Clinical Radiology 69 (2014) 163e171
Contents lists available at SciVerse ScienceDirect
Clinical Radiology journal homepage: www.clinicalradiologyonline.net
Diagnostic performance of dual-source CT coronary angiography with and without heart rate control: Systematic review and meta-analysis M. Li a, f, G.-M. Zhang b, f, J.-S. Zhao c, Z.-W. Jiang d, Z.-H. Peng a, Z.-T. Jin e, G. Sun a, * a
Department of Medical Imaging, Jinan Military General Hospital, Jinan, Shandong Province, China Department of Medical Cardiology, Jinan Military General Hospital, Jinan, Shandong Province, China c Department of Radiology, Qilu Children’s Hospital of Shandong University, Jinan, Shandong Province, China d Department of Health Statistics, School of Public Health, Fourth Military Medical University, Xi’an, Shaanxi, China e Department of Cardiology, General Hospital of the Second Artillery, Beijing, China b
art icl e i nformat ion Article history: Received 28 June 2013 Received in revised form 5 September 2013 Accepted 9 September 2013
AIM: To investigate the diagnostic accuracy of dual-source computed tomography (DSCT) coronary angiography with and without the application of a b-blocker. MATERIALS AND METHODS: An exact binomial rendition of the bivariate mixed-effects regression model was used to synthesize diagnostic test data. RESULTS: The pooled sensitivity at the patient level was 0.98 [95% confidence intervals (CI): 0.97e0.99], and specificity 0.88 (95% CI: 0.84e0.91). The results showed that without heart rate control, the sensitivity and specificity at the patient level did not decrease (p ¼ 0.27 and 0.56, respectively). At the artery level, no significant differences in sensitivity and specificity for studies with and without heart rate control were detected (p ¼ 0.04 and 0.05, respectively). At the segment level, the specificity decreased without heart rate control (p ¼ 0.03), whereas the sensitivity was not influenced (p ¼ 0.63). The median radiation exposure was 2.6 mSv, with 1.6 mSv and 8 mSv for heart rate-controlled studies and uncontrolled studies, respectively. CONCLUSIONS: DSCT coronary angiography without heart rate control has a similar excellent diagnostic performance at the patient level as that of heart rate control groups. However, controlling for heart rate to decrease radiation and to provide effective information for selecting the therapeutic strategy and risk stratification is recommended. Ó 2013 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.
Introduction Coronary artery disease (CAD) is a major cause of mortality and ill health.1 A precise evaluation of the presence of stenotic CAD is required for diagnosis and treatment. * Guarantor and correspondent: G. Sun, Department of Medical Imaging, Jinan Military General Hospital, No, 25, Shifan Road, Jinan, Shandong Province 250031, China. Tel.: þ86 531 51666277; fax: þ86 531 51666486. E-mail address: [email protected] (G. Sun). f The authors contributed equally to this work.
Although diagnostic invasive coronary angiography (ICA) can directly and precisely visualize the stenotic lumen of the coronary artery, it is an invasive and expensive examination, accompanied by the risk of major complications (0.1e0.2%).2 Therefore, a non-invasive test for the assessment of the coronary arteries is highly desirable. With the development of multidetector computed tomography (MDCT), CT angiography (CTA) has been applied widely as an clinical tool for evaluating CAD.3 However, motion artefact caused by rapid movement of the heart is a technical problem influencing its diagnostic performance.
0009-9260/$ e see front matter Ó 2013 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.crad.2013.09.008
164
M. Li et al. / Clinical Radiology 69 (2014) 163e171
With the emergence of 64-MDCT, CTA has been proven as a reliable imaging tool in assessing suspected CAD;4e6 however, the temporal resolution of 64-section CT is still limited. A low heart rate is generally required by applying a b-blocker in patients with a heart rate of >70 beats/min.4 With the improved temporal resolution of dual-source CT (DSCT), some studies have concluded that it has sufficient accuracy to evaluate suspected CAD without the application of a b-blocker.7e11 However, to date, no metaanalysis has systematically reviewed the clinical performance of DSCT angiography without heart rate control to verify this. The present study reviews the literature and evaluates the diagnostic performance of coronary DSCT angiography without a b-blocker, using DSCT angiography with heart rate control via a b-blocker as the reference.
Meta-Analyses (PRISMA) statement12 and the Standards for Reporting of Diagnostic Accuracy (STARD) statement.13
Materials and methods
Study eligibility
The present analysis was conducted according to the Preferred Reporting Items for Systematic Reviews and
The inclusion criteria were as follows: DSCT was applied as the diagnostic test for stenotic CAD; results were
Selection of studies PubMed and EMBASE were searched for all published studies that evaluate the diagnostic accuracy of DSCT angiography. The language was limited to English. The search terms were as follows: “computed tomography,” “multislice computed tomography,” “spiral computed tomography,” “multisection computed tomography,” “multidetector computed tomography,” “coronary angiography,” “coronary vessels,” and “coronary artery disease.” The literature search ranged from 2004 to 2012. The reference lists of published systematic reviews and meta-analyses were also screened. Two investigators examined the studies to exclude potential duplicate or overlapping data.
Figure 1 Flow chart of evidence search and selection.
