A Prospective Study for Comparison of MR and CT Imaging for Detection of Coronary Artery Stenosis

Transcription

1 JACC: CARDIOVASCULAR IMAGING VOL. 4, NO. 1, BY THE AMERICAN COLLEGE OF CARDIOLOGY FOUNDATION ISSN X/$36.00 PUBLISHED BY ELSEVIER INC. DOI: /j.jcmg A Prospective Study for Comparison of MR and CT Imaging for Detection of Coronary Artery Stenosis Ashraf Hamdan, MD,* Patrick Asbach, MD, Ernst Wellnhofer, MD,* Christoph Klein, MD,* Rolf Gebker, MD,* Sebastian Kelle, MD,* Harald Kilian, MD,* Alexander Huppertz, MD, Eckart Fleck, MD* Berlin, Germany; and Tel Aviv, Israel OBJECTIVES The purpose of the present study was to directly compare the diagnostic accuracy of magnetic resonance imaging (MRI) and multislice computed tomography (CT) for the detection of coronary artery stenosis. BACKGROUND Both imaging modalities have emerged as potential noninvasive coronary imaging modalities; however, CT unlike MRI exposes patients to radiation and iodinated contrast agent. METHODS One hundred twenty consecutive patients with suspected or known coronary artery disease prospectively underwent 32-channel 3.0-T MRI and 64-slice CT before elective X-ray angiography. The diagnostic accuracy of the 2 modalities for detecting significant coronary stenosis ( 50% luminal diameter stenosis) in segments 1.5 mm diameter was compared with quantitative invasive coronary angiography as the reference standard. RESULTS In the patient-based analysis MRI and CT angiography showed similar diagnostic accuracy of 83% (95% confidence interval [CI]: 75 to 87) versus 87% (95% CI: 80 to 92), p 0.38; sensitivity of 87% (95% CI: 76 to 93) versus 90% (95% CI: 80 to 95), p 0.16; and specificity of 77% (95% CI: 63 to 87) versus 83% (95% CI: 70 to 91), p 0.06, respectively. All cases of left main or 3-vessel disease were correctly diagnosed by MRI and CT angiography. In the patient-based analysis MRI and CT angiography were similar in their ability to identify patients who subsequently underwent revascularization: the area under the receiver-operator characteristic curve was 0.78 (95% CI: 0.69 to 0.87) for MRI and 0.82 (95% CI: 0.74 to 0.90) for CT angiography. CONCLUSIONS Thirty-two channel 3.0-T MRI and 64-slice CT angiography similarly identify significant coronary stenosis in patients with suspected or known coronary artery disease scheduled for elective coronary angiography. However, CT angiography showed a favorable trend toward higher diagnostic performance. (J Am Coll Cardiol Img 2011;4:50 61) 2011 by the American College of Cardiology Foundation From the *Department of Internal Medicine/Cardiology, Deutsches Herzzentrum Berlin, Berlin, Germany; Imaging Science Institute, Charité-Universitätsmedizin Berlin, Berlin, Germany; and the Heart Institute and Department of Diagnostic Imaging, Sheba Medical Center, Tel-Hashomer, Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel. The study was supported by the Foundation Deutsches Herzzentrum Berlin. Dr. Hamdan has received a grant from the European Society of Cardiology Working group for Cardiovascular Magnetic Resonance. All other authors report that they have no relationships to disclose. Manuscript received August 31, 2010; revised manuscript received September 28, 2010, accepted October 4, 2010.

2 JACC: CARDIOVASCULAR IMAGING, VOL. 4, NO. 1, Coronary artery disease (CAD) is the leading cause of death in the western world, and its prevalence is still increasing (1). The current gold standard for the diagnosis of obstructive CAD remains conventional coronary angiography; however, it is invasive and associated with risks, and a substantial number of the procedures are for diagnostic purposes only without the need for intervention (2). Thus, noninvasive, See page 62 low-risk, and cost-effective coronary angiography would represent important progress in the diagnosis of obstructive CAD. Magnetic resonance imaging (MRI) and multislice computed tomography (CT) have been proposed as noninvasive imaging modalities to determine the presence of coronary artery stenosis (3,4). Several studies have directly compared these 2 imaging approaches for the diagnosis of obstructive CAD (5 8). However, in recent years, noninvasive coronary imaging modalities have been further spectacularly developed. In the field of MRI the recent use of 3.0-T MRI systems (9,10) and 32-element coils (11,12) has allowed improvements in both the signal-to-noise ratio (SNR) and parallel imaging techniques, which enable further improvement of the spatial and temporal resolution of the 3-dimensional (3D) free-breathing whole-heart coronary imaging. By contrast, 64-slice CT technology has higher spatial and temporal resolution than the older CT generations, which results in improved image quality and great clinical reliability. In addition, a direct comparison of the ability of MRI and CT angiography to predict the need for revascularization has not been made. Therefore, we conducted a prospective 2-center study to determine the diagnostic accuracy of MRI and multislice CT angiography involving 32-channel 3.0-T MRI and 64-slice CT for the detection of clinically relevant coronary artery stenosis in patients with suspected or known CAD referred for invasive coronary angiography. Thus, the study was designed to determine the presence or absence of coronary artery stenosis in patients already at substantial risk for CAD who might require coronary revascularization. METHODS Study design. The study is a prospective, 2-center study. MRI and CT angiography were performed and evaluated at 2 different centers. The