Noninvasive Coronary Angiography

Abstract

WHEN SONES INADVERTENTLY PERFORMED THE first coronary angiogram in 1958, he could not anticipate the profound future implications of that event. Four years later, when he reported his experience with more than 1000 procedures, the technique was still considered experimental and was dismissed by many leading authorities in cardiovascular medicine. Yet, the introduction of coronary angiography started a new era, leading to the rapid development of coronary artery bypass graft surgery and percutaneous revascularization for the routine management of ischemic heart disease. Four decades later, more than 2 million angiographic procedures are performed annually in the United States alone. Although invasive coronary angiography clearly has led to improved outcomes, it also has contributed to greater expenses in health care cost, limiting its potential to become more widely available. This has led to a growing interest in the scientific community in the development of less expensive, noninvasive alternative methods for evaluating coronary anatomy. Multislice computed tomography (MSCT) has recently emerged above other competing technologies, such as electron-beam computed tomography and magnetic resonance imaging, as a practical alternative to invasive coronary angiography. Modern MSCT systems can provide electrocardiogram-gated acquisition with adequate temporal resolution (100-220 ms) and with the submillimeter spatial resolution needed to visualize with sufficient detail the lumen of the coronary arteries. Several studies have investigated the accuracy of MSCT in patients with known or suspected coronary artery disease (CAD). In these studies, analysis ofMSCThasbeenmostly limited tocoronary segmentsgreater than or equal to 1.5 or 2 mm in diameter, and up to 5% to 20% of all analyzable segments have been deemed nonevaluable due to motion artifacts, severe calcified plaques, and other technical imaging problems. The sensitivity and specificity of MSCT for detecting a 50% or greater diameter reduction in coronary segments has ranged between 72% and 95% and between 85% and 100%, respectively. Many of the published series have enrolled nonconsecutive patients, and only a few have reported the performance characteristics of MSCT using each patient as the unit of analysis. In this issue of JAMA, Hoffmann and colleagues evaluated the diagnostic accuracy of MSCT for the detection of obstructive CAD. Their study includes a large series of nonselected patients in whom very few (6.4%) coronary segments larger than 1.5 mm in diameter were excluded from analysis due to limited image quality. The majority (68%) of these excluded segments were considered nonevaluable due to imaging artifacts related to cardiac motion, most often in patients with resting heart rates greater than 80/min. Motion artifacts still represent an important limitation of current MSCT technology, even though most investigators now routinely administer -blockers to reduce resting heart rate. Extensive vessel calcifications also limit the interpretation of luminal stenosis by MSCT, often leading to overestimation of severity. These accounted for many of the falsepositive results observed by Hoffmann et al. Despite these limitations, their reported per-segment sensitivity (95%) and specificity (98%) are noteworthy, particularly when these performance characteristics are compared with those of other indirect methods used for the detection of obstructive CAD, such as nuclear scintigraphy or echocardiographic stress tests. The authors also report diagnostic characteristics according to a per-patient–based analysis. This is critically important, since the implications of detecting or missing the presence or absence of any significant coronary obstruction are more clinically relevant from the perspective of the individual patient. Even after accounting for nonevaluable segments, the high positive (90%) and negative (95%) predictive values strongly support the conclusion that, in this patient population, MSCT is a robust test for establishing the diagnosis of obstructive CAD. In addition to the conventional binomial analysis, the authors performed a quantitative comparison of percentage luminal stenosis as determined by MSCT and invasive angiography. Receiver operating characteristic curves were constructed to estimate discriminative power for identifying patients who might be candidates for revascularization. The area under the curve was 0.97 (95% confidence interval, 0.91-1.00) for detecting greater than 50% left main artery disease, greater than