| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiology, Northwestern University, 448 E Ontario St., Ste. 700,
Chicago, IL 60611.
2 Department of Radiology, PLA General Hospital, Beijing 100853, China.
3 Department of Preventive Medicine, Northwestern University, Chicago, IL.
4 Siemens Medical Solutions, Chicago, IL.
5 Department of Biomedical Engineering, Northwestern University, Chicago,
IL.
Received July 1, 2007;
accepted after revision September 4, 2007.
Address correspondence to J. C. Carr.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
50%) in patients with high
calcium scores. MATERIALS AND METHODS. Eighteen patients (12 men, six women; mean age, 56 y; age range, 38–77 y) who had at least one calcified plaque with a calcium score of > 100 underwent coronary MRA and conventional coronary angiography (CAG) within 2 weeks of MDCTA. Coronary MRA image quality of the calcified segments was assessed by two observers in consensus on a 4-point scale (1 = not visible, 2 = poor, 3 = good, 4 = excellent) using a 10-segment model from the modified American Heart Association classification. Three experienced radiologists, unaware of the results of conventional CAG, independently assessed for the presence of significant stenosis on MDCTA images and the corresponding MRA images. Receiver operating characteristic (ROC) curves were calculated for each reader using conventional CAG as the gold standard.
RESULTS. Thirty-three calcified plaques with a calcium score of > 100 were detected on MDCTA in the 18 patients. The coronary segments with nodal calcification (n = 17) showed a higher mean image quality score than the segments with diffuse calcification (n = 16) (3.47 ± 0.62 vs 2.94 ± 0.77, respectively; p < 0.05). Of the 33 coronary segments with calcification, 12 significant stenoses were identified on conventional CAG. The sensitivity, specificity, and area under the ROC curve (AUC) for MRA and MDCTA, respectively, were as follows: reader 1, 75%, 81%, 0.82 versus 75%, 48%, 0.68; reader 2, 83%, 71%, 0.82 versus 67%, 52%, 0.63; and reader 3, 83%, 71%, 0.85 versus 83%, 43%, 0.65, respectively. The average AUC of MRA for the three readers was significantly higher than that of MDCTA (p = 0.030).
CONCLUSION. Coronary MRA has higher image quality for coronary segments with nodal calcification than for coronary segments with diffuse calcification. Coronary MRA has better diagnostic performance than coronary MDCTA for the detection of significant stenosis in patients with high calcium scores.
Keywords: coronary angiography • coronary arteries • coronary artery disease • coronary artery stenosis • MDCT angiography • MR angiography
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Over the past decade, coronary MR angiography (MRA) and MDCT angiography (MDCTA) have emerged as possible alternative noninvasive coronary artery imaging techniques [2]. Previous comparative studies have shown that coronary MDCTA has a higher diagnostic accuracy than MRA in the detection of significant coronary artery stenosis [3–5]. Furthermore, coronary CT angiography (CTA) with 64-MDCT has become a routine clinical test because of its high diagnostic accuracy in identifying significant coronary stenosis [6, 7]. However, coronary MDCTA is limited in its assessment of coronary arteries with moderate and severe calcifications [8, 9]. In addition, the results of recent studies by Wang et al. [10] and Ramakrishna et al. [11] showed that approximately 50–70% of all coronary artery plaques are calcified in patients with asymptomatic or suspected CAD. This high prevalence of calcification further hinders coronary MDCTA from replacing invasive angiography in the foreseeable future.
Because MRI does not suffer from beam-hardening artifact caused by high-density calcification, coronary MRA can potentially depict the lumen of calcified coronary arteries. At present, coronary MRA with steady-state free precession (SSFP) is regarded as an increasingly advantageous approach to coronary MRA with its relatively high signal-to-noise ratio and good ability to image the coronary arteries [2, 12, 13]. To our knowledge, there have been no studies assessing the diagnostic capability of coronary MRA with SSFP for the detection of coronary stenosis in patients with high calcium scores. The purpose of this study was to compare the diagnostic performance of coronary MRA and MDCTA in the detection of significant stenosis caused by moderate and severe calcifications using conventional CAG as the gold standard.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Thirty-two eligible patients were recruited for the coronary MRA section of the study within 1–3 days after MDCTA examination. Twenty-seven patients (19 men and eight women; mean age, 58 y; age range, 38–79 y) successfully completed the MRA study. MRA studies could not be completed because of failed respiratory gating in three patients and failed ECG gating in two patients. Eighteen of the 27 patients subsequently underwent conventional CAG 3–10 days (mean, 6 days) after the MRA study.
