Noninvasive Detection of Coronary Artery Stenoses with Multislice Computed Tomography or Magnetic Resonance Imaging

Context A negative result on a noninvasive coronary artery imaging test could reduce the need for coronary angiography. Contribution The authors did 3 coronary artery imaging testsmultislice computed tomography (CT), magnetic resonance imaging (MRI), and conventional radiography (the reference standard)in 108 patients with clinically suspected coronary artery disease (CAD). The likelihood ratios for positive test results were 14.4 for multislice CT and 18.8 for MRI. For negative test results, they were 0.08 for multislice CT and 0.16 for MRI. Implications These tests' potential value is to rule out CAD. Multislice CT is superior to MRI for this purpose, but neither test reliably excludes CAD if the pretest odds are high, as in typical angina pectoris. The Editors Coronary artery disease is a major public health issue, affecting 13.2 million persons in the United States and causing more than 500000 deaths each year (1, 2). Diagnostic cardiac catheterization is performed more than 1.3 million times per year in U.S. hospitals (2) and is the best method available for detection of coronary stenoses; however, it is invasive and not without risk (3). Moreover, approximately 66% of all conventional coronary angiographies performed in Germany are for diagnostic purposes only and include no interventional procedure (4). From both the patient's point of view and a socioeconomic perspective, a noninvasive, painless, and cost-effective diagnostic approach that reduced the need for cardiac catheterization would be an important advance. Two techniques, multislice computed tomography (CT) and magnetic resonance imaging (MRI), have been investigated for this purpose. In recent studies, multislice CT has been reported to be fairly accurate in the diagnosis of coronary artery disease (510), while a multicenter study has shown that MRI of the coronary arteries permits diagnosis of left main coronary artery stenoses and exclusion of 3-vessel disease (11). The results of these studies have implied that the diagnostic accuracy of multislice CT might be superior to that of MRI. However, multislice CT, unlike MRI, involves the use of radiation and intravenous administration of a contrast agent; thus, it should be recommended only if shown to provide superior diagnostic accuracy. We therefore conducted a prospective study to compare the diagnostic performance of multislice CT and MRI in a consecutive series of patients scheduled to undergo conventional coronary angiography. This intraindividual design made it possible to compare the tests in the same patients (12) and to analyze whether multislice CT and MRI complement each other. Methods Study Design The study was a single-institution comparison of the diagnostic performance of multislice CT and MRI for the detection of coronary artery stenoses in a prospective patient cohort, with quantitative coronary angiography as the reference standard. All patients and all coronary arteries were included in each analysis, even if a study or a vessel was not interpretable (that is, an intention-to-diagnose design was used). By using this design, we could avoid overestimating diagnostic accuracy (1214). All 25 criteria of the Standards for Reporting of Diagnostic Accuracy (STARD) statement (13) can be found in this report. Multislice CT and MRI were both performed before conventional coronary angiography to avoid differential verification bias (15). The study protocol was approved by the institutional review board and the responsible federal authority (Federal Department for Radiation Protection). Study Group The study group consisted of consecutive eligible patients who were referred to our institution in Berlin, Germany, by outpatient centers and scheduled to undergo conventional coronary angiography within 14 days for clinically suspected coronary artery disease based on symptoms or results of diagnostic tests (for example, treadmill exercise test, myocardial scintigraphy, and echocardiography). Patients were eligible for the study if they were at least 40 years of age and were in sinus rhythm. As in previous studies (511), sinus rhythm was an inclusion criterion because a regular heartbeat is currently required for consistent successful application of noninvasive coronary imaging. The age limit of 40 years was set by the Federal Department for Radiation Protection because younger patients have increased susceptibility to ionizing radiation. The exclusion criteria were previous conventional coronary angiography, unstable angina or acute myocardial infarction, coronary artery bypass graft or stent, pregnancy or breastfeeding, or orthopnea. Patients were also excluded if they were under guardianship at the time of the study. Patients with contraindications to MRI (pacemaker, severe