[Skip to Content]
[Skip to Content Landing]
Figure 1.  Automated Process of Data Collection
Automated Process of Data Collection

Each of the 7 hospitals used the same software for routine clinical cardiac magnetic resonance image interpretation (CMR) and reporting. The clinical data were then automatically deidentified and sent to the cloud database. EHR indicates electronic health record; HL7, health level 7 communication standard; SSDI, Social Security Death Index.

Figure 2.  Kaplan-Meier Survival Curves for All Patients After Adjustment for Age, Sex, and Cardiac Risk Factors
Kaplan-Meier Survival Curves for All Patients After Adjustment for Age, Sex, and Cardiac Risk Factors

The thinner lines indicate 95% confidence intervals. Numbers at bottom indicate patients at risk.

Figure 3.  Kaplan-Meier Survival Curves for All Patients (N = 9151) in 8 Subpopulations After Adjustment for Patient Age, Sex, and Cardiac Risk Factors
Kaplan-Meier Survival Curves for All Patients (N = 9151) in 8 Subpopulations After Adjustment for Patient Age, Sex, and Cardiac Risk Factors

CAD indicates coronary artery disease; LVEF, left ventricular ejection fraction.

Table 1.  Baseline Patient Characteristics
Baseline Patient Characteristics
Table 2.  Multivariable Analysisa
Multivariable Analysisa
1.
Amsterdam  EA, Wenger  NK, Brindis  RG,  et al.  2014 AHA/ACC Guideline for the management of patients with non–ST-elevation acute coronary syndromes.  J Am Coll Cardiol. 2014;64(24):e139-e228. doi:10.1016/j.jacc.2014.09.017PubMedGoogle ScholarCrossref
2.
Yancy  CW, Jessup  M, Bozkurt  B,  et al; American College of Cardiology Foundation; American Heart Association Task Force on Practice Guidelines.  2013 ACCF/AHA Guideline for the management of heart failure.  J Am Coll Cardiol. 2013;62(16):e147-e239. doi:10.1016/j.jacc.2013.05.019PubMedGoogle ScholarCrossref
3.
Nandalur  KR, Dwamena  BA, Choudhri  AF, Nandalur  MR, Carlos  RC.  Diagnostic performance of stress cardiac magnetic resonance imaging in the detection of coronary artery disease: a meta-analysis.  J Am Coll Cardiol. 2007;50(14):1343-1353. doi:10.1016/j.jacc.2007.06.030PubMedGoogle ScholarCrossref
4.
Hamon  M, Fau  G, Née  G, Ehtisham  J, Morello  R, Hamon  M.  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-12-29PubMedGoogle ScholarCrossref
5.
Greenwood  JP, Maredia  N, Younger  JF,  et al.  Cardiovascular magnetic resonance and single-photon emission computed tomography for diagnosis of coronary heart disease (CE-MARC): a prospective trial.  Lancet. 2012;379(9814):453-460. doi:10.1016/S0140-6736(11)61335-4PubMedGoogle ScholarCrossref
6.
Schwitter  J, Wacker  CM, van Rossum  AC,  et al.  MR-IMPACT: comparison of perfusion-cardiac magnetic resonance with single-photon emission computed tomography for the detection of coronary artery disease in a multicentre, multivendor, randomized trial.  Eur Heart J. 2008;29(4):480-489. doi:10.1093/eurheartj/ehm617PubMedGoogle ScholarCrossref
7.
Schwitter  J, Wacker  CM, Wilke  N,  et al; MR-IMPACT Investigators.  MR-IMPACT II: Magnetic Resonance Imaging for Myocardial Perfusion Assessment in Coronary artery disease Trial: perfusion-cardiac magnetic resonance vs. single-photon emission computed tomography for the detection of coronary artery disease: a comparative multicentre, multivendor trial.  Eur Heart J. 2013;34(10):775-781. doi:10.1093/eurheartj/ehs022PubMedGoogle ScholarCrossref
8.
Heitner  JF, Klem  I, Rasheed  D,  et al.  Stress cardiac MR imaging compared with stress echocardiography in the early evaluation of patients who present to the emergency department with intermediate-risk chest pain.  Radiology. 2014;271(1):56-64. doi:10.1148/radiol.13130557PubMedGoogle ScholarCrossref
9.
U.S. Centers for Medicare and Medicaid Services. Medicare provider utilization and payment data. https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Medicare-Provider-Charge-Data/Physician-and-Other-Supplier.html. Accessed May 24, 2018.
10.
Fordyce  CB, Newby  DE, Douglas  PS.  Diagnostic strategies for the evaluation of chest pain: clinical implications from SCOT-HEART and PROMISE.  J Am Coll Cardiol. 2016;67(7):843-852. doi:10.1016/j.jacc.2015.11.055PubMedGoogle ScholarCrossref
11.
Lipinski  MJ, McVey  CM, Berger  JS, Kramer  CM, Salerno  M.  Prognostic value of stress cardiac magnetic resonance imaging in patients with known or suspected coronary artery disease: a systematic review and meta-analysis.  J Am Coll Cardiol. 2013;62(9):826-838. doi:10.1016/j.jacc.2013.03.080PubMedGoogle ScholarCrossref
12.
Gargiulo  P, Dellegrottaglie  S, Bruzzese  D,  et al.  The prognostic value of normal stress cardiac magnetic resonance in patients with known or suspected coronary artery disease: a meta-analysis.  Circ Cardiovasc Imaging. 2013;6(4):574-582. doi:10.1161/CIRCIMAGING.113.000035PubMedGoogle ScholarCrossref
13.
Greenwood  JP, Herzog  BA, Brown  JM,  et al.  Prognostic value of cardiovascular magnetic resonance and single-photon emission computed tomography in suspected coronary heart disease: long-term follow-up of a prospective, diagnostic accuracy cohort study.  Ann Intern Med. 2016;165:1-9. doi:10.7326/M15-1801PubMedGoogle ScholarCrossref
14.
Sammut  EC, Villa  ADM, Di Giovine  G,  et al.  Prognostic value of quantitative stress perfusion cardiac magnetic resonance.  JACC Cardiovasc Imaging. 2018;11(5):686-694. doi:10.1016/j.jcmg.2017.07.022PubMedGoogle ScholarCrossref
15.
21st Century Cures Act, HR 34, 114th Cong (2015).
16.
US Food and Drug Administration. Use of real-world evidence to support regulatory decision-making for medical devices, guidance for Industry and Food and Drug Administration Staff. https://www.fda.gov/downloads/medicaldevices/deviceregulationandguidance/guidancedocuments/ucm513027.pdf. Published August 31, 2017. Accessed February 1, 2019.
17.
US Social Security Death Index. https://www.ssdmf.com/. Accessed April 24, 2018.