There were no clinical events or medication changes recorded between examinations. Patients who had a heart rate of > 70 beats per minute (bpm) at rest received an oral ß-blocker (25–50 mg) or IV metoprolol (5 mg) to decrease their heart rate before both the MDCTA and MRA examinations. The difference in heart rate between the MDCTA and MRA studies was < 5 bpm. Sublingual or IV nitroglycerine (5 mg) was used in both coronary MDCTA and MRA to achieve maximal coronary vasodilation in all patients.
Coronary MDCTA
Coronary MDCTA was performed and calcium scores were obtained using a
64-MDCT scanner (Somatom Sensation Cardiac 64, Siemens Medical Solutions). An
initial unenhanced ECG-gated scan was obtained for coronary calcium scoring
(collimation, 24 x 0.6 mm; pitch, 0.2; tube current, 300 mAs; tube
voltage, 120 kV; slice thickness, 3 mm; threshold for calcium detection, 130
H; kernel, B30f). After the calcium scoring scan, a bolus of 82 ± 12 mL
(mean ± SD; range, 65–105 mL) of contrast agent (iohexol
[Omnipaque, GE Healthcare], 350 mg I/mL) was injected IV at a rate of 5 mL/s,
followed by a 50-mL saline chaser bolus. Scanning started automatically as
soon as the intensity in the ascending aorta reached a predefined threshold of
100 H. The detector collimation was 32 x 0.6 mm; tube rotation speed,
330 milliseconds per rotation; pitch, 0.2–0.3; tube voltage, 120 kV;
effective tube current, 750–850 mAs; and voxel size, 0.6 x 0.6
x 0.75 mm3. The temporal resolution was variable between 83
and 165 milliseconds according to the heart rate of the patient. Axial images
were reconstructed with a B25f kernel and a slice thickness of 0.75 mm
(increment, 0.4 mm) at the optimal ECG phases with minimum motion artifacts
for each main branch of the coronary arteries.
Coronary MRA
All coronary MRA studies were performed on a 1.5-T scanner (HD TwinSpeed,
GE Healthcare; or Magnetom Avanto, Siemens). A dedicated cardiac 8-element
phased-array coil and a body 6-element phased-array coil were used in the two
scanners, respectively. Eighteen patients were scanned on the GE Healthcare
scanner using a volume-targeted method in which the major coronary arteries
are imaged separately by several targeted, thin 3D slabs
[14]. The other nine patients
were scanned on the Siemens scanner using a whole-heart method in which the
whole coronary artery tree is covered by a thick 3D slab with one single
measurement [15]. All data
acquisitions were performed with an ECG-triggered 3D SSFP sequence with T2
preparation during free breathing.
Respiratory gating was performed using two spin-echo signals that were located parallel to the z-axis and intersected at the dome of the right hemidiaphragm. The width of the respiratory gating window was 6 mm. Volume-targeted coronary MRA was performed with a 3D slab of 16–20 slices of 2 mm thickness to cover one to two coronary arteries. Two or three slabs were acquired to cover the entire coronary artery tree. The scanning parameters were as follows: TR/TE, 4.9/2.4; flip angle, 65°; readout bandwidth, 125 kHz; field of view, 280 mm; matrix, 256 x 256; and voxel size, 1.1 x 1.1 x 2.0 mm3.
Whole-heart coronary MRA was performed with a 3D slab of 44 transverse slices (sync-interpolated into 88 slices of 1.3 mm thickness) to cover the whole volume of the heart. The imaging parameters were as follows: 3.7/1.7; flip angle, 90°; read-out bandwidth, 870 Hz per pixel; lines per heartbeat, 25–33; and voxel size, 0.9 x 0.9 x 1.3 mm3.
Conventional CAG
Selective conventional CAG was performed with a standard Judkins technique.
The images were interpreted by a cardiologist who was unaware of the results
of coronary MDCTA and MRA. The severity of coronary artery stenosis was
expressed as the percentage reduction in the luminal diameter determined using
the quantitative coronary analysis method. An 50% reduction in luminal
diameter was considered significant stenosis.
Image Analysis
All coronary MDCTA data were transferred to a computer workstation
(Leonardo, Siemens) for postprocessing. Maximum-intensity-projection (MIP),
curved multiplanar reformation (MPR), and volume-rendering (VR) images were
generated and used to evaluate the coronary arteries.
The coronary artery calcium scores were measured for each calcified plaque using a dedicated software program (Calcium Scoring, Leonardo, Siemens). Calcium scores were calculated according to the method of Agatston et al. [16]. A coronary calcification is usually classified as minimal (calcium score of 1–10), mild (11–100), moderate (101–400), or severe (> 400) according to the total Agatston calcium score in the whole coronary artery system [17]. Because the small size of a calcification does not affect the interpretation of coronary MDCTA, only a single calcified plaque with a calcium score of > 100 and a diffuse coronary calcification were included in the analysis. Calcified plaques were further characterized as having either a nodal or diffuse pattern. Diffuse coronary artery calcification was defined as a single arterial segment with more than three calcified plaques resulting in a combined calcium score of > 100 [18]. If diffuse calcifications extended to more than two segments, each segment was considered a separate calcified plaque.
