Author + information
- Received May 7, 2004
- Revision received September 3, 2004
- Accepted October 21, 2004
- Published online January 4, 2005.
- Axel Kuettner, MD*,
- Torsten Beck, MD†,
- Tanja Drosch, MS*,
- Klaus Kettering, MD†,
- Martin Heuschmid, MD*,
- Christof Burgstahler, MD†,
- Claus D. Claussen, MD*,
- Andreas F. Kopp, MD* and
- Stephen Schroeder, MD†,* ()
- ↵*Reprint requests and correspondence:
Dr. Stephen Schroeder, Department of Cardiology, Eberhard-Karls-University Tuebingen, Otfried-Mueller-Str. 10, 72076 Tuebingen, Germany.
Objectives The aim of our study was to evaluate the diagnostic accuracy of 16-multi-detector spiral computed tomography (MDCT) with 188 ms temporal resolution.
Background Because of rapid technical innovations, MDCT coronary imaging has significantly improved in the last five years. Recent results indicate a high diagnostic accuracy, especially in patients with clinical suspicion of coronary artery disease(CAD).
Methods A total of 72 consecutive patients (30 women, 42 men, age 64 ± 10 years) scheduled for invasive coronary angiography (ICA) because of suspected CAD were additionally studied by MDCT (Sensation 16 Speed 4D, Siemens, Forchheim, Germany). Thirty-seven of 72 patients (51%) received an additional beta-blockade before MDCT because of heart rates >65 beats/min. The MDCT scans were analyzed regarding the presence of coronary artery lesions. Results were compared with ICA.
Results All 72 scans showed diagnostic image quality (heart rate: 64.1 ± 9.2 beats/min, calcium mass: 86 ± 156 mg). Thirteen coronary segments were evaluated in each patient. Sixty-two of 936 (6.6%) segments showed a nondiagnostic image quality. All segments were included in the analysis. A total of 117 relevant lesions (diameter stenosis >50%) were detected using ICA, and 96 of 117 (82%) were detected by MDCT. Sensitivity, specificity, and positive and negative predictive values for the whole study group were as follows: 82%, 98%, 87%, and 97%, respectively. The correct clinical diagnosis of presence or absence of significant CAD was obtained in 65 of 72 (90%) patients. All stenoses were detected by MDCT in 52 of 72 (72%) patients.
Conclusions Our results indicate a high diagnostic accuracy of 16-slice MDCT with improved temporal resolution.
Since 1999, mechanical multi-detector spiral computed tomography (MDCT) systems have been available for a noninvasive detection of coronary artery disease (CAD). In autumn 2003, an MDCT generation using 16 slices for cardiac imaging, together with an improved gantry rotation time (16 × 0.75 mm, 375 ms), was introduced. The present study evaluated feasibility, image quality, and clinical accuracy of this MDCT generation in detecting coronary artery lesions in patients scheduled for invasive coronary angiography (ICA) because of the clinical suspicion of CAD.
A total of 72 consecutive patients scheduled for ICA were additionally studied by MDCT. The local ethics committee had approved the study protocol, and all patients gave informed consent. Exclusion criteria were known CAD, irregular heart rate, contraindications against iodinated contrast agents, and elevated serum creatinine levels (>1.5 mg/dl). Patients with heart rates >65/min received additional oral beta-blockade (50 to 100 mg metoprololtartrate) 45 min before MDCT angiography. A cutoff of 65 beats/min was based on previous findings revealing best image quality in heart rates <65 beats/min (1).
The MDCT datasets were acquired using Sensation 16 Speed 4D (Siemens Medical Solutions, Forchheim, Germany) with 375 ms gantry rotation time and 16 detector slices. A native scan was performed to determine coronary calcifications (16 × 1.5 mm, 3.8 mm/rotation, 133 mA at 120 kV). The contrast-enhanced scan (80 ml, Imeron 400) with electrocardiographic pulsing (2) was obtained after an extra scan (20 ml, Imeron 400; determination of contrast agent transit time) within one single breath-hold (15 to 20 s, 16 × 0.75 mm, 3.8 mm/rotation, 650 mA at 120 kV). Axial images were reconstructed at 60% RR-interval, and in case of motion artifacts at the time point with best image quality (slice thickness 1.0 mm, increment 0.5 mm, kernel B 30f).
