- Email: [email protected]
Intravenous three-dimensional coronary angiography using contrast enhanced electron beam computed tomography
Intravenous Three-Dimensional Coronary Angiography Using Contrast Enhanced Electron Beam Computed Tomography Matthew J. Budoff, MD, Ronald J. Oudiz, MD, Conrad P. Zalace, BS, Hamid Bakhsheshi, BS, Steven L. Goldberg, MD, William J. French, MD, Tapan G. Rami, and Bruce H. Brundage, MD Coronary angiography remains the diagnostic standard for establishing the presence, site, and severity of coronary artery disease (CAD). Electron beam computed tomography (EBCT), with its 3-dimensional capabilities, is an emerging technology with the potential for obtaining essentially noninvasive coronary arteriograms. The purpose of this study was to (1) test the accuracy of intravenous coronary arteriography using the EBCT to conventional coronary arteriographic images; (2) establish the inter-reader variability of this procedure; (3) determine the limitations due to location within the coronary tree; and (4) identify factors that contributed to improved image quality of the 3-dimensional EBCT angiograms. Fifty-two patients underwent both EBCT angiography and coronary angiography within 2 weeks. The coronary angiogram and the EBCT 3-dimensional images were analyzed by 2 observers blinded to the results of the other techniques. EBCT correctly identified 43 of 55 significantly stenosed arteries (sensitivity 78%), and cor-
rectly identified 118 of 130 of the nonobstructed arteries, yielding a specificity of 91% (p <0.001, chi-square analysis). The overall accuracy for EBCT angiography was 87%. Significantly more left main and anterior descending coronary arteries were adequately visualized than the circumflex and right coronary vessels (p 5 0.003). Overall, 23 of 208 (11%) major epicardial vessels were noninterpretable by the blinded EBCT readers, primarily due to motion artifacts caused by cardiac and respiratory motion and poor electrocardiographic gating. The inter-reader variability was similar to that of angiography, and its high accuracy makes this a clinically useful test. This study demonstrates, by using intravenous contrast enhancement, that EBCT can clearly depict the coronary artery anatomy and can permit identification of coronary artery stenosis. Q1999 by Excerpta Medica, Inc. (Am J Cardiol 1999;83:840 – 845)
oronary angiography is the diagnostic standard for establishing the presence, site, and severity of C coronary artery disease (CAD). Millions of coronary
spatial and temporal resolution and ability to trigger image acquisition to the electrocardiogram. The objectives of this study were to test the accuracy of intravenous coronary arteriography using the EBCT, by comparison to conventional coronary arteriographic images and establish the inter-reader variability of this procedure.
1
arteriograms, at an estimated cost of $3 to $5 billion dollars per year, are performed in the United States.2,3 This highly effective clinical tool is invasive, labor intensive, costly, and has well-established mortality (0.15%) and morbidity(1.5%),4,5 and requires at least a short hospital stay. Pathologic studies have demonstrated that the severity of coronary stenoses are underestimated during coronary angiography.6 –9 This may result from the limitations of image resolution with fluoroscopy and the 2-dimensional imaging inherent in coronary angiography. These limitations have led to the need for developing less invasive and less expensive tests.10 Electron beam computed tomography (EBCT), with its 3-dimensional (3-D) capabilities, is an emerging technology with the potential for obtaining essentially noninvasive coronary arteriograms. EBCT is well suited for the imaging of the coronary arteries with its unique combination of high From the Department of Medicine, Division of Cardiology, HarborUCLA Medical Center and Saint John’s Cardiovascular Research Center, Torrance, California. This study was supported by a grant from Imatron, Inc., San Francisco, California. Manuscript received September 3, 1998; revised manuscript received and accepted October 27, 1998. Address for reprints: Matthew J. Budoff, MD, Saint John’s Cardiovascular Research Center, 1124 West Carson Street, RB-2, Torrance, California 90502.
840
©1999 by Excerpta Medica, Inc. All rights reserved.
METHODS
Patient population: EBCT angiograms were performed in 52 patients (32 men and 20 women, mean age 53 6 9 years) and the results compared with coronary angiography. The 2 studies were performed within 2 weeks (mean 9 6 6 days) with no interval clinical event (myocardial infarction, revascularization). Patients were selected from consecutive outpatient cardiac catheterizations for the evaluation of CAD. Patients signed an informed, written consent approved by the Institutional Review Board at HarborUCLA Medical Center. Patients with abnormal baseline creatinine (.1.1 mg/dl) or iodine allergy were excluded from the study. Electron beam computed tomograpy scans: The EBCT studies were performed with an Imatron C-150XL Ultrafast computed tomographic scanner (San Francisco, California), in the high resolution volume mode using a 100-ms exposure time. A 512 3 512 matrix with the smallest possible field-of-view was employed, as previously described.11 The subjects 0002-9149/99/$–see front matter PII S0002-9149(98)01075-3
were positioned head first, scanned in the supine position without table tilt or slew. A noncontrast localization scan was performed, beginning 180 mm below the sternal notch to produce an anteroposterior view of the chest, to determine the superior and inferior margins of the heart. Electrocardiographic triggering was employed, corresponding to 80% of the RR interval, so that each image was obtained at the same point in late diastole. The slice thickness was 3 mm with table increments of 2 mm, to provide overlap for optimal resolution of the coronary tree. Nonionic contrast (iopamidol injection 76%) was administered through an 18-gauge catheter placed intravenously in an antecubital vein. The total contrast dose was calculated by a cardiologist to ensure adequate enhancement of the entire length of the coronary arteries. A simple calculation of the RR interval (measured in milliseconds) multiplied by the number of slices to be obtained, provided the total duration of scanning. This time, multiplied by 4 ml/s (the flow rate), yielded a total dose of contrast needed for that study. Circulation time: Initiation of scanning relative to contrast injection was timed based on each patient’s circulation time, determined with the use of an intravenous injection of indocyanine green, using ear densitometry to mark the appearance of indocyanine green dye (and contrast) in the systemic circulation. This was performed by mixing 2 mg of diluted indocyanine green dye (ICG, Cardio-Green, Baltimore, Maryland) with the contrast before the injection. By utilizing a D-402 densitometer (Waters Instruments, Rochester, Minnesota) attached to the earlobe, the appearance of green dye at the earlobe determined the circulation time. Commencement of scanning at the onset of green dye arrival at the earlobe assured optimal coronary contrast enhancement for the scanned images. A volume of 120 to 180 ml of nonionic contrast mixed with green dye was injected intravenously. Scanning was initiated just below the carina, and continued inferiorly to encompass the entire coronary tree. Forty contiguous cross-axial images were obtained, cephalad to caudad, using breath holding during deep inspiration. The cardiologist evaluated the axial images, and determined if additional imaging was necessary. If the entire coronary tree could not be imaged within the 40 slices, additional scans were obtained in 1 additional imaging sequence. The study was terminated when the apex of the heart was encompassed. Three-dimensional electron beam reconstruction.