All coronary MRA images were transferred to a computer workstation (ADW 4.3, GE Healthcare; or Leonardo, Siemens) for postprocessing. MIP and curved MPR images were reviewed for image analysis. VR images were not used for analysis of the MRA images.
|
|
|
Three radiologists experienced in cardiac image analysis as part of their
daily clinical practice for at least 3 years reviewed coronary MDCTA and MRA
images independently and separately. They were not involved in performing the
examinations and were unaware of the results of conventional CAG and of other
clinical information. Coronary MDCTA and MRA images were reviewed in random
order and 2 weeks apart. Each reader assessed the severity of coronary
stenosis to be 50% or < 50% at the site of each calcification on MDCTA
and corresponding site on MRA. The analysis of significant coronary stenosis
was performed visually by the three readers on the image workstations. To
estimate the likelihood that a significant stenosis was present, the readers
used a 5-point confidence scale: 5, definitely positive; 4, probably positive;
3, possibly positive; 2, probably negative; 1, definitely negative.
Statistical Analysis
To compare the diagnostic performance of coronary MRA and MDCTA, receiver
operating characteristic (ROC) curves were generated for the each of the two
techniques. The overall diagnostic accuracy of each imaging technique for each
reader was evaluated by calculating the area under the ROC curve (AUC)
[21]. An AUC value of >
0.80 was considered to have good diagnostic accuracy
[22]. The difference between
the imaging techniques in terms of the AUC for each reader and the mean AUC
across the three readers was analyzed using the bootstrap method (1,000
replicates), which accounts for potential intrasubject correlation
[23]. For each reader, the
number of significant stenoses correctly judged to be probably positive
(confidence level of 4) or definitely positive (confidence level of 5) was
considered the number of correctly diagnosed significant stenoses. The
sensitivity and specificity of each imaging technique were estimated and
compared between the two imaging techniques with the McNemar test
[24].
To assess interobserver agreement in image interpretation for coronary MDCTA and MRA for the three independent readers, kappa statistics were calculated between any two readers and among the three readers for each technique. The significance of the difference between the kappa values of the two techniques was tested using the bootstrap method.
All statistical analyses were performed using statistical software (SAS version 9.1, SAS Institute; and R-2.2.0, R Development Core Team). For all tests, except when indicated otherwise, a p value of 0.05 indicated a statistically significant difference.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Of the 33 calcified coronary segments on coronary MRA, six had an image quality score of 2, 14 had an image quality score of 3, and 13 had an image quality score of 4. For the six nonassessable segments that had an image quality score of 2, five were due to severe diffuse calcifications (calcium score > 400) and one was due to significant motion artifacts. The mean image quality score of the segments with nodal calcification was higher than that of the segments with diffuse calcification (3.47 ± 0.62 vs 2.94 ± 0.77, respectively; p <0.05).
Of the 33 calcified coronary segments, 12 showed significant stenoses at corresponding sites of calcification on conventional CAG. For the 12 lesions with significant stenosis identified on conventional CAG, four were caused by diffuse calcification, six were caused by mixed calcified plaque (Fig. 2A, 2B, 2C), and two were caused by nodal calcification located at the bifurcation of the LAD artery and a diagonal branch (Fig. 3A, 3B, 3C).
|
|
|
|
|
|
|
The interobserver agreement in image interpretation among the three
independent readers is summarized in Table
2. The agreement between any two readers for coronary MRA was
moderate (0.3 < 
< 0.5), whereas the agreement for coronary MDCTA
was poor between readers 1 and 2 and 1 and 3 ( < 0.1) and moderate
between readers 2 and 3 ( = 0.38). The kappa value that measures the
agreement between readers 1 and 2 was significantly higher in MRA than in
MDCTA ( = 0.45 vs 0.01, respectively; p = 0.007). In addition,
the overall kappa statistic that measures the agreement among the three
readers was significantly higher for coronary MRA than for coronary MDCTA
( = 0.39 vs 0.15, respectively; p = 0.014).
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In this study, coronary MRA showed a limitation for visualizing the coronary lumen in segments with severe diffuse calcification. This limitation is mainly caused by the limited spatial resolution of coronary MRA. In addition, the severe luminal narrowing and reduced blood flow within diffuse calcified arteries [25] may also impair the signal-to-noise ratio of the calcified segments on coronary MRA. However, coronary MRA showed a good capability for the assessment of coronary arteries affected by a nodal pattern of calcification.
