| Characteristics | All patients (n=792) | Prior CAD (n=273) |
|---|---|---|
| Age (years) | 56±14 | 61±12 |
| Female, n (%) | 319 (40) | 73 (27) |
| Body mass index, kg/m2 | 28±6 | 29±6 |
| Heart rate (beats/min) | 70±15 | 69±14 |
| Systolic blood pressure (mmHg) | 131±21 | 131±22 |
| Diastolic blood pressure (mmHg) | 72±12 | 71±13 |
| Angina, n (%) | 185 (24) | 115 (43) |
| History of heart failure, n (%) | 141 (18) | 67 (25) |
| Cardiac risk factors, n (%) | ||
| Hypercholesterolemia | 386 (49) | 196 (73) |
| Diabetes mellitus | 143 (18) | 75 (27) |
| Hypertension | 411 (52) | 179 (66) |
| Family history of CAD | 162 (22) | 61 (24) |
| Smoking | 128 (16) | 61 (23) |
| ≥3 risk factors | 182 (23) | 98 (36) |
| Medications, n (%) | ||
| Aspirin | 404 (52) | 220 (81) |
| β-adrenergic blockers | 412 (52) | 204 (75) |
| Statin | 390 (50) | 202 (75) |
| Calcium channel blockers | 115 (15) | 42 (16) |
| Nitroglycerin | 103 (13) | 82 (30) |
| ACE inhibitor | 346 (44) | 158 (59) |
| ECG abnormalities, n (%) | ||
| Left bundle branch block | 53 (7) | 17 (6) |
| Left ventricular hypertrophy | 68 (9) | 26 (10) |
| Resting ST segment changes | 116 (15) | 60 (22) |
| Pathologic Q wave | 110 (14) | 110 (41) |
| Ischemic ECG Abnormalities (Q waves, ST abnormality) | 196 (25) | 140 (51) |
| Stress CMR | ||
| LV ejection fraction, % | 56±15 | 51±16 |
| LVEDVI, ml/m2 | 89±31 | 96±35 |
| LVESVI, ml/m2 | 42±31 | 51±35 |
| LV mass index, g/m2 | 61±19 | 65±19 |
| RV ejection fraction, % | 54±9 | 54±10 |
| RVESVI, ml/m2 | 34±15 | 34±16 |
| Resting wall motion abnormality, n (%) | 206 (26) | 137 (51) |
| LGE, n (%) | 275 (35) | 175 (65) |
| LGE mass (g) | 5±12 | 11±16 |
| Inducible ischemia, n (%) | 210 (27) | 150 (55) |
| >10% ischemia, n (%) | 136 (17) | 106 (39) |
| Inducible ischemia score, median (IQR)* | 4.0 (4.0) | 4.0 (4.0) |
Data listed where available (N>=730 for categorical variables; N>= 708 for continuous variables).
Inducible ischemia score was calculated in patients with ischemia.
Abbreviations: ACE = angiotensin converting enzyme, CAD = coronary artery disease, ECG = electrocardiographic, IQR = interquartile range, LGE = late gadolinium enhancement, LV = left ventricular, LVEDVI = left ventricular end-diastolic volume index, LVESVI = left ventricular end-systolic volume index, RV = right ventricular, RVESVI = right ventricular end-systolic volume index
Univariable Associations with MACE and Event-free Survival
Univariable associations of selected clinical and stress CMR parameters with MACE were measured (Table 2). Among clinical, electrocardiographic, or stress CMR markers considered, presence and extent of inducible ischemia (>10% ischemia) demonstrated the highest association with MACE in the whole cohort (presence of ischemia HR 14.66, P<0.0001) and in patients with prior CAD (presence of ischemia HR 8.17, P=0.0001). In Kaplan-Meier analysis, patients with inducible ischemia experienced a substantial reduction in MACE-free survival as compared to patients without inducible ischemia in the whole cohort (Figure 1, P<0.0001) and in patients with prior CAD (P<0.0001). Cumulative hazard function (log-survival) plots of the first 5 years of follow-up after stress CMR illustrated that the rate of MACE was steady and markedly higher in patients with inducible ischemia than in those without ischemia (Figure 1). CAD patients without inducible ischemia qualitatively had an initially low accumulation of MACE in the 2 years after index CMR; after that, the rate of events started to increase. In the whole cohort and in patients with prior CAD, having inducible ischemia was associated with substantially higher mortality by Kaplan-Meier analysis than those without ischemia (Figure 2).
Table 2.