M. Li et al. / Clinical Radiology 69 (2014) 163e171 Table 1 Characteristics of included studies.
bECGMale/ Mean No. of HR female age gating sections (bpm) blocker (years)
First author
Year
Stolzmann et al.30 Moon et al.31 Achenbach et al.32 Bamberg et al.8 Yang et al.33 Xu et al.9 Tsiflikas et al.11 Scheffel et al.10 Scheffel et al.34 Ovrehus et al.35 Lin et al.36 Fang et al.37 Donati et al.38 Chen et al.39 Alkadhi et al.29 Alkadhi et al.29 Rixe et al.40 Reimann et al.7 Meng et al.41 Leschka et al.42 Stolzmann et al.43 Scheffel et al.44 Ravipati et al.45 Achenbach et al.46 Leschka et al.47 Brodoefel et al.48 Alkadhi et al.49 Leber et al.50 Weustink et la.51 Ropers et al.52 Johnson et al.53 Heuschmid et al.54 Scheffel et al.55
2011 1.4
68
P
64
58
Yes
2011 1.9
60.5
R
64
58.9
Yes
2011 2.1
59
P
128
71
Yes
2011 .
68.1
P
128
72.2
No
2010 3.6
62
.
64
59
Yes
2010 2.2 2010 2.7
64.6 64
R R
64 64
71.4 64
No No
2010 2.5
56
.
64
67.3
No
2010 3.8
64
P
64
61
No
2010 1.3
61
.
64
61
No
2010 3.4 2010 1.8 2010 4.2
61.2 59.6 64
R R R
64 64 64
66.9 88 .
No No No
2010 1.6
60.7
.
64
86.4
No
2010 3.2
62
P
128
58
Yes
2010 2.6
63
P
128
56
Yes
2009 1.6
65
R
64
68
No
2009 2.8
62
R
64
62.7
No
2009 1.7
63
.
64
71.8
No
2009 4.0
62
P
128
58
Yes
2008 0.7
65.8
P
64
60.7
Yes
2008 1.4
68
P
64
59
Yes
2008 1.9
66
.
64
.
Yes
2008 1.8
61
.
64
64.3
Yes
2008 2.1
61.9
R
64
67.7
No
2008 4.0
62
R
64
64.9
No
2008 2.2
62.9
R
64
68.5
No
2007 1.7
58
R
64
.
Yes
2007 3.8
61
R
64
68
No
2007 1.7
61
R
64
64
No
2007 2.2
60
R
64
68
No
2007 2.6
64
R
64
65
No
2006 4.0
63.1
R
64
70.3
No
165
presented in absolute numbers of true-positive, false-positive, true-negative, and false-negative results or the numbers could be calculated using the detailed data; ICA was used as the reference standard for the diagnosis of CAD; the cut-off criterion for diagnosis of CAD was >50% diameter stenosis. Studies were excluded if they included patients who had undergone coronary artery bypass graft surgery, percutaneous coronary intervention, or prior heart transplantation. Studies including patients with atrial fibrillation (AF) were also excluded.
Data extraction The following information was extracted from each study: first author, year of publication, country, number of patients, gender, age, heart rate, number of detector sections, ECG-gated protocol, radiation dose, usage of bblocker, and number of true-positive, true-negative, falsepositive, and false-negative values. True positive was defined as significant stenosis on a DSCT angiography confirmed by ICA, and true negative was defined as a negative DSCT angiographic finding proven by ICA. If stenosis at ICA was not detected using DSCT angiography, it was classified as false negative, and if a negative finding at ICA presented as a stenosis at CTA, it was classified as a false positive. Data at the level of segments, arteries, and patients were recorded separately. Study quality was evaluated according to Quality Assessment of Diagnostic Accuracy Studies (QUADAS).14 Two readers, who were not involved in any of the included studies, independently evaluated quality scores.
Data synthesis and statistical analysis
ECG, electrocardiograph; R, retrospective; P, prospective; HR, heart rate; TP, true positive; FP, false positive; FN, false negative; TN, true negative.
A Cohen k-test was conducted to test the interobserver agreement. A k-value >0.80 is considered good to excellent agreement. Publication bias was assessed using a regression line developed by Deeks et al. (p < 0.10 indicates significant asymmetry).15 The Cochran Q statistic and measured inconsistency test (I2) was applied to assess heterogeneity across studies.16 Meta regression was conducted to compare the diagnostic accuracy of DSCT coronary angiography with and without heart rate control. Diagnostic test data was synthesized using an exact binomial rendition of the bivariate mixed-effects regression model, which fits a two-level model, with independent binomial distributions for true-positive and true-negative conditionals and a bivariate normal model for the logit transforms of sensitivity and specificity between the included studies.17e19 The sensitivity, specificity, and 95% confidence intervals (CIs) were calculated. Positive and negative likelihood ratios (LRs) were also computed to eliminate the influence of the prevalence of disease. The standard formulas were demonstrated as follows: positive likelihood ratio ¼ sensitivity/(1especificity) and negative likelihood ratio ¼ specificity/ (1esensitivity). The post-test probability was calculated using likelihood ratios based on
166
M. Li et al. / Clinical Radiology 69 (2014) 163e171
Results
Bayes’ theorem to investigate the clinical utility of DSCT angiography.20 A hierarchical summary receiver operating characteristic (sROC) curve was constructed based on parameters estimated by the bivariate model, an area under the ROC curve (AUROC) serves as a global measure of test performance (low: 0.5 AUROC 0.7, moderate: 0.7 AUROC 0.9, or high: 0.9 AUROC 1).21 The sensitivity analysis was conducted by omitting each reference and re-analysing the data to test whether there were some studies influencing the final results significantly. The basic characteristics between references with and without heart rate control were compared using ManneWhitney U and c2 tests. STATA, version 12 (StataCorp, College Station, TX, USA) with MIDAS module, and SPSS, version 16.0 (SPSS, Chicago, Illinois), were used to analyse the data and construct the graphs.
