Author + information
- Received June 1, 1996
- Revision received March 10, 1997
- Accepted March 31, 1997
- Published online July 1, 1997.
- Dietrich Baumgart, MDA,*,
- Axel Schmermund, MDA,
- Guenter Goerge, MDA,
- Michael Haude, MDA,
- Junbo Ge, MDA,
- Michael Adamzik, MDA,
- Cornelia Sehnert, MDB,
- Klaus Altmaier, MDC,
- Dietrich Groenemeyer, MDB,
- Rainer Seibel, MDB and
- Raimund Erbel, MDA
- ↵*Dr. Dietrich Baumgart, Division of Cardiology, Center of Internal Medicine, University, Hufelandstrasse 55, D-45122 Essen, Germany. E-mail: firstname.lastname@example.org.
Objectives. This analysis compared the results of electron beam computed tomography (EBCT) with those of coronary angiography and intracoronary ultrasound (ICUS) for the in vivo detection of coronary atherosclerotic plaques.
Background. EBCT is a new imaging modality for identification of coronary calcifications. Coronary angiography depicts advanced changes in coronary morphology, whereas ICUS is an established diagnostic tool that detects the early stages of coronary artery disease.
Methods. In 57 patients (54 ± 9 years old), 267 coronary segments were analyzed with EBCT (3-mm slices, acquisition time 100 ms, threshold definition of coronary calcification at 130 Hounsfield units in an area ≥1 mm2, Agatston calcium score), coronary angiography and ICUS. The analysis was based on the number and extent of coronary calcifications on EBCT, coronary lumen reduction on coronary angiography and plaque formation with and without ultrasound signs of calcifications on ICUS.
Results. Compared with coronary angiography, EBCT yielded a sensitivity of 66%, a specificity of 78%, a positive predictive value of 39% and a negative predictive value of 91%. Compared with ICUS, EBCT yielded an overall sensitivity of 66%, a specificity of 88% and an overall accuracy of 81%. For plaques with and without ultrasound signs of calcifications, the sensitivity of EBCT was 97% and 47%, specificity 80% and 75% and overall accuracy 82% and 69%, respectively.
Conclusions. This in vivo correlation between ICUS and EBCT demonstrates that EBCT is a noninvasive method that helps to visualize the atherosclerotic process by localization and quantification of coronary artery calcifications. EBCT detects calcified plaques with high accuracy. Plaques without ultrasound signs of calcifications can be detected by EBCT but with lower sensitivity but equivalent specificity.
(J Am Coll Cardiol 1997;30:57–64)
Coronary artery disease still remains the primary cause of death in the western industrialized world (). Early diagnosis and risk stratification in patients with known coronary risk factors are major goals in preventional therapeutic strategies. Consequent control of these risk factors and effective lifestyle changes have been demonstrated to postpone the development of hemodynamically significant disease and to reduce cardiac mortality ([2, 3]).
Noninvasive tests are based on indirect signs of coronary atherosclerotic disease. Exercise tests, especially, are characterized by their limited diagnostic accuracy, with sensitivity values ranging between 44% and 84% (). Coronary calcium deposits as a sign of atherosclerosis are already present intracellularly and extracellularly in the early stage of the disease (). The presence of coronary calcifications on fluoroscopy or chest radiography has been associated with a significant increase in cardiac mortality ([6, 7]). However, sparse calcifications may remain unidentified by fluoroscopic techniques. A number of recent investigations have underlined the improved sensitivity of electron beam computed tomography (EBCT) for detection of coronary calcification ([8–11]).
Although coronary angiography yields high accuracy for the detection of hemodynamically significant stenoses, its sensitivity in demonstrating mild to moderate disease is inadequate. Discrepancies with regard to coronary anatomy, particularly in the main stem, have been previously reported ([12, 13]). One reason may be a remodeling process that compensates for lumen narrowing by the development of intimal plaques through vessel enlargement ([14, 15]). Angiographically, lumen narrowing occurs when the plaque is compromising at least 40% to 45% of the vessel lumen ([15, 16]). In contrast, intracoronary ultrasound (ICUS) gives morphologic information on the vessel wall and lumen size. Atheromatous plaque morphology and plaque composition can be quantified. Fair amounts of superficial as well as deep calcifications can be identified due to ultrasound shadowing with high accuracy. Given these characteristics, ICUS offers distinct advantages over angiography for the detection of atherosclerosis, especially in the early stages and thus has been considered the reference standard for detection of atherosclerotic plaques ().