Univariable Cox association of clinical, electrocardiographic, and stress cardiac magnetic resonance indices with MACE.
| Characteristic | All Patients (n=792)
|
||
|---|---|---|---|
| Hazard Ratio | 95% CI | P value | |
| Demographics | |||
| Age (per year) | 1.06 | 1.03–1.08 | <0.0001 |
| Female | 1.06 | 0.60–1.87 | 0.85 |
| Body mass index (per 1 kg/m2) | 0.98 | 0.93–1.03 | 0.38 |
| Cardiac risk factors | |||
| Hypercholesterolemia | 2.17 | 1.17–4.05 | 0.015 |
| Diabetes mellitus | 2.71 | 1.53–4.83 | 0.0007 |
| Hypertension | 3.18 | 1.63–6.23 | 0.0007 |
| Family history of CAD | 0.93 | 0.47–1.82 | 0.83 |
| Smoking | 2.54 | 1.40–4.61 | 0.002 |
| Prior MI | 3.52 | 2.00–6.20 | <0.0001 |
| Prior revascularization | 9.66 | 5.43–17.21 | <0.0001 |
| ≥3 coronary risk factors | 2.91 | 1.66–5.11 | 0.0002 |
| ECG abnormalities | |||
| Left bundle branch block | 2.18 | 0.93–5.13 | 0.07 |
| Pathologic Q wave | 2.99 | 1.63–5.52 | 0.0004 |
| Resting ST segment Abnormalities | 4.01 | 2.23–7.18 | <0.0001 |
| Stress CMR | |||
| LVEF (Δ = −5%) | 1.19 | 1.09–1.29 | 0.0001 |
| LVEDVI (Δ +10 ml/m2) | 1.12 | 1.04–1.20 | 0.003 |
| LVESVI (Δ = +10 ml/m2) | 1.13 | 1.06–1.21 | 0.0001 |
| LVMI (Δ = +10gm/m2) | 1.23 | 1.10–1.36 | 0.0002 |
| RVEF (Δ = −5%) | 1.21 | 1.05–1.39 | 0.008 |
| RVESVI (Δ = +10ml/m2) | 1.26 | 1.13–1.40 | <0.0001 |
| Resting WMA | 5.19 | 2.89–9.32 | <0.0001 |
| Presence of LGE | 3.15 | 1.76–5.66 | 0.0001 |
| LGE mass (per gram) | 1.02 | 1.00–1.04 | 0.015 |
| Inducible ischemia | 14.66 | 6.87–31.31 | <0.0001 |
| >10% ischemic myocardium | 11.83 | 6.43–21.76 | <0.0001 |
| Inducible ischemia score | 1.17 | 1.12–1.22 | <0.0001 |
Abbreviations: CI = confidence interval, HR = hazard ratio, CAD = coronary artery disease, ECG = electrocardiographic, LGE = late gadolinium enhancement, LVEF = left ventricular ejection fraction, LVEDVI = left ventricular end-diastolic volume index, LVESVI = left ventricular end-systolic volume index, LVMI = left ventricular mass index, RVEF = right ventricular ejection fraction, RVESVI = right ventricular end-systolic volume index, WMA = wall motion abnormality
Figure 1.
Kaplan-Meier MACE-free survival curves for the whole cohort (left upper panel) and patients with prior CAD (right upper panel). Follow-up period was truncated to 8 years. The bottom panels display the corresponding cumulative hazard function for illustration of rates of MACE accumulation in these respective patient groups during the first 5 years after stress CMR assessment. P-value was derived using the log-rank test. Abbreviations: ISCH = inducible ischemia by stress CMR.
Figure 2.
Kaplan-Meier all cause mortality-free survival curves for the whole cohort (A) and patients with prior CAD (B). Follow-up period was truncated to 8 years. P-value was derived using the log-rank test. Abbreviations: ISCH = inducible ischemia by stress CMR.
Annual Rates of MACE and Mortality
Over a median follow up of 1.9 years (range 1–8.5 years), 68 deaths (8.6%) and 24 (3%) nonfatal acute MIs occurred. There were 29 cardiac deaths, attributed to fatal acute MI (N=4), congestive heart failure (N=20), or lethal ventricular arrhythmia (N=5). Annual event rates by inducible ischemia for the whole cohort and stratified by the presence or absence of prior CAD are shown in Figure 3. The annual rate of MACE for the whole cohort was 3.1%. Patients with absence of inducible ischemia experienced low annual rates of MACE (0.6%, 1.3%, and 0.4%, in the whole cohort, prior CAD, no prior CAD, respectively). On the other hand, patients with inducible ischemia had a higher annual rate of MACE (9.4%, 11%, and 5.5%, in the whole cohort, prior CAD, no prior CAD, respectively). Annual rates of cardiac death and all-cause mortality were significantly lower in patients who did not have inducible ischemia.
Figure 3.
Annualized event rates of MACE (A), cardiac death (B), and all-cause mortality (C) in the whole cohort and stratified by the presence (red) or absence (blue) of inducible ischemia. Comparison P-values were calculated by chi-square tests.
