| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Institute of Diagnostic Radiology, Department of Medical Radiology, University
Hospital Zurich, Raemistrasse 100, 8091 Zurich, Switzerland.
2 Cardiovascular Center, University Hospital Zurich, Zurich, Switzerland.
3 Center for Integrative Human Physiology, University of Zurich, Zurich,
Switzerland.
Received February 1, 2007;
accepted after revision May 18, 2007.
Supported by the National Center of Competence in Research, Computer Aided
and Image Guided Medical Interventions of the Swiss National Science
Foundation, Switzerland.
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. CT and invasive coronary angiography were performed on 114 patients (mean age, 62 years) referred for known coronary artery disease (n = 26), atypical chest pain (n = 58), and presurgical exclusion of coronary artery disease before abdominal aortic (n = 14) or cardiac valve (n = 16) surgery. The population was divided into two groups depending on median average heart rate (60.0 beats/min) and median heart rate variability (2.7 beats/min) during scanning. Heart rate variability was calculated as SD from the mean heart rate. Two blinded observers using a 4-point scale independently assessed the quality of images of each coronary artery segment and classified each segment as being stenosed (luminal diameter narrowing > 50%) or not. Invasive coronary angiography was used as the reference standard.
RESULTS. In 71 (62.3%) of the patients, 241 significant coronary artery stenoses were identified with invasive coronary angiography. In 11 (9.7%) of the patients, 1.6% (26/1,672) of the segments were not evaluable with CT. Overall sensitivity, specificity, and positive and negative predictive values in a patient-based analysis were 97%, 81%, 90%, and 95%, respectively. Image quality was better (p < 0.05) in the low average heart rate group than in the high average heart rate group, but diagnostic accuracy was comparable for the two groups. In contrast, image quality and diagnostic accuracy were significantly better (p < 0.01) among patients in the low heart rate variability group than in the high heart rate variability group.
CONCLUSION. Lower heart rate variability is associated with higher diagnostic accuracy of 64-MDCT coronary angiography.
Keywords: 64-MDCT • coronary angiography • coronary artery disease • heart rate
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CT Protocol
All examinations were performed with a 64-MDCT scanner (Somatom Sensation
64, Siemens Medical Solutions). The scanning range covered the heart from the
level of tracheal bifurcation to the diaphragm. A bolus of 80 mL of iodixanol
(Visipaque 320, 320 mg/mL, GE Healthcare) followed by 30 mL of saline solution
was injected at a rate of 5 mL/s. Contrast administration was controlled with
bolus tracking. One radiologist placed a region of interest into the aortic
root, and image acquisition was started 5 seconds after attenuation reached
the predefined threshold of 140 H. Data acquisition was performed in a
craniocaudal direction at detector collimation, 32 x 0.6 mm; slice
collimation, 64 x 0.6 mm with a z-flying focal spot; gantry
rotation time, 330 milliseconds; pitch, 0.2; tube potential, 120 kV; effective
tube current-time product, 650 mAs. ECG pulsing for radiation dose reduction
[17] was used in all
patients.
CT Image Reconstruction
The adaptive cardiac volume approach was used for image reconstruction
[18]. Data sets were
retrospectively reconstructed in synchronization with ECG data during the mid-
to end-diastolic phase with reconstruction windows set at 60-70% of the R-R
interval. In the case of insufficient image quality, additional
reconstructions were performed in 5% steps of the R-R interval within the full
tube current window. Images were reconstructed from the CT data sets at slice
thickness, 0.75 mm; reconstruction increment, 0.5 mm; medium soft-tissue
convolution kernel (B30f). The reconstructed field of view was adjusted to
exactly encompass the heart (mean field of view, 149 ± 34 mm; range,
129-182 mm; matrix size, 512 x 512). All images were transferred to a
separate workstation (Wizard, Siemens Medical Solutions) equipped with cardiac
postprocessing software (Syngo Circulation, Siemens Medical Solutions).
