Available online at www.sciencedirect.com

Magnetic Resonance Imaging 29 (2011) 50–56

Cardiac magnetic resonance imaging: infarct size is an independentpredictor of mortality in patients with coronary artery disease

David Belloa,⁎, Arnold Einhorna, Rishi Kaushalb, Satish Kenchaiahc, Aidan Raneyc,David Fienod, Jagat Narulac, Jeffrey Goldbergere, Kalyanam Shivkumarb,

Haris Subaciuse, Alan KadisheaDivision of Cardiology, Orlando Regional Medical Center, Orlando, FL 32806, USA

bDivision of Cardiology, University of California at Los Angeles, David Geffen School of Medicine, Los Angeles, CA 90095, USAcDivision of Cardiology, University of California at Irvine Medical Center, Orange, CA 92868, USA

dDivision of Cardiology, Cedars Sinai Medical Center, Los Angeles, CA 90048, USAeDivision of Cardiology, Department of Medicine, Northwestern University, The Feinberg School of Medicine, Chicago, IL 60611, USA

Received 22 October 2009; revised 3 February 2010; accepted 5 March 2010

Abstract

Background: Cardiac magnetic resonance imaging (CMR) can accurately determine infarct size. Prior studies using indirect methods toassess infarct size have shown that patients with larger myocardial infarctions have a worse prognosis than those with smallermyocardial infarctions.Objectives: This study assessed the prognostic significance of infarct size determined by CMR.Methods: Cine and contrast CMR were performed in 100 patients with coronary artery disease (CAD) undergoing routine cardiacevaluation. Infarct size was determined by planimetry. We used Cox proportional hazards regression analyses (stepwise forward selectionapproach) to evaluate the risk of all-cause death associated with traditional cardiovascular risk factors, symptoms of heart failure,medication use, left ventricular ejection fraction, left ventricular mass, angiographic severity of CAD and extent of infarct size determinedby CMR.Results: Ninety-one patients had evidence of myocardial infarction by CMR. Mean follow-up was 4.8±1.6 years after CMR, during whichtime 30 patients died. The significant multivariable predictors of all-cause mortality were extent of myocardial infarction by CMR, extent ofleft ventricular systolic dysfunction, symptoms of heart failure, and diabetes mellitus (Pb.05). The presence of infarct greater than or equal to24% of left ventricular mass and left ventricular ejection fraction less than or equal to 30% were the most optimal cut-off points for theprediction of death with bivariate adjusted hazard ratios of 2.11 (95% confidence interval 1.02–4.38) and 4.06 (95% confidence interval1.73–9.54), respectively.Conclusions: The extent of myocardial infarction determined by CMR is an independent predictor of death in patients with CAD.© 2011 Published by Elsevier Inc.

Keywords: Cardiac magnetic resonance imaging; Coronary artery disease; Myocardial infarction; Myocardial infarction mortality predictor

1. Background

In spite of recent advances in therapies, coronary arterydisease (CAD) remains the leading cause of death in theUnited States [1]. Most patients with CAD die of sudden

⁎ Corresponding author. Mid-Florida Cardiology, Orlando, FL 32814USA.

E-mail address: David.Bello@orlandohealth.com (D. Bello).

0730-725X/$ – see front matter © 2011 Published by Elsevier Inc.doi:10.1016/j.mri.2010.03.031

,

cardiac death or congestive heart failure [1]. Depressedejection fraction has been identified as an important clinicalpredictor of both sudden cardiac death and progressive heartfailure. Patients with CAD and low ejection fractions arecandidates for therapy with an implantable cardioverter-defibrillator to prevent sudden cardiac death [2–6]. Anepidemiological paradox in patients with CAD and lowejection fractions is that, although they are at higher risk ofsudden cardiac death when compared to patients with CADand higher ejection fractions, they account for a minority of

51D. Bello et al. / Magnetic Resonance Imaging 29 (2011) 50–56

patients who die suddenly due to the fact that they comprise asmaller proportion of the population with CAD [7,8]. Thus,better risk-stratifiers are needed in the population with higherejection fractions, who account for the majority of patientsdying suddenly.

