Author + information
- Received February 17, 2017
- Revision received April 11, 2017
- Accepted April 18, 2017
- Published online July 19, 2017.
- Carlos Collet, MDa,
- Bernard Chevalier, MDb,
- Angel Cequier, PhDc,
- Jean Fajadet, MDd,
- Marcello Dominici, MDe,
- Steffen Helqvist, MDf,
- A.J. Van Boven, MDg,
- Dariusz Dudek, MDh,
- Dougal McClean, MDi,
- Manuel Almeida, MDj,
- Jan J. Piek, MD, PhDa,
- Erhan Tenekecioglu, MDk,
- Antonio Bartorelli, MD, PhDl,
- Stephan Windecker, MD, PhDm,
- Patrick W. Serruys, MD, PhDn,∗ ( and )
- Yoshinobu Onuma, MD, PhDk
- aAcademic Medical Center, University of Amsterdam, Amsterdam, the Netherlands
- bInstitut Jacques Cartier, Massy, France
- cBellvitge University Hospital, Idibell, Barcelona, Spain
- dClinique Pasteur, Toulouse, France
- eS. Maria University Hospital, Terni, Italy
- fRigshospitalet, University of Copenhagen, Copenhagen, Denmark
- gMedical Center Leeuwarden, Leeuwarden, the Netherlands
- hInstitute of Cardiology, Department of Interventional Cardiology, Jagiellonian University Medical College, Krakow, Poland
- iChristchurch Hospital, Christchurch, New Zealand
- jHospital Santa Cruz, Carnaxide, Portugal
- kErasmus Medical Center, Rotterdam, the Netherlands
- lCentro Cardiologico Monzino, University of Milan, Milan, Italy
- mUniversitätsklinik für Kardiologie, Inselspital, Bern, Switzerland
- nImperial College of London, London, United Kingdom
- ↵∗Address for correspondence:
Dr. Patrick W. Serruys, Imperial College of London, Westblaak 98, Entrance B, 6th Floor, Rotterdam 3012KM, the Netherlands.
Objectives To assess the diagnostic accuracy of coronary computed tomography angiography (CTA) for bioresorbable vascular scaffold (BVS) evaluation.
Background Coronary CTA has emerged as a noninvasive method to evaluate patients with suspected or established coronary artery disease. The diagnostic accuracy of coronary CTA to evaluate angiographic outcomes after BVS implantation has not been well established.
Methods In the ABSORB II (A Bioresorbable Everolimus-Eluting Scaffold Versus a Metallic Everolimus-Eluting Stent II) study, patients were randomized either to receive treatment with the BVS or everolimus-eluting metallic stent. At the 3-year follow-up, 238 patients (258 lesions) treated with BVS underwent coronary angiography with intravascular ultrasound (IVUS) evaluation and coronary CTA. The diagnostic accuracy of coronary CTA was assessed by the area under the receiver-operating characteristic curve with coronary angiography and IVUS as references.
Results The mean difference in coronary CTA-derived minimal luminal diameter was −0.14 mm (limits of agreement −0.88 to 0.60) with quantitative coronary angiography as reference, whereas the mean difference in minimal lumen area was 0.73 mm2 (limits of agreement −1.85 to 3.30) with IVUS as reference. The per-scaffold diagnostic accuracy of coronary CTA for detecting stenosis based on coronary angiography diameter stenosis of ≥50% revealed an area under the receiver-operating characteristic curve of 0.88 (95% confidence interval [CI]: 0.82 to 0.92) with a sensitivity of 80% (95% CI: 28 to 99) and a specificity of 100% (95% CI: 98 to 100), whereas diagnostic accuracy based on IVUS minimal lumen area ≤2.5 mm2 showed an area under the receiver-operating characteristic curve of 0.83 (95% CI: 0.77 to 0.88) with a sensitivity of 71% (95% CI: 44 to 90) and a specificity of 82% (95% CI: 75 to 87). The diagnostic accuracy of coronary CTA was similar to coronary angiography in its ability to identify patients with a significant lesion based on the IVUS criteria (p = 0.75).
Conclusions Coronary CTA has good diagnostic accuracy to detect in-scaffold luminal obstruction and to assess luminal dimensions after BVS implantation. Coronary angiography and coronary CTA yielded similar diagnostic accuracy to identify the presence and severity of obstructive disease. Coronary CTA might become the method of choice for the evaluation of patients treated with BVS.
Coronary computed tomography angiography (CTA) has emerged as a noninvasive method to evaluate patients with suspected or established coronary artery disease (CAD). The luminal angiographic evaluation combined with atherosclerotic plaque characterization has been shown to provide diagnostic and prognostic information (1,2). Furthermore, the adoption of coronary CTA into the work-up of patients with suspected CAD has been shown to improve the prescription of guideline-directed medical treatment (i.e., statins) (3,4).