CT Data Analysis
Coronary arteries were subdivided according to a 15-segment model proposed
by the American Heart Association
[19], and all segments with a
diameter of at least 1.5 mm at the origin were included. Diameter measurements
were performed with an electronic caliper tool. Axial source images,
multiplanar reformations, and thin-slab maximum intensity projections were
evaluated by two independent blinded readers. First, the two readers
semiquantitatively rated the image quality of each coronary segment on a
4-point scale as follows: 1, excellent (no artifacts, unrestricted
evaluation); 2, good (minor artifacts, good diagnostic quality); 3, adequate
(moderate artifacts but still acceptable and diagnostic); and 4, not
evaluative (severe artifacts impairing accurate evaluation). For each coronary
artery segment, both observers selected the individual optimal reconstruction
interval with the least motion artifacts. Second, both observers visually
estimated all coronary artery segments for the presence of significant
stenoses, defined as more than 50% narrowing of the luminal diameter. In the
event of disagreement at data analysis, consensus was achieved.
A third observer, who was not involved in image quality reading or stenosis assessment, used the recorded ECG information to document the heart rate of each cardiac cycle. For calculation of heart rate variability, the length of each heartbeat during CT acquisition was measured for each patient. The variability of heart rate during scanning was calculated as the SD from the average heart rate from this set of measurements, as previously described [14].
Invasive Coronary Angiography
All patients underwent invasive coronary angiography within 4 weeks after
CT (mean, 10.1 ± 8.9 days; range, 0-26 days). Invasive coronary
angiography was performed according to standard techniques, and multiple views
were stored on a CD-ROM. The angiograms were evaluated by one experienced
observer, who was blinded to the results of CT coronary angiography. Coronary
arteries were subdivided according to the model described for CT data analysis
[19]. With visual estimation,
each vessel segment was scored as being significantly stenosed (diameter
reduction > 50%) or not stenosed.
Statistical Analysis
Quantitative variables were expressed as mean ± SD, and categoric
variables as frequencies or percentages. The median of the average heart rate
and the median of the variability of the heart rate during scanning (i.e., the
SD of the mean heart rate) were used as cutoff points to subdivide the
patients into two average heart rate groups and two heart rate variability
groups. The Fisher's exact test was used to evaluate categoric data
(prevalence of coronary stenosis, sex distribution, and β-blocker use
among heart rate groups and variability groups). Wilcoxon's signed rank test
for paired samples was used to evaluate quantitative parameters stratified
across the heart rate groups (age, BMI, variability of heart rate) and
variability groups (age, BMI, average heart rate) and for comparison of image
quality between the heart rate groups and the heart rate variability groups.
Interobserver agreement for image quality readout and assessment of
significant coronary artery stenosis was calculated with kappa statistics. A
kappa value of 0.80 or higher was considered excellent agreement; between 0.40
and 0.80, moderate to substantial agreement; between 0.20 and 0.40, fair
agreement; and 0.20 or less, slight or poor agreement. Sensitivity,
specificity, positive predictive value, and negative predictive value were
calculated from results of chi-square tests of contingency, and 95% CI was
calculated from binomial expression. Invasive coronary angiography was
considered the standard of reference.
Nonevaluable segments were censored as positive findings in the vessel-based and patient-based analyses, reflecting the intention-to-diagnose nature of the study [20]. The statistics were calculated in segment-based and patient-based analyses (presence of at least one significant coronary artery stenosis or absence of any significant stenosis in each patient). Overall diagnostic accuracy for assessment of coronary artery stenosis was individually calculated for each patient as the sum of true-positive and true-negative findings divided with the individual number of segments per patient. Pearson correlation analysis was performed to compare the overall diagnostic accuracy for each patient with the average heart rate and the SD of the mean heart rate during CT. A value of p < 0.05 indicated a statistically significant difference. All statistical analysis was performed with SPSS statistical software (version 12.0, SPSS).
|
|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
= 0.75).