local institutional review board and the German Federal Department for Radiation Protection approved the study, and all patients gave written informed consent. Study population. The study group consisted of 120 consecutive patients who were referred between September 29, 2008 and May 3, 2009 to undergo invasive coronary angiography for suspected or known CAD. To avoid radiation exposure in younger patients, who have a higher lifetime attributable risk than older individuals receiving the same dose, patients enrolled in the study were at least 50 years of age (13). The exclusion criteria were atrial fibrillation, acute coronary syndrome, New York Heart Association functional class III or IV heart failure, previous coronary artery bypass graft operation, body mass index of more than 40 kg/m 2, pregnancy, and breastfeeding. Patients with contraindications to MRI (noncompatible implants or severe claustrophobia) or CT AND (impaired renal function with serum creatinine level 1.4 mg/dl or known allergy to iodinated contrast agents) were also not considered for inclusion in the study. Study protocol. Patients underwent MRI and CT angiography in random order before invasive coronary angiography. If no contraindications were present, each patient received sublingual isosorbide dinitrate (5 mg) immediately before MRI artery and CT angiography. Whenever the heart rate was 65 beats/min, the patient was given 50 mg of metoprolol orally 1 h before MRI and CT examinations and, additionally, intravenous metoprolol (up to 3 doses of 5 mg) if the heart rate was still 65 beats/min. 32-channel MRI coronary angiography. Magnetic resonance imaging was performed on a 3.0-T system (Achieva 3 Tesla, Philips, Best, the Netherlands) with a dedicated 32-element cardiac coil (4 4 anterior elements and 4 4 posterior elements) for data acquisition, as described recently (11,12). Cardiac synchronization was performed with vector electrocardiogram (ECG). A multislice survey using a segmented balanced steady-state free precession sequence allowed localization of the heart and diaphragm in the 3 standard planes (transversal, sagittal, and coronal). Subsequently, a reference 3D dataset was obtained to evaluate the individual coil sensitivities for subsequent sensitivity-encoding (SENSE) imaging. To determine the individual cardiac rest period a cine-scan with transversal slice orientation (balanced steady-state free precession, rep- ABBREVIATIONS ACRONYMS 3D 3-dimensional AUC area under the curve CAD coronary artery disease CT computed tomography ECG electrocardiogram LAD left anterior descending coronary artery LCX left circumflex coronary MRI magnetic resonance imaging RCA right coronary artery ROC receiver-operator characteristic SENSE sensitivity encoding SNR signal-to-noise ratio

3 52 JACC: CARDIOVASCULAR IMAGING, VOL. 4, NO. 1, 2011 Figure 1. Diagram Illustrating the Course of the 32-Channel MRI and 64-Slice CT Angiography Diagram illustrating the course of free-breathing navigator gated 32-channel 3.0-T magnetic resonance imaging (MRI) and 64-slice computed tomography (CT) coronary angiography. Contrast agent was administered for CT but not for MRI angiography. etition time/echo time/flip angle 3.7 ms/1.8 ms/45, acquired spatial resolution mm, retrospective gating, 50 phases/cardiac cycle) was performed to visually determine the optimal patient-specific triggerdelay time and the duration of data acquisition window per RR interval (14). For real time respiratory gating in the subsequent MRI angiography a pencilbeam prospective navigator was placed on the dome of the right hemidiaphragm to monitor the liver lung interface during free-breathing with an end expiratory acceptance window of 5 mm and a correction factor of 0.45 in cranio-caudal direction (15). A navigator-gated, ECG-triggered 3D wholeheart MRI angiography with 130 transversal slices covering the whole of the heart was acquired with a segmented turbo gradient echo sequence (repetition time/echo time/flip angle: 4.2 ms/1.3 ms/20 ) with at 2 preparation and a fat suppression pre-pulse. The spatial resolution of the MRI angiography was mm interpolated from mm. Data acquisition was accelerated by employing 2-dimensional parallel imaging with a SENSE factor of 1.5 in the feed-head and anterior-posterior direction. Contrast agent was not administered. A diagram showing the MRI procedure is given in Figure 1. For analysis, multiplanar reformatting of the 3D dataset was carried out with a previously described dedicated coronary analysis tool (16). Multislice CT coronary angiography. All scans were performed with a 64-slice CT scanner (SOMATOM Sensation 64, Siemens Healthcare, Erlangen, Germany) with a gantry rotation time of 330 ms, retrospective ECG gating, 120-kV tube voltage, and 850- to mas (effective) tube load. Computed tomography data were simultaneously acquired in 64 (32 2) datasets/ rotation with mm beam collimation. Scan direction was cranio-caudal, and scan volume ranged from the carina to below the diaphragmatic face of the heart. Pitch value was 0.2. In patients with a heart rate below 65 beats/min, ECG-gated tube current modulation was used. The window of full tube current was limited to 60% to 70% of the RR interval. After placement of an antecubital 18-G intravenous access, contrast agent transit time (iopromide, 370 mg of iodine/ml, Ultravist, Bayer Healthcare, Germany) was assessed by injecting a test bolus of 15 ml followed by a saline flush of 50 ml, both at a flow rate of 5 ml/s. Contrast agent transit time was defined as the time between the start of contrast injection and maximum enhancement in the ascending aorta at the level of the coronary ostia. For angiographic CT data acquisition, a delay 3 s longer than contrast agent transit time was used. The volume of contrast agent injected for the scan depended on the estimated scan duration. Contrast was injected at a flow rate of 5 ml/s for the same duration as data acquisition. Overall quantity varied from 75 to 100 ml. Contrast injection was followed by a 50-ml saline chaser bolus (5 ml/s). Figure 1 also shows the CT angiogram procedure. Half-scan reconstruction yielded a temporal resolution of 165 ms. The in-plane spatial resolution was mm with a slice thickness of 0.6 mm. For reconstruction, slice thickness of 0.75 mm,