Improvement of Risk Discrimination and Net Reclassification Improvement
The multivariable clinical risk model consisted of age, history of CAD, hypertension, significant smoking, resting ST segment abnormalities on ECG, and LV ejection fraction (Table 3). Neither presence nor extent of LGE was selected to enter this clinical multivariable model. This model was well calibrated to observed MACE (Model 1, Calibration χ2 13.95, P=0.12). When inducible ischemia was added to this clinical risk model, it improved the model discrimination (C-statistic increased from 0.81 of Model 1 to 0.86 of Model 2, P=0.04; Adjusted hazard ratio 7.37, P<0.0001) and goodness-of-fit (global χ2=68.67 without to 98.28 with ischemia, P<0.0001; Table 3). Model 2 was well calibrated (Supplemental Figure 1; Calibration χ2 6.53, P=0.69). Inducible ischemia also improved discrimination of MACE by C-statistic in multivariable models for cardiac and all-cause mortality. While both LGE presence and LGE mass (in grams) demonstrated significant univariable association with MACE, neither provided incremental association with MACE above inducible ischemia in the whole cohort. Neither LGE presence nor LGE mass provided additional model risk discrimination when added to a model that combined the clinical risk model and inducible ischemia.
Table 3.
Multivariable survival analysis for prediction of MACE. Model 1 represents the multivariable clinical risk Cox regression model for MACE. Models 2, 3, and 4 represent addition of inducible ischemia, >10% ischemia, and inducible ischemia score, respectively, to Model 1. C-statistics of all new models (Models 2 through 4) were compared to model 1 (ΔP).
| Model 1 | Model 2 | Model 3 | Model 4 | |||||
|---|---|---|---|---|---|---|---|---|
|
| ||||||||
| Statistic | P | Statistic | P | Statistic | P | Statistic | P | |
| Global χ2 | 68.67 (6 d.f.) | Referent | 98.28 (7 d.f.) | <0.0001Δ | 96.39 (7 d.f.) | <0.0001Δ | 79.03 (7 d.f.) | 0.001Δ |
| Calibration χ2 | 13.95 | 0.12 | 6.53 | 0.69 | 6.66 | 0.67 | 15.25 | 0.08 |
| C statistic (95% CI) | 0.81 (0.74–0.89) | Referent | 0.86 (0.80–0.92) | 0.04Δ | 0.85 (0.78–0.92) | 0.05Δ | 0.82 (0.75–0.90) | 0.17Δ |
| Covariate | Hazard Ratio (95%CI) | Hazard Ratio (95%CI) | Hazard Ratio (95%CI) | Hazard Ratio (95%CI) | ||||
| Age | 1.04 (1.02–1.07) | 0.002 | 1.03 (1.00–1.06) | 0.03 | 1.04 (1.01–1.07) | 0.009 | 1.04 (1.02–1.07) | 0.002 |
| History of CAD | 3.42 (1.71–6.83) | 0.0005 | 1.84 (0.89–3.79) | 0.10 | 2.23 (0.99–5.00) | 0.05 | 2.75 (1.35–5.62) | 0.006 |
| History of hypertension | 1.90 (0.95–3.81) | 0.07 | 1.81 (0.91–3.60) | 0.09 | 1.60 (0.77–3.33) | 0.20 | 1.62 (0.80–3.26) | 0.18 |
| History of smoking | 1.86 (1.01–3.46) | 0.048 | 1.64 (0.88–3.05) | 0.12 | 1.53 (0.81–2.90) | 0.19 | 1.69 (0.90–3.18) | 0.10 |
| Resting ST abnormalities | 2.64 (1.44–4.83) | 0.002 | 2.58 (1.39–4.83) | 0.003 | 2.70 (1.45–5.05) | 0.002 | 2.42 (1.30–4.49) | 0.005 |
| LV ejection fraction (Δ=−5%) | 1.05 (0.98–1.16) | 0.12 | 1.02 (0.92–1.12) | 0.74 | 1.01 (0.92–1.11) | 0.87 | 1.03 (0.94–1.13) | 0.50 |
| Inducible ischemia | - | - | 7.37 (3.23–16.83) | <0.0001 | - | - | - | - |
| >10% ischemia | - | - | - | - | 5.25 (2.54–10.86) | <0.0001 | - | - |
| Inducible ischemia score | - | - | - | - | - | - | 1.11 (1.05–1.18) | 0.0003 |
Abbreviations: d.f.=degrees of freedom, CAD = coronary artery disease, CI = confidence interval, LV = left ventricular. Calibration statistics performed as indicated in text.
We also assessed NRI by inducible ischemia above the multivariable clinical risk model for MACE across ACC/AHA practice guideline recommended annualized risk categories (low <1%, moderate 1–3%, high >3%; Figure 4). Categorical NRI was 0.229 (95% 0.063–0.391; 0.037 for patients with events; 0.193 for patients without events). Continuous NRI was 1.11 (95% CI 0.81–1.39). As shown in Figure 4, risk reclassification by inducible ischemia was most effective in patients at moderate pre-test risk with reclassification of 65.7% (140/213) of patients to low risk and 25.8% (55/213) of patients to high risk by inducible ischemia, with a low (0.3%/year) and a high (4.9%/year) annual rate of MACE, respectively. For patients at high pretest risk, inducible ischemia reclassified 31.6% (60/190) and 5.8% (11/190) into moderate and low risk, with low annual rates of MACE (2.1% and 0%, respectively), which were in contrast to high annual rate of MACE (14.3%) among patients who remained at high post-test risk (Table 4).