Advances in contrast-enhanced cardiac magnetic reso-nance imaging (CMR) have allowed accurate determinationof areas of scarring that represent myocardial infarction (MI)[9–14]. Among patients with CAD, an area of prior MIserves as a substrate for reentrant ventricular arrhythmias. Arecent CMR trial assessing the relationship between infarctsize and the inducibility of arrhythmias demonstrated that aninfarct size greater than 10% of left ventricular mass may bethe threshold substrate to sustain ventricular tachycardia[15]. Prior studies using indirect methods to determineinfarct size have shown that patients with larger infarcts havegreater mortality than those with smaller infarcts [16–31]. Inthe present study, we sought to assess the prognosticsignificance of infarct size determined by CMR in patientswith CAD.

2. Methods

2.1. Study population

One hundred consecutive patients with CAD wereprospectively enrolled. Evidence of CAD was defined byevidence of one of the following three criteria: (a) history ofhospitalization for MI, (b) prior revascularization (percuta-neous coronary intervention or coronary artery bypasssurgery and (c) significant stenosis of a major epicardialvessel (N50% proximal or 70% distal) by coronaryangiography. The research protocol was approved byInstitutional Review Board and all human participants gavewritten informed consent. Cine and contrast CMR wereperformed as part of a routine cardiac evaluation. Exclusioncriteria were primary valvular disease, constrictive, restric-tive, or hypertrophic cardiomyopathy; myocarditis; receivedor were likely to receive a cardiac transplant; or contra-indications to CMR (i.e., a pacemaker). Patients with ahistory of acute MI within the last month were also excluded.The primary endpoint was all-cause mortality.

2.2. CMR protocol

CMR images were acquired on a 1.5-T Siemens Sonatafrom 38 patients on a 1.0-T Siemens Harmony using aphased-array coil during repeated breath holds (≈8 s) usingmethods described previously [9–11]. Briefly, steady-statefree precession cine images were acquired in multiple short-axis (every 1 cm throughout the entire left ventricle) andtwo to three long-axis planes. Gadolinium (gadoteridol,0.15 mmol/kg) was administered intravenously, andcontrast-enhanced images were acquired after 10 min witha segmented inversion-recovery technique [11] in theidentical planes.

2.3. Analysis

Each CMR study was placed in random order after thesubject-identifying markers were removed. The cine andgadolinium-enhanced images were evaluated separately bythe consensus of two observers who were unaware of anypatient information. Hyperenhanced tissue on the delayedgadolinium-enhanced images was assumed to representscarred myocardium.

2.4. Global parameters

The myocardial borders were planimetered on all of theshort-axis cine images to determine left ventricular volumes,mass, ejection fraction, and infarct mass (assuming density=1.05 g/cm3) using methods described previously [15]. Thetotal area of infarct mass was then divided by the total area ofleft ventricular myocardial tissue to obtain infarct mass as apercentage of left ventricular mass.

2.5. Segmental model

Regional parameters were assessed using a 17-segmentmodel [32]. Segmental gadolinium enhancement withrespect to the extent of transmural infarct involvementfrom endocardium to epicardium was graded as follows: 0,0% hyperenhanced; 1, 1–25% hyperenhanced; 2, 26–50%hyperenhanced; 3, 51–75% hyperenhanced; and 4, 76–100% hyperenhanced.

2.6. Statistical analysis

Comparisons between patient groups were performedusing two-sample t-tests for continuous variables and the χ2

test for categorical variables. Continuous data are reported asmean ± standard deviation. A two-sided P value less than .05was considered statistically significant.

Pearson correlation co-efficient was estimated to assessthe correlation between infarct mass (IM) and left ventricularejection fraction (LVEF). Receiver operating characteristic(ROC) curve analysis was performed to aid the definition ofthreshold values for IM and LVEF. The curves wereevaluated graphically and cut-off points were defined basedon the separation of the curve from the diagonal line (Youdencriterion) in combination with clinical considerations.

