Author + information
- Received November 27, 2013
- Accepted January 14, 2014
- Published online May 20, 2014.
- Omar Wever-Pinzon, MD∗,
- Jorge Romero, MD†,
- Iosif Kelesidis, MD†,
- James Wever-Pinzon, MD∗,
- Carlos Manrique, MD†,
- Deborah Budge, MD∗,
- Stavros G. Drakos, MD, PhD∗,
- Ileana L. Piña, MD, MPH†,
- Abdallah G. Kfoury, MD∗,
- Mario J. Garcia, MD† and
- Josef Stehlik, MD, MPH∗∗ ()
- ∗U.T.A.H. Cardiac Transplant Program, University of Utah Health Sciences Center, Veterans Affairs Medical Center & Intermountain Medical Center, Salt Lake City, Utah
- †Montefiore Medical Center, Albert Einstein College of Medicine, Bronx, New York
- ↵∗Reprint requests and correspondence:
Dr. Josef Stehlik, University of Utah Health Sciences Center, Division of Cardiology, 30 North Medical Drive, 4A100 SOM, Salt Lake City, Utah 84132.
Objectives This study aimed to evaluate the diagnostic accuracy of coronary computed tomography angiography (CCTA) for detecting cardiac allograft vasculopathy (CAV) in comparison with conventional coronary angiography (CCAG) alone or with intravascular ultrasound (IVUS).
Background CAV limits long-term survival after heart transplantation, and screening for CAV is performed on annual basis. CCTA is currently not recommended for CAV screening due to the limited accuracy reported by early studies. Technological advances, however, might have resulted in improved test performance and might justify re-evaluation of this recommendation.
Methods A systematic review of Medline, Cochrane, and Embase for all prospective trials assessing CAV using CCTA was performed using a standard approach for meta-analysis for diagnostic test and a bivariate analysis.
Results Thirteen studies evaluating 615 patients (mean age 52 years, 83% male) and 9,481 segments fulfilled inclusion criteria. Patient-based analyses comparing CCTA versus CCAG for the detection of any CAV (> luminal irregularities) and significant CAV (stenosis ≥50%), showed mean weighted sensitivities of 97% and 94%, specificities of 81% and 92%, a negative predictive value (NPV) of 97% and 99%, a positive predictive value (PPV) of 78% and 67%, and diagnostic accuracies of 88% and 94%, respectively. There was a strong trend toward improved sensitivity (97% vs. 91%, p = 0.06) and NPV (99% vs. 97%, p = 0.06) to detect significant CAV with 64-slice compared with 16-slice CCTA. A patient-based analysis of 64-slice CCTA versus IVUS showed a mean weighted sensitivity and specificity of 81% and 75% to detect CAV (intimal thickening >0.5 mm), whereas the PPV and NPV were 93% and 50%, respectively.
Conclusions CCTA using currently available technology is a reliable noninvasive imaging alternative to coronary angiography with an excellent sensitivity, specificity, and NPV for the detection of CAV.
- cardiac allograft vasculopathy
- coronary computed tomography angiography
- heart transplantation
- transplant vasculopathy
Long-term survival following heart transplantation is limited by significant mortality rates (3% to 4%/year) beyond the first year after transplant, which have remained constant for the past 2 decades (1). Timely identification of factors contributing to the excess mortality observed in this time period is likely to result in decreased morbidity and improved survival. Cardiac allograft vasculopathy (CAV) affects over 50% of the recipients by 10 years after transplant and represents an important cause of mortality in recipients who survive the first year after transplant (1,2). Death or retransplantation 5 years after transplant occurs in 7% of the recipients who develop CAV of any degree and in 50% of the recipients who develop severe CAV (3,4).