18.
Klem  I, Heitner  JF, Shah  DJ,  et al.  Improved detection of coronary artery disease by stress perfusion cardiovascular magnetic resonance with the use of delayed enhancement infarction imaging.  J Am Coll Cardiol. 2006;47(8):1630-1638. doi:10.1016/j.jacc.2005.10.074PubMedGoogle ScholarCrossref
19.
Kim  RJ, Farzaneh-Far  A.  The diagnostic utility of cardiovascular magnetic resonance in patients with chest pain, elevated cardiac enzymes and non-obstructed coronary arteries.  Rev Esp Cardiol. 2009;62(9):966-971. doi:10.1016/S0300-8932(09)72093-7PubMedGoogle ScholarCrossref
20.
Kwong  RY, Farzaneh-Far  A.  Measuring myocardial scar by CMR.  JACC Cardiovasc Imaging. 2011;4(2):157-160. doi:10.1016/j.jcmg.2010.12.004PubMedGoogle ScholarCrossref
21.
Abbasi  SA, Ertel  A, Shah  RV,  et al.  Impact of cardiovascular magnetic resonance on management and clinical decision-making in heart failure patients.  J Cardiovasc Magn Reson. 2013;15:89. doi:10.1186/1532-429X-15-89PubMedGoogle ScholarCrossref
22.
Dandekar  VK, Bauml  MA, Ertel  AW, Dickens  C, Gonzalez  RC, Farzaneh-Far  A.  Assessment of global myocardial perfusion reserve using cardiovascular magnetic resonance of coronary sinus flow at 3 Tesla.  J Cardiovasc Magn Reson. 2014;16:24. doi:10.1186/1532-429X-16-24PubMedGoogle ScholarCrossref
23.
McGraw  S, Mirza  O, Bauml  MA, Rangarajan  VS, Farzaneh-Far  A.  Downstream clinical consequences of stress cardiovascular magnetic resonance based on appropriate use criteria.  J Cardiovasc Magn Reson. 2015;17:35. doi:10.1186/s12968-015-0137-xPubMedGoogle ScholarCrossref
24.
Framingham Heart Study, Risk Function for Cardiovascular Disease (10-year risk). https://www.framinghamheartstudy.org/fhs-risk-functions/cardiovascular-disease-10-year-risk/. Accessed May 3, 2018.
25.
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(2):157-172. doi:10.1002/sim.2929PubMedGoogle ScholarCrossref
26.
Steyerberg  EW, Pencina  MJ, Lingsma  HF, Kattan  MW, Vickers  AJ, Van Calster  B.  Assessing the incremental value of diagnostic and prognostic markers: a review and illustration.  Eur J Clin Invest. 2012;42(2):216-228. doi:10.1111/j.1365-2362.2011.02562.xPubMedGoogle ScholarCrossref
27.
Bothwell  LE, Greene  JA, Podolsky  SH, Jones  DS.  Assessing the gold standard: lessons from the history of RCTs.  N Engl J Med. 2016;374(22):2175-2181. doi:10.1056/NEJMms1604593PubMedGoogle ScholarCrossref
28.
Booth  CM, Tannock  IF.  Randomised controlled trials and population-based observational research: partners in the evolution of medical evidence.  Br J Cancer. 2014;110(3):551-555. doi:10.1038/bjc.2013.725PubMedGoogle ScholarCrossref
29.
Sherman  RE, Anderson  SA, Dal Pan  GJ,  et al.  Real-world evidence: what is it and what can it tell us?  N Engl J Med. 2016;375(23):2293-2297. doi:10.1056/NEJMsb1609216PubMedGoogle ScholarCrossref
30.
Hachamovitch  R, Rozanski  A, Shaw  LJ,  et al.  Impact of ischaemia and scar on the therapeutic benefit derived from myocardial revascularization vs medical therapy among patients undergoing stress-rest myocardial perfusion scintigraphy.  Eur Heart J. 2011;32(8):1012-1024. doi:10.1093/eurheartj/ehq500PubMedGoogle ScholarCrossref
31.
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(14):1509-1518. doi:10.1161/CIRCULATIONAHA.109.907659PubMedGoogle ScholarCrossref
32.
Hachamovitch  R, Berman  DS, Shaw  LJ,  et al.  Incremental prognostic value of myocardial perfusion single photon emission computed tomography for the prediction of cardiac death: differential stratification for risk of cardiac death and myocardial infarction.  Circulation. 1998;97(6):535-543. doi:10.1161/01.CIR.97.6.535PubMedGoogle ScholarCrossref
33.
Hachamovitch  R, Hayes  S, Friedman  JD,  et al.  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(8):1329-1340. doi:10.1016/S0735-1097(03)00125-6PubMedGoogle ScholarCrossref
34.
Klem  I, Shah  DJ, White  RD,  et al.  Prognostic value of routine cardiac magnetic resonance assessment of left ventricular ejection fraction and myocardial damage: an international, multicenter study.  Circ Cardiovasc Imaging. 2011;4(6):610-619. doi:10.1161/CIRCIMAGING.111.964965PubMedGoogle ScholarCrossref
35.
Stanton  T, Leano  R, Marwick  TH.  Prediction of all-cause mortality from global longitudinal speckle strain: comparison with ejection fraction and wall motion scoring.  Circ Cardiovasc Imaging. 2009;2(5):356-364. doi:10.1161/CIRCIMAGING.109.862334PubMedGoogle ScholarCrossref
36.
Lauer  MS, Blackstone  EH, Young  JB, Topol  EJ.  Cause of death in clinical research: time for a reassessment?  J Am Coll Cardiol. 1999;34(3):618-620. doi:10.1016/S0735-1097(99)00250-8PubMedGoogle ScholarCrossref
37.
Kim  SG, Fogoros  RN, Furman  S, Connolly  SJ, Kuck  KH, Moss  AJ.  Standardized reporting of ICD patient outcome: the report of a North American Society of Pacing and Electrophysiology Policy Conference, February 9-10, 1993.  Pacing Clin Electrophysiol. 1993;16(7, pt 1):1358-1362. doi:10.1111/j.1540-8159.1993.tb01728.xPubMedGoogle ScholarCrossref
38.
Hlatky  MA, Greenland  P, Arnett  DK,  et al; American Heart Association Expert Panel on Subclinical Atherosclerotic Diseases and Emerging Risk Factors and the Stroke Council.  Criteria for evaluation of novel markers of cardiovascular risk: a scientific statement from the American Heart Association.  Circulation. 2009;119(17):2408-2416. doi:10.1161/CIRCULATIONAHA.109.192278PubMedGoogle ScholarCrossref
39.
Buckert  D, Witzel  S, Steinacker  JM, Rottbauer  W, Bernhardt  P.  Comparing cardiac magnetic resonance–guided versus angiography-guided treatment of patients with stable coronary artery disease: results from a prospective randomized controlled trial.  JACC Cardiovasc Imaging. 2018;11(7):987-996. doi:10.1016/j.jcmg.2018.05.007PubMedGoogle ScholarCrossref
Original Investigation
February 8, 2019