Kaplan–Meier survival curves were compared for dichot-omized IM and LVEF variables. Cox proportional-hazardsregression analyseswere performedwith IM andLVEFused topredict survival times. Univariate and multivariable analyseswith both of the IM and LVEF in the model are reported.Likelihood ratio test was used. Subsequently, one-by-onebaseline covariates were added to the proportional hazardsmodel that already contained both IM and LVEF. Covariatesincluded in the analyseswere age (years), CAD (one-, two- andthree-vessel disease), gender, prior MI, prior coronary arterybypass graft surgery, percutaneous coronary intervention,diabetes mellitus, hyperlipidemia, hypertension, congestiveheart failure, implantable cardioverter defibrillator, angiotensin

Table 1Patient characteristics

Characteristic All patients (n=100) Alive (n=70) Dead (n=30) P a

Age-years 66±11 65±11 69±11 .098Male sex- no. (%) 87 (87) 60 (86) 27 (90) .56Prior myocardial infarctin, no. (%) 62 (62) 45 (64) 17 (57) .47Prior coronary artery bypass graft surgery, no. (%) 40 (40) 30 (43) 10 (33) .37Prior percutaneous coronary intervention, no. (%) 46 (46) 33 (47) 13 (43) .73Angiographic evidence of obstructive CAD, no. (%)1-Vessel disease 41 (41) 29 (41) 12 (40) .892-Vessel disease 25 (25) 19 (27) 6 (20) .453-Vessel disease 34 (34) 22 (31) 12 (40) .41Diabetes mellitus, no. (%) 23 (23) 13 (19) 10 (33) .11Hyperlipidemia, no. (%) 67 (67) 53 (76) 14 (47) .005Hypertension, no. (%) 49 (49) 36 (51) 13 (43) .46Congestive heart failure, no. (%) 63 (63) 36 (51) 27 (90) b.001ACE-Inhibitors/ARB, no. (%) 79 (79) 52 (74) 27 (90) .077beta-Blockers, no. (%) 85 (85) 56 (80) 29 (97) .032Potassium-sparing diuretics, no. (%) 16 (16) 12 (17) 4 (13) .63Lipid-lowering agents, no. (%) 65 (65) 49 (70) 16 (53) .11Antiplatelet agents, no. (%) 71 (71) 49 (70) 22 (73) .74CMR resultsCMR evidence of MI, no. (%) 91 (91) 61 (87) 30 (100)Infarct size, % of left ventricular mass±S.D. 19±14 17±13 24±14 .008Infarct transmurality (%±S.D.) 0.60±0.26 0.58±0.28 0.65±0.21 .18Transmural in ≥1 region, no. (%) 54 (54) 39 (56) 15 (51) .60Left ventricular ejection fraction (%±S.D.) 34±14 38±13 25±11 b.0001

CAD, coronary artery disease; ARB, angiotensin receptor blockers.a Alive vs. dead.

52 D. Bello et al. / Magnetic Resonance Imaging 29 (2011) 50–56

converting enzyme inhibitors or angiotensin receptor blockers,beta-blockers, potassium sparing diuretics, lipid-loweringagents and anti-platelet agents.

3. Results

3.1. Baseline characteristics

Characteristics of the study population are summarized inTable 1. A clinical history of prior MI as diagnosed byelectrocardiographic or enzyme criteria [33] was present in 62patients (62%), while 40 patients (40%) had a history ofcoronary artery bypass graft surgery, and 46 patients (46%)had a history of percutaneous transluminal coronary angio-plasty. All 100 patients had angiographically-confirmed CAD.

Fig. 1. Delayed contrast-enhanced MRI images from a patient with evidence of MI.

Twenty-four patients (24%) had an implantable cardioverter-defibrillator placed after their magnetic resonance imaging(MRI) scans.

The majority of patients were receiving optimal medicaltherapy at the outset of the study, with 79 (79%) treated withangiotensin-converting enzyme inhibitors, 85 (85%), beta-blockers, 65 (65%), lipid-lowering agents, and 71 (71%)antiplatelet therapy.

3.2. CMR results

CMR findings for the 100 patients are summarized inTable 1. Ninety-one patients (91%) had CMR-evidence ofMI. Fig. 1 shows representative images of a patient with MIvisualized by CMR. The mean infarct size as a percentage of

Fig. 2. Pearson correlation coefficient between infarct mass and leftventricular ejection fraction.

53D. Bello et al. / Magnetic Resonance Imaging 29 (2011) 50–56

left ventricular mass was 19% (5–32% or 19±14%). Themean ejection fraction was 34%.