4 JACC: CARDIOVASCULAR IMAGING, VOL. 4, NO. 1, Figure 2. Coronary Artery Segments According to the AHA Minor branches, such as the conus (CB), sinus node (SN), ventricular (V), acute marginal (AM), atrioventricular node (AV), and atrial circumflex (AC) branches, are indicated in the diagram only for general orientation. In addition to the illustrated 15 coronary artery segments we used the intermediate branch artery assigned to segment number 16. AHA American Heart Association; D1 first diagonal branch; D2 second diagonal branch; LAD left anterior descending coronary artery; LCX left circumflex coronary artery; LM left main coronary artery; OM obtuse marginal branch; PD posterior descending branch; PL posterolateral branch; RCA right coronary artery; RPD right posterior descending branch. Adapted from Austen et al. (18). increment of 0.5 mm, and standard (B25f) and sharp (B46f) convolution kernels were used. Initial reconstructions were obtained at 65% of the RR interval. If motion artifacts were present, additional reconstructions were performed in 5% increments and decrements and displayed on dedicated workstations (Leonardo; Siemens, Forchheim, Germany). Reviews of the axial source images, muliplanar reconstructions, and curved multiplanar reconstructions were used to evaluate the CT dataset. The effective dose of CT angiography was calculated with dedicated software (CT-Expo V , Medical University Hannover, Hannover, Germany) (17). Conventional coronary angiography. Selective coronary angiography was performed with the transfemoral Judkins approach with standard techniques after right and left intracoronary administration of 150 to 200 g glycerin trinitrate. Quantitative analysis of the coronary angiograms (CAAS 5.7, Pie Medical Imaging B.V., Maastricht, the Netherlands) was performed by an experienced reader without knowledge of the results of MRI and CT. At least 2 orthogonal projections were evaluated; after catheter-based image calibration and automated vessel contour detection the measurement was performed in the projection that showed the highest degree of stenosis. A significant coronary stenosis was defined as 50% luminal diameter narrowing in segments 1.5 mm diameter. MRI and CT data analysis. The MRI and CT datasets were interpreted by the consensus of 2 experienced observers in a blinded fashion at 2 different centers, without knowledge of the results of conventional coronary angiography or the clinical characteristics of the patients. Image quality was assessed on a 4-point scale, where 1 poor (nondiagnostic), 2 moderate (diagnostic with poor visibility of the anatomic details of the coronary arteries), 3 good (good visibility of the anatomic details of the coronary arteries), and 4 excellent (excellent visibility and differentiation of the anatomic details of the coronary arteries) (8). A 16-coronary-arterysegment model according to the American Heart Association (modified 15-segment model, with segment 16 being the intermediate branch of the left coronary artery) (Fig. 2) (18) constituted the basis for visual assessment of significant coronary artery stenosis in segments 1.5 mm diameter. For vessels with multiple stenosis the most severe stenosis determined the final vessel stenosis. Coronary arteries with at least 1 visible coronary segment were included, and coronary segments with prior stent implantation or segments that were not visible by MRI or CT angiography were excluded from the analysis. To assess the clinical relevance of the MRI and CT data, they were additionally analyzed in their ability to predict subsequent revascularization on the basis of location of the stenosis and vessel size. This was compared with the final decision of

5 54 JACC: CARDIOVASCULAR IMAGING, VOL. 4, NO. 1, 2011 Figure 3. Composition of Study Population Flow chart of the study population, coronary artery disease prevalence, and excluded coronary segments. RIM Ramus intermediate branch; other abbreviations as in Figures 1 and 2. the invasive cardiologist, on the basis of clinical and angiographic information, to revascularize or not. To assess interobserver variability for interpretation of MRI and CT angiography, 2 independent observers visually evaluated the datasets in a randomly selected sample of 50 studies. Statistical analysis. The quantitative conventional coronary angiography served as the reference standard. The sample size was calculated according to the method proposed by Connor (19). To calculate the sample size we assumed a difference in perpatient diagnostic accuracy of more than 10% between MRI and CT (8). We intended to give the study 80% power for an alpha level of We estimated that a sample of 120 patients would be needed, assuming 50% disease prevalence, 80% agreement between the 2 modalities on a perpatient basis, and 10% dropout