However, both coronary angiography and ICUS require invasive cardiac catheterization, which is still associated with complications. In addition, ICUS cannot be performed routinely in all patients. Thus, new diagnostic approaches are aimed at the noninvasive identification of coronary atherosclerosis in its early stage. Because none of the three technologies can be considered perfect, it was our aim to compare the diagnostic accuracy of EBCT against that of coronary angiography and ICUS for the detection of atherosclerotic coronary lesions.
1.1 Study Patients.
The study group included 57 consecutive patients (mean [±SD] age 54 ± 9 years, range 21 to 66; 45 men, 12 women) who underwent coronary angiography for routine clinical indication. Coronary artery disease was suspected in all patients on the basis of typical or atypical symptoms of angina pectoris. Previous angiographic examination excluded patients from the study group. Angina pectoris was classified Canadian Cardiovascular Society grade II in 41 patients and grade III in 16. Four patients had an anterior and two patients a posterior myocardial infarction >2 months before the investigation. Patients with an occlusion of one of the major coronary arteries or unstable angina pectoris were excluded. EBCT was performed 5 ± 6 days (range 0 to 15) before coronary angiography in 22 patients and 4 ± 8 days (range 1 to 16 days) after coronary angiography in 35. Informed consent to perform EBCT and ICUS was obtained from all patients.
1.2 Coronary Angiography.
Selective coronary angiography was performed by the Judkins technique. All coronary angiograms were reviewed by an observer (D.B., A.S., M.A.) who had no knowledge of the results of EBCT. The coronary artery system was analyzed in 15 coronary segments (). Classification of significant coronary artery disease was based on quantitative coronary angiography. Each segment was classified as normal, insignificantly stenosedor significantly stenosed. Normal segmentswere defined as those without any irregularities and were classified as “negative.” Insignificant stenoseswere diagnosed if lumen narrowing was <50% diameter stenosis. Lumen narrowing ≥50% was considered significant stenosis. Off-line caliper measurements (MEDIS, Reiber) were used to confirm categorization of lesions. Patients with significant stenoses in one, two or three major vessels were classified as having a one-, two-, or three-vessel coronary artery disease, respectively. Insignificantlyor significantly stenosedsegments were considered “positive” on coronary angiography.
ICUS was performed after the coronary angiography using catheters with 20- to 30-MHz transducers (Boston Scientific Corporation) guided by a 0.014-in. floppy guide wire (ACS, Temecula) and connected to a console (Hewlett-Packard). The imaging catheter was positioned under fluoroscopic guidance into the distal segment of the respective coronary artery. The catheter was then pulled back manually to the proximal part of the respective coronary artery in 3- to 4-mm steps. Each position of the ultrasound catheter was marked separately, and concomitantly a short fluoroscopic sequence with an additional injection of contrast medium was filmed. Each step was recorded separately and marked specifically. Thus, each fluoroscopic scene was matched by a specific ICUS scene. Finally, a slow continuous pullback was obtained from distal to proximal. All data were stored continuously on a videotape system (S-VHS, Sony, Tokyo, Japan) for playback and off-line analysis (). Coronary arteries were also analyzed according to the previously described segment analysis.
Coronary segments were classified as normal or as containing significant plaques ([19, 20]). A plaquewas considered significant when the plaque area exceeded 10% of the vessel area as measured with ICUS. Changes of <10% lumen narrowing were considered intimal thickeningand classified as normal. In each segment the most prominent lesion was considered for evaluation. Normal segments were classified as “negative,” whereas segments with significant plaques were considered “positive,” on ICUS. As is common for an analysis of ICUS results, plaques were further classified into plaques with and without ultrasound signs of calcification and are herein referred to as “hard” and “soft” plaques, respectively. Calcium was considered present when typical ultrasound shadowing could be detected (). Such shadowing on ultrasound requires the presence of a fair amount of calcium. Thus, this qualitative assessment of calcification might be subject to interobserver variability and is a limitation to the current analysis.
Nonenhanced fast computed tomographic scans were performed with an Imatron C100 CT scanner (Imatron Inc.) (n = 19) and an Evolution CT scanner (Siemens, Erlangen, Germany) (n = 38) in a high resolution mode with a 100-ms scan time and a 3-mm slice thickness using the single-slice technique without the application of contrast agents. The field of view was 26 cm2, with a resultant pixel size of 0.26 mm2. The position of the patient was checked and adjusted as necessary so that scanning began from the lower margin of the main pulmonary artery bifurcation; 28 to 36 slices were obtained during a single breath hold for ∼45 s. The slices were electrocardiographically triggered to 80% of the RR interval so as to obtain each image at the same point in diastole. From the established protocol (), the threshold for detection of calcification was set at a density of 130 Hounsfield units (HU) and >4 pixels (1.03 mm2). The lesion score was determined as previously described (). Of note is that this approach provides a quantitative analysis of calcium independent of any operator input.