Prognostic Value of Vasodilator Stress Cardiac Magnetic Resonance Imaging: A Multicenter Study With 48 000 Patient-Years of Follow-up

Author Affiliations
  • 1New York Presbyterian Brooklyn Methodist Hospital, New York, New York
  • 2Duke Cardiovascular Magnetic Resonance Center, Durham, North Carolina
  • 3Houston Methodist DeBakey Heart and Vascular Center, Houston, Texas
  • 4University of Illinois at Chicago
  • 5Piedmont Atlanta Hospital, Atlanta, Georgia
  • 6Weill Cornell Medical Center, New York
  • 7University of Minnesota Medical Center, Minneapolis
  • 8Heart Imaging Technologies, Durham, North Carolina
  • 9Northwestern University Feinberg School of Medicine, Chicago, Illinois
  • 10Editor, JAMA Cardiology
JAMA Cardiol. 2019;4(3):256-264. doi:10.1001/jamacardio.2019.0035
Key Points

Question  Can the results of vasodilator stress cardiac magnetic resonance imaging be associated with subsequent patient mortality?

Findings  In this multicenter study of 9151 patients followed up for up to 10 years, stress cardiac magnetic resonance imaging was strongly associated with patient mortality, both in the overall patient population as well as in a total of 14 different subpopulations.

Meaning  Stress cardiac magnetic resonance imaging may be another noninvasive indicator of prognosis.

Abstract

Importance  Stress cardiac magnetic resonance imaging (CMR) is not widely used in current clinical practice, and its ability to predict patient mortality is unknown.

Objective  To determine whether stress CMR is associated with patient mortality.

Design, Setting, and Participants  Real-world evidence from consecutive clinically ordered CMR examinations. Multicenter study of patients undergoing clinical evaluation of myocardial ischemia. Patients with known or suspected coronary artery disease (CAD) underwent clinical vasodilator stress CMR at 7 different hospitals. An automated process collected data from the finalized clinical reports, deidentified and aggregated the data, and assessed mortality using the US Social Security Death Index.

Main Outcomes and Measures  All-cause patient mortality.

Results  Of the 9151 patients, the median (interquartile range) patient age was 63 (51-70) years, 55% were men, and the median (interquartile range) body mass index was 29 (25-33) (calculated as weight in kilograms divided by height in meters squared). The multicenter automated process yielded 9151 consecutive patients undergoing stress CMR, with 48 615 patient-years of follow-up. Of these patients, 4408 had a normal stress CMR examination, 4743 had an abnormal examination, and 1517 died during a median follow-up time of 5.0 years. Using multivariable analysis, addition of stress CMR improved prediction of mortality in 2 different risk models (model 1 hazard ratio [HR], 1.83; 95% CI, 1.63-2.06; P < .001; model 2: HR, 1.80; 95% CI, 1.60-2.03; P < .001) and also improved risk reclassification (net improvement: 11.4%; 95% CI, 7.3-13.6; P < .001). After adjustment for patient age, sex, and cardiac risk factors, Kaplan-Meier survival analysis showed a strong association between an abnormal stress CMR and mortality in all patients (HR, 1.883; 95% CI, 1.680-2.112; P < .001), patients with (HR, 1.955; 95% CI, 1.712-2.233; P < .001) and without (HR, 1.578; 95% CI, 1.235-2.2018; P < .001) a history of CAD, and patients with normal (HR, 1.385; 95% CI, 1.194-1.606; P < .001) and abnormal left ventricular ejection fraction (HR, 1.836; 95% CI, 1.299-2.594; P < .001).

Conclusions and Relevance  Clinical vasodilator stress CMR is associated with patient mortality in a large, diverse population of patients with known or suspected CAD as well as in multiple subpopulations defined by history of CAD and left ventricular ejection fraction. These findings provide a foundational motivation to study the comparative effectiveness of stress CMR against other modalities.

Introduction

Noninvasive cardiac stress testing is a cornerstone in the clinical treatment of patients with known or suspected coronary artery disease (CAD).1,2 Numerous previous reports and meta-analyses have concluded that stress cardiac magnetic resonance imaging (CMR) is more accurate for the diagnosis of CAD compared with conventional stress imaging methods,3-8 yet CMR represents less than 0.1% of all imaging-based stress tests performed in the United States.9 Optimal patient treatment relies on both the diagnostic and prognostic information provided by noninvasive testing, and with escalating costs associated with use of advanced imaging methods, there has been a fundamental shift from a simple focus on test accuracy to a broader focus on patient outcome.10

The limited use of stress CMR is likely owing to multiple factors, including availability of good-quality laboratories, exclusion of patients with ferromagnetic devices, and a lack of data on patient outcome. The prognostic value of stress CMR has been reported in relatively small single-center studies, with few hard events and short follow-up times.11-14 Most previous outcome studies had fewer than 10 patient deaths in all groups combined, with the largest having fewer than 50 deaths.11-14 Accordingly, previous studies required the use of composite end points to demonstrate a link between CMR findings and patient outcome, and the association was driven primarily by soft events. At present, it is unknown whether stress CMR results are associated with mortality independent of soft events.

Collecting meaningful outcome data requires large numbers of patients and/or long follow-up times, which has only within the past decade been facilitated by the widespread use of electronic health records. In December 2016, the US Congress passed the 21st Century Cures Act,15 which required the US Food and Drug Administration to accept electronic health records as real-world evidence (RWE) in support of new drugs and devices.16 In this study, we used RWE to test the hypothesis that an abnormal stress CMR examination is independently associated with mortality in a large multicenter population of patients with known or suspected CAD.