Characteristics of studies A total of 804 potentially relevant references were identified. After screening for title and abstract, 135 studies were retrieved. After selecting full-text articles according to the inclusion and exclusion criteria mentioned above, 33 studies were included in the review. Fig 1 shows the flow chart of the literature search and selection process. The number of included patients ranged from 25 to 210 (median 82 patients). Twenty-eight studies reported using a 64 section DSCT and five using a 128-section DSCT system. b-Blockers were administered before the examination in 12 of the studies, during which seven studies applied prospective ECG gating, two studies applied retrospective ECG gating, and three studies were unclear. Twenty-one studies did not control heart rate using b-blockers, during which two studies
Table 2 Quality assessment of diagnostic accuracy studies. First author
Item 1
Item 2
Item 3
Item 4
Item 5
Item 6
Item 7
Item 8
Item 9
Item 10
Item 11
Item 12
Item 13
Item 14
Stolzmann et al.30 Moon et al.31 Achenbach et al.32 Bamberg et al.8 Yang et al.33 Xu et al.9 Tsiflikas et al.11 Scheffel et al.10 Scheffel et al.34 Ovrehus et al.35 Lin et al.36 Fang et al.37 Donati et al.38 Chen et al.39 Alkadhi et al.29 Rixe et al.40 Reimann et al.7 Meng et al.41 Leschka et al.42 Stolzmann et al.43 Scheffel et al.44 Ravipati et al.45 Achenbach et al.46 Leschka et al.47 Brodoefel et al.48 Alkadhi et al.49 Leber et al.50 Weustink et al.51 Ropers et al.52 Johnson et al.53 Heuschmid et al.54 Scheffel et al.55
Y Y Y Y Y Y Y N Y Y Y Y Y Y Y U U Y Y Y Y U Y Y Y Y U N Y Y N Y
Y Y Y Y Y N Y Y Y Y Y Y Y Y Y Y U Y Y Y Y N Y Y Y Y Y Y Y Y Y Y
Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y
Y U Y U U U Y U Y U Y Y Y Y Y U Y Y Y Y Y U U Y Y Y Y Y Y Y Y Y
Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y
Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y
Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y
Y U Y Y Y Y Y Y Y Y Y Y Y Y Y N Y Y Y Y Y U Y Y Y Y Y Y Y Y Y Y
Y Y N Y N Y N Y Y Y N N Y Y N Y N Y Y N U U N N Y N N N N U N N
Y Y Y Y U U Y U Y Y Y Y Y Y Y Y U Y Y Y Y U Y Y Y U U Y Y Y U U
Y Y Y U U Y Y Y Y Y Y Y Y Y Y N U U Y Y Y U U Y Y Y U U U U U Y
N U Y N U Y U U N N N U N U N N U N N N U U U N N N U U N U U U
Y N Y Y N N N N N N U Y Y U Y Y U Y U Y Y N Y N Y Y Y Y U Y Y Y
N Y Y Y Y N Y Y Y Y Y Y Y Y Y Y U N Y Y Y N N Y Y N N Y N N N N
The quality assessment of the studies was performed with a QUADAS tool. Items 1e14 represent the following questions, which were part of the tool. Item 1: Was the spectrum of patients representative of the patients who will receive the test in practice? Item 2: Were selection criteria clearly described? Item 3: Is the reference standard likely to correctly classify the target condition? Item 4: Is the time period between reference standard and index test short enough to be reasonably sure that the target condition did not change between the two tests? Item 5: Did the entire sample or a random selection of the sample receive verification using a reference standard of diagnosis? Item 6: Did patients receive the same reference standard regardless of the index test result? Item 7: Was the reference standard independent of the index test (i.e., the index test did not form part of the reference standard)? Item 8: Was the execution of the index test described in sufficient detail to permit replication of the test? Item 9: Was the execution of the reference standard described in sufficient detail to permit its replication? Item 10: Were the index test results interpreted without knowledge of the results of the reference standard? Item 11: Were the reference standard results interpreted without knowledge of the results of the index test? Item 12: Were the same clinical data available when test results were interpreted as would be available when the test is used in practice? Item 13: Were uninterpretable/intermediate test results reported? Item 14: Were withdrawals from the study explained? [14].
M. Li et al. / Clinical Radiology 69 (2014) 163e171
167
Figure 2 Forest plots showing sensitivity and specificity at the patient level.
applied prospective ECG gating, 15 studies applied retrospective ECG gating, and four studies were unclear. There was statistical differences of the imaging protocol between the two groups (p ¼ 0.004). The mean heart rates of b-blocker references ranged between 56 beats/min and 71 beats/min (median, 58.5 beats/min), which was significantly lower than that of non-b-blocker references [ranged between 61 beats/ min and 88 beats/min [median, 67 beats/min], p < 0.001]. No statistical differences were found for other characteristics (gender, p ¼ 0.22; age, p ¼ 0.81; detector, p ¼ 0.05). The average estimated effective radiation dose ranged from 0.76 to 16.1 mSv. The median radiation exposure was 2.6 mSv, with 1.6 and 8 mSv for heart rate-controlled studies and uncontrolled studies, respectively. The baseline characteristics of the study and the detailed diagnostic information are shown in Table 1.
Methodological quality The k-value of the interobserver agreement for assessing quality items was 0.85. Table 2 presents the quality score of
the included studies according to the QUADAS items. Overall, the qualities of enrolled study were moderate, with 20 of the included studies fulfilling 10 or more of the 14 criteria. However, none of them fulfilled all the items of the QUADAS tool.