Figure 4.
Risk reclassification improvement. Presence of inducible ischemia was added to the multivariable clinical risk model (Model 1 in Table 3) for risk reclassification across ACC/AHA practice guideline categories. Pie charts demonstrate proportion of patients reclassified by the addition of inducible ischemia across pre-test risk categories. Observed annualized rates of MACE for reclassified patients were displayed in bar graphs.
Table 4.
Reclassification table of risk of MACE from the multivariable clinical risk model (pre-test risk) with addition of inducible ischemia by stress CMR (post-test risk).
| Clinical model alone (pre-test risk) | Clinical model + inducible ischemia (post-test risk) | |||
|---|---|---|---|---|
| Patients without MACE | Low (<1%) | Moderate (1–3%) | High (>3%) | Total |
| Low (<1%) | 307.1 (92.5) | 22 (6.6) | 3 (0.90) | 332.1 |
| Moderate (1–3%) | 140 (67.6) | 18 (8.7) | 49.1 (23.7) | 207.1 |
| High (>3%) | 11 (7.4) | 55.6 (37.5) | 81.8 (55.1) | 148.4 |
| Total | 458.1 | 95.6 | 133.9 | 687.5 |
| Patients with MACE | ||||
| Low (<1%) | 3.9 (79.6) | 1 (20.4) | 0 | 4.9 |
| Moderate (1–3%) | 0 | 0 | 5.9 (100) | 5.9 |
| High (>3%) | 0 | 4.4 (10.6) | 37.2 (89.4) | 41.6 |
| Total | 3.9 | 5.4 | 43.1 | 52.4 |
Reclassification for censored data (to 3 years’ follow-up after index CMR) was performed by published methods (see text). Values in parentheses indicate the percentage of patients reclassified by inducible ischemia from the initial pre-test risk (row) to each post-test risk category (column). Number of events in each stratum is calculated based on Kaplan-Meier estimates (and are not necessarily whole numbers). NRI reported in text is not calculated directly from the reclassification table below: NRI was computed by pooling all upward reclassified subjects, calculating their Kaplan-Meier event rate, and performing the same calculation for all downward reclassified subjects.
We also built the multivariable clinical model without inclusion of any CMR parameters such as LV function, LV sizes, or LGE. This model consisted of age, history of CAD, diabetes, significant smoking, and resting ST segment changes. Adding inducible ischemia to the clinical model substantially improved the model discrimination of MACE (C-statistic 0.79 to 0.85, P=0.01; Adjusted hazard ratio 7.36, P<0.0001) and model goodness-of-fit (global χ2=67.95 without to 97.49 with ischemia, from 6 to 7 degrees of freedom, respectively, P<0.0001). Inducible ischemia also provided effective and robust risk reclassification to this multivariable clinical model: categorical NRI was 0.28 (95% CI 0.12–0.44) and continuous NRI was 1.16 (95% CI 0.88–1.43).
Similarly, inducible ischemia provided effective risk reclassification in patients with prior CAD. Absence of inducible ischemia reclassified 43.5% (30/69) of patients at moderate pre-test risk to low risk and these patients experienced low observed post-test annual rates of MACE (1.2%). In addition, absence of inducible ischemia reclassified 32.5% of high pre-test risk patients with prior CAD to moderate risk, and these patients also experienced a low observed annual rate of MACE (1.4%). In contrast, patients at high risk pre-test who remained high risk post-test experienced a 13.4% annual rate of MACE.
Association of Extent of Myocardial Ischemia with Outcome
Extent of ischemia, assessed either by >10% ischemic myocardium or inducible ischemia score, each provided strong and independent association with MACE (Table 1). More than 10% ischemic myocardium was observed in 136 patients (17%). The rate of MACE in patients without >10% ischemic myocardium was 1.1%, 2.5%, and 0.5% annually in patients from the whole cohort, those with prior CAD, and those without prior CAD, respectively. Conversely, the annual rate of MACE in patients with >10% ischemic myocardium was 12%, 13%, and 11%, in the overall cohort, with and without CAD, respectively. When adjusted to all covariates in the clinical risk model, the presence of >10% ischemia and inducible ischemia score each provided incremental prognosis to MACE (Model 3: HR 5.25, 95% CI 2.54–10.86, P<0.0001 and Model 4: HR 1.11, 95% CI 1.05–1.18, P=0.0003, respectively; Table 3). The presence of >10% ischemic myocardium reclassified risk above a multivariable clinical risk model for MACE across guideline-based annual risk categories (categorical NRI, 0.16; continuous NRI, 0.82 in the whole cohort). Among 213 patients at moderate pre-test risk, >10% ischemic myocardium reclassified 84 (39%) and 39 (18%) as low and high risk, respectively. A low 0.5% and high 6% annual rate of MACE were observed in these groups, respectively.