Methods
Multicenter Data Collection and Follow-up

Seven geographically diverse medical centers in the United States participated in this study, and the Duke Cardiovascular Magnetic Resonance Center served as the data coordinating center. The 7 enrolling centers were chosen because they were all using the software needed for collection of multicenter data and also were routinely performing CMR stress testing. Figure 1 depicts the infrastructure we developed and implemented to collect RWE. Construction of the system was funded in part by Small Business Technology Transfer grants from the National Institutes of Health (HL080843, HL106864, and HL117397).

In the first step of the overall system, each of the 7 participating centers used the same software for onsite clinical CMR image interpretation and reporting, as well as for cloud-based data aggregation (Precession and CloudCMR software, respectively; Heart Imaging Technologies). At each site, the structured data from the finalized clinical CMR reports were stored in an internal relational database. In the second step, the structured clinical report data and associated Digital Imaging and Communications in Medicine images from each of the 7 centers were automatically deidentified and transmitted to CloudCMR (Figure 1). In the final step, every 3 months, the 7 local systems automatically determined patient mortality by comparing local patient identifiers to the US Social Security Death Index,17 and any changes in vital status were automatically transmitted to CloudCMR (Figure 1). Where available, the Social Security Death Index mortality data were supplemented with information from other locally available data sources.

Study Population

Across the 7 participating centers, all consecutive patients undergoing stress CMR with a clinical indication to evaluate myocardial ischemia were included (Figure 1). Of the 9454 consecutive patients undergoing their first CMR stress test, 303 were missing data for 1 or more cardiac risk factors and were therefore excluded from the primary analysis. Accordingly, the study population consisted of a total of 9151 patients. The 303 excluded patients had a similar incidence of positive/negative stress test results compared with the 9151 included patients. Baseline demographics such as patient age, sex, height and weight, blood pressure, clinical risk factors, cardiac history, cardiac symptoms, and medications were obtained prospectively by local site investigators at the time of the clinical study (Table 1). Institutional review board approval and written patient consent were obtained at each center.

Cardiac Magnetic Resonance Image Acquisition

The CMR stress studies were performed as previously described by Klem et al18 on either a 1.5- or 3.0-T magnetic resonance imaging scanner. The stress CMR examination consisted of 3 components: (1) cine imaging to examine regional left ventricular (LV) wall motion; (2) stress and rest perfusion imaging to test for ischemia; and (3) late gadolinium enhancement (LGE) imaging to detect myocardial infarction and/or scar. In brief, steady state free-precession cine images were acquired in multiple short-axis locations (every centimeter from base to apex) and 3 long-axis locations (2, 3, and 4 chamber). After cine imaging was completed, adenosine or regadenoson was administered to cause coronary vasodilation under continuous electrocardiogram and blood pressure monitoring with contrast injection (0.065 mmol/kg of gadolinium contrast). A saturation-recovery, single-shot, gradient-echo sequence18 was typically set up for 4 short-axis slices (matched to cine locations excluding most basal and apical slices) per heartbeat. If the stress perfusion images were abnormal, rest perfusion images were acquired using the same protocol during a second 0.065–mmol/kg dose of gadolinium contrast at least 10 minutes after the effects of pharmacologic vasodilation had ended. Otherwise, the second contrast agent dose was administered without rest perfusion imaging (total dose, approximately 0.13 mmol/kg). The LGE was performed at least 5 minutes after the second dose of the contrast agent, using a segmented inversion-recovery technique in the identical views as cine imaging.

Cardiac Magnetic Resonance Image Analysis

As described previously,19-23 the presence of wall motion and LGE patterns were recorded in the clinical reports using the American Heart Association 17-segment model, and the presence of ischemia on stress perfusion imaging was recorded in 16 segments (apical segment not typically within the field of view). The CMR interpretation algorithm has been described previously.18 In brief, a stress perfusion defect was considered ischemia if there was no LGE or if the perfusion defect was larger than the region of LGE, unless there were matched stress and rest perfusion defects without evidence of LGE. The criterion for a perfusion defect was a persistent delay in first-pass myocardial enhancement in more than 3 consecutive images. In accordance with established practice,5 stress CMR was considered abnormal if any abnormality was recorded in the segmental scores for wall motion, ischemia, or LGE.

Statistical Analysis

Statistical analysis was performed using SAS, version 9.4 (SAS Institute). Continuous variables were expressed as median and interquartile range (IQR; 25th and 75th percentile) and were compared by the Wilcoxon rank-sum test. Comparisons of discrete variables were made using the χ2 test; Fisher exact test was used when the assumptions of the χ2 test were not met. Time to event was calculated as the period between the CMR study and death. Patients who did not experience the primary outcome were censored at the time of the most recent comparison with Social Security Death Index (automatically performed every 3 months). Cox proportional hazards analysis was used to evaluate the association between stress CMR and time to the primary outcome of all-cause mortality. Cardiac risk factors were defined as age, sex, history of diabetes, history of hypertension, history of hyperlipidemia, and current smoker. To assess the added prognostic value of stress CMR, multivariable model 1 was defined using the following a priori risk factors: age, sex, history of diabetes, history of hypertension, history of hyperlipidemia, current smoker, history of CAD, and history of myocardial infarction. In multivariable model 2, clinical variables associated with mortality at P less than .10 were considered as candidates for inclusion using Cox regression analysis with stepwise selection. In both models 1 and 2, the global χ2 statistic was calculated with and without stress CMR and compared using the likelihood ratio test. For patients without a history of CAD, Simplified Framingham Risk Score24 was defined as low, intermediate, and high as less than 10%, 10 to 20%, and more than 20%, respectively. Kaplan-Meier survival analysis was used to assess the predictive value of stress CMR, and adjustments for patient age, sex, and cardiac risk factors were made using the SAS PHREG procedure, which provided the P values. Formal risk reclassification analyses were conducted by calculating net reclassification improvement25,26 for all patients who were followed up for at least 4 years (or died prior to 4 years) using 3 probability of death categories: less than 10%, 10% to 20%, and more than 20%, chosen because these resulted in approximately equal numbers of patients in each category. A P value of less than .05 was considered statistically significant.

Results
Patient Characteristics

Table 1 summarizes baseline patient characteristics, including comparisons between patients with and without abnormal stress CMR. As shown, only 2496 of 9151 patients (27%) had a clinically documented history of obstructive CAD. Regarding cardiovascular risk factors, prevalence of diabetes, hypertension, hyperlipidemia, and current smoking were each approximately 1.5-fold higher among patients with vs without abnormal stress CMR (Table 1). Notably, 1949 patients were classified as low risk based on conventional clinical risk stratification (no history of CAD and Framingham risk score <10%).