The identification of each lesion on EBCT with respect to its location in the coronary tree was based on anatomic landmarks, such as diagonal or marginal side branches. In addition, the analysis was limited to major coronary arteries or major side branches and excluded small, unidentifiable vessels and side branches. This approach has proven very accurate in previous investigations (), where stents were used as additional landmarks for localization. For all three methods, recordings were viewed by two independent observers (C.S., K.A., D.G., R.S.). In case of disagreement between observers, the lesions were reviewed by a third observer in blinded manner, and the final categorization was determined by consensus. All segments with an EBCT score of 0 were classified as “negative.” All segments with an EBCT score >0 were classified as “positive.”
1.5 Data Analysis.
The analysis was based on the respective comparison between each technique. In addition, the acquired data were analyzed on the basis of patients and coronary segments. Results are presented as mean value ± SEM. One-way analysis of variance was performed with a commercial package (Sigma Stat Jandel Scientific). In case of significant differences, a Wilcoxon test was used to compare individual groups, including a Tipett-Holm correction for multiple comparison. Standard estimates of sensitivity, specificity and positive and negative predictive values were obtained by the use of two by two contingency tables. Differences between groups were evaluated by chi-square analysis for categoric variables. Regression analysis was performed using a Spearman test. A p value <0.05 was considered significant ().
Coronary angiography, EBCT and ICUS could be performed in all patients, with technically adequate images obtained for each analysis. Fig. 1shows a representative comparison of angiographic, EBCT and ICUS findings.
2.1 EBCT Versus Angiography.
There was no significant coronary artery disease in 28 patients (49%). Nineteen (33%) patients had one-vessel disease (left anterior descending coronary artery [LAD]: n = 13; right coronary artery [RCA]: n = 4; left circumflex coronary artery [LCx]: n = 2); 8 (14%) had two-vessel disease; and 2 (4%) had three-vessel disease. Patients without any detectable coronary artery disease by coronary angiography had a significantly lower EBCT score than patients with one-, two- or three-vessel disease (68.8 ± 4.6 vs. 494.5 ± 30.3, 546.9 ± 130.0 or 654.5 ± 85.0, respectively, p < 0.05). It was not possible to discriminate between patient groups with significant coronary artery disease on the basis of EBCT score (Fig. 2). The amount of calcifications detected by EBCT was significantly higher in the proximal LAD and LCx than in the middle and distal parts of these arteries. In contrast, calcifications of the RCA were more evenly distributed, with no significant differences between proximal, middle and distal segments (Fig. 3).
EBCT findings were negative in 6 (21%) of 28 patients and positive in 22 (79%) with negative coronary angiographic findings. Except for one patient with single-vessel disease, EBCT results were positive in all other patients with angiographically detected one- to three-vessel coronary artery disease. EBCT score was significantly lower in 220 angiographically normal segments (82%) than in the 6 segments (2%) with stenosis <50% or the 41 segments (16%) with stenosis >50% on angiographic analysis. There was no statistical difference between coronary segments with <50% or >50% coronary stenosis (Fig. 4).
There were 41 hemodynamically significant stenoses in 29 (51%) of 57 patients by angiography (Table 1); 26 stenoses (63%) showed ultrasound signs of calcification. Most of the stenoses in the RCA and LAD were calcified and were thus detected by EBCT. In contrast, the majority of the stenoses in the LCx did not show ultrasound signs of calcification. These stenoses were detected with lower sensitivity (Table 1). Only 1 (3%) of 29 patients with a significant angiographic stenosis had negative findings on EBCT. Statistical values for the detection of significant coronary artery stenoses are shown in Table 2. Of 47 (18%) of 267 segments that were positive on angiography, 31 (66%) were positive and 16 (34%) were negative on EBCT. Of the 220 segments that were negative on angiography, 49 (22%) were positive and 171 (78%) were negative on EBCT (Table 3). Statistical values for EBCT versus angiography are shown in Table 4.
2.2 EBCT Versus ICUS.
Only 4 (7%) of 57 patients showed no sign of atherosclerotic changes in the segments analyzed. EBCT and angiographic findings were negative in these patients.