On the basis of ROC analysis, this study revealed that all readers achieved an AUC value of > 0.80 for coronary MRA, indicating good diagnostic performance of MRA in the detection of significant stenosis in patients with high calcium scores. In this study, although six of 33 calcified segments were nonassessable on coronary MRA, all readers had an AUC value of > 0.80 because the nonassessable segments were mainly affected by severe diffuse calcification, which is commonly associated with significant coronary stenosis. The AUC values of MRA were greater than those of MDCTA for all readers. In addition, the average AUC value for MRA was statistically significantly greater than that for MDCTA among the three readers (0.831 vs 0.651, respectively; p = 0.030).
The sensitivity values of coronary MRA and MDCTA were similar across the three readers (p > 0.05). This result is different from previous comparative studies that showed coronary MDCTA had a higher sensitivity than MRA [1–4]. The possible reason for our results may be related to the use of a site-by-site design in our study. The site-by-site comparison between coronary MDCTA and MRA may make coronary MRA more sensitive for the detection of significant stenosis in the corresponding site of calcification. This may also explain the higher AUC value (mean AUC = 0.831) of coronary MRA in this study than in previous studies [26]. The average specificity value for coronary MRA was greater than that for MDCTA (0.746 vs 0.476, respectively; p < 0.05), which is consistent with the results from previous studies [2–4]. Coronary MDCTA is prone to allow overestimation of the severity of coronary stenosis due to the beam-hardening artifacts from calcification. Previous studies have also shown that severe coronary calcification may reduce the specificity of coronary MDCTA for the detection of significant stenoses in patients with high calcium scores [8, 27–30].
Regarding the interobserver agreement in image interpretation, coronary MRA showed moderate interobserver agreement among all readers. For coronary MDCTA, only readers 2 and 3 had moderate agreement, and the agreement was poor between readers 1 and 2 and between readers 1 and 3. Consequently, the overall kappa among all readers in the interpretation of MDCTA was only 0.15. However, the overall kappa value among the three readers was significantly higher for coronary MRA than for MDCTA (p = 0.014). This result indicates that coronary MRA allows more reliable image interpretation for evaluating significant coronary stenosis with moderate to severe calcification than MDCTA.
There were several limitations to this study. First, the study population in this preliminary study was small, and a multicenter study is necessary to further support the diagnostic value of coronary MRA for the detection of significant coronary stenosis with moderate and severe calcifications. Second, although an SSFP sequence was used in all studies, the available 3D coverage and spatial resolution of the pulse sequence from the two scanners were different. The impact of such a difference on image interpretation was out of the scope of this study and thus was not compared. Third, the design of site-by-site comparison between coronary MDCTA and MRA may have allowed a bias toward higher sensitivity with coronary MRA. Fourth, due to limited spatial resolution, coronary MRA does not allow the detection of stenoses in distal segments and side branches; thus, only calcifications located in the proximal and middle segments were assessed in our study.
In conclusion, coronary MRA has higher image quality for coronary segments with nodal calcification than that for coronary segments with diffuse calcification. Coronary MRA has better diagnostic performance than coronary MDCTA for the detection of significant stenosis in patients with high calcium scores.
Acknowledgments
We thank Nondas Leloudas, Department of Radiology, Northwestern University,
for help with coronary MRA data acquisition.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  Q. Yang, K. Li, X. Liu, X. Bi, Z. Liu, J. An, A. Zhang, R. Jerecic, and D. Li Contrast-Enhanced Whole-Heart Coronary Magnetic Resonance Angiography at 3.0-T: A Comparative Study With X-Ray Angiography in a Single Center J. Am. Coll. Cardiol., June 30, 2009; 54(1): 69 - 76. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. J. Francois, D. Tuite, V. Deshpande, R. Jerecic, P. Weale, and J. C. Carr Pulmonary Vein Imaging with Unenhanced Three-dimensional Balanced Steady-State Free Precession MR Angiography: Initial Clinical Evaluation Radiology, March 1, 2009; 250(3): 932 - 939. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Zhang, D. C. Levin, E. J. Halpern, D. Fischman, M. Savage, and P. Walinsky Accuracy of MDCT in Assessing the Degree of Stenosis Caused by Calcified Coronary Artery Plaques Am. J. Roentgenol., December 1, 2008; 191(6): 1676 - 1683. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. J. Duerinckx Can the Targeted Use of MR Angiography After CT Angiography Help Assess the Severity of Focal Calcific Coronary Lesions? Am. J. Roentgenol., December 1, 2007; 189(6): 1333 - 1334. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