3.3. Follow-up

All 100 patients were followed up for an average of4.8±1.6 years after CMR, during which time 30 patientsdied. All 30 patients who died had evidence of a prior MI onCMR. Those who died were more likely to have a lowerLVEF, a larger IM, and symptoms of congestive heart failure(Pb.05). In the 91 patients with CMR-evidence of infarct, the

Fig. 3. Receiver operating characteristic curve for Inf

transmural extent, location and the number of infarctedsegments that were fully transmural was not significantlydifferent between the two groups.

Twenty-four patients received an implantable cardi-overter-defibrillator during the follow-up period. Eight ofthe 24 patients (33%) receiving an implantable cardiover-ter-defibrillator died; all eight had an infarct mass greaterthan 10%.

3.4. Infarct mass and ejection fraction as predictors ofsurvival times

Fig. 2 shows the correlation between IM (per 1% of LVmass) and LVEF (per 1% decrease) of 0.42, Pb.001. Acoefficient of 0.42 also indicates substantial interdependenceamong the 2 measures. Both IM (HR=1.03, 95% CI 1.001–1.052; P=.044) and LVEF (HR=0.942, 95% CI 0.911–0.974; Pb.001) predicted the likelihood of death whenconsidered as continuous variables. Fig. 3 shows the resultsof the ROC curve analysis which was used to define the cut-off values for dichotomizing IM and LVEF. The area underthe curve (AUC) was significant for both variables(AUC=0.665, P=.010 and AUC=0.776, Pb.001 for IM andEF, respectively). IM was analyzed with two alternative cutpoints to achieve the following goals: first, the optimalbalance between sensitivity and specificity and, second,maximized sensitivity with acceptable level of specificity.To achieve these two goals, the IM was dichotomized atN10% and≥24% and was used in survival analysis to predictthe outcome. The presence of an IM greater than 10% had ahazard ratio of 1.90, but its 95% confidence interval crossedzero (95% CI 0.78–4.66; P=.160) (Fig. 4). When IM of 24%or greater was utilized the hazard ratio was 2.42 (95% CIfrom 1.17 to 5.02) and the P value was .017. However, IMcutoff of 24% only had 56% sensitivity and 74% specificityfor identifying patients likely to die, whereas IM of greaterthan 10% had a sensitivity of 80% and a specificity of 39%.A similar analysis was also performed for LVEF. Becausethe ROC curve showed a clear peak, a single value of

arct mass and left ventricular ejection fraction.

Fig. 4. Kaplan-Meier survival curves for patients with infarct mass ≤10%and N10% of left ventricular mass.

54 D. Bello et al. / Magnetic Resonance Imaging 29 (2011) 50–56

ejection fraction of ≤30% was utilized. LVEF≤30%conferred a 340% increase in the risk of death as comparedto patients with preserved heart function with sensitivity of77% and specificity of 71%, respectively.

LVEF remained a significant independent predictor in a two-variable model when IM and LVEF were used to predictsurvival times together. Independent predictive effect of IMwasnon-significant with IMN10% and LVEF≤30% cutoff points(HR=1.88, 95% CI 0.77–4.60; P=.144), but remainedsignificant with IM≥24% and EF ≤30% cutoff criteria(HR=2.11, 95%CI 1.02–4.38;P=.044).When patientsmeetingLVEF≤30% criterion are excluded, IM N10% identified 35(61.4%) and IM ≥24% identified 15 (26.3%) of the remainingpatients as being at risk for death. Six out of 35 (17.1%) and fourout of 15 (26.7%) actually died. By comparison, five out of 11(45.5%) of the patients who meet LVEF but not IM N10%criterion and 10 out of 23 (43.5%) of the patients who meetLVEF but not IM ≥24% criterion died. With IMN10% andLVEF≤30% criteria, 22 patients were not identified as being atrisk. Only one of these 22 died (4.5%). Risk was not indicatedfor 42 patients when IM criterion was raised to 24%. Three(7.1%) out of this cohort died in our sample.