The diagnosis of CAV has traditionally relied on the use of conventional coronary angiography (CCAG) (4). Although widely available and clinically accepted, it has the disadvantages of being invasive and only providing visualization of the vessel lumen. The development of intravascular ultrasound (IVUS) 2 decades ago brought greater understanding of the coronary wall anatomy and the disease process in CAV (5). Its high sensitivity and negative predictive value made IVUS valuable for the assessment of CAV in the investigational field. Nonetheless, the routine clinical use of IVUS as a screening tool for CAV is not currently recommended, as it is unclear whether IVUS can provide clinical incremental value over CCAG and functional data. Furthermore, the use of IVUS is invasive, costly, and can be time consuming (4). A technique that has been viewed with enthusiasm in the assessment of CAV is coronary computed tomography angiography (CCTA). It has the advantages of being noninvasive, allows for visualization of all coronary arteries, and provides information about the coronary artery wall. A small number of trials, with a limited number of patients, suggested that the diagnostic accuracy of CCTA in detecting CAV may be similar to that of CCAG (6–10). Prior reviews assessing the diagnostic accuracy of CCTA to detect CAV are limited by incomplete data and technical limitations (11,12). As a result of the limited data supporting the diagnostic accuracy of CCTA to detect CAV and concerns about radiation exposure, its use is not recommended for routine surveillance of CAV by the clinical practice guidelines (13). Our aim was to closely examine contemporary data on the accuracy of CCTA for the detection of CAV through a meta-analysis approach.
We conducted PubMed, Embase, and Cochrane Central Register of Clinical Trials searches for prospective studies in which CCTA was used to assess the presence and/or the severity of CAV using the terms: (CTA OR CCTA OR computed tomography OR coronary computed tomography angiography OR computed tomography coronary angiography OR cardiac computed tomography angiography OR coronary CT angiography OR coronary computed tomography OR multi-slice computed tomography OR multi-slice computed tomography angiography OR multi-detector computed tomography) AND (CAV OR cardiac allograft vasculopathy OR vasculopathy OR transplant vasculopathy OR graft vasculopathy OR coronary allograft vasculopathy OR coronary vasculopathy OR chronic allograft vasculopathy). We limited our search to humans and adults (>18 years of age) in peer-reviewed journals from 1966 to 2013. No language restriction was applied. The reference lists of bibliographies of identified papers were also reviewed. Trials in the abstract form without a published manuscript were excluded from this analysis.
Eligible trials had to fulfill the following criteria: 1) prospective study involving heart transplant recipients who underwent CCTA to assess for the presence of CAV in comparison to CCAG and/or IVUS; 2) study allowed for sensitivity, specificity, negative predictive value (NPV), and positive predictive value (PPV) calculations; and 3) there was use of standardized cutoffs for each test, or enough data were provided to calculate diagnostic and predictive accuracies using these cutoffs.
Data extraction and quality assessment
Three investigators (O.W.-P., J.R., C.M.) independently assessed trial eligibility and extracted data using standardized protocol and reporting forms. Disagreements were resolved by consensus. Characteristics of each trial, baseline demographics, methods, interval between transplantation and CCTA, prevalence of CAV, and data required to evaluate the diagnostic accuracy of CCTA were extracted. In instances where data were not readily available, the main investigator of that particular trial was approached to supply the relevant information. The assessment of quality of each study was done by evaluating 14 items considered relevant to the review topic, on the basis of the Quality Assessment of Diagnostic Accuracy Studies instrument (14). Two reviewers (J.R. and J.W.-P) independently assessed the quality items, and discrepancies were resolved by consensus.