Data synthesis and statistical analysis A total of 31,324 segments, 4697 vessels, and 1750 patients were analysed. No publication bias was significant at the patient, vessel, or segment level (p ¼ 0.10, 0.28, and 0.67, respectively). The pooled sensitivity at patient level 0.98 (95% CI: 0.97e0.99) and the pooled specificity was 0.88 (95% CI: 0.84e0.91; Fig 2). The pooled positive and negative LRs were 8.04 (95% CI: 6.17e10.48) and 0.02 (95% CI: 0.01e0.03), respectively. Table 3 shows that the sensitivity increased from segment, to vessel, and to patient level. In contrast, the specificity and positive and negative LRs were lowest at the patient level.
Table 3 Pooled summary results. TP
Total Patient-level Artery-level Segment-level HR controlled Patient-level Artery-level Segment-level HR uncontrolled Patient-level Artery-level Segment-level
FP
FN
TN
Sensitivity
Specificity
Positive LR
Negative LR
AUROC
(95%CI)
(95%CI)
(95% CI)
(95% CI)
(95% CI)
953 1307 3819
98 204 969
17 42 275
682 3144 26,261
0.98 (0.97e0.99) 0.97 (0.96e0.98) 0.95 (0.92e0.97)
0.88 (0.84e0.91) 0.94 (0.92e0.96) 0.97 (0.96e0.98)
8.04 (6.17e10.48) 16.39 (11.49e23.36) 35.19 (25.86e47.88)
0.02 (0.01e0.03) 0.03 (0.02e0.05) 0.05 (0.04e0.08)
0.99 (0.98e1.00) 0.99 (0.98e1.00) 0.99 (0.98e1.00)
494 537 1347
49 61 223
6 8 73
377 1428 10,711
0.99 (0.97e0.99) 0.99 (0.97e0.99) 0.95 (0.92e0.97)
0.89 (0.84e0.93) 0.96 (0.95e0.97) 0.98 (0.97e0.99)
8.89 (6.08e13.01) 24.01 (18.69e30.84) 55.35 (34.83e87.95)
0.01 (0.01e0.04) 0.02 (0.01e0.04) 0.05 (0.03e0.08)
0.99 (0.98e1.00) 0.98 (0.96e0.99) 0.99 (0.98e1.00)
459 770 2472
49 143 746
11 34 202
305 1716 15,550
0.98 (0.96e0.99) 0.96 (0.94e0.98) 0.94 (0.91e0.97)
0.86 (0.81e0.90) 0.92 (0.87e0.95) 0.96 (0.95e0.98)
7.16 (5.08e10.09) 11.79 (7.12e19.50) 26.54 (18.76e37.54)
0.03 (0.02e0.05) 0.04 (0.03e0.07) 0.06 (0.03e0.10)
0.98 (0.97e0.99) 0.98 (0.97e0.99) 0.99 (0.97e0.99)
TP, true-positive; FP, false-positive; TN, true-negative; FN, false-negative; LR, likelihood ratio; AUROC, area under the summary receiver operating characteristic curve; HR, heart rate.
168
M. Li et al. / Clinical Radiology 69 (2014) 163e171
Figure 3 Symmetric sROC at the patient level for accuracy of DSCT angiography for studies with and without heart rate control.
The application of a b-blocker was added as a factor for meta-regression analysis to compare the diagnostic performance for heart rate-controlled studies and heart rateuncontrolled studies. The results showed that without heart rate control the sensitivity and specificity required to detect significant stenosis at the patient level did not decrease (p ¼ 0.27 and 0.56, respectively). At the artery level, the sensitivity and the specificity of studies without heart rate control was similar to that with heart rate control (p ¼ 0.04 and 0.05, respectively). At the segment level, the specificity decreased without heart rate control (p ¼ 0.03), whereas sensitivity was not influenced (p ¼ 0.63). The AUROC was 0.99 (95% CI: 0.98e0.99) for the heart ratecontrolled group and 0.98 (95% CI: 0.97e0.99) for the heart rate-uncontrolled group on the patient level, as shown in Fig 3. Post-test probabilities were calculated with and without heart rate control according to the Bayes’ theorem to investigate the clinical utility of DSCT angiography (see Fig 4). The plot shows that coronary DSCT angiography with
and without the application of a b-blocker has perfect negative predictive value in patients with a low to intermediate pretest likelihood of CAD, whereas the positive predictive value was moderate. Heterogeneity between included studies was evaluated at the patient, artery, and segment levels. At the patient level, no statistically heterogeneity was found for sensitivity (Q ¼ 29.33; p ¼ 0.11; I2 ¼ 28.39%; 95% CI: 0.00e65.73%), whereas it was significant for specificity (Q ¼ 45.78; p < 0.001; I2 ¼ 54.13%; 95% CI: 31.99e76.26%). Conversely, heterogeneity was significant at the artery level for both sensitivity (Q ¼ 46.00; p < 0.001; I2 ¼ 63.04%; 95% CI: 44.3e81.78%) and specificity (Q ¼ 112.51; p < 0.001; I2 ¼ 84.89%; 95% CI: 78.84e90.94%) and at the segment level for both sensitivity (Q ¼ 503.37; p < 0.001; I2 ¼ 94.44%; 95% CI: 93.14e95.73%) and specificity (Q ¼ 683.01; p < 0.001; I2 ¼ 95.9%; 95% CI: 95.03e96.77%). Sensitivity analysis was performed at the patient level to investigate the influence of each study on the overall pooled results. Fig 5 indicates that no study influenced the pooled sensitivity and specificity by >0.03. Further analysis demonstrated that after the exclusion of six studies with a quality score of
Contents lists available at SciVerse ScienceDirect
Clinical Radiology journal homepage: www.clinicalradiologyonline.net
Diagnostic performance of dual-source CT coronary angiography with and without heart rate control: Systematic review and meta-analysis M. Li a, f, G.-M. Zhang b, f, J.-S. Zhao c, Z.-W. Jiang d, Z.-H. Peng a, Z.-T. Jin e, G. Sun a, * a