By multivariable analysis, when controlling for the effectsof LVEF and IM, the presence of diabetes (P=.017) andsymptoms of heart failure classification (P=.002) were theonly other variables independently associated with the risk ofdeath. Use of beta-blocker medication showed a trend(P=.080) and other baseline covariates were non-significant(PN.100). Other clinical and demographic variables used toadjust the model included age, gender, prior MI or bypasssurgery, prior percutaneous coronary intervention, the use ofangiotensin-converting enzyme (ACE) inhibitors, hypercho-lesterolemia, the use of lipid lowering agents or anti-plateletdrugs, hypertension, congestive heart failure, presence ofimplantable cardioverter-defibrillator (ICD), left ventricularmass and potassium spearing agents which were notindependent predictors of survival time.

4. Discussion

This study assessed the prognostic value of infarct sizedetermined by CMR in patients with known CAD. We foundthat the extent of infarct was an independent predictor ofdeath. In the present study, an IM greater than or equal to24% of left ventricular mass was the best cutoff value thatefficiently stratifies patients into low- and high-risk groups.For population studies to identify patients at risk for suddendeath it may be more appropriate to create a model that findsa balance between sensitivity and specificity.

4.1. Prognostic value of infarct size

Consistent with prior studies [7,16–22,34], in the presentstudy we found that among patients with CAD, depressedejection fraction, heart failure, and infarct size weresignificant predictors of death. When accounting for differenttraditional prognostic variables, multivariable analysisrevealed that an infarct size 24% or greater of left ventricularmass and LVEF ≤30% were the best predictors of death.Absence of diabetes and symptoms of heart failure wereindependently associated with greater survival. Traditionalcardiovascular risk factors have been demonstrated to beassociated with greater survival among heart failure patients,a finding known as “reverse epidemiology” [35]. In ourstudy, a large number of patients who died had heart failure.

More than one half of deaths after infarction are classifiedas cardiac, and half of these are believed to be arrhythmic inorigin [7,8]. Current markers of an increased risk of suddendeath from arrhythmia include conventional coronary riskfactors, ejection fraction and New York Heart Associationfunctional class, sustained and nonsustained ventriculartachycardia, electrocardiographic variables such as T-wavealternans, markers of autonomic nervous system functionand inducibility during electrophysiological testing [7].Although individual markers may be specific for identifyingan increased risk of arrhythmic death in certain high-riskgroups, no single marker has thus far proved to be highlyspecific for identifying an increased risk of sudden deathfrom arrhythmia in the general population, in which theabsolute number of sudden deaths is high despite a lowrelative risk.

Prior MI may serve as a substrate for reentrant tachyar-rhythmias which can lead to ventricular fibrillation andsudden cardiac death [22,36,37]. Therefore, the volume ofinfarct may be a crucial determinant in the development ofsuch arrhythmias related to sudden cardiac death, where thegreater the infarct, the larger the substrate for the develop-ment of subsequent arrhythmias [20,23,38]. Indeed, Bello etal. [15] identified a value of 10% of left ventricular mass to bethe critical substrate necessary to develop sustained ventric-ular arrhythmias in patients with CAD undergoing electro-physiological testing. Our study showed that aMI comprisinggreater than 10% of left ventricular mass carries a 2.0-foldrisk of death while maximizing sensitivity for the detection of

55D. Bello et al. / Magnetic Resonance Imaging 29 (2011) 50–56

mortality. Although an IM cutoff of 24% identified a higherrisk group, the sensitivity of this cutoff value was only 50%.Thus, for population studies to identify patients at risk forsudden death, a cutoff of 10% may have more utility despitethe lower specificity. This significant risk may be due to thisthreshold volume of infarct serving as a potentially fatalarrhythmogenic substrate. In addition to serving as a substratefor arrhythmias, larger infarcts are associated with leftventricular remodeling, which may lead to depressed leftventricular function and heart failure [39]. Thus, in additionto being at high risk for sudden cardiac death, patients withlarge infarcts may experience death secondary to progressiveheart failure.