Sensitivities (number of stenotic coronary segments estimated by the index test [CCTA] divided by the total number of stenotic segments identified by the reference standard [CCAG/IVUS]), specificities (number of normal segments estimated by the index test divided by the total number of segments without stenosis identified by the reference standard), PPV (stenotic segments identified by index test confirmed by reference standard divided by the total number of stenotic segments estimated by index test), and NPV (segments without stenosis identified by index test confirmed by reference standard divided by the total of nonstenotic segments estimated by index test) were calculated for every study. Summary sensitivity and specificity were estimated using both, a standard random effect model and a more recently developed bivariate random effects model (15,16). The bivariate approach assumed logit transforms of sensitivity and specificity from individual studies are from a bivariate normal distribution. The bivariate approach is considered to be a better approach as compared with the standard summary receiver operating characteristics (ROC) approach because (17): 1) it assesses heterogeneity across studies and provides a summary estimate of sensitivity and specificity; 2) it models sensitivity and specificity jointly so that a 95% confidence ellipse around the summary estimate can be calculated; 3) it allows a direct comparison of sensitivity and specificity between methods; and 4) it provides several choices to obtain a summary ROC curve (15,16). The summary ROC curve was obtained by transforming the regression line of logit sensitivity on logit specificity into ROC space (15). Publication bias was assessed for each analysis using Peter and Egger methods. We assessed between-study heterogeneity visually, by plotting sensitivity and specificity in the ROC curves. Summary ROC curves and confidence regions for summary sensitivity and specificity were also drawn (16,18). The analyses were conducted using standard software (Stata 12, Stata Corporation, College Station, Texas).
We further evaluated whether the performance of each technique depends on features of the technique and patient characteristics. A logistic regression for each technique was used to model the sensitivity on these factors.
We identified 2,830 papers, of which 1,672 abstracts were retrieved and reviewed for possible inclusion (Fig. 1). Thirteen studies, enrolling 615 patients (mean age 52 ± 8.5 years; 83% male) and a total of 9,481 coronary segments fulfilled the inclusion criteria and were included in the analysis (Table 1). These 13 studies allowed for 34 comparisons. Six studies were excluded from the final analysis because they did not meet the inclusion criteria: 1 study did not use CCTA and CCAG in all patients, CCAG was used only to confirm abnormal CCTA findings (19), another study compared CCTA versus dobutamine stress echocardiography and only used CCAG in patients with an abnormal result of either test (20), another study provided no data to calculate diagnostic accuracies (21), another study did not report demographic data, number of segments and the specific numbers compared with CCAG and IVUS (22), another study comparing 64-slice and 256-slice CCTA versus CCAG reported the diagnostic accuracy without differentiation between 64-slice and 256-slice CCTA (23); and the last study was conducted using 4-slice CCTA (24).
Of the 13 studies, 8 studies analyzing 410 patients (mean age 49 ± 7.6 years; 81% male) and 5,949 coronary segments evaluated CAV using 64-slice CCTA (7,8,10,25–29). Seven studies compared 64-slice CCTA with CCAG and 2 studies compared 64-slice CCTA with IVUS (7,10). Six studies analyzing 205 patients (mean age 50 ± 8.0 years; 83% male) and 3,532 coronary segments evaluated CAV using 16-slice CCTA, with all of them using CCAG for comparison and 2 of them also using IVUS for comparison (Table 1) (6,9,26,30–32).
The selected studies showed overall high-quality scores in all the 14 items of the Quality Assessment of Diagnostic Accuracy Studies instrument. Reporting was poor in 1 of the studies on item 13: “Were uninterpretable/intermediate test results reported?” Uninterpretable results are often not reported in diagnostic accuracy studies. However, this may lead to the biased assessment of the test characteristics. Bias will arise depending on the possible correlation between uninterpretable test results and the true disease status. If uninterpretable results occur randomly and are not related to the true disease status, these should have no effect on test performance. Two papers did not provide sufficient information regarding the execution of the reference standard, which could affect the reproducibility of the test results or its performance. Three papers did not explain whether the reference (2 papers) or index test (1 paper) were interpreted in a blinded fashion (review bias), which might have led to inflated measures of diagnostic accuracy (Figs. 2A and 2B).
Using Egger's test, there was no indication of publication bias for any of the analyzed techniques (p > 0.09 for all analyses). Likewise, the Peters test did not suggest presence of publication bias (p > 0.15 for all analyses).