Nonetheless, in patients with known CAD who had undergone percutaneous coronary revascularization with metallic stents, a lower diagnostic performance of coronary CTA has been reported compared with patients with de novo lesions, mainly due to the artifact created by the metal in the coronary lumen (5,6). The radiolucency of the polymeric backbone in the Absorb everolimus-eluting bioresorbable vascular scaffold (BVS) allows for coronary CTA imaging without beam-hardening artifacts. In the first-in-human study, coronary CTA was performed at 18 and 72 months after the implantation of the BVS. Qualitative and quantitative coronary luminal evaluation was found to be feasible in more than 80% of the patients at both follow-up periods (7). However, the absence of concurrent invasive imaging precluded validation of the noninvasive findings.
In the ABSORB II (A Bioresorbable Everolimus-Eluting Scaffold Versus a Metallic Everolimus-Eluting Stent II; NCT01425281) study, systematic invasive and coronary CTA evaluations were performed in all patients treated with BVS at the 3-year follow-up (8). We sought to investigate the diagnostic accuracy of coronary CTA to evaluate angiographic outcomes after BVS implantation with coronary angiography and intravascular ultrasound (IVUS) examination as references.
The ABSORB II trial was a prospective, randomized, active-controlled, single-blind, parallel 2-arm, multicenter clinical trial. A total of 501 subjects were randomized either to treatment with the everolimus-eluting BVS (Abbott Vascular, Santa Clara, California) or the everolimus-eluting metallic stent (Xience, Abbott Vascular) in a 2:1 ratio at 46 sites in Europe and New Zealand. The trial protocol allowed the treatment of up to 2 de novo noncomplex native coronary artery lesions (9). All patients provided informed consent before being included in the trial and all participating sites received medical ethics committee approval for the study. Subjects had clinical follow-up at 30 and 180 days and 1, 2, and 3 years. Invasive (i.e., coronary angiography and IVUS) and noninvasive (i.e., coronary CTA) evaluation was scheduled at the 3-year follow-up. Abbott Vascular funded the study.
Study device and bioresorption
The BVS is a balloon-expandable device consisting of a polymer backbone of poly-L-lactide coated with a thin layer of a 1:1 mixture of poly(L-lactide-co-D, L-lactide) polymer and the antiproliferative drug everolimus containing 100 μg everolimus/cm2 scaffold (10). The scaffold is radiolucent, but has 2 radiopaque platinum markers of 244 μm at each end that allows for easy visualization on computed tomography (CT) and other imaging modalities. At 3 years, both poly-L-lactide and poly(L-lactide-co-D, L-lactide) are resorbed.
Coronary CTA acquisition and analysis
Images were acquired from CT scanners of 64-slice and beyond (Online Table 1). Standard acquisition techniques were used, which included nitrates before image acquisition and beta-blockers in patients with a heart rate >65 beats/min, tube settings depending on patient body mass index (80 to 140 kV), and axial scan protocols for patients with lower heart rates to reduce radiation doses, all at the discretion of the sites. Images were reconstructed using thin slices (range 0.5 to 0.67 mm) and medium smooth reconstruction filters in different phases. All data were stored on a DVD for core laboratory evaluation.
Data from the coronary CTA was analyzed off-line by an independent core laboratory (Cardialysis BV, Rotterdam, the Netherlands) using a validated cardiovascular analysis package (Aquarius iNtuition software version 4.4, Terarecon, Inc. Foster City, California). Vessel cross-sections were reconstructed at 0.5-mm longitudinal increments, extending 5 mm proximally and distally beyond the BVS, using the platinum radiopaque markers as landmarks. Automatic segmentation of the vessel lumen and wall was performed to measure the areas and diameters; manual correction was allowed. The window display was set at a level of 750 Hounsfield units, width at 250 Hounsfield units; the window level was adjusted if necessary. The mean, minimal lumen area (MLA) and minimal lumen diameters (MLD) were determined for each scaffold. The reference lumen area was calculated as the average between the proximal and distal edge mean lumen areas. The luminal percentage area stenosis was calculated as the ratio between the MLA and the reference area as a percentage of the reference. A significant area of stenosis was defined as ≥75%, which approximates 50% diameter stenosis (11).