Department of Medical Imaging, Jinan Military General Hospital, Jinan, Shandong Province, China Department of Medical Cardiology, Jinan Military General Hospital, Jinan, Shandong Province, China c Department of Radiology, Qilu Children’s Hospital of Shandong University, Jinan, Shandong Province, China d Department of Health Statistics, School of Public Health, Fourth Military Medical University, Xi’an, Shaanxi, China e Department of Cardiology, General Hospital of the Second Artillery, Beijing, China b
art icl e i nformat ion Article history: Received 28 June 2013 Received in revised form 5 September 2013 Accepted 9 September 2013
AIM: To investigate the diagnostic accuracy of dual-source computed tomography (DSCT) coronary angiography with and without the application of a b-blocker. MATERIALS AND METHODS: An exact binomial rendition of the bivariate mixed-effects regression model was used to synthesize diagnostic test data. RESULTS: The pooled sensitivity at the patient level was 0.98 [95% confidence intervals (CI): 0.97e0.99], and specificity 0.88 (95% CI: 0.84e0.91). The results showed that without heart rate control, the sensitivity and specificity at the patient level did not decrease (p ¼ 0.27 and 0.56, respectively). At the artery level, no significant differences in sensitivity and specificity for studies with and without heart rate control were detected (p ¼ 0.04 and 0.05, respectively). At the segment level, the specificity decreased without heart rate control (p ¼ 0.03), whereas the sensitivity was not influenced (p ¼ 0.63). The median radiation exposure was 2.6 mSv, with 1.6 mSv and 8 mSv for heart rate-controlled studies and uncontrolled studies, respectively. CONCLUSIONS: DSCT coronary angiography without heart rate control has a similar excellent diagnostic performance at the patient level as that of heart rate control groups. However, controlling for heart rate to decrease radiation and to provide effective information for selecting the therapeutic strategy and risk stratification is recommended. Ó 2013 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.
Introduction Coronary artery disease (CAD) is a major cause of mortality and ill health.1 A precise evaluation of the presence of stenotic CAD is required for diagnosis and treatment. * Guarantor and correspondent: G. Sun, Department of Medical Imaging, Jinan Military General Hospital, No, 25, Shifan Road, Jinan, Shandong Province 250031, China. Tel.: þ86 531 51666277; fax: þ86 531 51666486. E-mail address: [email protected] (G. Sun). f The authors contributed equally to this work.
Although diagnostic invasive coronary angiography (ICA) can directly and precisely visualize the stenotic lumen of the coronary artery, it is an invasive and expensive examination, accompanied by the risk of major complications (0.1e0.2%).2 Therefore, a non-invasive test for the assessment of the coronary arteries is highly desirable. With the development of multidetector computed tomography (MDCT), CT angiography (CTA) has been applied widely as an clinical tool for evaluating CAD.3 However, motion artefact caused by rapid movement of the heart is a technical problem influencing its diagnostic performance.
0009-9260/$ e see front matter Ó 2013 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.crad.2013.09.008
164
M. Li et al. / Clinical Radiology 69 (2014) 163e171
With the emergence of 64-MDCT, CTA has been proven as a reliable imaging tool in assessing suspected CAD;4e6 however, the temporal resolution of 64-section CT is still limited. A low heart rate is generally required by applying a b-blocker in patients with a heart rate of >70 beats/min.4 With the improved temporal resolution of dual-source CT (DSCT), some studies have concluded that it has sufficient accuracy to evaluate suspected CAD without the application of a b-blocker.7e11 However, to date, no metaanalysis has systematically reviewed the clinical performance of DSCT angiography without heart rate control to verify this. The present study reviews the literature and evaluates the diagnostic performance of coronary DSCT angiography without a b-blocker, using DSCT angiography with heart rate control via a b-blocker as the reference.
Meta-Analyses (PRISMA) statement12 and the Standards for Reporting of Diagnostic Accuracy (STARD) statement.13
Materials and methods
Study eligibility
The present analysis was conducted according to the Preferred Reporting Items for Systematic Reviews and
The inclusion criteria were as follows: DSCT was applied as the diagnostic test for stenotic CAD; results were
Selection of studies PubMed and EMBASE were searched for all published studies that evaluate the diagnostic accuracy of DSCT angiography. The language was limited to English. The search terms were as follows: “computed tomography,” “multislice computed tomography,” “spiral computed tomography,” “multisection computed tomography,” “multidetector computed tomography,” “coronary angiography,” “coronary vessels,” and “coronary artery disease.” The literature search ranged from 2004 to 2012. The reference lists of published systematic reviews and meta-analyses were also screened. Two investigators examined the studies to exclude potential duplicate or overlapping data.
Figure 1 Flow chart of evidence search and selection.