CCTA versus CCAG
Twelve studies evaluated the presence of CAV by CCTA using CCAG as the reference test. CCTA was performed between 34 and 144 months (mean 88 months) after transplant. The mean heart rate at the time of CCTA was 84 beats/min (range 70 to 90 beats/min) and 5 studies used beta-blockers for heart rate reduction. The use of contrast media ranged from 60 to 115 ml and the radiation dose from 3 to 18 mSv. The incidence of contrast induced nephropathy was reported in most studies (n = 9), with no cases observed in the enrolled patients. Most studies (n = 10) used a ≥50% stenosis cutoff to determine the presence of significant CAV, except for 1 study that used a ≥70% stenosis cutoff. Six studies used a >0% stenosis cutoff to determine the presence of any CAV (Table 1).
Diagnostic accuracy of CCTA for the detection of any CAV
Any CAV was found in 87 patients by means of CCAG, which represents a prevalence of 43%. The combined overall weighted mean sensitivity, specificity, PPV, NPV, and accuracy for CCTA (64-slice and 16-slice CCTA) were 97% (95% confidence interval [CI]: 92% to 100%), 81% (95% CI: 74% to 88%), 78% (95% CI: 71% to 86%), 97% (95% CI: 93% to 100%), and 88% (95% CI: 84% to 93%), respectively (Fig. 3). Analyses of the different measures of diagnostic accuracy, stratified by the number of CCTA slices are shown in Table 2A. There were no significant differences in these measures when 64-slice and 16-slice CCTA were compared.
A total of 384 segments had evidence of any CAV by CCAG, which represents a prevalence of 13%. The combined overall weighted mean sensitivity, specificity, PPV, NPV, and accuracy for CCTA (64-slice and 16-slice) were 88% (95% CI: 85% to 92%), 89% (95% CI: 88% to 90%), 55% (95% CI: 51% to 59%), 98% (95% CI: 97% to 98%), and 90% (95% CI: 89% to 91%), respectively (Online Fig. 1). Additional analyses, stratified by the number of CCTA slices are shown in Table 2B. There were no significant differences in the measures of diagnostic accuracy between 64-slice and 16-slice CCTA.
Diagnostic accuracy of CCTA for the detection of significant CAV
Sixty-two patients had evidence of significant CAV by CCAG, which represents a prevalence of 16%. The combined overall mean weighted sensitivity, specificity, PPV, NPV, and accuracy for CCTA (64-slice and 16-slice) were 94% (95% CI: 87% to 100%), 92% (95% CI: 89% to 95%), 67% (95% CI: 59% to 76%), 99% (95% CI: 97% to 100%), and 94% (95% CI: 91% to 96%), respectively (Fig. 4). Analyses of diagnostic accuracy, stratified by the number of CCTA slices are shown in Table 3A. Although statistically there were no significant differences in the evaluated measures between 64-slice and 16-slice CCTA, there was a trend toward improved sensitivity (97% vs. 91%, p = 0.06), NPV (99% vs. 97%, p = 0.06), and lower PPV (63% vs. 72%, p = 0.06) with the use of 64-slice CCTA.
A total of 174 segments had significant CAV by CCAG, which represents a prevalence of 3%. The combined overall mean weighted sensitivity, specificity, PPV, NPV, and accuracy for CCTA (64-slice and 16-slice) were 86% (95% CI: 82% to 91%), 99% (95% CI: 98% to 99%), 73% (95% CI: 68% to 78%), 99% (95% CI: 99% to 100%), and 99% (95% CI: 98% to 99%), respectively (Online Fig. 2). Additional analyses, stratified by the number of CCTA slices are shown in Table 3B. There were no significant differences in the measures of diagnostic accuracy between 64-slice and 16-slice CCTA. However, there was a trend toward lower PPV with 64-slice versus 16-slice CCTA (60% vs. 84%, p = 0.06).