rate. The failure to falsify the null-hypothesis based on these assumptions is equivalent with the statement that the diagnostic accuracy of MRI and CT on a perpatient analysis differs by 10% (19). Statistical analysis was performed with a statistical software package (SPSS, version 17.0, SPSS, Chicago, Illinois). For all continuous parameters, data are given as mean SD. McNemar chi-square test was used to compare the diagnostic accuracy between MRI and CT angiography, and the area under the receiver-operator-characteristic (ROC) curve (AUC) (20) and DeLong and DeLong method (21) were used to compare the prediction of revascularization with both imaging modalities. Diagnostic accuracy, sensitivity, and specificity were calculated according to standard definitions. Agreement between observers was assessed with Cohen kappa statistics (22), and Wilcoxon test was used to compare the image quality of MRI and CT angiography. All tests were 2-sided, and a p value 0.05 was considered statistically significant. RESULTS Five patients could not be examined with MRI, because of claustrophobia or inadequate image quality (irregular breathing pattern); another 5 patients either cancelled the CT examination or had inadequate image quality, because of frequent extrasystoles or insufficient intravenous contrast. Thus, the final study cohort included 110 patients. Among the 330 arteries (right, left main left anterior descending [LAD], and intermediate branch left circumflex [LCX]), 5 arteries could not be evaluated by MRI, and another 3 could not be evaluated by CT, resulting in 322 vessels for the final analysis. Among the 1,561 coronary segments, 82 could not be evaluated by MRI, and another 50

6 JACC: CARDIOVASCULAR IMAGING, VOL. 4, NO. 1, could not be evaluated by CT; 60 further segments were stented, and 270 segments had a diameter 1.5 mm (Fig. 3). Representative examples of normal coronary angiogram and coronary stenosis that was detected by MRI and CT are shown in Figures 4 and 5. Demographic and clinical characteristics of the patients are shown in Table 1. The median time interval between the noninvasive tests and X-ray angiography was 1 day (mean, 0.8 day [range 0 to 3 days]), and MRI and CT were performed as sameday examinations in 85 patients (77%; mean interval, 0.1 day [range 0 to 3 days]). Mean heart rates during MRI and CT examination did not significantly differ ( beats/min vs beats/min, respectively; p 0.73). The MRI angiography was acquired during diastole in 100 patients (average heart rate, 62 7 beats/min; trigger delay, ms; acquisition duration, ms) and during systole in 10 patients (average heart rate, beats/min; trigger delay, ms; acquisition duration, ms). The mean effective scan time for magnetic resonance angiography was min. The optimal reconstruction window for the left coronary system (left main LAD, and LCX) in CT was found at 60% to 70% and 30% to 40% of the cardiac cycle in 105 and 5 patients, respectively, and for the right coronary artery (RCA) at 60% to 70% and 30% to 40% of the cardiac cycle in 101 and 9 patients, respectively. The mean time spent by the patient in the MRI and CT laboratories (including patient preparation) was min and min (p 0.001). Within 1 month after invasive coronary angiography, 53 patients underwent percutaneous revascularization (40 patients) or surgical revascularization (13 patients). Effective radiation dose for CT was msv for men and msv for women. Three patients had minor allergic reactions to contrast dye after CT angiography. Image quality. The image quality of the left main LAD was similar with MRI and CT angiography ( vs , respectively; p 0.89); for the RCA, MRI demonstrated higher image quality ( vs , respectively; p 0.001); however, for the LCX, CT showed higher image quality ( vs , respectively; p 0.001). The number of segments that could not be assessed was significantly (p 0.011) higher for MRI than for CT (Fig. 3); however, most of the coronary segments (48 of 82 [or 59%]) that could not be visualized by MRI were side Figure 4. Representative Example of Normal MRI, CT, and X-Ray Angiography Representative example of magnetic resonance imaging (MRI) and computed tomography (CT) volume rendering and corresponding invasive coronary angiography images, showing normal angiogram of left and right coronary artery systems. branch vessels (diagonal, marginal, or posterolateral). In contrast, only 19 of 50 (38%) segments that could not be visualized by CT were side branch vessels (Table 2). Patient-based analysis. The diagnostic performance of MRI and CT angiography on a per-patient basis is shown in Table 3 and Figure 6A. All cases of left main or 3-vessel disease (2 and 11 patients, respectively) were correctly diagnosed by MRI and CT angiography. Vessel-based analysis. Tables 3 and 4 provide direct comparisons of MRI and CT angiography with conventional coronary angiography for the analysis of coronary arteries. The diagnostic accuracy of MRI and CT angiography on a per-vessel basis was similar (Fig. 6C), with no significant differences among the right and left main LAD arteries. However, CT showed significantly higher diagnostic accuracy for the intermediate branch LCX than MRI. Prediction of revascularization. Table 5 and Figures 6B and 6D show similar ability of MRI and CT