M. Li et al. / Clinical Radiology 69 (2014) 163e171 Table 1 Characteristics of included studies.
bECGMale/ Mean No. of HR female age gating sections (bpm) blocker (years)
First author
Year
Stolzmann et al.30 Moon et al.31 Achenbach et al.32 Bamberg et al.8 Yang et al.33 Xu et al.9 Tsiflikas et al.11 Scheffel et al.10 Scheffel et al.34 Ovrehus et al.35 Lin et al.36 Fang et al.37 Donati et al.38 Chen et al.39 Alkadhi et al.29 Alkadhi et al.29 Rixe et al.40 Reimann et al.7 Meng et al.41 Leschka et al.42 Stolzmann et al.43 Scheffel et al.44 Ravipati et al.45 Achenbach et al.46 Leschka et al.47 Brodoefel et al.48 Alkadhi et al.49 Leber et al.50 Weustink et la.51 Ropers et al.52 Johnson et al.53 Heuschmid et al.54 Scheffel et al.55
2011 1.4
68
P
64
58
Yes
2011 1.9
60.5
R
64
58.9
Yes
2011 2.1
59
P
128
71
Yes
2011 .
68.1
P
128
72.2
No
2010 3.6
62
.
64
59
Yes
2010 2.2 2010 2.7
64.6 64
R R
64 64
71.4 64
No No
2010 2.5
56
.
64
67.3
No
2010 3.8
64
P
64
61
No
2010 1.3
61
.
64
61
No
2010 3.4 2010 1.8 2010 4.2
61.2 59.6 64
R R R
64 64 64
66.9 88 .
No No No
2010 1.6
60.7
.
64
86.4
No
2010 3.2
62
P
128
58
Yes
2010 2.6
63
P
128
56
Yes
2009 1.6
65
R
64
68
No
2009 2.8
62
R
64
62.7
No
2009 1.7
63
.
64
71.8
No
2009 4.0
62
P
128
58
Yes
2008 0.7
65.8
P
64
60.7
Yes
2008 1.4
68
P
64
59
Yes
2008 1.9
66
.
64
.
Yes
2008 1.8
61
.
64
64.3
Yes
2008 2.1
61.9
R
64
67.7
No
2008 4.0
62
R
64
64.9
No
2008 2.2
62.9
R
64
68.5
No
2007 1.7
58
R
64
.
Yes
2007 3.8
61
R
64
68
No
2007 1.7
61
R
64
64
No
2007 2.2
60
R
64
68
No
2007 2.6
64
R
64
65
No
2006 4.0
63.1
R
64
70.3
No
165
presented in absolute numbers of true-positive, false-positive, true-negative, and false-negative results or the numbers could be calculated using the detailed data; ICA was used as the reference standard for the diagnosis of CAD; the cut-off criterion for diagnosis of CAD was >50% diameter stenosis. Studies were excluded if they included patients who had undergone coronary artery bypass graft surgery, percutaneous coronary intervention, or prior heart transplantation. Studies including patients with atrial fibrillation (AF) were also excluded.
Data extraction The following information was extracted from each study: first author, year of publication, country, number of patients, gender, age, heart rate, number of detector sections, ECG-gated protocol, radiation dose, usage of bblocker, and number of true-positive, true-negative, falsepositive, and false-negative values. True positive was defined as significant stenosis on a DSCT angiography confirmed by ICA, and true negative was defined as a negative DSCT angiographic finding proven by ICA. If stenosis at ICA was not detected using DSCT angiography, it was classified as false negative, and if a negative finding at ICA presented as a stenosis at CTA, it was classified as a false positive. Data at the level of segments, arteries, and patients were recorded separately. Study quality was evaluated according to Quality Assessment of Diagnostic Accuracy Studies (QUADAS).14 Two readers, who were not involved in any of the included studies, independently evaluated quality scores.
Data synthesis and statistical analysis
ECG, electrocardiograph; R, retrospective; P, prospective; HR, heart rate; TP, true positive; FP, false positive; FN, false negative; TN, true negative.
A Cohen k-test was conducted to test the interobserver agreement. A k-value >0.80 is considered good to excellent agreement. Publication bias was assessed using a regression line developed by Deeks et al. (p < 0.10 indicates significant asymmetry).15 The Cochran Q statistic and measured inconsistency test (I2) was applied to assess heterogeneity across studies.16 Meta regression was conducted to compare the diagnostic accuracy of DSCT coronary angiography with and without heart rate control. Diagnostic test data was synthesized using an exact binomial rendition of the bivariate mixed-effects regression model, which fits a two-level model, with independent binomial distributions for true-positive and true-negative conditionals and a bivariate normal model for the logit transforms of sensitivity and specificity between the included studies.17e19 The sensitivity, specificity, and 95% confidence intervals (CIs) were calculated. Positive and negative likelihood ratios (LRs) were also computed to eliminate the influence of the prevalence of disease. The standard formulas were demonstrated as follows: positive likelihood ratio ¼ sensitivity/(1especificity) and negative likelihood ratio ¼ specificity/ (1esensitivity). The post-test probability was calculated using likelihood ratios based on
166
M. Li et al. / Clinical Radiology 69 (2014) 163e171
Results
Bayes’ theorem to investigate the clinical utility of DSCT angiography.20 A hierarchical summary receiver operating characteristic (sROC) curve was constructed based on parameters estimated by the bivariate model, an area under the ROC curve (AUROC) serves as a global measure of test performance (low: 0.5 AUROC 0.7, moderate: 0.7 AUROC 0.9, or high: 0.9 AUROC 1).21 The sensitivity analysis was conducted by omitting each reference and re-analysing the data to test whether there were some studies influencing the final results significantly. The basic characteristics between references with and without heart rate control were compared using ManneWhitney U and c2 tests. STATA, version 12 (StataCorp, College Station, TX, USA) with MIDAS module, and SPSS, version 16.0 (SPSS, Chicago, Illinois), were used to analyse the data and construct the graphs.