With use of the individual reconstruction interval with the best image quality
for each segment, image quality was graded excellent (score 1) for 45.9%
(768/1,672) of the coronary artery segments, good (score 2) for 37.8%
(633/1,672), and adequate (score 3) for 14.7% (245/1,672). Image quality was
graded not evaluative (score 4) for 1.6% (26/1,672) of the segments (six right
coronary artery segments, six left anterior descending coronary artery
segments, 14 left circumflex artery segments) even when the best individual
reconstruction interval was used. As many as five nonevaluable coronary artery
segments (mean, 2.4 ± 1.3 segments; range, 1-5 segments) were present
in 9.7% (11/114) of the patients. In patients who had nonevaluable segments,
mean image quality score was 2.6 ± 0.5 (range, 1.93-3.21), average
heart rate was 63.9 ± 12.5 beats/min (range, 50-82 beats/min), and
variability of heart rate was 8.1 ± 4.2 beats/min (range, 2.7-15.5
beats/min).
Overall Diagnostic Accuracy in Assessment of Coronary Artery Stenosis
A total of 241 coronary artery stenoses with luminal diameter narrowing
greater than 50% were identified with invasive coronary angiography in 71
(62.3%) of the patients. Single-vessel disease was present in 14.0% (16/114),
two-vessel disease in 14.0% (16/114), and three-vessel disease in 34.3%
(39/114) of the patients. Significant coronary artery stenosis was absent in
37.7% (43/114) of the patients. The kappa value for detection of coronary
artery stenosis with CT was 0.86, indicating excellent interobserver
agreement. CT coronary angiography correctly depicted 218 of 241 significant
stenoses detected with invasive coronary angiography (Figs.
1A and
1B). Among the 23 lesions
missed by both readers, image quality was rated other than excellent for 20
(87.0%) of the segments because of motion artifacts. In 48 segments, lesions
were incorrectly graded as stenotic on CT. If all 26 nonevaluable segments are
considered false-positive findings, a total of 74 false-positive ratings were
made. Thus the overall sensitivity of CT coronary angiography in segment-based
analysis was 90.5% (218/241; 95% CI, 86.0-93.9), the specificity was 94.8%
(1,357/1,431; 95% CI, 93.6-95.9), the positive predictive value was 74.7%
(218/292; 95% CI, 69.3-79.5), and the negative predictive value was 98.3%
(1,357/1,380; 95% CI, 97.5-98.9).
CT correctly depicted at least one significant coronary artery stenosis in 97.2% (69/71) of patients with significant CAD on invasive coronary angiography; the diagnosis was missed in two patients (2.8%). In six (5.3%) of the 114 patients, CAD was not found at invasive coronary angiography, but CT evidence suggested the presence of significant stenosis. False ratings with CT led to the diagnosis of multivessel disease instead of one-vessel disease in 10 patients. The misdiagnosis of one-vessel disease was made with CT in the cases of two patients in whom multivessel disease was found on invasive coronary angiography. In two of the 11 patients with not completely evaluative CT angiograms, coronary stenosis was absent on invasive coronary angiography, and therefore the findings had to be considered false-positive (Figs. 2A and 2B). In the other nine patients with nonevaluable segments, coronary artery stenosis was identified in at least one evaluable segment, correctly indicating the need for invasive coronary angiography. In patient-based analysis, the overall sensitivity of CT coronary angiography was 97.2% (69/71; 95% CI, 90.2-99.7), the specificity was 81.4% (35/43; 95% CI, 66.6-91.6), the positive predictive value was 89.6% (69/77; 95% CI, 80.6-95.4), and the negative predictive value was 94.6% (35/37; 95% CI, 81.8-99.3).
|
|
Diagnostic Accuracy in Relation to Average Heart Rate
The median of the average heart rate among all 114 patients was 60.0
beats/min; thus 57 (50.0%) of the patients had a heart rate of 60.0 beats/min
or less (mean, 52.4 ± 4.7 beats/min), and 57 (50.0%) a heart rate
greater than 60.0 beats/min (mean, 71.8 ± 9.8 beats/min). There were no
statistically significant differences in patient age (p = 0.92), BMI
(p = 0.66), or heart rate variability (p = 0.77), but there
were significantly (p < 0.05) fewer women in the low average heart
rate group than in the high average heart rate group. There was no significant
difference in use of β-blocker medication for the two heart rate
subgroups (p = 0.24). Tables
1 and
2 summarize the demographic
data, image quality scoring results, and diagnostic accuracy in the two heart
rate subgroups.