EBCT images were transferred to a Pentium PC-based computer (IBM, Triangle Park, North Carolina). Editing of the 2-dimensional axial images were performed before 3-D reconstruction. Chest wall, spinal column, pulmonary vessels, and other noncardiac structures were manually edited from each image. Three-dimensional images of the coronary arteries were then reconstructed by a commercially available software package (NeoImagery Industries, City of Industry, California). A combination of 2 spatial reconstruction algorithms were used in this study, maximum intensity projections and shaded-surface render-
FIGURE 1. An EBCT reconstruction of a patient with a proximal stenosis of the left anterior descending artery (L) (arrow). A second stenoses of the artery can be seen just above the L. These stenoses were confirmed by coronary angiography. The 3-dimensional image can be rotated to multiple views to evaluate the entire coronary tree. The left anterior descending (L) and the circumflex artery (C) are both seen in this view.
FIGURE 2. Sensitivities, specificities, and accuracies of the EBCT results compared with angiography in the evaluation of significant lesions. One hundred eighty-five of 208 coronary vessels were analyzed by both methods. A vessel-by-vessel analysis is provided for the left anterior descending (LAD), left circumflex (LCx), and right coronary artery (RCA) distributions.
ing. The maximum intensity projections technique relies on identifying all of the pixels above a certain threshold. The shaded-surface rendering technique discards all pixels below a certain Hounsfield unit (HU) threshold, and the remaining pixels are shaded according to depth and lighting. This allows for visualization of the entire surface of the heart, including the inner vessel lumens of the epicardial vessels (see Figure 1). In this study, the 3-D computer program was set to use a combination of both techniques (60% maximum intensity projections and 40% shaded-surface rendering) to improve the resolution of the images, because we found that this provides the best
CORONARY ARTERY DISEASE/EBCT CORONARY ANGIOGRAPHY
841
representation of the coronary arteries. Two interpreters, who were blinded to the data, analyzed the reconstructed and axial EBCT images, by visual analysis, for stenosis location and the maximum percent stenosis in each vessel in 1 of 6 categories: none, luminal irregularities (1% to 49%), 50% to 74%, 75% to 99%, 100% occlusion, or nonanalyzable. Coronary angiography: Coronary arteriograms were obtained in the standard fashion with $2 orthogonal views of each coronary artery. The coronary angiograms were analyzed by visual assessment by 2 experienced readers unaware of the EBCT findings. Each coronary vessel (left main, left anterior descending, circumflex, and right coronary artery) was assessed, and the visual estimation of the percent luminal reduction for each lesion was reported. Multiple projections were acquired to discern the maximal coronary artery luminal narrowing. Two blinded readers recorded the maximum percent stenosis in each vessel in 1 of 5 categories: 0% (no stenosis), 1% to 49% (luminal irregularities), 50% to 74%, 75% to 99%, and 100% (complete occlusion). Comparison of electron beam computed tomography and coronary angiography: Ensuring that the stenoses
identified on the EBCT images were compared with the same region of the angiogram required a thorough narrative description of location. To improve correlation between the location of stenosis by EBCT with the angiographic equivalent, the location of each lesion was meticulously recorded on a standard diagram.12 All lesions were defined with reference to angiographic anatomic landmarks such as vessel bifurcations and side branches, readily identified from the coronary angiogram and 3-D EBCT study. Significant CAD was defined as $50% luminal diameter stenosis in that vessel for both techniques. The coronary angiogram and the EBCT 3-D images were analyzed by 2 observers blinded to the results of the other technique and disagreements resolved by consensus with a third investigator. Statistical analysis: The ability of the EBCT angiogram to correlate with angiographically confirmed coronary stenosis was evaluated by calculating sensitivity, specificity, positive and negative predictive values, and accuracy using 2 3 2 contingency tables. Statistical significance was calculated using the chisquare test. Maximum percent stenosis as determined by EBCT were compared with the same values obtained by coronary angiography by the paired t test. Results by coronary vessel distribution or groups of patients were compared using paired or unpaired t test, when appropriate. Cohen’s Kappa statistic13 was used to compare the interobserver agreement for EBCT and coronary angiography.
RESULTS The scan time for the acquisition of the EBCT images was 30 to 50 seconds, dependent upon the heart rate. With patient instruction, 51 of 52 patients (98%) were able to hold their breath throughout the scan. Total contrast per patient was 175 6 53 ml. Of the 52 patients studied, 26 were found to have mul842 THE AMERICAN JOURNAL OF CARDIOLOGYT
VOL. 83
TABLE I Comparison Between Contrast-Enhanced Electron Beam Computed Tomography and Coronary Angiography in 208 Coronary Vessels Coronary Angiography Electron Beam CT Normal 1%–49% 50%–74% 75%–99% 100% Noninterpretable
Normal 1%–49% 50%–74% 75%–99% 100% 93 4 2 2 4 8
8 13 1 1 2 3
5 1 10 4 2 4
2 2 4 10 3 5
— 2 — 3 7 3
tivessel disease, 7 with 1-vessel disease, and 19 patients with nonobstructive disease or normal coronaries by angiography. Coronary stenoses were compared in 185 of 208 (89%) vessels by both techniques. The number of coronary vessels technically adequate to evaluate by EBCT were: 52 of 52 (100%) of left main coronary arteries, 48 of 52 (92%) of left anterior descending arteries, 44 of 52 (85%) of the left circumflex arteries, and 41 of 52 (79%) of the right coronary arteries. Significantly more left main and left anterior descending coronary arteries were adequately visualized than the left circumflex and right coronary arteries (p 5 0.003). EBCT correctly identified 43 of 55 significantly stenosed arteries (luminal stenosis $50%) for a sensitivity of 78%, and correctly identified 118 of 130 of the nonobstructed arteries yielding a specificity of 91% (p ,0.001). Overall accuracy for EBCT angiography was 87%. Table I demonstrates the consensus reading between intravenous EBCT evaluation and coronary angiography (stenosis severity) in the 208 coronary vessels. The degree of the relation between EBCT results and coronary angiography was high (contingency coefficient 0.78, p ,0.001). Vessel-by-vessel analysis is displayed in Figure 2. There were no stenoses in the LMCA distribution by catheterization or EBCT angiography (specificity of 100%). Of the 28 lesions in the left anterior descending artery distribution, EBCT underestimated 5, for a sensitivity of 82%, and overestimated 2 stenoses for a specificity of 90%. There were 15 significant stenoses in the left circumflex artery distribution, and 5 were underestimated by EBCT, for a sensitivity of 67%. In the remaining 29 patients, 4 lesions were overestimated, for a specificity of 86%. Of the 41 right coronary artery vessels analyzable, 2 lesions were underestimated (deemed ,50% by EBCT) and 6 were overestimated (deemed .50% by EBCT) for sensitivity and specificity of 83% and 80% respectively. There was no significant differences between the sensitivity, specificity, or accuracy of the individual vessels (p 5 NS). Image quality: The images were sorted into 3 groups based on interpretation by the EBCT readers into good (group 1), fair (group 2) or poor (group 3) quality images. Fifteen patients in group 1 were compared with 14 patients in group 3 for peak left anterior descending contrast enhancement, peak myocardial MARCH 15, 1999
occurred more frequently in distal vessels (p 5 0.03) than the proximal half of the vessels. For coronary angiography, the Cohen’s kappa statistic was 0.91, which is not significantly different than EBCT.