CCTA versus IVUS
Four studies evaluated the presence of any CAV by CCTA using IVUS as the reference test. The CCTA was performed between 34 and 88 months (mean 64 months) after transplant. The mean heart rate at the time of testing was 84 beats/min (range 77 to 90 beats/min). The use of contrast media ranged from 60 to 100 ml and the mean radiation dose was 10 mSv. No cases of contrast induced nephropathy occurred in the enrolled patients. Three studies used an intimal thickness >0.5 mm to determine the presence of CAV, whereas 1 study used the presence of “proliferative changes” to determine the presence of CAV (Table 1).
Sixteen patients were found to have CAV by IVUS, representing a prevalence of 80%. The weighted mean sensitivity and specificity of 64-slice CCTA were 81% (95% CI: 57% to 93%) and 75% (95% CI: 30% to 95%), whereas the PPV was 93% (95% CI: 68% to 99%) and NPV was 50% (95% CI: 19% to 81%). This technique had a mean weighted accuracy of 80% (95% CI: 58% to 92%). There were no studies evaluating the diagnostic accuracy of 16-slice CCTA using a patient-based analysis.
A total of 204 segments had CAV by IVUS, representing a prevalence of 34%. The combined overall weighted mean sensitivity, specificity, PPV, NPV and accuracy for CCTA (64-slice and 16-slice, n = 836 segments) were 89% (95% CI: 86% to 93%), 89% (95% CI: 87% to 92%), 79% (95% CI: 74% to 83%), 94% (95% CI: 92% to 96%), and 90% (95% CI: 88% to 92%), respectively (Fig. 5). The weighted mean sensitivity and specificity of 64-slice (n = 229 segments) versus 16-slice CCTA (n = 607 segments) were 78% (95% CI: 70% to 86%) versus 94% (95% CI: 91% to 98%, p = 0.08) and 88% (95% CI: 83% to 94%) versus 90% (95% CI: 87% to 92%, p = 0.90), whereas the PPV was 82% (95% CI: 74% to 90%) versus 77% (95% CI: 71% to 82%, p = 0.43), and NPV was 83% (95% CI: 77% to 90%) versus 98% (95% CI: 96% to 99%, p = 0.07), respectively. There was no significant difference in mean weighted accuracy between 64-slice 83% (95% CI: 78% to 88%) and 16-slice CCTA 91% (95% CI: 89% to 93%, p = 0.58).
In the present meta-analysis, we focused on the diagnostic performance of the newest generation of CCTA (≥16 slices) to assess the extent and severity of CAV compared with the clinical standard of CCAG, and with the highly sensitive IVUS. The main result of our study is the demonstration of high sensitivity, specificity and NPV of CCTA for the diagnosis of CAV when compared with the clinical standard CCAG. The high sensitivity and NPV suggest that a negative CCTA can accurately exclude CAV in the absolute majority of heart transplant recipients undergoing this test. As expected, the sensitivity and NPV of CCTA to detect CAV were lower when the highly sensitive IVUS was used as the standard reference test.
CAV is a progressive disorder that is characterized by diffuse concentric intimal hyperplasia involving both epicardial and intramyocardial arteries (33,34). It is difficult to diagnose in the early stages as it is typically silent in the denervated transplanted heart, and ischemia or graft dysfunction are usually not evident until the disease is advanced and manifests as heart failure, arrhythmias or sudden death. Yearly screening to detect CAV before graft injury results is therefore recommended (13). Noninvasive stress testing such as exercise electrocardiography, stress echocardiography, and single-photon emission computed tomography have been used, however, the modest diagnostic accuracy is a major limitation of these tests (35), and thus clinical guidelines recommend CCAG as a screening tool for CAV (13). Although CCAG has an important role in the clinical assessment of the severity, progression, and classification of CAV, it often underestimates the extent and severity of CAV as a consequence of the vascular remodeling process (36,37). Histopathologic studies have shown that as many as 73% of angiographically normal segments have mild to moderate fibrous intimal thickening by light microscopy (38). Addition of coronary IVUS allows for detection and quantification of intimal hyperplasia by imaging of the vessel wall structure, making it the most sensitive test for detecting CAV. At 4 years after transplantation, CCAG detects CAV in one-third of recipients, whereas diagnosis of CAV is made in 55% by IVUS (37). However, the drawbacks of IVUS are its invasiveness, limitation to assessment of only proximal epicardial arteries, risk of serious complications and cost (39). Although the rate of progression and CAV severity as assessed by IVUS has been shown to have prognostic implications as far as the risk of myocardial infarction, heart failure, retransplantation and death (40,41), the detailed information provided by IVUS does not trigger changes in clinical care that would alter patient outcomes. Therefore, addition of IVUS to screening CCAG has not become the standard approach to CAV screening.