ICUS was performed in 267 segments. Forty-three main stem segments (16%), 141 LAD segments (53%) (43 patients), 57 LCx segments (21%) (26 patients) and 26 RCA (10%) segments (9 patients) were analyzed. On the basis of these ICUS images, 178 segments (67%) in 53 patients were classified as normal, 55 segments (21%) in 30 patients had atherosclerotic plaques without ultrasound signs of calcification (“soft” plaques), and 34 segments (13%) in 17 patients showed plaques with ultrasound signs of calcification (“hard” plaques). The EBCT score was significant higher in segments containing soft and hard plaques than in normal segments (Fig. 5).
EBCT results were positive in 33 (97%) of 34 segments with hard plaques. In the 233 segments that were negative for hard plaques on ICUS, 47 (20%) were positive and 186 (80%) were negative on EBCT (Table 3). EBCT results were positive in 26 (47%) of 55 segments with soft plaques. In the 212 segments that were negative for soft plaques on ICUS, 54 (25%) were positive and 158 (75%) were negative on EBCT (Table 3). When segments with hard and soft plaques were considered together, 59 (66%) of 89 segments with plaques were positive on EBCT. In the 178 segments that were negative for plaques on ICUS, 21 (12%) were positive and 157 (88%) were negative on EBCT (Table 3). Statistical values for EBCT for detection of hard, soft and overall plaque are shown in Table 4. Further analysis of soft plaques did not show any correlation between soft plaque area and EBCT score (Fig. 6). A positive EBCT score was most often seen in plaques with minor lumen narrowing that ranged between 20% and 35% of plaque area. There was no significant difference in EBCT score between concentric and eccentric plaques (51.7 ± 11.3 vs. 64.7 ± 4.8, p = 0.325).
2.3 ICUS Versus Angiography.
Of 89 (33%) of 267 segments that were positive by ICUS, 47 (53%) were positive and 42 (47%) were negative by angiography. All 178 segments that were negative by ICUS were also negative by angiography (Table 5). Table 4shows the statistical values for ICUS versus angiography.
To our knowledge, this is the first in vivo comparison of EBCT with ICUS and coronary angiography. The value of EBCT for the detection of coronary atherosclerosis can be summarized as follows: EBCT is the first noninvasive method for visualizing, localizing and quantifying the atherosclerotic process. Plaques with ultrasound signs of calcification are detected with the highest sensitivity and specificity compared with that for coronary angiography and ICUS. Early signs of atherosclerosis (i.e., soft plaques without apparent ultrasound signs of calcification on ICUS) can also be detected by EBCT but with lower sensitivity than for calcified plaques but equivalent specificity. In the individual patient, a negative EBCT finding excludes significant coronary artery disease with excellent accuracy. However, the extent of calcification does not correlate with stenosis severity. The discrepancy between EBCT and coronary angiography in the detection rate of atherosclerotic plaques can be explained by our ICUS studies.
3.1 EBCT Versus ICUS.
To our knowledge, this is the first report to show very reliable in vivo detection and localization of plaques with ultrasound signs of calcification regardless of the degree of lumen narrowing by EBCT. These data support the contention derived from autopsy studies ([5, 23, 24]) that there is a close link between coronary calcium and atherosclerosis.
However, the most important clinical implication is the high negative predictive value of EBCT. Thus, a negative EBCT score can exclude calcified plaques with high accuracy. Histologic findings have demonstrated that calcium deposits are already present in the early stages of the atherosclerotic process, beginning with Stary type III lesions (), present in the second decade of life. Calcium is regularly found in more advanced lesions, such as atheroma (Stary type IV) or fibroatheroma (Stary type V), which prevail from the third decade of life (). During the progression of atherosclerotic plaque formation, the amount of calcium is increased because calcified matrix vesicles result from dead fibrocytes () and smooth muscle cells (). This finding is supported by histopathologic studies in 13 perfusion-fixed hearts scanned with a 3-mm slice thickness (). A direct correlation between EBCT-detected calcium and atherosclerotic plaque area was observed. Thus, EBCT is able to detect preclinical signs of coronary artery disease before lumen narrowing is found.