DISCUSSION
FIGURE 3. CAD in a 56-year-old man with a complete occlusion of the proximal left anterior descending artery. Contrast-enhanced EBCT images reveal the stenosis using 3-dimensional reconstruction (A) and corresponding coronary angiography demonstrate the stenosis (arrow) in the left anterior oblique view (B). Three-dimensional reconstruction allows for viewing from any angle.
contrast enhancement, time to left anterior descending and myocardial peak enhancement, and total amount of contrast given. The mean peak left anterior descending enhancement in group 1 was 436 6 119 HU and in group 3 was 361 6 151 HU (p ,0.05). The mean difference between peak left anterior descending and myocardial enhancement, time to peak left anterior descending or myocardial enhancement and total contrast given showed no significant differences between groups. The only predictor of good image quality was peak left anterior descending enhancement. Reanalysis with exclusion of the poor studies raises the sensitivity to 89%, specificity to 93%, and accuracy to 91%. In most cases, image quality was good in the left main coronary artery and proximal and middle segments of the left anterior descending arteries (Figures 1 and 3). It was more difficult to discern stenoses in the left circumflex and right coronary artery distributions. Spatial misregistration, due to cardiac and respiratory motion, caused 21% of right coronary arteries and 15% of left circumflex arteries to be unanalyzable. Inter-reader variability: Table II compares the results of 2 experienced readers of the EBCT studies. Disagreements between the EBCT interpreters as to which vessels contained $50% stenosis occurred in 19% of cases. The Cohen’s kappa statistic for interreader variability was 0.86 6 0.02 for segmental classification (grouping normals with luminal stenoses), and 0.66 6 0.03 for the entire cohort for the 5 subgroups. Although some variability was noted with regard to severity of stenosis, only 13 lesions were deemed significant by 1 reader and nonsignificant (luminal irregularity or normal) by the other. Most of these cases (77%) were stenoses in the left circumflex or right coronary artery distribution (p 5 NS). Fourteen lesions were deemed nonassessable by 1 reader yet assessable by the other. Eleven of the 14 lesions (79%) were in the distal portion of the coronary vessels and 10 of 14 (71%) were in the left circumflex or right coronary artery distributions. The disagreements
This study of EBCT angiography and coronary angiography represents the second largest reported series, and is the first to characterize inter-reader variability and contrast factors that contribute to image quality of the EBCT angiogram. Several studies have reported using EBCT images to reconstruct a 3-D representation of the proximal coronary arteries using intravenous contrast enhancement.14 –17 Measurement of coronary artery diameters in EBCT and quantitative coronary angiography yielded a correlation of 0.83.18 Achenbach et al14 reported good visualization of the left anterior descending and left main coronary arteries in 27 patients undergoing coronary and EBCT angiography by this technique, with a sensitivity of 88% and a specificity of 100%. Multiple studies have reported that rapid vessel movements during EBCT imaging in the left circumflex and right coronary arteries caused diminished sensitivity for identifying stenoses.15,17 In reported studies, EBCT angiography had technically adequate images for evaluation in about 80% to 90% of coronary vessels investigated. A report of 28 patients undergoing EBCT and coronary angiography found 88% of coronary segments visualized, with an overall accuracy of 90%.19 Achenbach et al20 recently reported 125 cases, with sensitivities of 92% and specificities of 94%, with 75% of vessels evaluated. The results of our study reflect a slightly lower accuracy (87%) than the results of Achenbach et al 20 (accuracy of 93%), and is more consistent with those of Schmermund et al19 (90%). This is most likely due to the higher percentage of analyzed vessels (87% in this study compared with 75% in Achenbach et al’s study20). Exclusion of studies deemed poor by the EBCT reader increases sensitivity to 89% and specificity to 93%. This reflects the contribution of motion artifacts, poor technique, and poor contrast enhancement. Calcification at the site of stenosis, an important limitation in 1 study,19 did not contribute significantly to the interpretation in this study. Underestimation and overestimation of the stenosis severity (compared with angiography) occurred with equal frequency in this study. The largest series of EBCT angiography could only evaluate 70% of right coronary and left circumflex arteries.20 In our study, it was more difficult to discern stenoses in the left circumflex and right coronary arteries than the left main coronary and left anterior descending arteries (p 5 0.003), because the right coronary artery and left circumflex artery lie in close proximity to the atria, which move during the diastolic imaging time, making interpretation of stenoses more difficult. Faster scan times and imaging earlier in the cardiac cycle (where atrial motion is less) might lead to improved resolution of these vessels. In this study, scans with good image quality had
CORONARY ARTERY DISEASE/EBCT CORONARY ANGIOGRAPHY
843
been identified. Overall, 23 of 208 major epicardial vessels were noninterpretable by the blinded EBCT readers (11 right coronary arteries, 8 left circumflex, 4 left anterior deNoninterpretable scending, and 0 left main coronary — arteries), primarily due to motion ar2 tifacts caused by cardiac and respira— 3 tory motion and poor electrocardio— graphic gating. Electrocardiographic 15 triggering was performed at each level at 80% of the RR interval. Scanning at this diastolic time has been shown to minimize cardiac motion by decreasing ventricular contraction. However, the heart is still moving anteriorly and the atria are contracting, causing motion artifacts in right coronary and left circumflex arteries.29 It is proposed that optimal timing of image acquisition within the cardiac cycle could reduce motion artifacts. The value of 80% of the RR interval used in this and other studies might not be optimal for imaging of the coronary segments near the right or left atria, because atrial contraction during end-diastole causes rapid movement of the base of the heart.16 Interference caused by the contrast-enhanced cardiac and venous structures, including atrial appendages and coronary veins, reduced recognition of the left circumflex and right coronary arteries. Also, imaging of the caudal portion of the heart occurred near the end of the contrast infusion; thus, these segments were less distinguishable because of increasing myocardial contrast enhancement. Modifying the duration and/or rate of contrast administration may alleviate this limitation in future studies. Limited spatial resolution becomes more important for smaller coronary vessels for all imaging modalities, and has been demonstrated to be a limiting factor for evaluating smaller coronary vessels.19 The development of an accurate, cost effective, and noninvasive coronary arteriogram could revolutionize the diagnosis of CAD, by improving diagnostic accuracy, and reducing cost and procedural complications. The results of this prospective study of EBCT angiography are promising. Recent improvements, including more reliable electrocardiographic gating as well as greater experience with this technique, should improve accuracy.