|
|
With use of the individual reconstruction interval with best image quality for each coronary artery segment, mean image quality was significantly better in the cases of patients with heart rates of 60.0 beats/min or less (score, 1.62 ± 0.60) than in those of patients with a heart rate greater than 60.0 beats/min (score, 1.82 ± 0.55) (p < 0.05). In contrast, the percentages of nonevaluable segments were comparable for the two heart rate subgroups at 1.4% (12/844) and 1.7% (14/828).
Significant coronary artery stenosis was present in 20.4% (172/844) of segments in patients with a heart rate of 60.0 beats/min or less and in 8.3% (69/828) of segments in patients with a heart rate greater than 60.0 beats/min (p < 0.05). At a heart rate of 60.0 beats/min or less, 15 false-negative and 41 false-positive ratings (29 because of misclassification, 12 because of inclusion of nonevaluable segments) were made with CT. At a heart rate greater than 60.0 beats/min, eight false-negative and 33 false-positive ratings (19 because of misclassification, 14 because of inclusion of nonevaluable segments) were made with CT. There was no tendency toward decreased diagnostic accuracy in association with elevated heart rate in per-segment or per-patient analysis (Tables 1 and 2). Pearson correlation analysis revealed no significant correlation between average heart rate and overall diagnostic accuracy per patient (r = -0.05, p = 0.79) (Fig. 3).
|
|
|
|
2.7 beats/min), only one (0.1%) of 841 segments was considered
nonevaluable. Greater heart rate variability (> 2.7 beats/min) was
associated with a higher rate (3.0% [25/831]) of nonevaluable coronary
segments.
As identified with invasive coronary angiography, significant coronary
artery stenosis was present in 16.4% (138/841) of the segments in patients
with a heart rate variability of 2.7 beats/min or less, whereas 12.4%
(103/831) of the segments were stenosed in patients with a heart rate
variability greater than 2.7 beats/min (p = 0.09). In the lower
variability group, seven false-negative and 15 false-positive ratings (14
because of misclassification, one because of inclusion of nonevaluable
segments) were made with CT. At a heart rate variability greater than 2.7
beats/min, 16 false-negative and 59 false-positive ratings (34 because of
misclassification, 25 because of inclusion of nonevaluable segments) were made
with CT. The highest diagnostic accuracy was observed in the group with
less-variable heart rhythm ( 2.7 beats/min) at a sensitivity of 94.9%,
specificity of 97.9%, positive predictive value of 89.7%, and negative
predictive value of 99.0% in per-segment analysis. With heart rate variability
greater than 2.7 beats/min, there was substantial deterioration in accuracy,
sensitivity declining to 84.5%, specificity to 92.0%, and positive predictive
value to 59.6%; negative predictive value remained high at 97.7%. Similar
results were obtained in a patient-based analysis. Sensitivity decreased from
100% to 94.1%, specificity from 85.0% to 78.3%, positive predictive value from
92.5% to 86.5%, and negative predictive value from 100% to 90.0% among
patients with heart rate variability greater than 2.7 beats/min compared with
the lower heart rate variability group. Pearson correlation analysis revealed
a significant correlation between heart rate variability during scanning and
diagnostic accuracy per patient (r = -0.61, p < 0.01)
(Fig. 4).
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
In our study, the overall sensitivity of 97% was comparable with that in previous 64-MDCT studies [1-7]; the specificity of 81% was lower than in other studies. This finding can be explained by our approach of including nonevaluable coronary segments as false-positive ratings on an intent-to-diagnose basis [20]. In contrast, in the group of patients with a regular heart rate, the sensitivity of 91% and specificity of 94% were within the range of those in previous 64-MDCT studies [1-7].