Effects of Early Coronary Revascularization
Early coronary revascularization (within 90 days following stress CMR) occurred in 76 patients (9.6%) and was performed in patients with a greater extent of ischemia by CMR (ischemia score 4.9±4.2 and 0.8±2.3, did and did not receive early revascularization, respectively, P<0.0001). The interaction term between early revascularization and ischemia score was significant (P = 0.02) when included in the multivariable clinical model for MACE, indicating effect modification by early revascularization on the association between ischemia score and MACE. There were 17 patients who underwent early revascularization and experienced MACE. A greater extent of ischemia was strongly associated with increased hazard of MACE in patients who did not undergo early revascularization (32 patients with MACE, HR=1.18, 95% CI 1.13–1.24; P<0.0001), but this association was not significant in patients who received early revascularization (HR=1.06, 95% CI 0.95–1.18; P=0.30). Similarly among patients with prior CAD, a greater extent of ischemia score demonstrated increased hazards for MACE for those who did not received early revascularization (N=211, 24 MACE, HR=1.11, 95% CI 1.05–1.18; P=0.0007), but such association was not significant in those who received revascularization (N=62, 14 MACE; HR=1.04, 95% CI 0.92–1.18; P = 0.51).
When patients who underwent early coronary revascularization were censored on the day of the procedure, inducible ischemia, inducible ischemia score, and >10% ischemia all maintained strong association with MACE (HR 13.66, 1.19, and 12.09, respectively; all P<0.0001). Inducible ischemia also improved discrimination for MACE beyond the multivariable clinical model (consisted of age, prior CAD, ST segment change, and LVEF) (C-statistic 0.82 to 0.87; P=0.04, Adjusted hazard ratio 7.16, P<0.0001) and improved the model goodness-of-fit (global χ2=51.96 without to 76.88 with ischemia, from 4 to 5 degrees of freedom, respectively, P<0.0001). For patients with moderate pre-test clinical risk, addition of inducible ischemia reclassified 89% of patients (57% to low risk; 32% to high risk) with corresponding changes in the observed event rates (0.4%/year and 5.4%/year, for low and high risk post-test, respectively). Categorical and continuous NRI were 0.208 (95% CI 0.001–0.414) and 1.21 (95% CI 0.87–1.57), respectively.
Discussion
We found that stress CMR provides highly effective patient risk reclassification for cardiac death and non-fatal MI beyond clinical covariates in both patients with suspected and established CAD. The overall rate of MACE in our population was 3.1%/year, which is comparable to large registry data from the nuclear scintigraphy literature.13 Inducible ischemia by stress CMR was the strongest univariable and multivariable predictor of cardiac death and MI. An absence of inducible ischemia identified a subgroup of patients with excellent cardiac prognosis (<1% risk of cardiac death) even in patients with known CAD who are at high risk. Most importantly, we found that stress CMR improved risk reclassification for the vast majority of patients at intermediate pre-test risk in both the whole cohort as well as in patients with known CAD. These results were supported by a low observed rate of MACE (0.3%/year) in those intermediate pre-test risk patients reclassified to low risk by the absence of inducible ischemia. On the other hand, intermediate pre-test risk patients reclassified to high risk by the presence of inducible ischemia had very high rates of MACE (~5%/year), substantially higher than ACC/AHA proposed high-risk category. Patients without inducible ischemia had a very low rate of accumulation of cardiac events for 2 years following stress CMR. With 98% of studies achieving diagnostic quality, stress CMR is a practical and robust method, which can discriminate and reclassify patient cardiac risk and potentially guide clinical decision making for further invasive investigation.
Current clinical risk algorithms and stress testing in patients with prior CAD remains challenging. Despite excellent long-term prognosis with a negative nuclear stress test in an overall referral population (<1% annual rate of MACE),14–16 there remains significant heterogeneity of risk ascribed to patient-specific cardiac risk factors and the presence and extent of CAD.17 In a registry study of 7,376 patients, Hachamovitch et al. demonstrated higher rates of MACE in patients with prior CAD versus those without, in setting of normal nuclear perfusion imaging in both groups (1.4% vs. 0.4%/year, respectively).18 While non-CAD patients with a normal test had stable hazard of MACE over time, hazard of MACE increased for every 6 months after index testing in patients with prior CAD.18 Furthermore, Rozanski et al.19 and others20, 21 showed that patients who were unable to perform exercise stress and had to undergo vasodilator stress testing, experienced higher rates of MACE (approaching 3.9%/yr) despite a normal nuclear perfusion imaging. These large, well-designed studies reaffirmed the inherent Bayesian limitations in further risk stratifying patients already at moderate to high pre-test clinical risk. Our study results showed that stress CMR is a promising tool for risk stratification in these common clinical settings.