Quantitative coronary angiography analysis
Quantitative coronary angiography (QCA) was performed at an independent angiographic core laboratory (Cardialysis BV), using a validated cardiovascular analysis package (CAAS 5.10 system, Pie Medical, Maastricht, the Netherlands) blinded to the coronary CTA results. The MLD was calculated as the average of the MLD in 2 projections. Angiographic binary restenosis was defined as a diameter stenosis of ≥50%. For the calculation of luminal areas, the videodensitometric method was used (12).
Grayscale IVUS was acquired with a 3.2-F, 45-MHz rotational IVUS catheter (Revolution 45 MHz, Volcano Corporation, Rancho Cordova, California), using automated pullbacks at 0.5 mm/s and 30 frames/s. All pullbacks were analyzed off-line at 1 mm longitudinal intervals and analyzed by an independent core laboratory (Cardialysis BV) using a validated software (QIvus 2.2, Medis, Leiden, the Netherlands). The methods of quantitative IVUS have been reported previously (13). The scaffold segments were identified by the first and the last cross-sectional IVUS frame in which the scaffold struts could be identified and/or where the proximal or distal metallic markers could be identified. The post-procedural region of interest was the segment beginning at 5 mm distal of the scaffold segment extending to the proximal 5 mm of the scaffold segment. For this analysis, significant lesions were defined as an MLA of ≤2.5 mm2 this has been shown to correlate with invasive fractional flow reserve of <0.80 in previous studies (14).
Binary variables are presented as percentages and continuous variables as mean ± SD or median (interquartile range [IQR]), as appropriate. The area under the curve (AUC) was used to assess the diagnostic accuracy of the test. The DeLong method was used for AUC comparison (15). The analyses are presented at the scaffold (lesion) level. The correlation between methods was assessed using the Pearson product-momentum coefficient (r). Correlations were compared using the Fisher r-to-z transformation method. Agreement between image modalities was assessed using the Passing-Bablok and the Bland–Altman method (16,17). Predictors of accuracy were also explored. For this purpose, accuracy was defined as a difference of ≤0.20 mm in the MLD between coronary CTA and QCA (18). Linear regression was performed to assess the variables associated with decreased accuracy. Logistic regression was performed to explore the predictors of accuracy. Statistical analyses were executed with SPSS Software version 24.0 (IBM, Armonk, New York) and MedCalc Software version 14.12 (Ostend, Belgium). Data are presented in agreement with Standards for Reporting of Diagnostic Accuracy Studies guidelines (19).
The primary endpoint was to determine the diagnostic accuracy of coronary CTA in detecting in-scaffold obstruction and assessing in-scaffold luminal dimensions with coronary angiography and IVUS as references. The secondary endpoint was to compare the diagnostic accuracy of coronary CTA and coronary angiography with IVUS as a reference.
Two hundred thirty-eight patients (258 lesions) treated with Absorb BVS underwent coronary CTA at 3 years. Coronary angiography with IVUS and coronary CTA were acquired with a median difference of 7 days (IQR: 1 to 13 days). The prevalence of obstructive lesions based on QCA diameter stenosis of >50% was 2.5% (5 lesions) and 8.7% (17 lesions) based on IVUS MLA <2.5 mm2. Quantitative coronary CTA analysis was feasible in 87% (208 patients, 220 lesions). In 6 patients, quantitative coronary CTA analysis was not performed due to target lesion revascularization with a metallic stent before the scheduled follow-up (Figure 1). Patient and procedural characteristics are shown in Table 1.
Coronary CTA analysis
The in-scaffold mean MLD and MLA were 2.09 ± 0.5 mm and 3.62 ± 1.6 mm2 corresponding with a diameter and area stenosis of 14 ± 13% and 24 ± 23%, respectively. Mean luminal diameter and area were 2.48 ± 0.5 mm and 5.13 ± 2.0 mm2, respectively. Four lesions (1.8%) presented with angiographic in-scaffold restenosis.
Comparison between coronary CTA and QCA analysis
In 201 lesions (189 patients), the mean difference between coronary CTA and QCA in MLD and MLA were −0.14 mm (limits of agreement [LOA]: −0.88 to 0.60) and −0.72 mm2 (LOA: −3.35 to 1.91), respectively. A significant correlation was found in all measurements derived from coronary CTA and QCA. The Passing-Bablok regression analysis disclosed comparable measurements in MLD (Figure 2). The AUC for coronary CTA area of stenosis was 0.88 (95% confidence interval [CI]: 0.82 to 0.92) for the diagnosis of a scaffold diameter stenosis of ≥50% assessed by QCA. The sensitivity for detecting obstructive stenosis was 80% (95% CI: 28 to 99), and the specificity was 100% (95% CI: 98 to 100). The positive and negative predictive values were 100% (95% CI: 29 to 100) and 99% (95% CI: 96.4 to 99.9), respectively. A sensitivity analysis with a threshold of QCA diameter stenosis of ≥70% yielded similar diagnostic accuracy (Online Table 3). The best cutpoints to identify significant lesions are shown in Table 2.
