Intravenous Coronary Angiography by Electron Beam Computed Tomography
A Clinical Evaluation
Background—Noninvasive detection of coronary stenoses with electron beam CT (EBCT) after intravenous injection of contrast medium has recently emerged. We sought to determine the diagnostic accuracy of EBCT angiography in the clinical setting using conventional coronary angiography as the “gold standard.”
Methods and Results—Thirty-seven patients (30 men) were investigated. After intravenous injection of 150 mL of contrast medium, 40 to 60 consecutive transaxial tomograms, covering the proximal and middle parts of the coronary arteries, were obtained with ECG triggering at end diastole during breath-holding. Three-dimensional reconstructions of the proximal and middle parts of the arteries were compared with the conventional angiograms. Of the 259 proximal and middle coronary segments, 211 (81%) were analyzable by EBCT. Of the left anterior descending coronary artery (LAD) segments, 95% were assessable. Right coronary artery (RCA) and left circumflex artery (LCx) segments were assessable in 66% and 76%, respectively. Overall sensitivity and specificity to detect a >50% diameter stenosis were 77% and 94%, respectively. This was 82% and 92% for the LAD, 60% and 97% for the RCA, and 83% and 89% for the LCx (all figures based on assessable lesions).
Conclusions—Intravenous EBCT coronary angiography is a promising coronary imaging technique. The technique is not yet robust enough to be an alternative to conventional coronary angiography. It can detect and rule out significant coronary artery disease of the left main proximal and mid portions of the LAD with good accuracy.
Recently, electron beam CT (EBCT) has emerged as an imaging technique that allows visualization of epicardial coronary arteries after intravenous injection of contrast material.1 2 3 4 5 Few data concerning the diagnostic accuracy of EBCT coronary angiography are available at the moment. The latest upgrading of the scanner allows improved resolution in the scanning direction. Slice thickness (collimation) and table feed after each scan can be decreased to 1.5 mm. The purpose of the present study was to compare noninvasive EBCT coronary angiography with conventional coronary angiography in the clinical setting and to determine the diagnostic accuracy of EBCT angiography.
Thirty-seven patients who underwent diagnostic coronary angiography were asked to participate in the study and gave informed consent. The study was approved by the Institutional Review Board. Two patient slots per week for cardiac research were available at the EBCT site. Therefore, our patient population is not strictly consecutive. All patients who met the inclusion criteria were approached until the 2 slots for that week were filled. Exclusion criteria were previous bypass operation or stent implantation, severe lung disease or comorbidity that made breath-holding difficult, renal failure, nonsinus rhythm, and unstable clinical condition.
The EBCT scanner (Siemens Evolution) allows the acquisition of high-resolution tomograms in 100 ms, which is fast enough to prevent cardiac motion artifacts. A description of scanner specifications can be found elsewhere.6 Contrast transit time was determined by injection of 10 mL of contrast medium (Iopromide, Schering) at a rate of 4 mL/s through an antecubital vein and visualization of the passage of the contrast through the ascending aorta by 20 consecutive tomograms. The time from contrast injection to peak density of the aorta was considered the transit time. Image acquisition started with the injection of 150 mL contrast medium at 4 mL/s. At transit time, tomography commenced just proximal to the takeoff of the left main (LM) coronary artery after an ECG trigger at 80% of the RR interval. The table increment after each tomogram was 1.5 mm. A total of 40 to 60 tomograms were acquired. Because of patient limitations (maximal breath-hold time) and scanner limitations (1 scan per heart cycle and 1.5-mm tomogram thickness), only the proximal and middle parts of the coronary arteries could be visualized consistently. To decrease breath-holding time, atropine 0.5 to 1.0 mL was administered if heart rate was <60 bpm. Radiation dose was estimated to be <20 mGy.
The 2-dimensional tomograms were transferred to a Silicon Graphics workstation, where they were stacked and interpolated to form a 3-dimensional (3D) volume by use of Voxel-View volume-rendering software. Postprocessing techniques were applied to better visualize the coronary arteries between the other contrast-enhanced cardiac structures.7 8 Average postprocessing time was 15 to 20 minutes (Figure 1⇓). To selectively visualize the coronary lumen, renderings were made with a lower threshold of 110 Hounsfield units. Two cardiologists unaware of the results of the selective coronary angiogram or the EBCT angiograms independently evaluated either EBCT angiograms (the 3D reconstructions and the individual tomograms) or the selective coronary angiograms. The coronary tree was divided into proximal, middle, and distal segments according to AHA guidelines.9 The proximal and middle segments were graded as assessable or nonassessable. Then they were graded as having either no significant disease (<50% diameter stenosis) or significant disease (≥50% diameter stenosis). In case of disagreement, a third cardiologist decided.
The selective angiogram served as the “gold standard” for determination of the diagnostic value of the noninvasive EBCT coronary angiogram. The diagnostic accuracy of EBCT angiography is expressed as sensitivity, specificity, and positive and negative predictive value.