Although a growing number of clinical studies have demonstrated strong prognostic association of stress CMR findings with major cardiovascular events,22–25 the ability of CMR to impact clinical management via risk reclassification is unknown. We show that stress CMR improved risk reclassification in the majority of patients at intermediate-to-high pre-test risk, regardless of status of prior CAD, across ACC/AHA practice guideline recommended strata that guide current clinical management decisions. Our study also provided observational supportive evidence that knowledge of ischemia extent provided to clinicians guiding their decisions to perform early coronary revascularization, may have altered patient outlook. In the context of current guideline recommendations for coronary revascularization in patients at high risk for cardiac events, our results suggest that stress CMR effectively defines and reclassifies those patients who may benefit from these invasive investigations.
Our study has several limitations. The sample size and number of adverse events allowed assessment of the prognostic value of ischemia by CMR with adjustments to only a limited number of clinical covariates. We included CMR-based LV functional parameters and used a backward elimination strategy to build a robust clinical model against which inducible ischemia by CMR still emerged to provide incremental patient risk reclassification. Whether CMR can guide decision in utilizing invasive approach in addition to medical therapy requires prospectively evaluation. Third, we assessed the NRI of inducible ischemia for MACE (composite cardiac death and acute MI) due to the limited sample size and number of cases of cardiac death in our cohort; whether these results are applicable to risk reclassification for cardiac mortality alone require further evaluation.
In conclusion, stress CMR provides effective risk reclassification across ACC/AHA guideline-recommended categories, most notably in patients with intermediate pre-test risk. Future investigations involving stress CMR as a part of a comprehensive strategy to guide clinical decision-making for invasive angiography and mechanical revascularization in addition to medical therapy merits prospective evaluation.
Supplementary Material
SHORT COMMENTARY.
Recent results from the COURAGE trial suggest that an initial strategy of mechanical reperfusion may not improve cardiac outcomes beyond optimized medical therapy in patients with stable coronary artery disease (CAD). Nevertheless, a subgroup of patients with significant myocardial ischemia may still benefit from coronary revascularization, and current practice guidelines recommend CAD patients at high risk for adverse events be considered for revascularization. Given increasing concern over effective resource utilization in cardiac imaging, establishing evidence indicating how cardiac imaging successfully influences clinical decision-making is imperative. We studied 815 consecutive patients referred for evaluation of myocardial ischemia by stress cardiac magnetic resonance imaging (CMR), finding that inducible ischemia had the strongest association with major adverse cardiovascular events that include cardiac death or myocardial infarction, after adjustment to clinical predictors, prior CAD, and LV ejection fraction. Absence of inducible ischemia by CMR was associated with low annual rate of events in the entire population and in a subgroup with CAD. Adding inducible ischemia to a clinical risk model reclassified over 90% of patients at moderate pre-test risk, with corresponding observed event rates of 0.3%/year and 4.9%/year, for low and high risk post-test, respectively. These results demonstrate that CMR effectively reclassifies risk beyond clinical risk predictors, specifically in the moderate pre-test risk subgroup and in patients with prior CAD. Stress CMR offers an effective strategy to safely manage some patients without the need for invasive angiography.
Acknowledgments
Funding Sources: Dr. Shah is supported by an American Heart Association Post-Doctoral Fellowship Award (11POST000002) and a training grant from the Heart Failure NIH Clinical Research Network (U01-HL084877). Dr. Heydari is supported by a Clinical Fellowship Award from the Alberta Heritage Foundation for Medical Research. Dr. Murthy and Dr. Neilan were supported by NIH training grant T32-HL094301.
Footnotes
Conflict of Interest Disclosures: Dr. Kwong is supported by a NIH grant R01-HL091157 and a research grant from Astellas Pharmaceuticals. All other authors have no financial disclosures relevant to the content of this manuscript.