TABLE II Analysis of Inter-Reader Variability for the Intravenous Electron Beam Computed Tomographic Images Observer 1 Observer 2 Normal 1%–49% 50%–74% 75%–99% 100% Noninterpretable
Normal 1%–49% 50%–74% 75%–99% 100% 98 8 2 2 1 2
3 11 1 — — 2
— 4 12 — — —
1 1 2 16 3 3
1 — — 4 10 1
higher sensitivities and specificities than those that were fair or poor by the EBCT investigators. We attempted to discern the factors that contributed to good image quality and better 2- and 3-D representations of the coronary tree. The only predictor of good image quality was peak left anterior descending enhancement. Larger contrast doses did not improve image resolution or accuracy of the EBCT angiograms. In our study, the inter-reader variability of coronary angiography was 13%, whereas other studies ranged from 18.65%21 to 36%.22 Zir et al22 reported interobserver variability up to 65%, with lowest agreement in the right coronary and left circumflex arteries. In our study, the probability of disagreement with respect to a vessel by EBCT with $50% stenosis was 19%. Similar to angiographic studies,23 nonagreement between the investigators was equally divided between over- and under-reading, with the highest disagreement among EBCT angiograms in the distal vessels and poor studies. Distal vessels had significantly higher disagreements than proximal vessels, as did vessels of angiograms deemed poor quality by the readers, a finding reported in angiographic trials.24,25 We did not have enough power to evaluate differences in the inter-reader variability among different coronary vessels. This study suggests that lesions in the proximal portions of the coronary trunks can be read with acceptable accuracy and interobserver variability, whereas lesions in the more distal coronary circulations need cautious interpretation. In this study, EBCT angiography had similar inter-reader variability (0.86) to coronary angiography (0.91, no significant difference from EBCT). This study demonstrates that EBCT, with intravenous contrast enhancement, can clearly depict the coronary anatomy and can permit identification of coronary artery stenosis. To date, the clinical application of EBCT in the investigation of CAD has primarily been a screening tool for coronary calcifications.26,27 The application of EBCT could thus be extended from screening for coronary calcium, a measurement of plaque burden,28 to the actual visualization of coronary stenoses using EBCT angiography. Study limitations: Studies of these initial 52 patients shed insight into the current limitations and difficulties of developing this new technology. The results are promising, with a 78% sensitivity, 91% specificity, and 87% accuracy for identifying significant CAD. Despite these results, a number of limitations have 844 THE AMERICAN JOURNAL OF CARDIOLOGYT
VOL. 83
1. Marcus ML, Skorton DJ, Johnson MR, Collins SM, Harrison DG, Kerber RE. Visual estimates of percent diameter coronary stenosis: “a battered gold standard.” J Am Coll Cardiol 1988;11:882– 885. 2. Levin DC. Invasive evaluation (coronary arteriography) of the coronay artery disease patient: clinical, economic, and social issues. Circulation 1982;66(suppl. III):71–79. 3. American Heart Association. Heart and Stroke Facts: 1995 statistical supplement. Dallas, TX: American Heart Association, 1995. 4. Adams DF, Fraser DB, Abrams HL. The complications of coronary arteriography. Circulation 1973;48:609 – 618. 5. Kennedy JW. Complications associated with cardiac catheterization and angiography. Cath Cardiovasc Diagn 1982;8:5–11. 6. Marcus ML, Armstrong MD, Heistad DD, Eastham CL, MarkAL. Comparison of three methods of evaluating coronary obstructive lesions: postmortem arteriography, pathologic examination and measurement of regional myocardial perfusion during maximal vasodilation. Am J Cardiol 1982;49:1688 –1706. 7. Grondin CM. Dyrda I. Pasternac A, Campeau L, Bourassa MG, Lesperance J. Discrepancies between cineangiographic and postmortem findings in patients
MARCH 15, 1999
with coronary artery disease and recent myocardial revascularization. Circulation 1974;49:703-708. 8. Thomas AC, Daview MJ, Dilly S, Dilly N, Franc F. Potential errors in estimation of coronary arterial stenoses from clinical coronary arteriography with reference to the shape of the coronary arterial lumen. Br Heart J 1986;55:129 – 139. 9. Mintz GS, Painter JA, Pichard AD, Kent KM, Satler LF, Popma JJ, Chuang YC, Bucher TA, Sokoloqicz LE, Leon MB. Atherosclerosis in angiographically “normal” coronary artery reference segments: an intravascular ultrasound study with clinical correlations. J Am Coll Cardiol 1995;25:1479 –1485. 10. Gersh BJ. Noninvasive imaging in acute coronary disease: a clinical perspective. Circulation 1991;84:1140 –1147. 11. Moshage WEL, Achenbach S, Seese B, Bachmann K, Kirchgeorg M. Coronary artery stenoses: three-dimensional imaging with electrocardiographically triggered, contrast agent-enhanced, electrom beam CT. Radiology 1995;196:707– 714. 12. Principal Investigators of CASS. National heart, lung, and blood institute coronary artery surgery study (CASS). Circulation 1981;63:1– 40. 13. Cohen J. A coefficient of agreement for nominal scales. Educ Pysch Meas 1960;20:37– 43. 14. Achenbach S, Moshage W, Ropers D, Nossen J, Daniel WG. Value of electron beam computed tomography for the noninvasive detection of high-grade coronary-artery stenoses and occlusions (see comments). N Engl J Med 1998; 339:1964 –1971. 15. Thomas PJ, McCollough CH, Ritman EL. An electron-beam CT approach for transvenous coronary angiography. J Comput Assist Tomogr 1995;19:383–389. 16. Achenbach S, Moshage W, Bachmann K. Noninvasive coronary angiography by contrast-enhanced electron beam computed tomography. Clin Cardiol 1998; 21:323–330. 17. Chernoff DM, Ritchie CJ, Higgins CB. Evaluation of electron beam CT coronary angiography in healthy subjects. Am J Roentgenol 1997;169:93–99. 18. Achenbach S, Moshage W, Ropers D, Bachmann K. Comparison of vessel diameters in electron beam tomography and quantitative coronary angiography. Int J Cardiac Imaging 1998;14:1–7. 19. Schmermund A, Rensing BJ, Sheedy PF, Bell MR, Rumberger JA. Intrave-
nous electron-beam computed tomographic coronary angiography for segmented analysis of coronary artery stenoses. J Am Coll Cardiol 1998;31:1547–1554. 20. Achenbach S, Moshage W, Ropers D, Nossen J, Bachmann K. Comparison of contrast-enhanced electron beam CT and coronary angiography in 125 patients. J Am Coll Cardiol 1998;31:128A. 21. Fisher LD, Judkins MP, Lesperance J, Cameron A, Swaye P, Ryan T, Maynard C, Bourassa M, Kennedy JW, Gosselin A, Kemp H, Faxon D, Wexler L, Davis KB. Reproducibility of coronary arteriographic reading in the Coronary Artery Surgery Study (CASS). Cathet Cardiovasc Diagn 1982; 8:565–571. 22. Zir LM, Miller SW, Dinsmore RE, Gilbert JP, Harthorne JW. Inter-observer variability in coronary angiography. Circulation 1976;53:627– 632. 23. Kemp HG. Interobserver variability in interpretation of coronary arteriograms. Clev Clin Quart 1980;47:137–139. 24. Detre KM, Wright E, Murphy ML, Takaro T. Observer agreement in evaluating coronary angiograms. Circulation 1975;52:979 –986. 25. DeRouen TA, Murray JA, deSoyza N. Coronary angiogram interpretation. Interobserver variability. Circulation 1977;55:324 –329. 26. Wexler L, Brundage BH, Crouse J, Detrano R, Fuster V, Maddahi J, Rumberger J, Stanford W, White R, Taubert K. Coronary artery calcification: pathophysiology, epidemiology, imaging methods and clinical implications. Circulation 1996;94:1175–1192. 27. Budoff MJ, Georgiou D, Brody A, Agatston AS, Kennedy JM, Wolfkiel C, Stanford W, Shields P, Lewis RJ, Janowitz WR, Rich S, Brundage BH. Ultrafast computed tomography as a diagnostic modality in the detection of coronary artery disease: a multicenter study. Circulation 1996;93:898 –904. 28. Rumberger JA, Simons DB, Fitzpatrick LA, Sheedy PF, Schwartz RS. Coronary artery calcium areas by electron beam computed tomography and coronary atherosclerotic plaque area: a histopathologic correlative study. Circulation 1995;92:2157–2162. 29. Kaufmann RB, Sheedy PF, Breen JF, Kelzenberg JR, Kruger BL, Schwartz RS, Moll PP. Detection of heart calcification with electron beam CT: interobserver and intraobserver reliability for scoring quantification. Radiology 1994;190:347–352.