CCTA is a noninvasive technique that permits quantification of lumen size and wall vessel structure. Its diagnostic accuracy has been extensively studied in large study groups of nontransplanted patients (42). However, the small number of studies evaluating the accuracy of CCTA to detect CAV has included only limited number of patients (6–9,25,26,30–32). The aim of our meta-analysis was to determine whether data pooled from prospectively conducted studies could provide a stronger basis for clinical application of CCTA in transplant recipients. Although the results were consistent when analyzed as individual coronary segments, or when assessed on the basis of an overall test result in each patient, the main clinical implications can be drawn from the patient-based data. Using CCAG as the reference test, CCTA has an excellent sensitivity and NPV for the diagnosis of CAV, being 94% and 99% for the detection of significant CAV and 97% for the presence of any CAV. Compared with 16-slice CCTA, the use of 64-slice CCTA was associated with a strong trend toward improved sensitivity and NPV for the detection of significant CAV, likely the result of better spatial and temporal resolution. These results are clinically important as indicate that CCTA could be used effectively as a gatekeeper to CCAG, thus reducing the number of downstream tests and procedures, CCAG-related complications and costs, as prior experience in the general population suggests (43,44). The specificity and PPV of CCTA for the evaluation of the presence of any CAV were modest, a finding that can be explained by the low sensitivity of CCAG, related to its inability to evaluate the vessel wall. As expected, the specificity of CCTA to detect significant CAV was higher than for the evaluation of the presence of any CAV. Although the specificity of CCTA to detect significant CAV was excellent, the PPV remain modest likely as a consequence of the low prevalence of significant CAV lesions.
The segment-based results are important from the technical perspective. These results showed an excellent sensitivity and NPV of CCTA to detect both, the presence of any CAV and the presence of significant CAV, confirming the high diagnostic accuracy of this technique.
The routine use of CCTA for the assessment of CAV is not supported by current clinical guidelines (13), as a result of not only the limited data on its diagnostic accuracy, but also due to limited data evaluating its prognostic value on post-transplant survival. More recently, Rohnean et al. (19) prospectively evaluated 62 heart transplant recipients by means of serial CCTA and showed that this technology provides a safe and reliable alternative to CCAG for the long-term evaluation of CAV. After 5 years, 30% of patients with a normal CCTA at baseline progressed to develop wall thickening (24%) or significant stenosis (6%), whereas 22% of patients with wall thickening at baseline progressed to develop significant stenosis. More importantly, there were no major coronary events during the study period, suggesting that no patients were classified as false-negatives by CCTA and further supporting its high sensitivity and NPV (19).
An added diagnostic advantage of CCTA over CCAG is the ability to visualize both the lumen and the vessel wall. In our analysis, CCTA showed a good diagnostic accuracy in comparison with IVUS, the most sensitive and specific method for detecting early coronary artery disease and CAV. It is possible that these findings might have important clinical implications as use of certain immunosuppressant agents (e.g., proliferative signal inhibitors) have been associated with reduced incidence and slower progression of CAV (45). Further, vessel wall characterization may also provide mechanistic insights into the disease process underlying CAV. Our group has recently reported histopathologic differences in vascular remodeling between native coronary artery disease and CAV, showing that outward vessel remodeling, a phenotype more frequently associated with vulnerable lesions, was reduced in CAV compared with native coronary artery disease, likely the result of both accelerated intimal ingrowth and limited outward remodeling in CAV (46).