CMR Results

Overall median LV ejection fraction (LVEF) was 60.0% (IQR, 50.0-65.0). Patients with a normal stress CMR examination had a median LVEF of 61% (IQR, 60-65), and those with an abnormal examination had a median LVEF of 50% (IQR, 39-60) (P < .001). Abnormal wall motion was observed in 3282 of 9151 patients (36%), ischemia on stress perfusion was observed in 2388 (26%), and LGE was observed in 3445 (38%). The vasodilator used during stress CMR was adenosine in 6197 patients (68%), and regadenoson in 2954 (32%).

Patient Mortality

There were a total of 48 615 patient-years of follow-up. Of the 9151 patients in the study, 1517 died during a median follow-up time of 5.0 years (IQR, 2.3-7.8 years). The overall annual mortality rate was 3.1% per year. For patients without a history of CAD and with a normal stress CMR test result, the annual mortality rates for patients at low, intermediate, and high Framingham risk were 0.8%, 1.4%, and 2.7% per year, respectively, compared with 2.7%, 4.0%, and 4.9% for patients with an abnormal stress CMR, respectively.

Multivariable Analysis and Net Reclassification

Table 2 shows the results of multivariable analysis. For an a priori clinical model consisting of age, sex, cardiac risk factors, history of CAD, and history of myocardial infarction, the addition of stress CMR increased χ2 from 581.8 to 687.4, and the difference was statistically significant (P < .001). For a Cox proportional hazards model with risk factors determined programmatically in a stepwise manner, the addition of stress CMR increased χ2 from 620.7 to 721.1, and the difference was also statistically significant (P < .001). When LVEF by CMR was included in multivariable models 1 and 2 after the clinical variables, stress CMR remained independently associated with mortality for both models (P < .001).

Net reclassification analysis revealed that of the 6235 patients who were followed up for 4 years, for patients who did not survive, stress CMR resulted in a net improvement in prediction in 6.5% (95% CI, 2.7-8.6; P < .001). For patients who did survive, the net improvement was 4.9% (95% CI, 3.3-5.7; P < .001). Overall, the net reclassification improvement was 11.4% (95% CI, 7.3-13.6; P < .001).

Survival Analysis

Figure 2 shows the Kaplan-Meier survival analysis for all patients, after adjustment for age, sex, and cardiac risk factors. Stress CMR examination results (normal vs abnormal) showed a strong association with all-cause mortality (all patients, HR 1.883; 95% CI, 1.680-2.112; P < .001, Figure 2).

The upper row of Figure 3 shows Kaplan-Meier survival analyses after adjustment for age, sex, and cardiac risk factors for all patients without (Figure 3A) and with (Figure 3B) a history of CAD. Similarly, Figure 3C and D show patients with normal and abnormal LVEF, Figure 3E and Figure 3F show patients without and with LGE, and Figure 3G and H show patients without and with chest pain. In all 8 patient subpopulations in Figure 3, an abnormal stress CMR was strongly associated with mortality (history of CAD HR, 1.955; 95% CI, 1.712-2.233; P < .001; no history of CAD HR, 1.578; 95% CI, 1.235-2.2018; P < .001; normal LVEF HR, 1.385; 95% CI, 1.194-1.606; P < .001; abnormal LVEF HR 1.836; 95% CI, 1.299-2.594; P < .001; typical chest pain present HR, 2.074; 95% CI, 1.518-2.834; P < .001; typical chest pain absent HR, 1.860; 95% CI, 1.643-2.105; P < .001; LGE present HR, 1.447; P < .001; 95% CI, 1.151-1.819; P < .002; LGE absent HR, 1.676; 95% CI, 1.437-1.956; P < .001).

eFigure 1 in the Supplement shows results for patients without a history of CAD (n = 6655). The upper row shows Kaplan-Meier survival analyses for patients with low, intermediate, and high Framingham risk scores. The lower row shows Kaplan-Meier survival analyses for patients younger than 50 years, 50 to 65 years, and older than 65 years. In all 6 patient subpopulations in eFigure 1 in the Supplement, an abnormal stress CMR was strongly associated with mortality.

eFigure 2 in the Supplement shows the results for all patients by sex. For both male and female patients, an abnormal stress CMR was strongly associated with mortality. eFigure 3 in the Supplement shows results for increasing spatial extent of the abnormality for each of the 3 CMR examination components (wall motion, infarct/scar, and ischemia). Increasing spatial extent was significantly associated with all 3 examination components (Table 2).

An abnormal stress CMR was associated with mortality at both 1.5-T and 3-T magnetic field strengths (1.5-T HR, 1.885; 95% CI, 1.666-2.132; P < .001; 3-T HR, 1.634; 95% CI, 1.191-2.242; P < .002 for each).

Discussion

Previous studies have established that stress CMR is one of the most accurate techniques to detect functionally significant CAD.3-8 However, despite evidence of improved diagnostic accuracy, according to the US Centers for Medicare and Medicaid Services, a total of 2447 stress CMR examinations were reported in 2016, compared with 370 677 stress echocardiograms and 2 091 654 stress nuclear examinations.9 This study demonstrates that vasodilator stress CMR is significantly associated with mortality in a large multicenter population of patients with known or suspected CAD followed up for up to 10 years. To our knowledge, the 1517 patient deaths reported here are more than 30 times larger than any previous study of stress CMR mortality (less than 50 deaths).11-14

Randomized clinical trials are characterized by intrinsic scientific strengths and are widely recognized as the gold standard that has reshaped medical knowledge and practice.27,28 However, randomized clinical trials may not be generalizable to routine clinical practice because narrow entrance criteria limit the diversity of randomized clinical trial patient populations.28,29 Conversely, in this study, the use of RWE allowed us to examine the predictive value of stress CMR in a broad spectrum of patients composed of 16 different subpopulations (Figure 3; eFigures 1 and 2 in the Supplement). Importantly, each of the 16 patient subpopulations comprised thousands of patients and hundreds of deaths. These findings suggest that stress CMR has broad prognostic significance regardless of patient age, symptoms, or risk factors. Accordingly, while RWE studies may in general be more susceptible to patient selection bias compared with randomized clinical trial studies, the consistent Kaplan-Meier findings across 16 different cross-sections of a heterogeneous multicenter population suggest that patient selection bias is unlikely to affect the primary conclusion of our study: that stress CMR is significantly associated with mortality.