4.2. Detection of MI — the role of CMR

Technical improvements in CMR have allowed theidentification of MI with great precision [9–14]. Infarctsize measures by human manual contouring and intensitythresholding have been shown to significantly overestimatethe infarct area in a laboratory animal model [40].Nonetheless, previous studies have validated the usefulnessof contrast-enhanced MRI to detect cardiac scar tissue andhave demonstrated good correlation between infarct size oncontrast enhanced MRI and peak release of creatinine kinase-MB and troponin I [9,11,13,40–43]. Similar to prior studies,we found that the vast majority (93%) of patients with leftventricular dysfunction and CAD had CMR-evidence of MI[44,45]. Soriano et al. found that 81% of patients withangiographically-proven CAD and left ventricular dysfunc-tion had CMR-evidence of MI [44], and McCrohon et al.found that 100% of their study group with CAD and leftventricular dysfunction had evidence of MI [37]. In thepresent study, the size of MI and LVEF, regardless of itsclinical manifestation, was found to be predictive ofmortality. Roes et al. [28] followed 231 patients with CMRafter MI for 1.7 years, employing a qualitative method forinfarct size assessment. In concordance with our study, theirresults showed that infarct size on contrast-enhanced MRI isa good predictor for long-term mortality in patients withhealed MI.

4.3. Clinical implications

Several recent clinical trials have established the implant-able cardioverter-defibrillator as an important therapeuticmodality for primary and secondary prevention of mortalityin post-infarction patients with severe left ventricular dysfunc-tion [2–4,6]. However, any strategy for primary preventionbased on ejection fraction alone has major limitations. Forinstance, the vast majority of patients who die suddenlyfollowing a MI have ejection fractions greater than 30% [7,8].Although these patients are at lower risk for sudden cardiacdeath, they account for the majority of patients who diesuddenly. In addition, not all patients with low ejectionfractions will experience arrhythmic death. Thus, treating allpatients with a depressed ejection fraction with an implantable

cardioverter-defibrillator may not be cost-effective. It there-fore, becomes necessary to identify a risk-stratifier other thanejection fraction to direct ICD therapy. The present studyshows that IM is an independent predictor of sudden cardiacdeath. IM, consequently, could serve as a risk-stratifier amongthe patient population with an LVEF greater than 30%. Inaddition, stratification based on IM may reduce the number ofimplantable cardioverter-defibrillators implanted in the popu-lationwith low ejection fractions, who, if proven,may not be atrisk of arrhythmic death.

In summary, the present study offers outcome data of astrong relationship between mortality and IM using CMR;whether IM is a better predictor of sudden cardiac death thanLVEF remains to be tested. The defibrillators to reduce riskby MagnetIc ResoNance Imaging Evaluation (DETER-MINE) trial will address this question in a large multicentertrial, which is expected to enroll over 10000 patients over thenext 7 years. In the DETERMINE trial patients will berandomized to implantable defibrillator therapy based oninfarct size by CMR.

4.4. Limitations

In the present study, the primary endpoint was overallmortality. Secondary endpoints such as cardiac death andarrhythmic death were not analyzed due to the smallsample size.

References

[1] Lloyd-Jones D, Adams RJ, Brown TM, et al. Heart disease and strokestatistics–2010 update: a report from the American Heart Association.Circulation 2010;121(7):e46–215.

[2] Moss A, Hall W, Cannom D, et al. Improved survival with animplanted defibrillator in patients with coronary artery disease at highrisk for ventricular arrhythmia. N Engl J Med 1996;335:1933–40.

[3] Moss A, Zareba W, Hall W, et al. Prophylactic implantation of adefibrillator in patients with myocardial infarction and reduced ejectionfraction. N Engl J Med 2002;346:877–83.

[4] Buxton A, Lee K, Fisher J, et al. A randomized study of the preventionof sudden death in patients with coronary artery disease. N Engl J Med1999;341:1882–90.

[5] Kadish A, Dyer A, Daubert JP, et al. Prophylactic defibrillatorimplantation in patients with nonischemic dilated cardiomyopathy.N Engl J Med 2004;350:2151–8.

[6] Bardy G, Lee K, Mark D, et al. Amiodarone or an implantablecardioverter-defibrillator for congestive heart failure. N Engl J Med2005;352:225–37.

[7] Huikuri HV, Castellanos A, Myerburg RJ. Sudden death due to cardiacarrhythmias. N Engl J Med 2001;345(20):1473–82.

[8] Zipes DP, Wellens HJ. Sudden cardiac death. Circulation 1998;98(21):2334–51.

[9] Wu E, Judd R, Vargas J, Klocke F, Bonow R, Kim R. Visualisation ofpresence, location, and transmural extent of healed Q-wave and non-Q-wave myocardial infarction. Lancet 2001;357:21–8.