CORONARY ARTERY DISEASE/EBCT CORONARY ANGIOGRAPHY
845
rectly identified 118 of 130 of the nonobstructed arteries, yielding a specificity of 91% (p <0.001, chi-square analysis). The overall accuracy for EBCT angiography was 87%. Significantly more left main and anterior descending coronary arteries were adequately visualized than the circumflex and right coronary vessels (p 5 0.003). Overall, 23 of 208 (11%) major epicardial vessels were noninterpretable by the blinded EBCT readers, primarily due to motion artifacts caused by cardiac and respiratory motion and poor electrocardiographic gating. The inter-reader variability was similar to that of angiography, and its high accuracy makes this a clinically useful test. This study demonstrates, by using intravenous contrast enhancement, that EBCT can clearly depict the coronary artery anatomy and can permit identification of coronary artery stenosis. Q1999 by Excerpta Medica, Inc. (Am J Cardiol 1999;83:840 – 845)
oronary angiography is the diagnostic standard for establishing the presence, site, and severity of C coronary artery disease (CAD). Millions of coronary
spatial and temporal resolution and ability to trigger image acquisition to the electrocardiogram. The objectives of this study were to test the accuracy of intravenous coronary arteriography using the EBCT, by comparison to conventional coronary arteriographic images and establish the inter-reader variability of this procedure.
1
arteriograms, at an estimated cost of $3 to $5 billion dollars per year, are performed in the United States.2,3 This highly effective clinical tool is invasive, labor intensive, costly, and has well-established mortality (0.15%) and morbidity(1.5%),4,5 and requires at least a short hospital stay. Pathologic studies have demonstrated that the severity of coronary stenoses are underestimated during coronary angiography.6 –9 This may result from the limitations of image resolution with fluoroscopy and the 2-dimensional imaging inherent in coronary angiography. These limitations have led to the need for developing less invasive and less expensive tests.10 Electron beam computed tomography (EBCT), with its 3-dimensional (3-D) capabilities, is an emerging technology with the potential for obtaining essentially noninvasive coronary arteriograms. EBCT is well suited for the imaging of the coronary arteries with its unique combination of high From the Department of Medicine, Division of Cardiology, HarborUCLA Medical Center and Saint John’s Cardiovascular Research Center, Torrance, California. This study was supported by a grant from Imatron, Inc., San Francisco, California. Manuscript received September 3, 1998; revised manuscript received and accepted October 27, 1998. Address for reprints: Matthew J. Budoff, MD, Saint John’s Cardiovascular Research Center, 1124 West Carson Street, RB-2, Torrance, California 90502.
840
©1999 by Excerpta Medica, Inc. All rights reserved.
METHODS
Patient population: EBCT angiograms were performed in 52 patients (32 men and 20 women, mean age 53 6 9 years) and the results compared with coronary angiography. The 2 studies were performed within 2 weeks (mean 9 6 6 days) with no interval clinical event (myocardial infarction, revascularization). Patients were selected from consecutive outpatient cardiac catheterizations for the evaluation of CAD. Patients signed an informed, written consent approved by the Institutional Review Board at HarborUCLA Medical Center. Patients with abnormal baseline creatinine (.1.1 mg/dl) or iodine allergy were excluded from the study. Electron beam computed tomograpy scans: The EBCT studies were performed with an Imatron C-150XL Ultrafast computed tomographic scanner (San Francisco, California), in the high resolution volume mode using a 100-ms exposure time. A 512 3 512 matrix with the smallest possible field-of-view was employed, as previously described.11 The subjects 0002-9149/99/$–see front matter PII S0002-9149(98)01075-3
were positioned head first, scanned in the supine position without table tilt or slew. A noncontrast localization scan was performed, beginning 180 mm below the sternal notch to produce an anteroposterior view of the chest, to determine the superior and inferior margins of the heart. Electrocardiographic triggering was employed, corresponding to 80% of the RR interval, so that each image was obtained at the same point in late diastole. The slice thickness was 3 mm with table increments of 2 mm, to provide overlap for optimal resolution of the coronary tree. Nonionic contrast (iopamidol injection 76%) was administered through an 18-gauge catheter placed intravenously in an antecubital vein. The total contrast dose was calculated by a cardiologist to ensure adequate enhancement of the entire length of the coronary arteries. A simple calculation of the RR interval (measured in milliseconds) multiplied by the number of slices to be obtained, provided the total duration of scanning. This time, multiplied by 4 ml/s (the flow rate), yielded a total dose of contrast needed for that study. Circulation time: Initiation of scanning relative to contrast injection was timed based on each patient’s circulation time, determined with the use of an intravenous injection of indocyanine green, using ear densitometry to mark the appearance of indocyanine green dye (and contrast) in the systemic circulation. This was performed by mixing 2 mg of diluted indocyanine green dye (ICG, Cardio-Green, Baltimore, Maryland) with the contrast before the injection. By utilizing a D-402 densitometer (Waters Instruments, Rochester, Minnesota) attached to the earlobe, the appearance of green dye at the earlobe determined the circulation time. Commencement of scanning at the onset of green dye arrival at the earlobe assured optimal coronary contrast enhancement for the scanned images. A volume of 120 to 180 ml of nonionic contrast mixed with green dye was injected intravenously. Scanning was initiated just below the carina, and continued inferiorly to encompass the entire coronary tree. Forty contiguous cross-axial images were obtained, cephalad to caudad, using breath holding during deep inspiration. The cardiologist evaluated the axial images, and determined if additional imaging was necessary. If the entire coronary tree could not be imaged within the 40 slices, additional scans were obtained in 1 additional imaging sequence. The study was terminated when the apex of the heart was encompassed. Three-dimensional electron beam reconstruction.
EBCT images were transferred to a Pentium PC-based computer (IBM, Triangle Park, North Carolina). Editing of the 2-dimensional axial images were performed before 3-D reconstruction. Chest wall, spinal column, pulmonary vessels, and other noncardiac structures were manually edited from each image. Three-dimensional images of the coronary arteries were then reconstructed by a commercially available software package (NeoImagery Industries, City of Industry, California). A combination of 2 spatial reconstruction algorithms were used in this study, maximum intensity projections and shaded-surface render-
FIGURE 1. An EBCT reconstruction of a patient with a proximal stenosis of the left anterior descending artery (L) (arrow). A second stenoses of the artery can be seen just above the L. These stenoses were confirmed by coronary angiography. The 3-dimensional image can be rotated to multiple views to evaluate the entire coronary tree. The left anterior descending (L) and the circumflex artery (C) are both seen in this view.