The MDCT image interpretation
Two readers, blinded to the results of ICA and to all clinical information, evaluated the MDCT scans in a joint reading on an offline workstation (Leonardo, Siemens, Germany). Coronary calcifications were assessed and quantified on native scans (calcium mass = mg calcium hydroxyapatite). Contrast-enhanced scans were investigated using axial slices and 4-mm thin slab maximum intensity projections (Vitrea 2, Vital Images, Plymouth, Minnesota). Image quality was classified as: 1 = excellent, (no motion artifacts); 2 = good (minor motion artifacts); 3 = moderate (substantial motion artifacts, but luminal assessment possible), 4 = heavily calcified (vessel lumen obscured by calcification); and 5 = blurred (no luminal assessment possible). An image quality score (IQS) was calculated for each patient as the average of the classification of the 13 segments ([Σ segments 1 to 13]/13). The readers assessed relevant lesions with a diameter stenosis ≥50%. Results were documented separately for 13 coronary segments (right coronary artery [RCA]: segments 1 to 4; left main [LM]: segment 5; left anterior descending coronary artery [LAD]: segments 6 to 10; left circumflex artery [LCx]: segments 11 to 13 ).
The clinical diagnosis based on MDCT was considered to be correct if the detection/exclusion of any relevant lesion >50% corresponded with the findings of ICA.
Quantitative coronary angiography
The ICA was considered the reference standard for the detection of relevant lesions (>50% diameter stenosis) and was performed according to standard techniques. All angiograms were evaluated by one blinded independent observer using quantitative coronary analysis (Philips Medical Systems, Einthoven, the Netherlands).
Categorical data are presented with absolute frequencies and percentages, continuous variables as their means ± standard deviations. Unpaired ttests were used to compare categorical data. Chi-square tests were performed to evaluate gender differences. Values of p > 0.05 were considered to reveal statistically significant differences. Segments could be viewed as nested within vessels that are nested within patients. The influence of nested effects was evaluated by correlating their sensitivities and specificities. All analyses were conducted using Jump 5.1 Software (SAS, Cary, North Carolina).
Cardiovascular risk factors were distributed as follows: hyperlipoproteinemia 52 of 72 (72%), hypertension 51 of 72 (71%), smoking 23 of 72 (32%), diabetes 13 of 72 (18%), and family history 11 of 72 (15%). The patient characteristics are summarized in Table 1.Thirty-seven of 72 (51%) received additional beta-blockade because of heart rates >65 beats/min, 12 of 72 (17%) patients did not receive any beta-blockade (5 of 30 female, 7 of 42 male, p = 1.0).
All 72 scans showed sufficient image quality, and no patient or segment was excluded from the analysis. A total of 936 (72 × 13) segments were evaluated. Image quality was as follows: excellent = 330 (35.3%) segments; good = 323 (34.5%) segments; moderate = 158 (16.9%) segments; heavily calcified = 63 (6.7%) segments; blurred = 62 (6.6%) segments. The IQS was 2.13 ± 0.75 (1 to 3.8, median 2.04). Detailed results are summarized in Table 2.To evaluate the influence of heart rate on image quality, the patient cohort was subdivided into two groups: group 1 = ≤65 beats/min (n = 47, 59 ± 4.7 [43 to 65]); group 2 = >65 beats/min (n = 25, 74 ± 7.1 [66 to 88]). A significant influence on the overall IQS was found (1.97 ± 0.76 vs. 2.42 ± 0.67; p = 0.015), but the number of segments without diagnostic image quality was not significantly different (35 of 611 [5.7%] vs. 27 of 325 [8.3%]; p = 0.16).