[10] Simonetti O, Kim R, Fieno D, et al. An improved MR imagingtechnique for the visualization of myocardial infarction. Radiology2001;218:215–23.

[11] Kim R, Wu E, Rafael A, et al. The use of contrast-enhanced magneticresonance imaging to identify reversible myocardial dysfunction.N Engl J Med 2000;343:1445–53.

56 D. Bello et al. / Magnetic Resonance Imaging 29 (2011) 50–56

[12] Fieno D, Kim R, Chen E, Lomasney J, Klocke F, Judd R. Contrast-enhanced magnetic resonance imaging of myocardium at risk:distinction between reversible and irreversible injury through infarcthealing. J Am Coll Cardiol 2000;36:1985–91.

[13] Kim R, Fieno D, Parrish T, et al. Relationship of MRI delayed contrastenhancement to irreversible injury, infarct age, and contractilefunction. Circulation 1999;100:1992–2002.

[14] Ricciardi M, Wu E, Davidson C, et al. Visualization of discretemicroinfarction after percutaneous coronary intervention associatedwith mild creatine kinase-MB elevation. Circulation 1999;103:2780–3.

[15] Bello D, Fieno D, Kim R, et al. Infarct morphology identifies patientswith substrate for sustained ventricular tachycardia. J Am Coll Cardiol2005;45(7):1104–8.

[16] Geltman EM. Infarct size as a determinant of acute and long-termprognosis. Cardiol Clin 1984;2:95–103.

[17] Miller TD, Christian TF, Hopfenspirger MR, Hodge DO, Gersh BJ,Gibbons RJ. Infarct size after acute myocardial infarction measured byquantitative tomographic 99mTc sestamibi imaging predicts subse-quent mortality. Circulation 1995;92:334–41.

[18] Miller TD, Hodge DO, Sutton JM, et al. Technetium-99m sestamibiinfarct size predicts mortality. Am J Cardiol 1998;81:1491–3.

[19] Burns RJ, Gibbons RJ, Yi O, et al. The relationships of leftventricular ejection fraction, end-systolic volume index and infarctsize to six-month mortality after hospital discharge followingmyocardial infarction treated by thrombolysis. J Am Coll Cardiol2002;39:30–6.

[20] Wilber D, Lynch J, Montgomery D, Lucchesi B. Postinfarction suddendeath: significance of inducible ventricular tachycardia and infarct sizein a conscious canine model. Am Heart J 1985;109:8–18.

[21] Sobel BE, Bresnahan GF, Shell WE, Yoder RD. Estimation of infarctsize in man and its relation to prognosis. Circulation 1972;46:640–8.

[22] Geltman EM, Ehsani AA, Campbell MK, Schechtman K, Roberts R,Sobel BE. The influence of location and extent of myocardialinfarction on long-term ventricular dysrhythmia and mortality.Circulation 1979;60:805–14.

[23] Jones-Collins B, Patterson R. Quantitative measurement of electricalinstability as a function of myocardial infarct size in the dog. Am JCardiol 1981;48:858–63.

[24] Hurrell DG, Milavetz J, Hodge DO, Gibbons RJ. Infarct sizedetermination by technetium 99m sestamibi single-photon emissioncomputed tomography predicts survival in patients with chroniccoronary artery disease. Am Heart J 2000;140:61–6.

[25] Antman EM, Tanasijevic MJ, Thompson B, et al. Cardiac-specifictroponin I levels to predict the risk of mortality in patients with acutecoronary syndromes. N Engl J Med 1996;335:1342–9.

[26] Gibbons RJ, Valeti US, Araoz PA, Jaffe AS. The quantification ofinfarct size. J Am Coll Cardiol 2004;44:1533–42.

[27] Smith LR, Harrell FEJ, Rankin JS, et al. Determinants of early versuslate cardiac death in patients undergoing coronary artery bypass graftsurgery. Circulation 1991;84(Suppl III):III-245–53.

[28] Roes SD, Kelle S, Kaandorp AM, et al. Comparison of myocardialinfarct size assessed with contrast-enhanced magnetic resonanceimaging and left ventricular function and volumes to predict mortalityin patients with healed myocardial infarction. Am J Cardiol 2007;100:930–6.