7 56 JACC: CARDIOVASCULAR IMAGING, VOL. 4, NO. 1, 2011 Figure 5. MRI and CT in a Patient With 2-Vessel Coronary Artery Disease Typical examples of volume rendering (left panels), and reformatted images (center panels) of magnetic resonance imaging (MRI) and computed tomography (CT) angiography and corresponding invasive coronary angiography images (right panels) of the left and right coronary artery systems. (Top panels) Two-vessel disease involving the distal LAD (pink arrows) and the distal LCX (yellow arrows). (Bottom panels) Normal distal right coronary arteries by MRI, CT, and invasive coronary angiography. Note the excellent visualization of the coronary artery including distal segments and side branches. angiography to predict coronary intervention at 1 month, on the patient- and vessel-based analysis. On the basis of quantitative coronary angiography, 10 of Table 1. Baseline Characteristics of Study Population Characteristic Value Age, yrs Age range, yrs Sex, male 77 (70) BMI, kg/m BMI 30 kg/m 2 32 (29) Risk factors Hypertension 78 (70.9) Diabetes 28 (25.4) Hypercholesterolemia 67 (60.9) Current cigarette smoking 22 (20) Family history of CAD 55 yrs 55 (50) Known CAD 33 (30) Previous myocardial infarction 18 (16.3) Prior percutaneous coronary intervention 22 (20) Beta-blocker premedication 58 (53) Clinical presentation Typical angina 35 (31.8) Atypical angina 34 (30.9) Nonspecific chest pain 20 (18.2) No chest pain 21 (19.1) Distribution of disease by conventional coronary angiography None 48 (44) 1-vessel 34 (31) 2-vessel 17 (15) 3-vessel 11 (10) Data are n (%) unless otherwise indicated, mean SD when appropriate. BMI body mass index; CAD coronary artery disease. the 62 patients with stenosis 50% were not revascularized due to small vessels (n 4) or because the lesion was not regarded as hemodynamically relevant (n 6). On the vessel-based analysis of the 103 vessels, 29 were not revascularized. This was because the lesion was not regarded as hemodynamically relevant in 15 cases, the lesion was located in a small vessel or in distal segments in 7 cases, intervention was performed in a second session more than 1 month after the first intervention in 3 cases, the lesions were regarded as not suitable for intervention in 2 cases, the patient received bypass surgery and the distal LCX was not graftable in 1 case, and chronic total occlusion in 1 patient. Interobserver agreement. In 50 randomly selected patients, the patient-based interobserver agreement was kappa 0.80 (95% confidence interval [CI]: 0.64 to 0.96) for MRI and kappa 0.84 (95% CI: 0.69 to 0.99) for CT. On the vessel basis the interobserver agreement was kappa 0.78 (95% CI: 0.67 to 0.89) for MRI and kappa 0.82 (95% CI: 0.70 to 0.94) for CT. DISCUSSION The present study demonstrates similar diagnostic accuracy of MRI and CT angiography with 32- channel 3.0-T MRI and 64-slice CT for the detection of significant coronary artery stenosis in patients with suspected or known CAD scheduled for elective coronary angiography. However, CT angiography showed a favorable but nonsignificant

8 JACC: CARDIOVASCULAR IMAGING, VOL. 4, NO. 1, Table 2. Evaluation of Segments by Invasive Angiography, 32-Channel MRI, and 64-Slice CT Angiography 32-Channel MRI 64-Slice CT Not Evaluable False Positive False Negative Not Evaluable False Positive False Negative LM LAD Proximal Mid Distal Diagonal branches LCX Proximal Distal Marginal branches Intermediate branch RCA Proximal Mid Distal RPD/PL Total CT computed tomography; LAD left anterior descending artery; LCX left circumflex; LM left main; MRI magnetic resonance imaging; RCA right coronary artery; RPD/PL right posterior descending branch/posterolateral branch. trend toward higher diagnostic performance and better prediction of subsequent revascularization. Previous studies comparing MRI and CT angiography have yielded variable results with 1.5-T MRI and different generations of multislice CT (5 8). The relatively new CT generations have been demonstrated to outperform MRI angiography (7,8). However, in recent years MRI angiography has also shown substantial progress, and the 3.0-T MRI system (9,10) and 32-element coil (11,12) have been introduced for coronary imaging. Our results (Table 3) are in close agreement with those of a recently published meta-analysis (23) that directly compares MRI and CT angiography in 5 studies (325 patients) and shows similar sensitivity (87% vs. 87%) but higher specificity (77% vs. 70%). The differences might be related to the use of the 32-channel coil and 3.0-T MRI, both resulting in higher SNR. The only study that used 3.0-T MRI for coronary angiography resulted in slightly higher sensitivity (94% vs. 87%), specificity (82% vs. 77%), and diagnostic accuracy (88% vs. 83%) (10), compared with our results. The use of contrast agent, younger patient age, lower body mass index, and Table 3. Diagnostic Accuracy of 32-Channel MRI and 64-Slice CT Angiography for Patient- and Vessel-Based Detection of Coronary Stenosis >50% 32-Channel MRI 64-Slice CT p Value Patient-based analysis Sensitivity 54/62 (87 [76 93]) 56/62 (90 [80 95]) 0.16 Specificity 37/48 (77 [63 87]) 40/48 (83 [70 91]) 0.06 Positive predictive value 54/65 (83 [72 90]) 56/64 (88 [77 93]) 0.62 Negative predictive value 37/45 (82 [69 91]) 40/46 (87 [74 94]) 0.57 Diagnostic accuracy 91/110 (83 [75 87]) 96/110 (87 [80 92]) 0.38 Vessel-based analysis Sensitivity 83/103 (81 [72 87]) 87/103 (85 [76 90]) 0.52 Specificity 183/219 (84 [78 88]) 191/219 (87 [82 91]) 0.2 Positive predictive value 83/119 (70 [61 67]) 87/115 (76 [67 83]) 0.38 Negative predictive value 183/203 (90 [85 93]) 191/207 (92 [87 95]) 0.48 Diagnostic accuracy 266/322 (83 [78 86]) 278/322 (86 [82 90]) 0.09 Values are n/n (% [95% confidence interval]). CT computed tomography; MRI magnetic resonance imaging.