Comparison between coronary CTA and IVUS analysis
In 196 lesions (187 patients), the mean difference between coronary CTA and IVUS in MLD and MLA were 0.22 mm (LOA: −0.50 to 0.93) and 0.73 mm2 (LOA: −1.85 to 3.30). A significant correlation between measurements derived from coronary CTA and IVUS was observed. Passing-Bablok regression analysis exposed a systematic difference in the assessment of MLA (Figure 3). The AUC for coronary CTA MLA was 0.83 (95% CI: 0.77 to 0.88) for the diagnosis of a scaffold MLA ≤2.5 mm2 assessed with IVUS. The sensitivity for detecting obstructive stenosis based on IVUS criteria was 71% (95% CI: 44 to 90), and the specificity was 82% (95% CI: 75 to 87). The positive and negative predictive values were 27% (95% CI: 19 to 36) and 97% (95% CI: 93 to 98), respectively. The best cutpoints to identify significant lesions are shown in Table 2.
Comparison between coronary CTA, QCA, and IVUS
In 189 lesions (180 patients), the 3 modalities were available for analysis. Compared with IVUS, coronary CTA and QCA underestimated in-scaffold MLD by 10% and 16%, and MLA by 16% and 34%, respectively (Figure 4). On the basis of IVUS, a similar diagnostic performance between coronary CTA and QCA was found (AUC for MLD coronary CTA: 0.82 [95% CI: 0.76 to 0.87] vs. AUC for QCA: 0.81 [95% CI: 0.74 to 0.86]; difference in the AUC 0.013 [95% CI: −0.06 to 0.09], p = 0.756; and AUC for MLA coronary CTA 0.82 [95% CI: 0.76 to 0.87] vs. AUC for QCA 0.80 [95% CI: 0.74 to 0.86]; difference in the AUC 0.02 [95% CI: −0.041 to 0.076]; p = 0.565) (Figure 5). With IVUS as a reference, the strength of the correlation was higher for coronary angiography compared with coronary CTA (MLD rcoronary CTA = 0.76 vs. rQCA = 0.87 [p < 0.001] and MLA rcoronary CTA = 0.63 vs. rQCA = 0.78 [p = 0.003]) (Figure 6).
Predictors of coronary CTA accuracy
The univariate analysis showed that heart rate during coronary CTA acquisition (B coefficient: 0.005; p = 0.09) and the number of detectors of the CT scanner (B coefficient: −0.279; p = 0.015) were predictors of improved accuracy; however, after adjustment for vessel diameter and body mass index, no independent predictor of accuracy was found.
The main findings of this study can be summarized as follows: 1) coronary CTA is accurate for detecting the presence and the severity of in-scaffold obstructions with QCA and IVUS as references; 2) coronary CTA and coronary angiography underestimate in-scaffold luminal dimensions compared with IVUS; and 3) coronary CTA and QCA yielded similar diagnostic accuracy for in-scaffold evaluation with IVUS as reference.
Contemporary studies using multidetector CT have shown high sensitivity, specificity, and negative predictive value to exclude obstructive lesions in patients with suspected CAD (20,21). However, in patients with previous percutaneous coronary interventions and metallic stent, the performance of coronary CTA is hampered by the blooming artifact generated by the metal, limiting the use of this technology in this subset of patients (5). One of the advantages of the polymeric scaffold is that it allows for noninvasive coronary CTA imaging without beam-hardening artifacts. For the first time, in the ABSORB II study, invasive and noninvasive angiographic images were acquired at the same time point. Coronary CTA was shown to be accurate for the detection of significant lesions and for quantitative in-scaffold luminal evaluation with QCA and IVUS as references. At the scaffold level, the sensitivity, specificity, and negative predictive values were 71%, 82%, and 97%, respectively, with IVUS as reference. This finding suggests that coronary CTA can be used in the clinical setting to exclude in-scaffold restenosis and in the research setting to assess angiographic outcomes after BVS implantation.