FIGURE 2. Sensitivities, specificities, and accuracies of the EBCT results compared with angiography in the evaluation of significant lesions. One hundred eighty-five of 208 coronary vessels were analyzed by both methods. A vessel-by-vessel analysis is provided for the left anterior descending (LAD), left circumflex (LCx), and right coronary artery (RCA) distributions.
ing. The maximum intensity projections technique relies on identifying all of the pixels above a certain threshold. The shaded-surface rendering technique discards all pixels below a certain Hounsfield unit (HU) threshold, and the remaining pixels are shaded according to depth and lighting. This allows for visualization of the entire surface of the heart, including the inner vessel lumens of the epicardial vessels (see Figure 1). In this study, the 3-D computer program was set to use a combination of both techniques (60% maximum intensity projections and 40% shaded-surface rendering) to improve the resolution of the images, because we found that this provides the best
CORONARY ARTERY DISEASE/EBCT CORONARY ANGIOGRAPHY
841
representation of the coronary arteries. Two interpreters, who were blinded to the data, analyzed the reconstructed and axial EBCT images, by visual analysis, for stenosis location and the maximum percent stenosis in each vessel in 1 of 6 categories: none, luminal irregularities (1% to 49%), 50% to 74%, 75% to 99%, 100% occlusion, or nonanalyzable. Coronary angiography: Coronary arteriograms were obtained in the standard fashion with $2 orthogonal views of each coronary artery. The coronary angiograms were analyzed by visual assessment by 2 experienced readers unaware of the EBCT findings. Each coronary vessel (left main, left anterior descending, circumflex, and right coronary artery) was assessed, and the visual estimation of the percent luminal reduction for each lesion was reported. Multiple projections were acquired to discern the maximal coronary artery luminal narrowing. Two blinded readers recorded the maximum percent stenosis in each vessel in 1 of 5 categories: 0% (no stenosis), 1% to 49% (luminal irregularities), 50% to 74%, 75% to 99%, and 100% (complete occlusion). Comparison of electron beam computed tomography and coronary angiography: Ensuring that the stenoses
identified on the EBCT images were compared with the same region of the angiogram required a thorough narrative description of location. To improve correlation between the location of stenosis by EBCT with the angiographic equivalent, the location of each lesion was meticulously recorded on a standard diagram.12 All lesions were defined with reference to angiographic anatomic landmarks such as vessel bifurcations and side branches, readily identified from the coronary angiogram and 3-D EBCT study. Significant CAD was defined as $50% luminal diameter stenosis in that vessel for both techniques. The coronary angiogram and the EBCT 3-D images were analyzed by 2 observers blinded to the results of the other technique and disagreements resolved by consensus with a third investigator. Statistical analysis: The ability of the EBCT angiogram to correlate with angiographically confirmed coronary stenosis was evaluated by calculating sensitivity, specificity, positive and negative predictive values, and accuracy using 2 3 2 contingency tables. Statistical significance was calculated using the chisquare test. Maximum percent stenosis as determined by EBCT were compared with the same values obtained by coronary angiography by the paired t test. Results by coronary vessel distribution or groups of patients were compared using paired or unpaired t test, when appropriate. Cohen’s Kappa statistic13 was used to compare the interobserver agreement for EBCT and coronary angiography.
RESULTS The scan time for the acquisition of the EBCT images was 30 to 50 seconds, dependent upon the heart rate. With patient instruction, 51 of 52 patients (98%) were able to hold their breath throughout the scan. Total contrast per patient was 175 6 53 ml. Of the 52 patients studied, 26 were found to have mul842 THE AMERICAN JOURNAL OF CARDIOLOGYT
VOL. 83
TABLE I Comparison Between Contrast-Enhanced Electron Beam Computed Tomography and Coronary Angiography in 208 Coronary Vessels Coronary Angiography Electron Beam CT Normal 1%–49% 50%–74% 75%–99% 100% Noninterpretable
Normal 1%–49% 50%–74% 75%–99% 100% 93 4 2 2 4 8
8 13 1 1 2 3
5 1 10 4 2 4
2 2 4 10 3 5
— 2 — 3 7 3
tivessel disease, 7 with 1-vessel disease, and 19 patients with nonobstructive disease or normal coronaries by angiography. Coronary stenoses were compared in 185 of 208 (89%) vessels by both techniques. The number of coronary vessels technically adequate to evaluate by EBCT were: 52 of 52 (100%) of left main coronary arteries, 48 of 52 (92%) of left anterior descending arteries, 44 of 52 (85%) of the left circumflex arteries, and 41 of 52 (79%) of the right coronary arteries. Significantly more left main and left anterior descending coronary arteries were adequately visualized than the left circumflex and right coronary arteries (p 5 0.003). EBCT correctly identified 43 of 55 significantly stenosed arteries (luminal stenosis $50%) for a sensitivity of 78%, and correctly identified 118 of 130 of the nonobstructed arteries yielding a specificity of 91% (p ,0.001). Overall accuracy for EBCT angiography was 87%. Table I demonstrates the consensus reading between intravenous EBCT evaluation and coronary angiography (stenosis severity) in the 208 coronary vessels. The degree of the relation between EBCT results and coronary angiography was high (contingency coefficient 0.78, p ,0.001). Vessel-by-vessel analysis is displayed in Figure 2. There were no stenoses in the LMCA distribution by catheterization or EBCT angiography (specificity of 100%). Of the 28 lesions in the left anterior descending artery distribution, EBCT underestimated 5, for a sensitivity of 82%, and overestimated 2 stenoses for a specificity of 90%. There were 15 significant stenoses in the left circumflex artery distribution, and 5 were underestimated by EBCT, for a sensitivity of 67%. In the remaining 29 patients, 4 lesions were overestimated, for a specificity of 86%. Of the 41 right coronary artery vessels analyzable, 2 lesions were underestimated (deemed ,50% by EBCT) and 6 were overestimated (deemed .50% by EBCT) for sensitivity and specificity of 83% and 80% respectively. There was no significant differences between the sensitivity, specificity, or accuracy of the individual vessels (p 5 NS). Image quality: The images were sorted into 3 groups based on interpretation by the EBCT readers into good (group 1), fair (group 2) or poor (group 3) quality images. Fifteen patients in group 1 were compared with 14 patients in group 3 for peak left anterior descending contrast enhancement, peak myocardial MARCH 15, 1999
occurred more frequently in distal vessels (p 5 0.03) than the proximal half of the vessels. For coronary angiography, the Cohen’s kappa statistic was 0.91, which is not significantly different than EBCT.