[29] Wu E, Ortiz JT, Tejedor P. Infarct size by contrast enhanced cardiacmagnetic resonance is a stronger predictor of outcomes than leftventricular ejection fraction or end-systolic volume index: prospectivecohort study. Heart 2008;94:730–6.

[30] Kwong RY, Chan AK, Brown KA, et al. Impact of unrecognizedmyocardial scar detected by cardiac magnetic resonance imaging onevent-free survival in patients presenting with signs or symptoms ofcoronary artery disease. Circulation 2006;113:2733–43.

[31] Yan AT, Shayne AJ, Brown KA, et al. Characterization of the peri-infarct zone by contrast-enhanced cardiac magnetic resonance imagingis a powerful predictor of post-myocardial infarction mortality.Circulation 2006;114:32–9.

[32] Cerqueira MD, Weissman NJ, Dilsizian V, et al. Standardizedmyocardial segmentation and nomenclature for tomographic imagingof the heart. Circulation 2002;105:539.

[33] Alpert JS, Thygesen K, Antman EM, Bassand JP. Myocardialinfarction redefined — a consensus document of the Joint EuropeanSociety of Cardiology/American College of Cardiology Committee forthe redefinition of myocardial infarction. J Am Coll Cardiol 2000;36:959–69.

[34] Bart BA, Shaw LK, McCants CB. Clinical determinants of mortality inpatients with angiographically diagnosed ischemic or nonischemiccardiomyopathy. J Am Coll Cardiol 1997;30:1002–8.

[35] Kalantar-Zadeh K, Block G, Horwich T, Fonarow GC. Reverseepidemiology of conventional cardiovascular risk factors in patientswith chronic heart failure. JACC 2004;43(8):1439–44.

[36] Bayes de Luna A, Coumel P, Leclercq JF. Ambulatory sudden cardiacdeath: mechanisms of production of fatal arrhythmia on the basis ofdata from 157 cases. Am Heart J 1989;117:151–9.

[37] Bolick D, Hackel D, Reimer K, Ideker R. Quantitative analysis ofmycocardial infarct structure in patients with ventricular tachycardia.Circulation 1986;74:1266–79.

[38] Cohn J, Ferrari R, Sharpe N. Remodeling oboaIFoC. Cardiacremodeling — concepts and clinical implications: a consensus paperfrom an international forum on cardiac remodeling. J Am Coll Cardiol2000;35:569–82.

[39] Bello D, Shah DJ, Farah GM. Gadolinium cardiovascular magneticresonance predicts reversible myocardial dysfunction and remodelingin patients with heart failure undergoing beta-blocker therapy.Circulation 2003;108:1945–53.

[40] Hsu LY, Natanzon A, Kellman P, Hirsch GA, Aletras AH, Arai AE.Quantitative myocardial infarction on delayed enhancement MRI. PartI: animal validation of an automated feature analysis and combinedthresholding infarct sizing algorithm. J Magn Reson Imaging 2006;23:298–308.

[41] Kaandorp TA, Bax JJ, Lamb HJ, Viergever EP, Boersma E,Poldermans D, et al. Which parameters on magnetic resonanceimaging determine Q waves on the electrocardiogram? Am J Cardiol2005;95:925–9.

[42] Engelstein ED, Zipes DP. Sudden cardiac death. In: Alexander W,editor. The Heart, Arteries and Veins. 9th ed. New York: McGraw-Hill; 1998. p. 1081–112.

[43] Choi KM, Kim RJ, Gubernikoff G, Vargas JD, Parker M, Judd RM.Transmural extent of acute myocardial infarction predicts long-termimprovement in contractile function. Circulation 2001;104:1101–7.

[44] Soriano CJ, Ridocci F, Estornell J, Jimenez J, Martinez V, DeVelasco JA. Noninvasive diagnosis of coronary artery disease inpatients with heart failure and systolic dysfunction of uncertainetiology, using late gadolinium-enhanced cardiovascular magneticresonance. J Am Coll Cardiol 2005;45:743–8.

[45] McCrohon JA, Moon JCC, Prasad SK, et al. Differentiation of heartfailure related to dilated cardiomyopathy and coronary artery diseaseusing gadolinium-enhanced cardiovascular magnetic resonance.Circulation 2003;108:54–9.