claustrophobia, or intracranial or intra-auricular metallic implants) or multislice CT (renal insufficiency [creatinine level >132.6 mol/L (>1.5 mg/dL)] or allergy to iodinated contrast agents) were not examined with the contraindicated method and were excluded from the analysis. Enrollment took place between 5 November 2003 and 16 September 2004. All patients gave written informed consent. If no contraindications to nitroglycerine (such as aortic stenosis) were present, each patient received sublingual isosorbide dinitrate (5 mg) immediately before multislice CT and MRI. Sixty-five patients undergoing multislice CT were receiving long-term oral -blocker therapy, but no additional oral or intravenous -blockers were given before CT because such therapy might have favored CT over MRI (a negative correlation has been shown between heart rate and image quality for multislice CT [16]). Multislice CT Protocol Imaging was performed during 1 breath-hold by using a 16-slice CT scanner with 0.5-mm detector collimation (Aquilion 16, Toshiba Medical Systems, Otawara, Japan), as described elsewhere (17), after intravenous injection of a nonionic, iso-osmolar contrast agent (iodixanol, 320 mg of iodine per mL [Visipaque, GE Healthcare Biosciences, Buckinghamshire, United Kingdom]) at a rate of 3.5 mL/s. Further details of the multislice CT protocol are given in the Appendix. MRI Protocol Magnetic resonance imaging was performed on a 1.5-T system (Magnetom Sonata, Siemens Medical Solutions, Erlangen, Germany) equipped with a high-performance gradient subsystem (maximum amplitude, 40 mT/m, and minimum rise time, 200 microseconds) using a dedicated cardiac 12-element phased-array coil as described recently (18). Further details of the MRI protocol are given in the Appendix. Conventional Coronary Angiography Selective coronary angiography was performed with the transfemoral Judkins approach by using standard techniques after right and left intracoronary administration of 100 to 150 g of isosorbide dinitrate. Interpretation of Imaging Results Multislice CT and magnetic resonance angiograms were processed and interpreted independently in random order by 2 readers on a workstation (Vitrea 2, version 3.3, Vital Images, Plymouth, Minnesota) in a blinded fashion, without knowledge of the results of conventional coronary angiography and clinical characteristics. For multislice CT, all 15 coronary artery segments (according to the classification of the American Heart Association; see the Appendix Figure) (19) constituted the basis for evaluation. In contrast, for MRI only the 8 proximal and middle segments, namely segments 1, 2, and 3 (of the right coronary artery); segment 5 (the left main coronary artery); segments 6 and 7 (of the left anterior descending coronary artery); and segments 11 and 13 (of the left circumflex coronary artery) were evaluated because the limited spatial resolution of MRI does not allow detection of coronary stenoses in distal segments and side branches (11). With axial slices, 3-dimensional reconstructions, and curved multiplanar reformations, all coronary arteries were classified quantitatively as having clinically significant disease (50% diameter reduction) or no significant disease (<50%) on multislice CT by using a semiautomatic vessel analysis tool (17) and on MRI by using visual quantification as described elsewhere (18). Coronary arteries that could not be adequately interpreted because of poor image quality were classified as not interpretable. Appendix Figure. American Heart Association classification of coronary artery segmental anatomy. Coronary artery segments are numbered 1 through 15. 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. These branches may or may not be visualized in the individual patient. Those whose origins can vary widely are shown unattached to the parent artery. Circ = left circumflex coronary artery; D1 = first diagonal branch; D2 = second diagonal branch; LAD = left anterior descending coronary artery; main LCA = 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. (19). Quantitative analysis of the coronary angiograms (Integris 3000, Philips Medical Systems, Best, the Netherlands) was performed by an experienced reader without knowledge of the results of multislice CT and MRI. At least 2 orthogonal projections were evaluated; the measurement was performed in the projection that showed the highest degree of stenosis. The diameter of the reference vessel on conventional coronary angiography had to measure at least 1.5 mm for a stenosis to be included in the analysis, thus covering all stenoses that might be targets for revascularization (20). Secondary Outcomes The total room time required for each diagnostic method and the amount of contrast agent administered for multislice CT and conventional coronary angiography were secondary outcomes. In a consecutive