Thirty-seven patients (30 men) underwent both conventional and EBCT coronary angiography. Mean age was 58±6 years (range, 42 to 82 years). Mean weight was 70±10 kg, and mean height was 171±12 cm. Mean time interval between both examinations was 10±13 days. At conventional angiography, 9 patients had no significant coronary stenosis, 12 had 1-vessel disease, 11 had 2-vessel disease, and 5 had 3-vessel disease. LM disease was present in 1 patient. Mean heart rate during the EBCT angiography was 72±10 bpm (60 to 105 bpm). Mean breath-holding time was 36 seconds. After mild hyperventilation and instruction, all patients were able to hold their breath for ≥40 tomograms. Atropine 0.5 to 1 mg was administered in 25 patients. Conventional angiography showed 1 significant LM lesion, 15 right coronary artery (RCA) lesions, 16 left anterior descending coronary artery (LAD) lesions, and 8 left circumflex (LCx) lesions in the proximal and middle portions of the coronary tree. Of the 259 proximal and middle coronary artery segments, 211 (81%) were assessable by EBCT angiography (Table 1⇓). Five (33%) of the proximal and middle RCA lesions and 2 (25%) of the proximal and middle LCx lesions were located in segments deemed not assessable by EBCT angiography. Table 2⇓ summarizes diagnostic accuracy parameters of EBCT angiography. The causes for the inability to assess 46 proximal and middle coronary segments with EBCT are summarized in Table 3⇓. The causes for false-negatives or false-positives are given in Table 4⇓. The major cause of false-positive classification was poor opacification in small distal coronary arteries. Calcification of the vessel wall obscuring a luminal narrowing was the major cause of false-negative classification. The only reported side effect of atropine was oral dryness in 5 patients. Angina occurred once, with reversal after oral nitroglycerin.
Others1 2 3 4 5 and we have shown that EBCT coronary angiography is feasible and safe. It allows visualization of the proximal and middle coronary arteries in a majority of patients. With special computer software, the individual tomograms can be stacked and 3D renderings of the coronary arteries constructed.7 8
Limitations of the Technique
The inability to assess coronary anatomy in a large proportion of mid RCA and LCx lesions requires further improvement of the technique. Conversely, all but 1 of the LM arteries and 95% of the proximal and mid LAD segments were assessable by current EBCT with a good diagnostic accuracy (Table 2⇑). Breath-holding limitations and single tomogram acquisition per heart cycle restrict the volume of the heart that can be scanned during 1 contrast injection. Therefore, distal coronary arteries are only rarely visualized. Although one could argue whether stenoses in small distal coronary arteries have important symptomatic or prognostic meaning per se, visualization of the posterior descending artery is important and should be possible. A second contrast injection can visualize the distal part of the coronary trajectory, but at the expense of doubling the contrast volume (Figure 2⇓). To accurately visualize coronary arteries <2 mm (the major cause of false-positive results and nonassessability), spatial resolution needs improvement.
A recently introduced new detector array is expected to improve in-plane resolution by 30%. This will probably increase assessability and diagnostic accuracy in small coronary vessels and might improve discrimination of overlapping contrast-filled structures. At the moment, the scanner can make only 1 tomogram per heart cycle. Ideally, the complete heart should be scanned within a few heartbeats, thereby shortening breath-holding time, decreasing the deleterious effect of arrhythmias on imaging, and reducing the total amount of contrast medium necessary for opacification. For this, the scanner has to be modified to allow acquisition of multiple simultaneous parallel tomograms and thus become more of a volume scanner. Finally, to prevent end-diastolic motion artifacts of the RCA, tomogram acquisition time <100 ms and true end-diastolic ECG (R wave) triggering are necessary. A solution to the problem of circular calcification of the vessel wall (a major source of false-negatives) might be to visualize the coronary artery from the inside. A postprocessing technique called “fly-through” tracks the contrast-enhanced lumen of the artery on the individual tomograms. After stacking and interpolation, the computer constructs a movie that gives the illusion of traveling through the artery.
EBCT coronary angiography is a technique under development and is currently not an alternative to conventional coronary angiography. At the present time, it can detect and rule out significant coronary artery disease of the proximal and middle portions of the LAD with good accuracy and thus may provide us with important prognostic information.
- Received August 18, 1998.
- Revision received October 5, 1998.
- Accepted October 8, 1998.
- Copyright © 1998 by American Heart Association
Achenbach S, Moshage W, Bachmann K. Detection of high-grade restenosis after PTCA using contrast-enhanced electron beam CT. Circulation. 1997;96:2785–2788.
Gould RG. Principles of ultrafast computed tomography: historical aspects, mechanisms and scanner characteristics. In: Stanford W, Rumberger JA, eds. Ultrafast Computed Tomography in Cardiac Imaging: Principles and Practice. Mt Kisco, NY: Futura; 1993:1–16.
Van Ooijen PMA, de Feyter PJ, Oudkerk M. An introduction to three dimensional cardiac image rendering and processing. Cardiology. 1997;4:312–319.
American Heart Association Committee Report. A reporting system on patients evaluated for coronary artery disease. Circulation. 1975;51:7–34.The diagnostic accuracy of intravenous electron beam CT (EBCT) coronary angiography was determined. Thirty-seven patients underwent both EBCT and conventional coronary angiography. Of the proximal and middle left anterior descending coronary artery (LAD) segments, 95% were assessable. This was 66% for the right coronary artery (RCA) and 76% for the circumflex artery (LCx). Overall sensitivity and specificity to detect a >50% diameter stenosis were 77% and 94%. For the LAD, this was 82% and 92%; the RCA, 60% and 97%; and the LCx, 83% and 89%. EBCT imaging can detect and rule out significant coronary artery disease of the left main, proximal, and mid portion of the LAD with good accuracy.