EBCT also detected 50% of the soft plaques without apparent calcification on ICUS. Comparable to in vitro results (), the sensitivity of EBCT for detection of soft plaques was lower than that for hard plaques. However, EBCT visualized the atherosclerotic process in a number of coronary segments with minor lumen narrowing (plaque area between 20% and 35%). Thus, detection of coronary atherosclerosis was earlier than that by conventional coronary angiography. Detection of soft plaques by EBCT may be due to microcalcifications or densification of the plaque in the process of plaque maturation that is not visible to ICUS. Although ICUS has a spatial resolution of 250 μm, the detection of calcium with ICUS is based on the phenomenon of ultrasound shadowing, which requires a fair amount of calcium with a lateral extension substantially >250 μm. This conjecture is strongly supported by the thorough investigations of Friedrich et al. () and Peters et al. (). In their studies the detection of intralesional calcium by ICUS was crucially dependent on the histologic pattern of calcium distribution. Plaques with dense calcifications were detected with high sensitivity (90%), whereas plaques with only microcalcifications were detected with substantially lower sensitivity (64% to 77%). Specificity remained at 100% in all subgroups. When this potential limitation is taken into account, the respective specificity and positive predictive values for EBCT in the present study might have been even higher.
Only 3 (20%) of 15 soft plaques that reduced the lumen area to >50% had a positive EBCT score. Although detection of relevant soft stenoses for the single site was low, overall only one of these patients with stenosis had a total EBCT score that was negative. Thus, for the individual patient, the chances for identification of the atherosclerotic process are high.
Although plaque area may correlate with clinical signs of stable angina pectoris, plaque area is not necessarily an indicator for acute cardiac events (i.e. myocardial infarction) due to plaque rupture of unstable plaques ([31, 32]). Because it is impossible to identify unstable plaques with EBCT, such risk stratification for future major cardiac events seems problematic. In contrast, extensive coronary calcifications have been identified by previous investigators as a predictor for such major events ([6, 7]). These apparently contradictory statements can be reconciled because the likelihood of major cardiac events increases with the severity of coronary artery disease. Prognosis is more closely related to the overall magnitude of atherosclerotic plaque “burden” within the coronary system than to single or multiple discrete lumen narrowing defined by quantitative coronary angiography (). The present study supports the observation that overall coronary artery calcifications increase with severity of the disease.
The positive predictive value of 41% was low because 47 ICUS-negative segments (i.e., no sign of any atherosclerotic plaque) were positive on EBCT, with a low score of 12.3 ± 1.9. This finding is explained by a lack of ICUS for detecting discrete, speckled amounts of calcium in the adventitia rather than by an oversensitivity of EBCT for detecting calcific deposits. EBCT yielded high accuracy in a patient cohort with early signs of atherosclerosis (i.e., normal findings on coronary angiography but positive findings on ICUS).
3.2 EBCT Versus Coronary Angiography.
Early detection of patients with significant coronary artery disease and reliable exclusion of coronary artery disease in those with normal coronary arteries by noninvasive means are a major goal of EBCT. Sensitivity for the detection of significant coronary artery disease in the present study compared well with that of previous studies ([7, 10, 11, 33, 34]). With a sensitivity between 60% and 90%, EBCT is within the range of other noninvasive tests (e.g., exercise electrocardiography or stress echocardiography) for the detection of significant coronary artery disease. Specificity, in contrast, showed a broad scatter of values in previous studies, ranging from as low as 36% to 91%. Specificity depends on the underlying definition of vessel wall changes, the age of the patient cohort and the cutoff score for the detection of calcifications. Using a cutoff score of 160 HU, Arad et al. () demonstrated the highest values for sensitivity (89%), specificity (82%) and negative predictive value (100%) compared with cutoff scores of 100 and 680 HU with respect to the prognosis of future cardiac events in patients with coronary artery disease. Because beginning coronary atherosclerosis often remains undetected by angiography, a comparison between coronary angiography and EBCT has to yield discrepant results with respect to the presence of coronary artery disease. As has been demonstrated by the present study, specificity for the detection of coronary plaques is increased from 78% to 88%, when EBCT results are compared with those of ICUS instead of coronary angiography.
Positive predictive values in the present study were low because ∼50% of our patients did not exhibit any significant coronary artery disease. Similar findings were reported by Loecker et al. (). However, when our EBCT results were analyzed according to patients and not segments, 28 (97%) of 29 patients with significant coronary artery had a positive EBCT score. These results may be explained by the presence of calcifications at other sites with plaques in different stages of disease. In vitro as well as in vivo studies ([33, 36]) have demonstrated that increasing EBCT scores are associated with an increasing likelihood of more advanced atherosclerotic involvement. Although mean values helped to differentiate patients with normal and diseased coronary arteries in the present study, an individual cutoff value for the presence of significant coronary artery disease could not be detected. However, given the high negative predictive value, the absence of coronary calcifications at any site is highly specific for the absence of obstructive coronary artery disease.