The most important finding in our study is that regularity of heart rate during scanning significantly correlates with diagnostic accuracy per patient. An investigation of the influence of average heart rate and heart rate variability on image quality [14] showed that an average heart rate had no significant effect on image quality whereas a variable heart rate during scanning significantly degraded visualization of coronary arteries. In addition, there was only a minor difference in the number of nonevaluable segments between the groups with low and high heart rates. This finding was most probably due to the fact that heart rate variability was not increased in patients with high heart rates.
With intercycle variability of heart rate, the commonly used relative ECG-gated image reconstruction technique (reconstructions at a certain percentage of the R-R interval) does not generate images in exactly corresponding cardiac phases. The functionally different parts within one cardiac cycle are shortened or prolonged with different degrees of lack of proportion of the heart rate [23]. We used an optional two-segment reconstruction algorithm. This algorithm was initially designated to compensate with increased temporal resolution for coronary motion at a higher heart rate. However, in two- or multiple-segment reconstruction algorithms, periodic motion over n consecutive cardiac cycles is assumed.
Inconsistencies between heart cycles due to arrhythmia result in merging of slightly different cardiac phases despite acquisition of data from consecutive cardiac cycles at the same percentage in relation to the R wave. This merging of data introduces blurring of the reconstructed image. For images reconstructed with monosegment techniques, acquisition of data from slightly different cardiac phases results in a timing shift from image to image, giving 2D or 3D reformats a stair-step appearance. However, for images reconstructed with multisegment techniques, the timing shift is incorporated into individual images, reducing the stair-step appearance of reformats at the expense of increased blurring within individual images [24]. Thus irregular heart rate has resulted in impairment of image quality in an increasing number of nonevaluable segments. It also has resulted in a decline of diagnostic accuracy for stenosis detection.
Besides the heart rate decreasing effect of β-blockers, the major mechanism of β-blockers in improving the image quality of CT coronary angiography is considered its ability to reduce heart rate variability [14]. Consequently, administration of β-blockers can be recommended for imaging of patients with only slightly increased but highly irregular heart rates. In our study, low heart rate variability, best image quality, lowest rate of nonevaluable segments, and highest diagnostic accuracy were found in the group of patients taking β-receptor antagonists.
There were several limitations to this study. First, the patient cohort was heterogeneous, including patients with suspected and known CAD and who had undergone coronary surgery or intracoronary stenting. Second, the degree of variability in heart rate was higher among patients with a heart rate less than 60 beats/min than in the high heart rate group. This effect was most probably due to the indications for β-blocker treatment or to the individual response to β-blocker medication in our population [25]. In addition, the prevalence of coronary stenosis differed in the two mean heart rate subgroups, which might have affected the sensitivity and negative predictive value. Third, stenosis assessment with CT coronary angiography was restricted to visual estimation; the degree of stenosis was not quantified. Fourth, factors such as the influence of intracoronary attenuation of contrast agent on the diagnostic accuracy of CT coronary angiography [26] were not studied. Fifth, we did not use the most recent CT scanner technology to assess the effect of heart rate variables on diagnostic accuracy [27]. Sixth, because of the lack of a reference standard, we did not assess the effect of heart rate variability on identifying and characterizing coronary plaques. Finally, we did not use the absolute ECG gating technique to investigate the effect of heart rate parameters on the image quality of CT coronary angiography. It has been suggested [28] that this technique is advantageous in imaging of patients with variable heart rates.
The diagnostic accuracy of 64-MDCT coronary angiography in the detection of significant coronary artery stenosis is high but decreases in imaging of patients with an irregular heart rate. Regularity of heart rate appears to be more important than average heart rate for optimal image quality and high diagnostic accuracy. The results of this study support the administration of β-receptor antagonists to imaging patients with irregular heart rates, even when the average heart rate is relatively low. In contrast, patients with high but regular heart rates can undergo CT coronary angiography without premedication, and the image quality will be diagnostic for identification of coronary artery stenosis.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