Randomized clinical trials typically involve a core laboratory for data analysis, thereby providing standardized interpretation of the data. However, a disadvantage of the use of a core laboratory is that centralized expert review may not reflect what actually happens in routine clinical practice. In this study, the CMR images were interpreted during routine clinical evaluation at each of the 7 participating sites. The results demonstrate that stress CMR as it is performed in real-world clinical practice is associated with patient mortality.

The 9151 patients reported here are 3.7 times greater than the 2447 stress CMR examinations reported annually to CMS for the entire United States.9 The ability to assess predictive value in a broad spectrum of patient subpopulations while maintaining statistical significance was likely owing to the use of an automated process to collect RWE from large numbers of patients. The technique reported here requires little or no human input other than routine clinical workflow (Figure 1). The approach described here may serve as a model for the evaluation of other diagnostic procedures.

We found an overall annual mortality rate of 3.1% per year, which is similar to the rate reported by Hachamovitch et al30 in a large population of patients with similar demographics (3.3% per year). Similarly, in another study of different patients from a different health care system, Bingham et al31 reported an all-cause mortality rate of 2.8% per year (66 deaths in 908 patients followed up for 2.6 years). Also similar to previous reports,32,33 for patients with a normal stress test result and without a history of CAD, we found annual mortality rates of 0.8%, 1.4%, and 2.7% per year for patients with low, intermediate, and high Framingham risk, respectively.

Limitations

Baseline demographics were obtained by local site investigators at the time of the clinical study and were limited to the prespecified variables presented in this article, which do not represent a comprehensive list of all possible prognostic markers and do not account for changes in these variables during the follow-up period. Our automated data collection system did not track symptoms, such as atypical chest pain, syncope, dizziness, nausea, or palpitations, nor did it track the number of patients with an equivocal echocardiogram or stress electrocardiogram. Information regarding specific cardiovascular outcomes, such as myocardial infarction, sudden death, implantable defibrillator placement, transplantation, or hospitalization, were not available.

Follow-up data in this study were limited to the primary end point of all-cause death, and the cause of death was not known. Thus, not all deaths were necessarily owing to cardiac causes. However, many have argued that all-cause mortality is an important and appropriate study end point because it is objective, clinically relevant, and unbiased, which is often not the case for cardiac mortality or softer outcomes such as revascularization or hospitalization.34-37

In this study, we were unable to determine whether patients were revascularized after the CMR stress test. However, if anything, one would anticipate that revascularization was performed more commonly in patients with an abnormal CMR stress test result and that revascularization would not result in an increase in mortality.

Our study achieves only 1 of several key phases recommended by the American Heart Association for the evaluation of a novel marker of cardiovascular risk,38 namely that stress CMR provides incremental information beyond that of standard risk markers in a large and diverse multicenter patient population. The 2018 MAGnet study39 addressed another key phase, namely the effect of stress CMR on patient management decisions.39 An important additional American Heart Association–recommended phase would be to systematically study the comparative effectiveness of CMR vs other modalities used in the assessment of patients with known or suspected CAD.

Conclusions

In this large, multicenter study, vasodilator stress CMR as performed in clinical practice was found to be independently associated with mortality, incremental to common clinical risk factors as well as history of CAD and LVEF. These findings provide a foundational motivation to study the comparative effectiveness of stress CMR against other modalities used in the assessment of patients with known or suspected CAD.

Back to top
Article Information

Corresponding Author: Robert M. Judd, PhD, Duke Cardiovascular Magnetic Resonance Center, Duke University Medical Center, PO Box 3934, Durham, NC 27710 (robert.judd@duke.edu).

Accepted for Publication: January 4, 2019.

Correction: This article was corrected on April 10, 2019, to fix an error in the Conflict of Interest Disclosures section.

Published Online: February 8, 2019. doi:10.1001/jamacardio.2019.0035

Author Contributions: Dr Judd and Ms Parker had full access to all data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Heitner, R. Kim, H. Kim, Shah, Debs, Farzaneh-Far, Polsani, Weinsaft, Judd.

Acquisition, analysis, or interpretation of data: R. Kim, H. Kim, Klem, Shah, Debs, Farzaneh-Far, Polsani, J. Kim, Weinsaft, Shenoy, Hughes, Cargile, Ho, Bonow, Jenista, Parker, Judd.

Drafting of the manuscript: Heitner, R. Kim, H. Kim, Farzaneh-Far, Polsani, Cargile, Parker, Judd.

Critical revision of the manuscript for important intellectual content: Heitner, R. Kim, H. Kim, Klem, Shah, Debs, Farzaneh-Far, J. Kim, Weinsaft, Shenoy, Hughes, Ho, Bonow, Jenista, Judd.

Statistical analysis: H. Kim, Parker, Judd.

Obtained funding: Shah, Judd.

Administrative, technical, or material support: Heitner, R. Kim, Shah, Debs, Farzaneh-Far, Polsani, J. Kim, Weinsaft, Cargile, Ho, Jenista, Judd.

Supervision: Heitner, R. Kim, Shah, Weinsaft, Judd.

Contributed to interpretation of the data and major revisions in the manuscript: Bonow.

Conflict of Interest Disclosures: Dr R. Kim reported grants from the National Heart, Lung, and Blood Institute during the conduct of the study and equity interest in Heart Imaging Technologies. Dr Weinsaft reported personal fees from Adverum Pharmaceuticals outside the submitted work and direction of a cardiac magnetic resonance imaging core lab for analysis of imaging data acquired as part of an industry sponsored study focused on Friedreich Ataxia. Dr Cargile reported grants and other support from National Institutes of Health during the conduct of the study and support from Heart Imaging Technologies outside the submitted work. Dr Jenista reported grants from Siemens during the conduct of the study and personal fees from HeartIT outside the submitted work. Dr Judd reported grants from National Institutes of Health during the conduct of the study and equity interest in Heart Imaging Technologies. All other authors have no conflicts of interest.

Funding/Support: Funded in part by grants from the National Heart, Lung, and Blood Institute and the National Institutes of Health (grants R42 HL080843, R42 HL106864, R42 HL117397, R01 HL128278, and K23 HL132011).

Role of the Funder/Sponsor: The funding sources 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.

Meeting Presentation: This paper was presented at the Society for Cardiovascular Magnetic Resonance 22nd Annual Scientific Sessions; February 8, 2019; Bellevue, Washington.

Disclaimer: Dr Bonow is Editor of JAMA Cardiology, but he was not involved in any of the decisions regarding review of the manuscript or its acceptance.