9 58 JACC: CARDIOVASCULAR IMAGING, VOL. 4, NO. 1, 2011 Figure 6. MRI and CT Diagnostic Performance and Prediction of Revascularization (A) The receiver-operator characteristic (ROC) curve describing the patient-based diagnostic performance of magnetic resonance imaging (MRI) and computed tomography (CT) angiography. The area under the curve (AUC) was 0.82 (95% confidence interval [CI]: 0.73 to 0.90) for MRI and 0.87 (95% CI: 0.79 to 0.94) for CT; p (B) The ROC curves for MRI (AUC 0.78; 95% CI: 0.69 to 0.87), CT (AUC 0.82; 95% CI: 0.74 to 0.90), and invasive angiography (AUC 0.91; 95% CI: 0.85 to 0.97) for prediction of coronary revascularization. Both curves were compared with the reference standard: patients who underwent subsequent revascularization and those who did not. On the patient basis, prediction of revascularization with MRI and CT was similar p 0.27; however, invasive angiography predicted coronary revascularization significantly better than MRI, p 0.005, and CT, p (C) The ROC curve on the vessel basis for MRI and CT diagnostic performance. The AUC was 0.82 (95% CI: 0.77 to 0.87) for MRI and 0.86 (95% CI: 0.81 to 0.90) for CT; p (D) The ROC curves on the vessel basis for prediction of revascularization: MRI (AUC 0.80; 95% CI: 0.74 to 0.86), CT (AUC 0.83; 95% CI: 0.79 to 0.89), and invasive angiography (AUC 0.94; 95% CI: 0.92 to 0.97). Prediction of revascularization with MRI and CT was similar, p 0.15; however, invasive angiography predicted coronary revascularization significantly better than MRI, p 0.001, and CT, p exclusion of patients with previous angioplasty might account for these differences. For 64-slice CT angiography the per-patient sensitivity and specificity ranged from 91% to 99% and 74% to 96%, respectively, among single-center studies (24) and from 85% to 99% and 64% to 90%, respectively, among multicenter studies (4,25), which is in agreement with our data. Compared with the data in the recently published meta-analysis (23) the sensitivity and specificity of our CT data are comparable but slightly lower: 90% versus 97% and 83% versus 87%, respectively. The differences might be because previous studies were often made in selected patients after elimination or imputation of lesions in a substantial number of segments that could not be evaluated. Moreover, previous studies were often performed in patients with suspected CAD or in populations with a low prevalence of CAD. Three-dimensional, free-breathing MRI angiography is a valuable noninvasive tool for the evaluation of CAD (3,26,27), without the use of radiation and potentially nephrotoxic contrast agents. However, the diagnostic performance of this technique at 1.5-T has not reached that of CT angiography (7,8). This is probably attributable to the lower SNR, resulting in inferior spatial and temporal resolution than in CT. The 3.0-T MRI systems, however, might overcome these shortcomings due to improved SNR as a result of increased strength of the static magnetic field (9). Further progress in the field of MRI angiography is the use of 32-element coils, which alleviate noise amplification to some extent and therefore potentially increase the SNR and enable the use of parallel imaging in the form of 2-dimensional SENSE (11,12). The advantage of using 32-channel 3.0-T MRI angiography was translated in the present study into high image quality of the coronary arteries as shown by visual assessment and high diagnostic performance, which was comparable to that of CT angiography. However, the LCX had lower image

10 JACC: CARDIOVASCULAR IMAGING, VOL. 4, NO. 1, Table 4. Diagnostic Accuracy of 32-Channel MRI and 64-Slice CT Angiography for Detection of Coronary Stenosis >50% in Different Vessels 32-Channel MRI 64-Slice CT p Value RCA Sensitivity 28/32 (88 [72 95]) 25/32 (78 [61 89]) 0.20 Specificity 63/75 (84 [74 91]) 65/75 (87 [77 93]) 0.77 Positive predictive value 28/40 (70 [55 82]) 25/35 (71 [55 84]) 1.0 Negative predictive value 63/67 (94 [86 98]) 65/72 (90 [81 95]) 0.53 Diagnostic accuracy 89/107 (83 [75 89]) 90/107 (84 [76 90]) 1.0 LM-LAD Sensitivity 35/42 (83 [69 92]) 37/42 (88 [75 95]) 0.19 Specificity 56/68 (82 [72 90]) 57/68 (84 [73 91]) 1.0 Positive predictive value 35/47 (75 [60 85]) 37/48 (77 [63 87]) 0.81 Negative predictive value 56/63 (89 [79 95]) 57/62 (92 [82 97]) 0.76 Diagnostic accuracy 91/110 (83 [75 89]) 94/110 (85 [78 91]) 0.48 LCX-intermediate branch Sensitivity 20/29 (69 [51 82]) 25/29 (86 [69 95]) 0.07 Specificity 64/76 (84 [74 90]) 69/76 (91 [82 95]) Positive predictive value 20/32 (63 [45 77]) 25/32 (78 [61 89]) 0.27 Negative predictive value 64/73 (88 [78 93]) 69/73 (95 [87 98]) 0.24 Diagnostic accuracy 84/105 (80 [71 86]) 94/105 (90 [82 94]) Values are n/n (% [95% confidence interval]). Abbreviations as in Table 2. quality and lower diagnostic accuracy as assessed by MRI compared with CT. This might be due to the relatively small caliber and posterior location of the LCX, which results in a lower SNR because of the increased distance from the artery to the receiver coils. By contrast, MRI provides higher image quality for the RCA. This is possibly related to the higher temporal resolution (acquisition duration), because the rest period of the RCA during the cardiac cycle is shorter than that of the left coronary artery system (28). This, however, did not result in a statistically relevant diagnostic advantage compared with CT. Conversely, compared with MRI, CT has higher spatial resolution, most probably resulting in a significantly higher number of segments that could