DISCUSSION
FIGURE 3. CAD in a 56-year-old man with a complete occlusion of the proximal left anterior descending artery. Contrast-enhanced EBCT images reveal the stenosis using 3-dimensional reconstruction (A) and corresponding coronary angiography demonstrate the stenosis (arrow) in the left anterior oblique view (B). Three-dimensional reconstruction allows for viewing from any angle.
contrast enhancement, time to left anterior descending and myocardial peak enhancement, and total amount of contrast given. The mean peak left anterior descending enhancement in group 1 was 436 6 119 HU and in group 3 was 361 6 151 HU (p ,0.05). The mean difference between peak left anterior descending and myocardial enhancement, time to peak left anterior descending or myocardial enhancement and total contrast given showed no significant differences between groups. The only predictor of good image quality was peak left anterior descending enhancement. Reanalysis with exclusion of the poor studies raises the sensitivity to 89%, specificity to 93%, and accuracy to 91%. In most cases, image quality was good in the left main coronary artery and proximal and middle segments of the left anterior descending arteries (Figures 1 and 3). It was more difficult to discern stenoses in the left circumflex and right coronary artery distributions. Spatial misregistration, due to cardiac and respiratory motion, caused 21% of right coronary arteries and 15% of left circumflex arteries to be unanalyzable. Inter-reader variability: Table II compares the results of 2 experienced readers of the EBCT studies. Disagreements between the EBCT interpreters as to which vessels contained $50% stenosis occurred in 19% of cases. The Cohen’s kappa statistic for interreader variability was 0.86 6 0.02 for segmental classification (grouping normals with luminal stenoses), and 0.66 6 0.03 for the entire cohort for the 5 subgroups. Although some variability was noted with regard to severity of stenosis, only 13 lesions were deemed significant by 1 reader and nonsignificant (luminal irregularity or normal) by the other. Most of these cases (77%) were stenoses in the left circumflex or right coronary artery distribution (p 5 NS). Fourteen lesions were deemed nonassessable by 1 reader yet assessable by the other. Eleven of the 14 lesions (79%) were in the distal portion of the coronary vessels and 10 of 14 (71%) were in the left circumflex or right coronary artery distributions. The disagreements
This study of EBCT angiography and coronary angiography represents the second largest reported series, and is the first to characterize inter-reader variability and contrast factors that contribute to image quality of the EBCT angiogram. Several studies have reported using EBCT images to reconstruct a 3-D representation of the proximal coronary arteries using intravenous contrast enhancement.14 –17 Measurement of coronary artery diameters in EBCT and quantitative coronary angiography yielded a correlation of 0.83.18 Achenbach et al14 reported good visualization of the left anterior descending and left main coronary arteries in 27 patients undergoing coronary and EBCT angiography by this technique, with a sensitivity of 88% and a specificity of 100%. Multiple studies have reported that rapid vessel movements during EBCT imaging in the left circumflex and right coronary arteries caused diminished sensitivity for identifying stenoses.15,17 In reported studies, EBCT angiography had technically adequate images for evaluation in about 80% to 90% of coronary vessels investigated. A report of 28 patients undergoing EBCT and coronary angiography found 88% of coronary segments visualized, with an overall accuracy of 90%.19 Achenbach et al20 recently reported 125 cases, with sensitivities of 92% and specificities of 94%, with 75% of vessels evaluated. The results of our study reflect a slightly lower accuracy (87%) than the results of Achenbach et al 20 (accuracy of 93%), and is more consistent with those of Schmermund et al19 (90%). This is most likely due to the higher percentage of analyzed vessels (87% in this study compared with 75% in Achenbach et al’s study20). Exclusion of studies deemed poor by the EBCT reader increases sensitivity to 89% and specificity to 93%. This reflects the contribution of motion artifacts, poor technique, and poor contrast enhancement. Calcification at the site of stenosis, an important limitation in 1 study,19 did not contribute significantly to the interpretation in this study. Underestimation and overestimation of the stenosis severity (compared with angiography) occurred with equal frequency in this study. The largest series of EBCT angiography could only evaluate 70% of right coronary and left circumflex arteries.20 In our study, it was more difficult to discern stenoses in the left circumflex and right coronary arteries than the left main coronary and left anterior descending arteries (p 5 0.003), because the right coronary artery and left circumflex artery lie in close proximity to the atria, which move during the diastolic imaging time, making interpretation of stenoses more difficult. Faster scan times and imaging earlier in the cardiac cycle (where atrial motion is less) might lead to improved resolution of these vessels. In this study, scans with good image quality had
CORONARY ARTERY DISEASE/EBCT CORONARY ANGIOGRAPHY
843
been identified. Overall, 23 of 208 major epicardial vessels were noninterpretable by the blinded EBCT readers (11 right coronary arteries, 8 left circumflex, 4 left anterior deNoninterpretable scending, and 0 left main coronary — arteries), primarily due to motion ar2 tifacts caused by cardiac and respira— 3 tory motion and poor electrocardio— graphic gating. Electrocardiographic 15 triggering was performed at each level at 80% of the RR interval. Scanning at this diastolic time has been shown to minimize cardiac motion by decreasing ventricular contraction. However, the heart is still moving anteriorly and the atria are contracting, causing motion artifacts in right coronary and left circumflex arteries.29 It is proposed that optimal timing of image acquisition within the cardiac cycle could reduce motion artifacts. The value of 80% of the RR interval used in this and other studies might not be optimal for imaging of the coronary segments near the right or left atria, because atrial contraction during end-diastole causes rapid movement of the base of the heart.16 Interference caused by the contrast-enhanced cardiac and venous structures, including atrial appendages and coronary veins, reduced recognition of the left circumflex and right coronary arteries. Also, imaging of the caudal portion of the heart occurred near the end of the contrast infusion; thus, these segments were less distinguishable because of increasing myocardial contrast enhancement. Modifying the duration and/or rate of contrast administration may alleviate this limitation in future studies. Limited spatial resolution becomes more important for smaller coronary vessels for all imaging modalities, and has been demonstrated to be a limiting factor for evaluating smaller coronary vessels.19 The development of an accurate, cost effective, and noninvasive coronary arteriogram could revolutionize the diagnosis of CAD, by improving diagnostic accuracy, and reducing cost and procedural complications. The results of this prospective study of EBCT angiography are promising. Recent improvements, including more reliable electrocardiographic gating as well as greater experience with this technique, should improve accuracy.