References
- 1.Boden WE, O’Rourke RA, Teo KK, Hartigan PM, Maron DJ, Kostuk WJ, Knudtson M, Dada M, Casperson P, Harris CL, Chaitman BR, Shaw L, Gosselin G, Nawaz S, Title LM, Gau G, Blaustein AS, Booth DC, Bates ER, Spertus JA, Berman DS, Mancini GB, Weintraub WS. Optimal medical therapy with or without pci for stable coronary disease. N Engl J Med. 2007;356:1503–1516. doi: 10.1056/NEJMoa070829. [DOI] [PubMed] [Google Scholar]
- 2.Shaw LJ, Berman DS, Maron DJ, Mancini GB, Hayes SW, Hartigan PM, Weintraub WS, O’Rourke RA, Dada M, Spertus JA, Chaitman BR, Friedman J, Slomka P, Heller GV, Germano G, Gosselin G, Berger P, Kostuk WJ, Schwartz RG, Knudtson M, Veledar E, Bates ER, McCallister B, Teo KK, Boden WE. Optimal medical therapy with or without percutaneous coronary intervention to reduce ischemic burden: Results from the clinical outcomes utilizing revascularization and aggressive drug evaluation (courage) trial nuclear substudy. Circulation. 2008;117:1283–1291. doi: 10.1161/CIRCULATIONAHA.107.743963. [DOI] [PubMed] [Google Scholar]
- 3.Hachamovitch R, Hayes SW, Friedman JD, Cohen I, Berman DS. Comparison of the short-term survival benefit associated with revascularization compared with medical therapy in patients with no prior coronary artery disease undergoing stress myocardial perfusion single photon emission computed tomography. Circulation. 2003;107:2900–2907. doi: 10.1161/01.CIR.0000072790.23090.41. [DOI] [PubMed] [Google Scholar]
- 4.Gibbons RJ, Abrams J, Chatterjee K, Daley J, Deedwania PC, Douglas JS, Ferguson TB, Jr, Fihn SD, Fraker TD, Jr, Gardin JM, O’Rourke RA, Pasternak RC, Williams SV, Alpert JS, Antman EM, Hiratzka LF, Fuster V, Faxon DP, Gregoratos G, Jacobs AK, Smith SC., Jr Acc/aha 2002 guideline update for the management of patients with chronic stable angina--summary article: A report of the american college of cardiology/american heart association task force on practice guidelines (committee on the management of patients with chronic stable angina) Circulation. 2003;107:149–158. doi: 10.1161/01.cir.0000047041.66447.29. [DOI] [PubMed] [Google Scholar]
- 5.Greenwood JP, Maredia N, Younger JF, Brown JM, Nixon J, Everett CC, Bijsterveld P, Ridgway JP, Radjenovic A, Dickinson CJ, Ball SG, Plein S. Cardiovascular magnetic resonance and single-photon emission computed tomography for diagnosis of coronary heart disease (cemarc): A prospective trial. Lancet. 2012;379:453–460. doi: 10.1016/S0140-6736(11)61335-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Hamon M, Fau G, Nee G, Ehtisham J, Morello R. Meta-analysis of the diagnostic performance of stress perfusion cardiovascular magnetic resonance for detection of coronary artery disease. J Cardiovasc Magn Reson. 2010;12:29. doi: 10.1186/1532-429X-12-29. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Thygesen K, Alpert JS, White HD. Universal definition of myocardial infarction. J Am Coll Cardiol. 2007;50:2173–2195. doi: 10.1016/j.jacc.2007.09.011. [DOI] [PubMed] [Google Scholar]
- 8.Coelho-Filho OR, Seabra LF, Mongeon FP, Abdullah SM, Francis SA, Blankstein R, Di Carli MF, Jerosch-Herold M, Kwong RY. Stress myocardial perfusion imaging by cmr provides strong prognostic value to cardiac events regardless of patient’s sex. JACC Cardiovasc Imaging. 2011;4:850–861. doi: 10.1016/j.jcmg.2011.04.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Kwong RY, Chan AK, Brown KA, Chan CW, Reynolds HG, Tsang S, Davis RB. Impact of unrecognized myocardial scar detected by cardiac magnetic resonance imaging on event-free survival in patients presenting with signs or symptoms of coronary artery disease. Circulation. 2006;113:2733–2743. doi: 10.1161/CIRCULATIONAHA.105.570648. [DOI] [PubMed] [Google Scholar]
- 10.Hosmer DW, Hosmer T, Le Cessie S, Lemeshow S. A comparison of goodness-of-fit tests for the logistic regression model. Stat Med. 1997;16:965–980. doi: 10.1002/(sici)1097-0258(19970515)16:9<965::aid-sim509>3.0.co;2-o. [DOI] [PubMed] [Google Scholar]
- 11.Pencina MJ, D’Agostino RB, Sr, D’Agostino RB, Jr, Vasan RS. Evaluating the added predictive ability of a new marker: From area under the roc curve to reclassification and beyond. Stat Med. 2008;27:157–172. doi: 10.1002/sim.2929. discussion 207–112. [DOI] [PubMed] [Google Scholar]