The effective radiation dose and total amount of iodinated contrast agent administered during CCTA testing could be of concern if implemented as a repeated annual assessment. In this meta-analysis, the reported dose of radiation was 3 to 18 mSv and use of contrast media was 60 to 115 ml, which are somewhat higher compared with CCAG. The development of newer techniques including acquisition protocols using high-pitch spiral scan (Flash Spiral mode), step and shoot protocols targeting the systolic phase of the cardiac cycle and reduced-dose low voltage CCTA with sonogram-affirmed iterative reconstruction are expected to reduce radiation and contrast dose to approximately 1 mSv and 10 ml per study, respectively (47–49). It is also conceivable that in those patients with none or minimal vasculopathy, on the basis of the high sensitivity and NPV of CCTA, the interval between screening tests could be increased. In addition, contrast induced nephropathy was not observed in any of the evaluated patients, although this might have been the result of a selection bias in the analyzed studies.
The results of our study are limited to a select group of heart transplant recipients. Patients with chronic kidney disease were excluded from the majority of the studies and no patients with a serum creatinine >2.8 mg/dl were included. Nonetheless, this remains a limitation as well for the reference test CCAG. Patients with cardiac arrhythmias were also excluded. Further, the ability of CCTA to evaluate small distant branches is limited. Recognizing this limitation, none of the evaluated studies included segments <15 mm in diameter, where CAV is commonly recognized in its early stages.
In this meta-analysis, CCTA demonstrated to be a robust technique for the diagnosis of CAV. Its excellent sensitivity and NPV suggest that CAV can be accurately excluded with this noninvasive test, thus avoiding invasive procedures. In light of these results, CCTA should be considered as an alternative to CCAG for CAV screening in the routine care of heart transplant recipients.
For supplemental figures, please see the online version of this article.
This work was presented in part at the American Transplant Congress 2012, Boston, Massachusetts. The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- cardiac allograft vasculopathy
- conventional coronary angiography
- coronary computed tomography angiography
- confidence interval
- intravascular ultrasound
- negative predictive value
- positive predictive value
- receiver-operating characteristic
- Received November 27, 2013.
- Accepted January 14, 2014.
- American College of Cardiology Foundation
- Schmauss D.,
- Weis M.
- Tsutsui H.,
- Ziada K.M.,
- Schoenhagen P.,
- et al.
- Romeo G.,
- Houyel L.,
- Angel C.Y.,
- Brenot P.,
- Riou J.Y.,
- Paul J.F.
- Sigurdsson G.,
- Carrascosa P.,
- Yamani M.H.,
- et al.
- Khan R.,
- Jang I.K.
- Arends L.R.,
- Hamza T.H.,
- van Houwelingen J.C.,
- Heijenbrok-Kal M.H.,
- Hunink M.G.,
- Stijnen T.
- Littenberg B.,
- Moses L.E.
- Harbord R.M.,
- Deeks J.J.,
- Egger M.,
- Whiting P.,
- Sterne J.A.
- Carrascosa P.,
- Capuñay C.,
- Carrascosa J.,
- et al.
- Johnson D.E.,
- Alderman E.L.,
- Schroeder J.S.,
- et al.
- Hausmann D.,
- Erbel R.,
- Alibelli-Chemarin M.J.,
- et al.
- Hamon M.,
- Biondi-Zoccai G.G.,
- Malagutti P.,
- Agostoni P.,
- Morello R.,
- Valgimigli M.
- Shaw L.J.,
- Hausleiter J.,
- Achenbach S.,
- et al.
- Catalan P.,
- Callejo D.,
- Blasco J.A.
- Miller D.V.,
- German C.,
- Oberton S.B.,
- et al.