Additional Contributions: We thank Myles Lefkowitz, BS, Michael P. Lunney, MPH, Piedmont Atlanta Hospital, and Saadi Ahmed, Heart Imaging Technologies, for their help with technical aspects of this study, each of whom has given written permission to include their names in this article. Mr Ahmed is an employee of Heart Imaging Technologies and received compensation.

References
1.
Amsterdam  EA, Wenger  NK, Brindis  RG,  et al.  2014 AHA/ACC Guideline for the management of patients with non–ST-elevation acute coronary syndromes.  J Am Coll Cardiol. 2014;64(24):e139-e228. doi:10.1016/j.jacc.2014.09.017PubMedGoogle ScholarCrossref
2.
Yancy  CW, Jessup  M, Bozkurt  B,  et al; American College of Cardiology Foundation; American Heart Association Task Force on Practice Guidelines.  2013 ACCF/AHA Guideline for the management of heart failure.  J Am Coll Cardiol. 2013;62(16):e147-e239. doi:10.1016/j.jacc.2013.05.019PubMedGoogle ScholarCrossref
3.
Nandalur  KR, Dwamena  BA, Choudhri  AF, Nandalur  MR, Carlos  RC.  Diagnostic performance of stress cardiac magnetic resonance imaging in the detection of coronary artery disease: a meta-analysis.  J Am Coll Cardiol. 2007;50(14):1343-1353. doi:10.1016/j.jacc.2007.06.030PubMedGoogle ScholarCrossref
4.
Hamon  M, Fau  G, Née  G, Ehtisham  J, Morello  R, Hamon  M.  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-12-29PubMedGoogle ScholarCrossref
5.
Greenwood  JP, Maredia  N, Younger  JF,  et al.  Cardiovascular magnetic resonance and single-photon emission computed tomography for diagnosis of coronary heart disease (CE-MARC): a prospective trial.  Lancet. 2012;379(9814):453-460. doi:10.1016/S0140-6736(11)61335-4PubMedGoogle ScholarCrossref
6.
Schwitter  J, Wacker  CM, van Rossum  AC,  et al.  MR-IMPACT: comparison of perfusion-cardiac magnetic resonance with single-photon emission computed tomography for the detection of coronary artery disease in a multicentre, multivendor, randomized trial.  Eur Heart J. 2008;29(4):480-489. doi:10.1093/eurheartj/ehm617PubMedGoogle ScholarCrossref
7.
Schwitter  J, Wacker  CM, Wilke  N,  et al; MR-IMPACT Investigators.  MR-IMPACT II: Magnetic Resonance Imaging for Myocardial Perfusion Assessment in Coronary artery disease Trial: perfusion-cardiac magnetic resonance vs. single-photon emission computed tomography for the detection of coronary artery disease: a comparative multicentre, multivendor trial.  Eur Heart J. 2013;34(10):775-781. doi:10.1093/eurheartj/ehs022PubMedGoogle ScholarCrossref
8.
Heitner  JF, Klem  I, Rasheed  D,  et al.  Stress cardiac MR imaging compared with stress echocardiography in the early evaluation of patients who present to the emergency department with intermediate-risk chest pain.  Radiology. 2014;271(1):56-64. doi:10.1148/radiol.13130557PubMedGoogle ScholarCrossref
9.
U.S. Centers for Medicare and Medicaid Services. Medicare provider utilization and payment data. https://www.cms.gov/Research-Statistics-Data-and-Systems/Statistics-Trends-and-Reports/Medicare-Provider-Charge-Data/Physician-and-Other-Supplier.html. Accessed May 24, 2018.
10.
Fordyce  CB, Newby  DE, Douglas  PS.  Diagnostic strategies for the evaluation of chest pain: clinical implications from SCOT-HEART and PROMISE.  J Am Coll Cardiol. 2016;67(7):843-852. doi:10.1016/j.jacc.2015.11.055PubMedGoogle ScholarCrossref
11.
Lipinski  MJ, McVey  CM, Berger  JS, Kramer  CM, Salerno  M.  Prognostic value of stress cardiac magnetic resonance imaging in patients with known or suspected coronary artery disease: a systematic review and meta-analysis.  J Am Coll Cardiol. 2013;62(9):826-838. doi:10.1016/j.jacc.2013.03.080PubMedGoogle ScholarCrossref
12.
Gargiulo  P, Dellegrottaglie  S, Bruzzese  D,  et al.  The prognostic value of normal stress cardiac magnetic resonance in patients with known or suspected coronary artery disease: a meta-analysis.  Circ Cardiovasc Imaging. 2013;6(4):574-582. doi:10.1161/CIRCIMAGING.113.000035PubMedGoogle ScholarCrossref
13.
Greenwood  JP, Herzog  BA, Brown  JM,  et al.  Prognostic value of cardiovascular magnetic resonance and single-photon emission computed tomography in suspected coronary heart disease: long-term follow-up of a prospective, diagnostic accuracy cohort study.  Ann Intern Med. 2016;165:1-9. doi:10.7326/M15-1801PubMedGoogle ScholarCrossref
14.
Sammut  EC, Villa  ADM, Di Giovine  G,  et al.  Prognostic value of quantitative stress perfusion cardiac magnetic resonance.  JACC Cardiovasc Imaging. 2018;11(5):686-694. doi:10.1016/j.jcmg.2017.07.022PubMedGoogle ScholarCrossref
15.
21st Century Cures Act, HR 34, 114th Cong (2015).
16.
US Food and Drug Administration. Use of real-world evidence to support regulatory decision-making for medical devices, guidance for Industry and Food and Drug Administration Staff. https://www.fda.gov/downloads/medicaldevices/deviceregulationandguidance/guidancedocuments/ucm513027.pdf. Published August 31, 2017. Accessed February 1, 2019.
17.
US Social Security Death Index. https://www.ssdmf.com/. Accessed April 24, 2018.
18.
Klem  I, Heitner  JF, Shah  DJ,  et al.  Improved detection of coronary artery disease by stress perfusion cardiovascular magnetic resonance with the use of delayed enhancement infarction imaging.  J Am Coll Cardiol. 2006;47(8):1630-1638. doi:10.1016/j.jacc.2005.10.074PubMedGoogle ScholarCrossref
19.