Characteristics of studies A total of 804 potentially relevant references were identified. After screening for title and abstract, 135 studies were retrieved. After selecting full-text articles according to the inclusion and exclusion criteria mentioned above, 33 studies were included in the review. Fig 1 shows the flow chart of the literature search and selection process. The number of included patients ranged from 25 to 210 (median 82 patients). Twenty-eight studies reported using a 64 section DSCT and five using a 128-section DSCT system. b-Blockers were administered before the examination in 12 of the studies, during which seven studies applied prospective ECG gating, two studies applied retrospective ECG gating, and three studies were unclear. Twenty-one studies did not control heart rate using b-blockers, during which two studies
Table 2 Quality assessment of diagnostic accuracy studies. First author
Item 1
Item 2
Item 3
Item 4
Item 5
Item 6
Item 7
Item 8
Item 9
Item 10
Item 11
Item 12
Item 13
Item 14
Stolzmann et al.30 Moon et al.31 Achenbach et al.32 Bamberg et al.8 Yang et al.33 Xu et al.9 Tsiflikas et al.11 Scheffel et al.10 Scheffel et al.34 Ovrehus et al.35 Lin et al.36 Fang et al.37 Donati et al.38 Chen et al.39 Alkadhi et al.29 Rixe et al.40 Reimann et al.7 Meng et al.41 Leschka et al.42 Stolzmann et al.43 Scheffel et al.44 Ravipati et al.45 Achenbach et al.46 Leschka et al.47 Brodoefel et al.48 Alkadhi et al.49 Leber et al.50 Weustink et al.51 Ropers et al.52 Johnson et al.53 Heuschmid et al.54 Scheffel et al.55
Y Y Y Y Y Y Y N Y Y Y Y Y Y Y U U Y Y Y Y U Y Y Y Y U N Y Y N Y
Y Y Y Y Y N Y Y Y Y Y Y Y Y Y Y U Y Y Y Y N Y Y Y Y Y Y Y Y Y Y
Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y
Y U Y U U U Y U Y U Y Y Y Y Y U Y Y Y Y Y U U Y Y Y Y Y Y Y Y Y
Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y
Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y
Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y
Y U Y Y Y Y Y Y Y Y Y Y Y Y Y N Y Y Y Y Y U Y Y Y Y Y Y Y Y Y Y
Y Y N Y N Y N Y Y Y N N Y Y N Y N Y Y N U U N N Y N N N N U N N
Y Y Y Y U U Y U Y Y Y Y Y Y Y Y U Y Y Y Y U Y Y Y U U Y Y Y U U
Y Y Y U U Y Y Y Y Y Y Y Y Y Y N U U Y Y Y U U Y Y Y U U U U U Y
N U Y N U Y U U N N N U N U N N U N N N U U U N N N U U N U U U
Y N Y Y N N N N N N U Y Y U Y Y U Y U Y Y N Y N Y Y Y Y U Y Y Y
N Y Y Y Y N Y Y Y Y Y Y Y Y Y Y U N Y Y Y N N Y Y N N Y N N N N
The quality assessment of the studies was performed with a QUADAS tool. Items 1e14 represent the following questions, which were part of the tool. Item 1: Was the spectrum of patients representative of the patients who will receive the test in practice? Item 2: Were selection criteria clearly described? Item 3: Is the reference standard likely to correctly classify the target condition? Item 4: Is the time period between reference standard and index test short enough to be reasonably sure that the target condition did not change between the two tests? Item 5: Did the entire sample or a random selection of the sample receive verification using a reference standard of diagnosis? Item 6: Did patients receive the same reference standard regardless of the index test result? Item 7: Was the reference standard independent of the index test (i.e., the index test did not form part of the reference standard)? Item 8: Was the execution of the index test described in sufficient detail to permit replication of the test? Item 9: Was the execution of the reference standard described in sufficient detail to permit its replication? Item 10: Were the index test results interpreted without knowledge of the results of the reference standard? Item 11: Were the reference standard results interpreted without knowledge of the results of the index test? Item 12: Were the same clinical data available when test results were interpreted as would be available when the test is used in practice? Item 13: Were uninterpretable/intermediate test results reported? Item 14: Were withdrawals from the study explained? [14].
M. Li et al. / Clinical Radiology 69 (2014) 163e171
167
Figure 2 Forest plots showing sensitivity and specificity at the patient level.
applied prospective ECG gating, 15 studies applied retrospective ECG gating, and four studies were unclear. There was statistical differences of the imaging protocol between the two groups (p ¼ 0.004). The mean heart rates of b-blocker references ranged between 56 beats/min and 71 beats/min (median, 58.5 beats/min), which was significantly lower than that of non-b-blocker references [ranged between 61 beats/ min and 88 beats/min [median, 67 beats/min], p < 0.001]. No statistical differences were found for other characteristics (gender, p ¼ 0.22; age, p ¼ 0.81; detector, p ¼ 0.05). The average estimated effective radiation dose ranged from 0.76 to 16.1 mSv. The median radiation exposure was 2.6 mSv, with 1.6 and 8 mSv for heart rate-controlled studies and uncontrolled studies, respectively. The baseline characteristics of the study and the detailed diagnostic information are shown in Table 1.
Methodological quality The k-value of the interobserver agreement for assessing quality items was 0.85. Table 2 presents the quality score of
the included studies according to the QUADAS items. Overall, the qualities of enrolled study were moderate, with 20 of the included studies fulfilling 10 or more of the 14 criteria. However, none of them fulfilled all the items of the QUADAS tool.