A total of 117 relevant lesions were detected using ICA, and 96 of 117 (82%) were detected by MDCT (Fig. 1). In 52 of 72 (72%) patients, all relevant lesions were detected by MDCT. The majority of missed lesions were located in distal segments and side branches (Table 3).Sensitivity, specificity, and positive and negative predictive values for all patients and segments were 82%, 98%, 87% and 97%, respectively; after the exclusion of diagonal and marginal branches (segments 9, 10, and 13) values were 88%, 98%, 91% and 98%, respectively; after exclusion of patients with a calcium mass >300 mg (n = 5 patients, 526 ± 202, 388 to 852, median 408), as well as for patients with heart rates >70 beats/min (n = 14 patients, 77.7 ± 6.9, 70 to 88, median 78), values were 86%, 98%, 85%, and 98%, respectively. The correct clinical diagnosis was obtained in 65 of 72 (90%) patients. The influence of nesting was calculated as 96 of 117 to 94 of 115 = 0.3%. Thus, sensitivity and specificity could not be underestimated by more than 0.3%. They would have been overestimated had negative findings shown a higher correlation than the positive ones. But this bias was not observed.
Influence of gender on diagnostic accuracy
Seventy-six of 90 lesions were correctly detected in men, and 20 of 27 lesions in women. Sensitivity, specificity, and positive and negative predictive values were 84%, 98%, 89%, and 97%, respectively, in men, and 74%, 98%, 76%, and 98%, respectively, in women. The IQS was comparable in both groups (2.13 ± 0.77 vs. 2.12 ± 0.76; p = 0.96). The prevalence of CAD was 59% (25 of 42) in men and 33% (10 of 30) in women (p = 0.2). A correct clinical diagnosis could be obtained in 38 of 42 (90.5%) men and in 27 of 30 (90%) women (p = 0.95).
The important finding of the present study is that image quality and diagnostic accuracy of noninvasive MDCT imaging could be further improved by the use of 16 slices together with a faster gantry rotation time.
In our study evaluating the entire coronary tree, only 6.6% of coronary segments, in majority side branches and distal vessel segments, showed no diagnostic image quality. However, although MDCT imaging is becoming more accurate, a complete visualization of the entire coronary tree can still not be expected, and further technical improvements are required until MDCT might challenge ICA as reference imaging modality in patients with suspicion of CAD (4).
The previously reported promising results on non-invasive lesion detection (5–9) could be reproduced in our study. The most relevant clinical information obtained by MDCT imaging, the correct clinical diagnosis, could be achieved in our study in 90% of patients, which was comparable with recently reported data (78% , 85% , 93% , 100% ). However, the results from all of these single-center studies need to be confirmed by multi-center trials.
Despite lack of statistically significant differences in prevalence of CAD, patient characteristics, or image quality, there was a trend to a higher diagnostic accuracy in men compared with women. This observation is surprising and requires further evaluation because our study was not designed to elucidate this question. It is, however, in keeping with other noninvasive tests, for example, exercise electrocardiography or radionuclide perfusion imaging (10).
Limitations of MDCT are radiation exposure (1.5 to 2 mSv for calcium scoring, 4 to 6 mSv for the contrast enhanced scan ), the need for iodinated contrast agents, and the fact that a reduction of heart rate by using beta-blockade is still recommendable. However, MDCT gives information on vessel lumen, atherosclerotic plaque composition, and functional parameters by one single examination (11).
In conclusion, noninvasive MDCT imaging is becoming more and more accurate. However, further improvements of spatial and temporal resolution are still required to challenge diagnostic invasive coronary angiography.
- Abbreviations and Acronyms
- coronary artery disease
- cardiovascular disease
- invasive coronary angiography
- image quality score
- left anterior descending coronary artery
- left circumflex artery
- left main (segment 5)
- multi-detector spiral computed tomography
- right coronary artery
- Received May 7, 2004.
- Revision received September 3, 2004.
- Accepted October 21, 2004.
- American College of Cardiology Foundation
- Scanlon P.J.,
- Faxon D.P.,
- Audet A.M.,
- et al.
- Achenbach S.
- Nieman K.,
- Cademartiri F.,
- Lemos P.A.,
- Raaijmakers R.,
- Pattynama P.M.,
- de Feyter P.J.
- Ropers D.,
- Baum U.,
- Pohle K.,
- et al.
- Martuscelli E.,
- Romagnoli A.,
- D'Eliseo A.,
- et al.
- Mollet N.R.,
- Cademartiri F.,
- Nieman K.,
- et al.
- Kuettner A.,
- Trabold T.,
- Schroeder S.,
- et al.