TABLE II Analysis of Inter-Reader Variability for the Intravenous Electron Beam Computed Tomographic Images Observer 1 Observer 2 Normal 1%–49% 50%–74% 75%–99% 100% Noninterpretable
Normal 1%–49% 50%–74% 75%–99% 100% 98 8 2 2 1 2
3 11 1 — — 2
— 4 12 — — —
1 1 2 16 3 3
1 — — 4 10 1
higher sensitivities and specificities than those that were fair or poor by the EBCT investigators. We attempted to discern the factors that contributed to good image quality and better 2- and 3-D representations of the coronary tree. The only predictor of good image quality was peak left anterior descending enhancement. Larger contrast doses did not improve image resolution or accuracy of the EBCT angiograms. In our study, the inter-reader variability of coronary angiography was 13%, whereas other studies ranged from 18.65%21 to 36%.22 Zir et al22 reported interobserver variability up to 65%, with lowest agreement in the right coronary and left circumflex arteries. In our study, the probability of disagreement with respect to a vessel by EBCT with $50% stenosis was 19%. Similar to angiographic studies,23 nonagreement between the investigators was equally divided between over- and under-reading, with the highest disagreement among EBCT angiograms in the distal vessels and poor studies. Distal vessels had significantly higher disagreements than proximal vessels, as did vessels of angiograms deemed poor quality by the readers, a finding reported in angiographic trials.24,25 We did not have enough power to evaluate differences in the inter-reader variability among different coronary vessels. This study suggests that lesions in the proximal portions of the coronary trunks can be read with acceptable accuracy and interobserver variability, whereas lesions in the more distal coronary circulations need cautious interpretation. In this study, EBCT angiography had similar inter-reader variability (0.86) to coronary angiography (0.91, no significant difference from EBCT). This study demonstrates that EBCT, with intravenous contrast enhancement, can clearly depict the coronary anatomy and can permit identification of coronary artery stenosis. To date, the clinical application of EBCT in the investigation of CAD has primarily been a screening tool for coronary calcifications.26,27 The application of EBCT could thus be extended from screening for coronary calcium, a measurement of plaque burden,28 to the actual visualization of coronary stenoses using EBCT angiography. Study limitations: Studies of these initial 52 patients shed insight into the current limitations and difficulties of developing this new technology. The results are promising, with a 78% sensitivity, 91% specificity, and 87% accuracy for identifying significant CAD. Despite these results, a number of limitations have 844 THE AMERICAN JOURNAL OF CARDIOLOGYT
VOL. 83
1. Marcus ML, Skorton DJ, Johnson MR, Collins SM, Harrison DG, Kerber RE. Visual estimates of percent diameter coronary stenosis: “a battered gold standard.” J Am Coll Cardiol 1988;11:882– 885. 2. Levin DC. Invasive evaluation (coronary arteriography) of the coronay artery disease patient: clinical, economic, and social issues. Circulation 1982;66(suppl. III):71–79. 3. American Heart Association. Heart and Stroke Facts: 1995 statistical supplement. Dallas, TX: American Heart Association, 1995. 4. Adams DF, Fraser DB, Abrams HL. The complications of coronary arteriography. Circulation 1973;48:609 – 618. 5. Kennedy JW. Complications associated with cardiac catheterization and angiography. Cath Cardiovasc Diagn 1982;8:5–11. 6. Marcus ML, Armstrong MD, Heistad DD, Eastham CL, MarkAL. Comparison of three methods of evaluating coronary obstructive lesions: postmortem arteriography, pathologic examination and measurement of regional myocardial perfusion during maximal vasodilation. Am J Cardiol 1982;49:1688 –1706. 7. Grondin CM. Dyrda I. Pasternac A, Campeau L, Bourassa MG, Lesperance J. Discrepancies between cineangiographic and postmortem findings in patients
MARCH 15, 1999
with coronary artery disease and recent myocardial revascularization. Circulation 1974;49:703-708. 8. Thomas AC, Daview MJ, Dilly S, Dilly N, Franc F. Potential errors in estimation of coronary arterial stenoses from clinical coronary arteriography with reference to the shape of the coronary arterial lumen. Br Heart J 1986;55:129 – 139. 9. Mintz GS, Painter JA, Pichard AD, Kent KM, Satler LF, Popma JJ, Chuang YC, Bucher TA, Sokoloqicz LE, Leon MB. Atherosclerosis in angiographically “normal” coronary artery reference segments: an intravascular ultrasound study with clinical correlations. J Am Coll Cardiol 1995;25:1479 –1485. 10. Gersh BJ. Noninvasive imaging in acute coronary disease: a clinical perspective. Circulation 1991;84:1140 –1147. 11. Moshage WEL, Achenbach S, Seese B, Bachmann K, Kirchgeorg M. Coronary artery stenoses: three-dimensional imaging with electrocardiographically triggered, contrast agent-enhanced, electrom beam CT. Radiology 1995;196:707– 714. 12. Principal Investigators of CASS. National heart, lung, and blood institute coronary artery surgery study (CASS). Circulation 1981;63:1– 40. 13. Cohen J. A coefficient of agreement for nominal scales. Educ Pysch Meas 1960;20:37– 43. 14. Achenbach S, Moshage W, Ropers D, Nossen J, Daniel WG. Value of electron beam computed tomography for the noninvasive detection of high-grade coronary-artery stenoses and occlusions (see comments). N Engl J Med 1998; 339:1964 –1971. 15. Thomas PJ, McCollough CH, Ritman EL. An electron-beam CT approach for transvenous coronary angiography. J Comput Assist Tomogr 1995;19:383–389. 16. Achenbach S, Moshage W, Bachmann K. Noninvasive coronary angiography by contrast-enhanced electron beam computed tomography. Clin Cardiol 1998; 21:323–330. 17. Chernoff DM, Ritchie CJ, Higgins CB. Evaluation of electron beam CT coronary angiography in healthy subjects. Am J Roentgenol 1997;169:93–99. 18. Achenbach S, Moshage W, Ropers D, Bachmann K. Comparison of vessel diameters in electron beam tomography and quantitative coronary angiography. Int J Cardiac Imaging 1998;14:1–7. 19. Schmermund A, Rensing BJ, Sheedy PF, Bell MR, Rumberger JA. Intrave-
nous electron-beam computed tomographic coronary angiography for segmented analysis of coronary artery stenoses. J Am Coll Cardiol 1998;31:1547–1554. 20. Achenbach S, Moshage W, Ropers D, Nossen J, Bachmann K. Comparison of contrast-enhanced electron beam CT and coronary angiography in 125 patients. J Am Coll Cardiol 1998;31:128A. 21. Fisher LD, Judkins MP, Lesperance J, Cameron A, Swaye P, Ryan T, Maynard C, Bourassa M, Kennedy JW, Gosselin A, Kemp H, Faxon D, Wexler L, Davis KB. Reproducibility of coronary arteriographic reading in the Coronary Artery Surgery Study (CASS). Cathet Cardiovasc Diagn 1982; 8:565–571. 22. Zir LM, Miller SW, Dinsmore RE, Gilbert JP, Harthorne JW. Inter-observer variability in coronary angiography. Circulation 1976;53:627– 632. 23. Kemp HG. Interobserver variability in interpretation of coronary arteriograms. Clev Clin Quart 1980;47:137–139. 24. Detre KM, Wright E, Murphy ML, Takaro T. Observer agreement in evaluating coronary angiograms. Circulation 1975;52:979 –986. 25. DeRouen TA, Murray JA, deSoyza N. Coronary angiogram interpretation. Interobserver variability. Circulation 1977;55:324 –329. 26. Wexler L, Brundage BH, Crouse J, Detrano R, Fuster V, Maddahi J, Rumberger J, Stanford W, White R, Taubert K. Coronary artery calcification: pathophysiology, epidemiology, imaging methods and clinical implications. Circulation 1996;94:1175–1192. 27. Budoff MJ, Georgiou D, Brody A, Agatston AS, Kennedy JM, Wolfkiel C, Stanford W, Shields P, Lewis RJ, Janowitz WR, Rich S, Brundage BH. Ultrafast computed tomography as a diagnostic modality in the detection of coronary artery disease: a multicenter study. Circulation 1996;93:898 –904. 28. Rumberger JA, Simons DB, Fitzpatrick LA, Sheedy PF, Schwartz RS. Coronary artery calcium areas by electron beam computed tomography and coronary atherosclerotic plaque area: a histopathologic correlative study. Circulation 1995;92:2157–2162. 29. Kaufmann RB, Sheedy PF, Breen JF, Kelzenberg JR, Kruger BL, Schwartz RS, Moll PP. Detection of heart calcification with electron beam CT: interobserver and intraobserver reliability for scoring quantification. Radiology 1994;190:347–352.
CORONARY ARTERY DISEASE/EBCT CORONARY ANGIOGRAPHY
845