not be assessed by MRI. However, the most common false positive CT results are found in the proximal segments (Table 2), which might be because calcified obstructions which might cause blooming effects and therefore might result in obscured visualization of the underlying coronary lumen are more often located in the proximal segments. This can be regarded Table 5. Diagnostic Accuracy of 32-Channel MRI and 64-Slice CT Angiography for Patient- and Vessel-Based Prediction of Coronary Revascularization 32-Channel MRI 64-Slice CT Invasive Angiography p Value* Patient-based analysis Sensitivity 46/52 (89 [77 95]) 48/52 (92 [82 97]) 52/52 (100 [93 100]) 0.39 Specificity 39/58 (67 [54 78]) 42/58 (72 [59 82]) 48/58 (83 [71 90]) 0.58 Positive predictive value 46/65 (71 [58 80]) 48/64 (75 [63 84]) 52/62 (84 [72 91]) 0.69 Negative predictive value 39/45 (87 [74 93]) 42/46 (91 [79 97]) 48/48 (100 [92 100]) 0.52 Diagnostic accuracy 85/110 (77 [69 84]) 90/110 (82 [73 88]) 100/110 (91 [84 95]) 0.38 Vessel-based analysis Sensitivity 62/74 (84 [73 91]) 65/74 (88 [78 94]) 74/74 (100 [95 100]) 0.15 Specificity 191/248 (77 [71 82]) 198/248 (80 [74 84]) 219/248 (88 [83 92]) 0.32 Positive predictive value 62/119 (52 [43 61]) 65/115 (57 [47 65]) 74/103 (72 [62 80]) 0.51 Negative predictive value 191/203 (94 [89 97]) 198/207 (96 [91 98]) 219/219 (100 [98 100]) 0.50 Diagnostic accuracy 253/322 (79 [73 83]) 263/322 (82 [77 86]) 293/322 (91 [87 94]) 0.21 Values are n/n (%); values in brackets are 95% confidence intervals. *Comparison between computed tomography (CT) and magnetic resonance imaging (MRI) angiography.

11 60 JACC: CARDIOVASCULAR IMAGING, VOL. 4, NO. 1, 2011 as a drawback of the technique, and data on prognosis needs to clarify its clinical relevance. Because the coronary anatomy of the patient besides symptoms and functional tests, as determined by conventional coronary angiography is crucial for deciding on the indication for revascularization procedure (29), we additionally compared the ability of MRI and CT angiography to predict the need for percutaneous or surgical coronary revascularization. The 2 imaging modalities had similar ability to identify patients who underwent coronary revascularization on the basis of clinical and angiographic information; however, invasive angiography predicted coronary revascularization significantly better than MRI and CT (Fig. 6). Because luminal narrowing alone did not provide the physiological effects of stenosis and functional imaging offers different and complementary information, the combined use of noninvasive coronary angiography and functional imaging in the same study protocol might be necessary to adequately predict the need for revascularization and might improve the relatively low specificity of both modalities (Table 5). Because MRI stress testing has been intensely researched in recent years and its high diagnostic value in the absence of radiation and potentially nephrotoxic contrast agents has been demonstrated (30), a combined use of MRI stress testing and coronary angiography in the same imaging session would provide incremental and crucial benefit for the evaluation of CAD. In addition, MRI is a major aid in diagnosing heart disease, particularly when examining ventricular function and myocardial viability. Study limitations. We used contrast medium for CT angiography only, but gadolinium-based contrast agent might further improve the results of MRI angiography (10); however, the use of contrast medium might also enhance coronary veins and impair the depiction of coronary arteries. In addition, gadolinium-based contrast agent results in additional study cost and has potential side effects, particularly for patients with impaired renal function. It is also difficult to repeat the scan in the same imaging session if the acquisition is aborted for some reason. We excluded segments with coronary stents from our analysis, because they cause artifacts mainly in MRI but also in CT angiography and make a reliable analysis impossible. In symptomatic patients after stent implantation, in-stent restenosis cannot be excluded, and a functional test might be more suitable in this patient population. In the present study we used a high-end research MRI system and compared it with a clinically available 64-slice CT machine. More recently, 320- or 256- slice CT has been introduced. This new generation of CT scanners, which were not available at the participating centers, might have better diagnostic accuracy for the detection of CAD. Finally, the present study was performed in patients with clinical indications for cardiac catheterization and a high prevalence of CAD. The study findings might not necessarily be extrapolated to patients with less severe CAD. CONCLUSIONS In this study we have demonstrated that 32-channel 3.0-T MRI and 64-slice CT angiography similarly identify clinically relevant coronary stenosis and similarly predict subsequent revascularization in patients with suspected or known CAD scheduled for elective coronary angiography. However, CT angiography shows a trend toward higher diagnostic performance. Acknowledgments The authors thank Dr. Matthias Stuber from the Johns Hopkins University School of Medicine, Baltimore, Maryland, and Dr. Bernhard Schnackenburg from Philips Clinical Science, Hamburg, Germany, for optimization of the magnetic resonance angiography sequence. They also thank the technicians of the 3 laboratories for assistance in performing the study and Anne Gale (ELS) for editorial support. Reprint requests and correspondence: Dr. Ashraf Hamdan, Department of Internal Medicine/Cardiology, German Heart Institute Berlin, Augustenburger Platz 1, Berlin, Germany. hamdashraf@gmail.com. REFERENCES 1. Lloyd-Jones D, Adams R, Carnethon M, et al. Heart disease and stroke statistics 2009 update: a report from the American Heart Association Statistics Committee and Stroke Statistics Subcommittee. Circulation 2009; 119:e Dissmann W, de Ridder M. The soft science of German cardiology. Lancet 2002;359: Kim WY, Danias PG, Stuber M, et al. Coronary magnetic resonance angiography for the detection of coronary stenoses. N Engl J Med 2001;345: Miller JM, Rochitte CE, Dewey M, et al. Diagnostic performance of coronary angiography by 64-row CT. N Engl J Med 2008;359:

12 JACC: CARDIOVASCULAR IMAGING, VOL. 4, NO. 1, Gerber BL, Coche E, Pasquet A, et al. Coronary artery stenosis: direct comparison of four-section multi-detector row CT and 3D navigator MR imaging for detection initial results. Radiology 2005;234: Kefer J, Coche E, Pasquet A, et al. Head to head comparison of multislice coronary CT and 3D navigator MRI for the detection of coronary artery stenosis. J Am Coll Cardiol 2005;46: Dewey M, Teige F, Schnapauff D, et al. Noninvasive detection of coronary artery stenoses with multislice computed tomography or magnetic resonance imaging. Ann Intern Med 2006;145: Pouleur AC, le Polain de Waroux JB, Keefer J, Pasquet A, Vanoverschelde JL, Gerber BL. Direct comparison of whole-heart navigator-gated magnetic resonance coronary angiography and 40- and 64-slice multidetector row computed tomography to detect the coronary artery stenosis in patients scheduled for conventional coronary angiography. Circ Cardiovasc Imaging 2008;1: Stuber M, Botnar RM, Fischer SE, et al. Preliminary report on in vivo coronary MRA at 3 Tesla in humans. Magn Reson Med 2002;48: Yang Q, Li K, Liu X, et al. Contrastenhanced whole-heart coronary magnetic resonance angiography at 3.0-T. A comparative study with X-ray angiography in a single center. J Am Coll Cardiol 2009;54: Nehrke K, Börnert P, Mazurkewitz P, Winkelmann R, Grässlin I. Freebreathing whole-heart coronary MR angiography on a clinical scanner in four minutes. J Magn Reson Imaging 2006;23: Niendorf T, Hardy CJ, Giaquinto RO, et al. Toward single breath-hold whole-heart coverage coronary MRA using highly accelerated parallel imaging with a 32-channel MR system. Magn Reson Med 2006;56: Einstein AJ, Henzlova MJ, Rajagopalan S. Estimating risk of cancer associated with radiation exposure from 64- slice computed tomography coronary angiography. JAMA 2007;298: Weber OM, Martin AJ, Higgins CB. Whole-heart steady-state free precession coronary artery magnetic resonance angiography. Magn Reson Med 2003;50: Jahnke C, Nehrke K, Paetsch I, et al. Improved bulk myocardial motion suppression for navigator-gated coronary magnetic resonance imaging. J Magn Reson Imaging 2007;26: Etienne A, Botnar RM, van Muiswinkel AM, Boesiger P, Manning WJ, Stuber M. Soap-bubble visualization and quantitative analysis of 3D coronary magnetic resonance angiograms. Magn Reson Med 2002;48: Stamm G, Nagel HD. [CT-expo a novel program for dose evaluation in CT]. Rofo 2002;174: Austen WG, Edwards JE, Frye RL, et al. A reporting system on patients evaluated for coronary artery disease. Report of the Ad Hoc Committee for Grading of Coronary Artery Disease, Council on Cardiovascular Surgery, American Heart Association. Circulation 1975;51: Connor RJ. Sample size for testing differences in proportions for the paired sample design. Biometrics 1987;43: Zou KH, O Malley AJ, Mauri L. Receiver operating characteristic analysis for evaluating diagnostic tests and predictive models. Circulation 2007;115: DeLong ER, DeLong DM, Clarke- Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988;44: Cohen J. A coefficient of agreement for nominal scales. Educ Psychol Meas 1960;20: Schuetz GM, Zacharopoulou NM, Schlattmann P, Dewey M. Metaanalysis: noninvasive coronary angiography using computed tomography versus magnetic resonance imaging. Ann Intern Med 2010;152: Min JK, Shaw LJ, Berman DS. The present state of coronary computed tomography angiography a process in evolution. J Am Coll Cardiol 2010;55: Meijboom WB, Meijs MF, Schuijf JD, et al. Diagnostic accuracy of 64- slice computed tomography coronary angiography: a prospective, multicenter, multivendor study. J Am Coll Cardiol 2008;52: Jahnke C, Paetsch I, Nehrke K, et al. Rapid and complete coronary arterial tree visualization with magnetic resonance imaging: feasibility and diagnostic performance. Eur Heart J 2005; 26: Sakuma H, Ichikawa Y, Chino S, Hirano T, Makino K, Takeda K. Detection of coronary artery stenosis with whole-heart coronary magnetic resonance angiography. J Am Coll Cardiol 2006;48: Jahnke C, Paetsch I, Achenbach S, et al. Coronary MR imaging: breathhold capability and patterns, coronary artery rest periods, and beta-blocker use. Radiology 2006;239: Smith SC Jr., Feldman TE, Hirshfeld JW Jr., et al. ACC/AHA/SCAI 2005 guideline update for percutaneous coronary intervention summary article: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (ACC/AHA/SCAI Writing Committee to Update the 2001 Guidelines for Percutaneous Coronary Intervention). J Am Coll Cardiol 2006;47: Gebker R, Jahnke C, Manka R, Hamdan A, Schnackenburg B, Fleck E, Paetsch I. Additional value of myocardial perfusion imaging during dobutamine stress magnetic resonance for the assessment of coronary artery disease. Circ Cardiovasc Imaging 2008; 1: Key Words: 3.0-T MRI y coronary angiography y CT.