Kim  RJ, Farzaneh-Far  A.  The diagnostic utility of cardiovascular magnetic resonance in patients with chest pain, elevated cardiac enzymes and non-obstructed coronary arteries.  Rev Esp Cardiol. 2009;62(9):966-971. doi:10.1016/S0300-8932(09)72093-7PubMedGoogle ScholarCrossref
20.
Kwong  RY, Farzaneh-Far  A.  Measuring myocardial scar by CMR.  JACC Cardiovasc Imaging. 2011;4(2):157-160. doi:10.1016/j.jcmg.2010.12.004PubMedGoogle ScholarCrossref
21.
Abbasi  SA, Ertel  A, Shah  RV,  et al.  Impact of cardiovascular magnetic resonance on management and clinical decision-making in heart failure patients.  J Cardiovasc Magn Reson. 2013;15:89. doi:10.1186/1532-429X-15-89PubMedGoogle ScholarCrossref
22.
Dandekar  VK, Bauml  MA, Ertel  AW, Dickens  C, Gonzalez  RC, Farzaneh-Far  A.  Assessment of global myocardial perfusion reserve using cardiovascular magnetic resonance of coronary sinus flow at 3 Tesla.  J Cardiovasc Magn Reson. 2014;16:24. doi:10.1186/1532-429X-16-24PubMedGoogle ScholarCrossref
23.
McGraw  S, Mirza  O, Bauml  MA, Rangarajan  VS, Farzaneh-Far  A.  Downstream clinical consequences of stress cardiovascular magnetic resonance based on appropriate use criteria.  J Cardiovasc Magn Reson. 2015;17:35. doi:10.1186/s12968-015-0137-xPubMedGoogle ScholarCrossref
24.
Framingham Heart Study, Risk Function for Cardiovascular Disease (10-year risk). https://www.framinghamheartstudy.org/fhs-risk-functions/cardiovascular-disease-10-year-risk/. Accessed May 3, 2018.
25.
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(2):157-172. doi:10.1002/sim.2929PubMedGoogle ScholarCrossref
26.
Steyerberg  EW, Pencina  MJ, Lingsma  HF, Kattan  MW, Vickers  AJ, Van Calster  B.  Assessing the incremental value of diagnostic and prognostic markers: a review and illustration.  Eur J Clin Invest. 2012;42(2):216-228. doi:10.1111/j.1365-2362.2011.02562.xPubMedGoogle ScholarCrossref
27.
Bothwell  LE, Greene  JA, Podolsky  SH, Jones  DS.  Assessing the gold standard: lessons from the history of RCTs.  N Engl J Med. 2016;374(22):2175-2181. doi:10.1056/NEJMms1604593PubMedGoogle ScholarCrossref
28.
Booth  CM, Tannock  IF.  Randomised controlled trials and population-based observational research: partners in the evolution of medical evidence.  Br J Cancer. 2014;110(3):551-555. doi:10.1038/bjc.2013.725PubMedGoogle ScholarCrossref
29.
Sherman  RE, Anderson  SA, Dal Pan  GJ,  et al.  Real-world evidence: what is it and what can it tell us?  N Engl J Med. 2016;375(23):2293-2297. doi:10.1056/NEJMsb1609216PubMedGoogle ScholarCrossref
30.
Hachamovitch  R, Rozanski  A, Shaw  LJ,  et al.  Impact of ischaemia and scar on the therapeutic benefit derived from myocardial revascularization vs medical therapy among patients undergoing stress-rest myocardial perfusion scintigraphy.  Eur Heart J. 2011;32(8):1012-1024. doi:10.1093/eurheartj/ehq500PubMedGoogle ScholarCrossref
31.
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(14):1509-1518. doi:10.1161/CIRCULATIONAHA.109.907659PubMedGoogle ScholarCrossref
32.
Hachamovitch  R, Berman  DS, Shaw  LJ,  et al.  Incremental prognostic value of myocardial perfusion single photon emission computed tomography for the prediction of cardiac death: differential stratification for risk of cardiac death and myocardial infarction.  Circulation. 1998;97(6):535-543. doi:10.1161/01.CIR.97.6.535PubMedGoogle ScholarCrossref
33.
Hachamovitch  R, Hayes  S, Friedman  JD,  et al.  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(8):1329-1340. doi:10.1016/S0735-1097(03)00125-6PubMedGoogle ScholarCrossref
34.
Klem  I, Shah  DJ, White  RD,  et al.  Prognostic value of routine cardiac magnetic resonance assessment of left ventricular ejection fraction and myocardial damage: an international, multicenter study.  Circ Cardiovasc Imaging. 2011;4(6):610-619. doi:10.1161/CIRCIMAGING.111.964965PubMedGoogle ScholarCrossref
35.
Stanton  T, Leano  R, Marwick  TH.  Prediction of all-cause mortality from global longitudinal speckle strain: comparison with ejection fraction and wall motion scoring.  Circ Cardiovasc Imaging. 2009;2(5):356-364. doi:10.1161/CIRCIMAGING.109.862334PubMedGoogle ScholarCrossref
36.
Lauer  MS, Blackstone  EH, Young  JB, Topol  EJ.  Cause of death in clinical research: time for a reassessment?  J Am Coll Cardiol. 1999;34(3):618-620. doi:10.1016/S0735-1097(99)00250-8PubMedGoogle ScholarCrossref
37.
Kim  SG, Fogoros  RN, Furman  S, Connolly  SJ, Kuck  KH, Moss  AJ.  Standardized reporting of ICD patient outcome: the report of a North American Society of Pacing and Electrophysiology Policy Conference, February 9-10, 1993.  Pacing Clin Electrophysiol. 1993;16(7, pt 1):1358-1362. doi:10.1111/j.1540-8159.1993.tb01728.xPubMedGoogle ScholarCrossref
38.
Hlatky  MA, Greenland  P, Arnett  DK,  et al; American Heart Association Expert Panel on Subclinical Atherosclerotic Diseases and Emerging Risk Factors and the Stroke Council.  Criteria for evaluation of novel markers of cardiovascular risk: a scientific statement from the American Heart Association.  Circulation. 2009;119(17):2408-2416. doi:10.1161/CIRCULATIONAHA.109.192278PubMedGoogle ScholarCrossref
39.
Buckert  D, Witzel  S, Steinacker  JM, Rottbauer  W, Bernhardt  P.  Comparing cardiac magnetic resonance–guided versus angiography-guided treatment of patients with stable coronary artery disease: results from a prospective randomized controlled trial.  JACC Cardiovasc Imaging. 2018;11(7):987-996. doi:10.1016/j.jcmg.2018.05.007PubMedGoogle ScholarCrossref
×