Data synthesis and statistical analysis A total of 31,324 segments, 4697 vessels, and 1750 patients were analysed. No publication bias was significant at the patient, vessel, or segment level (p ¼ 0.10, 0.28, and 0.67, respectively). The pooled sensitivity at patient level 0.98 (95% CI: 0.97e0.99) and the pooled specificity was 0.88 (95% CI: 0.84e0.91; Fig 2). The pooled positive and negative LRs were 8.04 (95% CI: 6.17e10.48) and 0.02 (95% CI: 0.01e0.03), respectively. Table 3 shows that the sensitivity increased from segment, to vessel, and to patient level. In contrast, the specificity and positive and negative LRs were lowest at the patient level.
Table 3 Pooled summary results. TP
Total Patient-level Artery-level Segment-level HR controlled Patient-level Artery-level Segment-level HR uncontrolled Patient-level Artery-level Segment-level
FP
FN
TN
Sensitivity
Specificity
Positive LR
Negative LR
AUROC
(95%CI)
(95%CI)
(95% CI)
(95% CI)
(95% CI)
953 1307 3819
98 204 969
17 42 275
682 3144 26,261
0.98 (0.97e0.99) 0.97 (0.96e0.98) 0.95 (0.92e0.97)
0.88 (0.84e0.91) 0.94 (0.92e0.96) 0.97 (0.96e0.98)
8.04 (6.17e10.48) 16.39 (11.49e23.36) 35.19 (25.86e47.88)
0.02 (0.01e0.03) 0.03 (0.02e0.05) 0.05 (0.04e0.08)
0.99 (0.98e1.00) 0.99 (0.98e1.00) 0.99 (0.98e1.00)
494 537 1347
49 61 223
6 8 73
377 1428 10,711
0.99 (0.97e0.99) 0.99 (0.97e0.99) 0.95 (0.92e0.97)
0.89 (0.84e0.93) 0.96 (0.95e0.97) 0.98 (0.97e0.99)
8.89 (6.08e13.01) 24.01 (18.69e30.84) 55.35 (34.83e87.95)
0.01 (0.01e0.04) 0.02 (0.01e0.04) 0.05 (0.03e0.08)
0.99 (0.98e1.00) 0.98 (0.96e0.99) 0.99 (0.98e1.00)
459 770 2472
49 143 746
11 34 202
305 1716 15,550
0.98 (0.96e0.99) 0.96 (0.94e0.98) 0.94 (0.91e0.97)
0.86 (0.81e0.90) 0.92 (0.87e0.95) 0.96 (0.95e0.98)
7.16 (5.08e10.09) 11.79 (7.12e19.50) 26.54 (18.76e37.54)
0.03 (0.02e0.05) 0.04 (0.03e0.07) 0.06 (0.03e0.10)
0.98 (0.97e0.99) 0.98 (0.97e0.99) 0.99 (0.97e0.99)
TP, true-positive; FP, false-positive; TN, true-negative; FN, false-negative; LR, likelihood ratio; AUROC, area under the summary receiver operating characteristic curve; HR, heart rate.
168
M. Li et al. / Clinical Radiology 69 (2014) 163e171
Figure 3 Symmetric sROC at the patient level for accuracy of DSCT angiography for studies with and without heart rate control.
The application of a b-blocker was added as a factor for meta-regression analysis to compare the diagnostic performance for heart rate-controlled studies and heart rateuncontrolled studies. The results showed that without heart rate control the sensitivity and specificity required to detect significant stenosis at the patient level did not decrease (p ¼ 0.27 and 0.56, respectively). At the artery level, the sensitivity and the specificity of studies without heart rate control was similar to that with heart rate control (p ¼ 0.04 and 0.05, respectively). At the segment level, the specificity decreased without heart rate control (p ¼ 0.03), whereas sensitivity was not influenced (p ¼ 0.63). The AUROC was 0.99 (95% CI: 0.98e0.99) for the heart ratecontrolled group and 0.98 (95% CI: 0.97e0.99) for the heart rate-uncontrolled group on the patient level, as shown in Fig 3. Post-test probabilities were calculated with and without heart rate control according to the Bayes’ theorem to investigate the clinical utility of DSCT angiography (see Fig 4). The plot shows that coronary DSCT angiography with
and without the application of a b-blocker has perfect negative predictive value in patients with a low to intermediate pretest likelihood of CAD, whereas the positive predictive value was moderate. Heterogeneity between included studies was evaluated at the patient, artery, and segment levels. At the patient level, no statistically heterogeneity was found for sensitivity (Q ¼ 29.33; p ¼ 0.11; I2 ¼ 28.39%; 95% CI: 0.00e65.73%), whereas it was significant for specificity (Q ¼ 45.78; p < 0.001; I2 ¼ 54.13%; 95% CI: 31.99e76.26%). Conversely, heterogeneity was significant at the artery level for both sensitivity (Q ¼ 46.00; p < 0.001; I2 ¼ 63.04%; 95% CI: 44.3e81.78%) and specificity (Q ¼ 112.51; p < 0.001; I2 ¼ 84.89%; 95% CI: 78.84e90.94%) and at the segment level for both sensitivity (Q ¼ 503.37; p < 0.001; I2 ¼ 94.44%; 95% CI: 93.14e95.73%) and specificity (Q ¼ 683.01; p < 0.001; I2 ¼ 95.9%; 95% CI: 95.03e96.77%). Sensitivity analysis was performed at the patient level to investigate the influence of each study on the overall pooled results. Fig 5 indicates that no study influenced the pooled sensitivity and specificity by >0.03. Further analysis demonstrated that after the exclusion of six studies with a quality score of