- 12.Fihn SD, Gardin JM, Abrams J, Berra K, Blankenship JC, Dallas AP, Douglas PS, Foody JM, Gerber TC, Hinderliter AL, King SB, 3rd, Kligfield PD, Krumholz HM, Kwong RY, Lim MJ, Linderbaum JA, Mack MJ, Munger MA, Prager RL, Sabik JF, Shaw LJ, Sikkema JD, Smith CR, Jr, Smith SC, Jr, Spertus JA, Williams SV, Anderson JL. 2012 accf/aha/acp/aats/pcna/scai/sts guideline for the diagnosis and management of patients with stable ischemic heart disease: A report of the american college of cardiology foundation/american heart association task force on practice guidelines, and the american college of physicians, american association for thoracic surgery, preventive cardiovascular nurses association, society for cardiovascular angiography and interventions, and society of thoracic surgeons. Circulation. 2012;126:e354–471. doi: 10.1161/CIR.0b013e318277d6a0. [DOI] [PubMed] [Google Scholar]
- 13.Hachamovitch R, Berman DS, Kiat H, Cohen I, Friedman JD, Shaw LJ. Value of stress myocardial perfusion single photon emission computed tomography in patients with normal resting electrocardiograms: An evaluation of incremental prognostic value and cost-effectiveness. Circulation. 2002;105:823–829. doi: 10.1161/hc0702.103973. [DOI] [PubMed] [Google Scholar]
- 14.Shaw LJ, Iskandrian AE. Prognostic value of gated myocardial perfusion spect. J Nucl Cardiol. 2004;11:171–185. doi: 10.1016/j.nuclcard.2003.12.004. [DOI] [PubMed] [Google Scholar]
- 15.Machecourt J, Longere P, Fagret D, Vanzetto G, Wolf JE, Polidori C, Comet M, Denis B. Prognostic value of thallium-201 single-photon emission computed tomographic myocardial perfusion imaging according to extent of myocardial defect. Study in 1,926 patients with follow-up at 33 months. J Am Coll Cardiol. 1994;23:1096–1106. doi: 10.1016/0735-1097(94)90597-5. [DOI] [PubMed] [Google Scholar]
- 16.Hachamovitch R, Berman DS, Kiat H, Cohen I, Cabico JA, Friedman J, Diamond GA. Exercise myocardial perfusion spect in patients without known coronary artery disease: Incremental prognostic value and use in risk stratification. Circulation. 1996;93:905–914. doi: 10.1161/01.cir.93.5.905. [DOI] [PubMed] [Google Scholar]
- 17.Shaw LJ, Hage FG, Berman DS, Hachamovitch R, Iskandrian A. Prognosis in the era of comparative effectiveness research: Where is nuclear cardiology now and where should it be? J Nucl Cardiol. 2012 Oct;19:1026–43. doi: 10.1007/s12350-012-9593-y. [DOI] [PubMed] [Google Scholar]
- 18.Hachamovitch R, Hayes S, Friedman JD, Cohen I, Shaw LJ, Germano G, Berman DS. Determinants of risk and its temporal variation in patients with normal stress myocardial perfusion scans: What is the warranty period of a normal scan? J Am Coll Cardiol. 2003;41:1329–1340. doi: 10.1016/s0735-1097(03)00125-6. [DOI] [PubMed] [Google Scholar]
- 19.Rozanski A, Gransar H, Hayes SW, Friedman JD, Hachamovitch R, Berman DS. Comparison of long-term mortality risk following normal exercise vs adenosine myocardial perfusion spect. J Nucl Cardiol. 2010;17:999–1008. doi: 10.1007/s12350-010-9300-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Navare SM, Mather JF, Shaw LJ, Fowler MS, Heller GV. Comparison of risk stratification with pharmacologic and exercise stress myocardial perfusion imaging: A meta-analysis. J Nucl Cardiol. 2004;11:551–561. doi: 10.1016/j.nuclcard.2004.06.128. [DOI] [PubMed] [Google Scholar]
- 21.Johnson NP, Schimmel DR, Jr, Dyer SP, Leonard SM, Holly TA. Survival by stress modality in patients with a normal myocardial perfusion study. Am J Cardiol. 2011;107:986–989. doi: 10.1016/j.amjcard.2010.11.022. [DOI] [PubMed] [Google Scholar]
- 22.Jahnke C, Nagel E, Gebker R, Kokocinski T, Kelle S, Manka R, Fleck E, Paetsch I. Prognostic value of cardiac magnetic resonance stress tests: Adenosine stress perfusion and dobutamine stress wall motion imaging. Circulation. 2007;115:1769–1776. doi: 10.1161/CIRCULATIONAHA.106.652016. [DOI] [PubMed] [Google Scholar]
- 23.Bingham SE, Hachamovitch R. Incremental prognostic significance of combined cardiac magnetic resonance imaging, adenosine stress perfusion, delayed enhancement, and left ventricular function over preimaging information for the prediction of adverse events. Circulation. 2011;123:1509–1518. doi: 10.1161/CIRCULATIONAHA.109.907659. [DOI] [PubMed] [Google Scholar]
- 24.Bodi V, Sanchis J, Lopez-Lereu MP, Nunez J, Mainar L, Monmeneu JV, Husser O, Dominguez E, Chorro FJ, Llacer A. Prognostic value of dipyridamole stress cardiovascular magnetic resonance imaging in patients with known or suspected coronary artery disease. J Am Coll Cardiol. 2007;50:1174–1179. doi: 10.1016/j.jacc.2007.06.016. [DOI] [PubMed] [Google Scholar]
- 25.Steel K, Broderick R, Gandla V, Larose E, Resnic F, Jerosch-Herold M, Brown KA, Kwong RY. Complementary prognostic values of stress myocardial perfusion and late gadolinium enhancement imaging by cardiac magnetic resonance in patients with known or suspected coronary artery disease. Circulation. 2009;120:1390–1400. doi: 10.1161/CIRCULATIONAHA.108.812503. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.