In concordance with previous histopathologic ([36–38]) and angiographic studies (), calcifications are predominantly located at the proximal part of the left coronary arteries. The RCA follows a different pattern, with a more even distribution of coronary atherosclerosis that also involves the distal parts of the RCA beyond the crux cordis.
3.3 Study Limitations.
The present correlation between techniques was based on a segmental analysis of coronary atherosclerotic changes rather than on a lesion by lesion analysis; the nature of the atherosclerotic process seldomly presents as a strictly local lesion but instead involves a regional segment with changes ranging from intimal proliferation to calcified plaque. Thus, the slicing angle of each tomographic image is crucial for the discrimination between single lesions. A more horizontal cut of the respective coronary artery increases the likelihood that a single lesion will be detected twice on two contiguous tomograms. Such a limitation is more likely to occur with lesions seen in the proximal LAD than in the more perpendicularly coursing LCx or RCA. Again, segmental analysis, as in the present study, may circumvent these methodologic problems.
3.4 Clinical Perspective.
Although the diagnosis of obstructive coronary artery disease by EBCT is the subject of controversy (), its ability to accurately detect calcified coronary plaques is of paramount importance. EBCT represents the only currently available noninvasive method for the identification of subclinical coronary atherosclerosis, which has been clearly demonstrated to be a strong predictor of cardiac mortality and morbidity (). Early identification of coronary plaques using EBCT in high risk subjects could allow identification of patients in whom aggressive risk factor modification should be undertaken and might result in a significant reduction of future cardiac events.
Asymptomatic patients as well as patients with atypical angina pectoris who have EBCT-negative findings should not be subjected to coronary angiography unless one of the exercise tests unequivocally demonstrates a positive result. Additional use of contrast medium in conjunction with new technical developments in three-dimensional reconstruction of the coronary artery tree may have important implications for the potential noninvasive detection of coronary artery disease.
We greatly appreciate the help of Silke Lange, PhD, Institution for Development and Research of Microtechnology, University Bochum, in the statistical analysis of the data.
- electron beam computed tomography
- Hounsfield units
- intracoronary ultrasound
- left anterior descending coronary artery
- left circumflex coronary artery
- right coronary artery
- Received June 1, 1996.
- Revision received March 10, 1997.
- Accepted March 31, 1997.
- The American College of Cardiology
- Friedewald ST,
- Wyngaarden JB,
- Smith LH
- Brown BG,
- Zhao XQ,
- Sacco DE,
- Albers JJ
- ↵Dennis CA, Pool PE, Perrins EJ, et al. Stress testing with closed-loop arbutamine as an alternative to exercise. J Am Coll Cardiol 1995;261151–261158
- Loecker TH,
- Schwartz RS,
- Cotta CW,
- Hickman JR
- Reinmüller R,
- Lipton MJ
- Agatston AS,
- Janowitz WR,
- Hildner FJ,
- Zusmer NR,
- Viamonte M,
- Detrano R
- Mintz GS,
- Painter JA,
- Pichard AD,
- et al.
- ↵Wallerr BF, Pinkerton CA, Slack JD. Intravascular ultrasound: a histologic study of vessels during life. The new “gold standard” for vascular imaging. Circulation 1992;852305–852310
- American Heart Association Committee Report
- Ge J,
- Erbel R,
- Gerber T,
- Görge G,
- Koch L,
- Haude M,
- Meyer J
- ↵Schmermund A, Haude M, Baumgart D, et al. Non-invasive assessment of coronary Palmaz-Schatz stents by contrast enhanced electron beam computed tomography. Eur Heart J 1996;171546–171553
- Winer BJ
- Simons DB,
- Schwartz RS,
- Edwards WD,
- Sheedy P,
- Breen J,
- Rumberger J
- Rumberger JA,
- Simons DB,
- Fitzpatrick LA,
- Sheedy P,
- Schwartz RS
- Stary HC
- Kearney P,
- Erbel R,
- Ge J,
- Zamorano J,
- Koch L,
- Görge G,
- Meyer J
- Fallavollita J,
- Brody A,
- Bunnell I,
- Kumar K,
- Canty JM
- Arad Y,
- Sparado LA,
- Goodman K,
- et al.
- McCarthy JH,
- Palmer FJ
- Rumberger JA,
- Yacyshyn V,
- Sheedy P,
- et al.
- Emond M,
- Mock MB,
- Davis KB