Author + information
- Maros Ferencik, MD, PhD⁎ (, )
- Koen Nieman, MD, PhD and
- Stephan Achenbach, MD
- ↵⁎Massachusetts General Hospital, Harvard Medical School, 165 Cambridge Street, Suite 400, Boston, Massachusetts 02114
To the Editor:
Contrast-enhanced multi-detector computed tomography (MDCT) coronary angiography permits reliable visualization of coronary arteries. Recent studies showed a high sensitivity and specificity of 16-slice MDCT for the detection of hemodynamically significant coronary stenosis (1,2). In addition to luminal narrowing, it was early recognized that MDCT also visualizes the diseased coronary arterial wall (3). Two recent studies with 16-slice MDCT showed sensitivity of 75% and 82% and specificity of 88% and 92% to detect coronary plaque using intravascular ultrasound as the standard of reference (4,5). The goal of our study was to establish the interobserver reproducibility of coronary plaque detection by 16-slice MDCT.
We enrolled 45 consecutive patients (37 men; mean age, 56 ± 9 years; mean heart rate during the scan, 55 ± 6 beats/min) who underwent contrast-enhanced MDCT before coronary angiography as part of a research protocol. Patients were scheduled for a clinically indicated invasive coronary angiography. Only patients with a heart rate <65 beats/min were included in the evaluation. Patients with allergic reaction to iodinated contrast, atrial fibrillation, or creatinine serum level >1.5 mg/ml were excluded from the study. The institutional review boards approved the study protocol, and patients’ written informed consents were obtained.
Examinations were performed with a 16-slice MDCT scanner (Siemens, Forchheim, Germany; tube current, 550 mA; tube voltage, 120 kVp; collimation, 16 × 0.75 mm; gantry rotation, 375 ms; contrast, 90 ml iopromide 370 mgI/ml; slice thickness, 1 mm at 0.5-mm intervals; kernel B35f). An oral beta-blocker (atenolol 100 mg) was administered one hour before the scan to 38 (84%) patients with a heart rate >60 beats/min. Sublingual nitroglycerin (glycerol trinitrate 0.8 mg) was administered immediately before the scan. Prospectively, electrocardiographically triggered tube current modulation was applied during all examinations.
The MDCT datasets (axial, multiplanar reformatted, and maximum-intensity projection images) were evaluated by two investigators. Images were initially analyzed at a predefined window and level setting (window, 700 HU; level, 200 HU). Adjustments of the image display setting were performed on the discretion of an observer if deemed necessary.
Investigators visually evaluated images for the presence of plaque in all 17 coronary segments defined by the American Heart Association (6). Previously defined criteria for the evaluation of atherosclerotic plaque in contrast-enhanced MDCT coronary angiography were applied (Fig. 1)(4,5). Coronary artery segments were identified in a consensus reading via side branches. Subsequently, the investigators separately and independently analyzed each coronary segment. They graded the image quality for every segment as excellent (no artifacts), good (minimal motion artifacts and/or increased image noise, but evaluable), or nonevaluable. For each coronary segment with excellent or good image quality, the investigators then decided whether calcified plaque, noncalcified plaque, both, or neither were present in MDCT. Kappa statistic was calculated to determine interobserver agreement.
A total of 735 coronary segments were identified. Fifty coronary segments (46 side branches, 2 mid segments, and 2 distal segments) were deemed nonevaluable by one or both observers. Segments were excluded because of small vessel caliber (n = 45), motion artifacts (n = 4), or incomplete visualization of the posterior descending coronary artery (n = 1).
Coronary plaque was detected in 237 of 685 (34.6%) segments by observer #1 and in 238 of 685 (34.7%) segments by observer #2. Calcified plaque was detected in 158 of 685 (23.1%) segments both by observer #1 and observer #2. Noncalcified plaque was detected in 207 of 685 (30.2%) segments by observer #1 and in 209 of 685 (30.5%) segments by observer #2.
In the segment-based analysis, interobserver agreement was very good for the detection of any coronary plaque (93.1% agreement, κ = 0.85), calcified plaque (97.7%, κ = 0.93), and noncalcified plaque (92.4%, κ = 0.82). Similar results were observed in the per vessel and per patient analysis (Table 1).
Interobserver agreement for plaque detection in the proximal coronary segments (proximal right, left main, proximal left anterior descending, and proximal left circumflex coronary arteries) and distal coronary segments (all other segments) was similar: any coronary plaque (91.7%, κ = 0.83, vs. 93.7%, κ = 0.84), calcified plaque (96.1%, κ = 0.92, vs. 98.2%, κ = 0.94), and noncalcified plaque (91.7%, κ = 0.83, vs. 92.7%, κ = 0.79).
The potential value of noninvasive detection, quantification, and characterization of coronary plaque for risk stratification and disease progression monitoring is currently being explored. For these applications, MDCT assessment of coronary plaque would need to be consistent among different observers.
We found very high interobserver reproducibility for the detection of coronary plaque by MDCT. Using four-slice MDCT, Leber et al. (7) reported an interobserver agreement of 79%, 88%, and 70% for the detection of any plaque, calcified plaque, and noncalcified plaque. In our study, corresponding results were 93%, 98%, and 92%. Good agreement was found for both calcified and noncalcified plaque, and could be maintained in both proximal and distal branches.
Limitations of the study include the study population that was pre-selected with stringent exclusion criteria. Also, scans were performed and read by very experienced investigators. Results may therefore not be generalized to less experienced centers and a broader range of patients. The segmental analysis used in the present study is prone to overestimate the diagnostic performance of MDCT. However, it provides the most reliable method for the comparison of readings of two independent observers.
In conclusion, very good interobserver agreement was shown for the detection of both calcified and noncalcified coronary plaque by 16-slice MDCT.
Please note: Dr. Nieman was supported by the Interuniversity Cardiology Institute of the Netherlands, Utrecht, the Netherlands.
- American College of Cardiology Foundation
- Nieman K.,
- Cademartiri F.,
- Lemos P.A.,
- Raaijmakers R.,
- Pattynama P.M.,
- de Feyter P.J.
- Ropers D.,
- Baum U.,
- Pohle K.,
- et al.
- Schroeder S.,
- Kopp A.F.,
- Baumbach A.,
- et al.
- Achenbach S.,
- Moselewski F.,
- Ropers D.,
- et al.
- Leber A.W.,
- Knez A.,
- Becker A.,
- et al.