With IVUS as reference, in-scaffold luminal diameters were underestimated by 16% with QCA and by 10% with coronary CTA. IVUS has been shown to overestimate lumen dimensions compared with an imaging gold standard such as optical coherence tomography; and both coronary CTA and coronary angiography have been shown to underestimate in-scaffold luminal diameters compared with optical coherence tomography (22–24). Similarly, Voros et al. (25) have shown that, in patients with intermediate coronary stenosis, coronary CTA significantly underestimates MLD by 21% compared with IVUS (p = 0.0001), whereas coronary CTA underestimates MLD by 10% compared with coronary angiography (p = 0.12). Moreover, in the FIGURE-OUT (Functional Imaging criteria for GUiding REview of invasive coronary angiOgraphy, intravascular Ultrasound, and coronary computed Tomographic angiography) study, coronary CTA was shown to underestimate MLD compared with QCA (26). In the present study, no systematic or proportional differences in the evaluation of MLD between coronary CTA and QCA were found. The clinical implications of the underestimation of luminal dimension with angiographic methods and the overestimation of intravascular sound-based methods requires further investigation.
Although the polymeric backbone of the BVS is radiolucent, the platinum radiopaque markers localized at each edge of the scaffold might interfere with luminal segmentation (24). In this cohort, a post hoc analysis revealed 7 patients with differences in MLA of more than 3 mm2 between coronary CTA and IVUS. The blooming artifact from the marker (2 cases), calcifications in the scaffolded region (3 cases), and insufficient image quality (2 cases) generated an overestimation of the degree of stenosis. Moreover, the accuracy of coronary CTA has been found to correlate with the quality of the image (5). Several factors may influence image quality. In the present study, 88% of CT scanners were equipped with at least 128-row detector that have increased spatial resolution and improved image quality compared with the 64-row detector scanners (27). Also, heart rate during coronary CTA acquisition, which is a key factor in determining image quality was comparable with previous reports (i.e., 59 beats/min [IQR: 54 to 65 beats/min]) (1). Heart rate during CT scan, the number of detectors of the CT scanner, body mass index, and severity of vessel calcifications have been shown to be predictors of image quality. Although in the univariate analysis the heart rate during the CT scan and CT scanner type were shown to be predictors of accuracy, the multivariate logistic regression analysis did not show any independent predictor.
Since the introduction of percutaneous coronary interventions, coronary angiography has been the preferred method to assess angiographic outcomes. Late lumen loss and restenosis are widely accepted endpoints to evaluate the efficacy of new devices. With unparalleled technological development, coronary CTA has been shown to be an alternative tool for angiographic investigation. Recently, Budoff et al. (20) have shown that, in nontreated coronary vessels, coronary CTA and coronary angiography have a similar diagnostic accuracy to detect obstructive lesions. The present study extends these findings to patients treated with BVS. Indeed, coronary CTA and coronary angiography yielded similar diagnostic accuracy for the detection of in-scaffold stenosis. In the clinical arena, however, a higher prevalence of in-scaffold restenosis is expected during a clinically driven investigation. By means of the Bayes’ theorem, assuming a higher disease prevalence (e.g., 25%) coronary CTA shows a high negative predictive value (92% [95% CI: 85 to 96]), reassuring its value in ruling out in-scaffold restenosis (28).
First, the efficacy of drug elution on neointimal hyperplasia formation has decreased restenosis rate; in this study, the prevalence of disease (i.e., restenosis) was low (i.e., 2.5% based on QCA diameter stenosis >50%). The low prevalence of restenosis could have contributed to the high accuracy observed in this study. Moreover, referral bias might be present because symptomatic patients, in whom the prevalence of restenosis is expected to be higher, were referred directly to coronary angiography. Second, the use of anatomic (QCA and IVUS-derived luminal dimensions) instead of functional assessment (i.e., fractional flow reserve) as clinical references precludes making a strong conclusion regarding the use of coronary CTA to assess the need of reintervention. Third, given the noncomplex nature of the lesion included in this study, these results cannot be extrapolated to specific lesion subsets such as aorto-ostial lesions, bifurcations, vessels with excessive tortuosity, or heavy calcifications. Fourth, adverse reactions and complications after CT scan or coronary angiography were not collected systematically; for this reason, no conclusion regarding the advantages in terms of safety of a noninvasive method can be made. Fifth, due the fact that coronary CTA was acquired at one time point (i.e., 3 years), we were unable to calculate late lumen loss. Sixth, the CT scanners used were neither uniform nor of the latest technology (e.g., 320-row detector CT scanner was used in 5% of the cases). Nevertheless, the achievement of the final results reinforces the validity of coronary CTA in the BVS evaluation.
Coronary CTA has a good diagnostic accuracy to detect in-scaffold luminal obstruction and to assess luminal dimensions after BVS implantation. Coronary CTA and coronary angiography yielded similar diagnostic accuracy to identify the presence and severity of obstructive disease. Thus, coronary CTA might become the method of choice for the evaluation of patients treated with bioresorbable scaffolds.
COMPETENCY IN MEDICAL KNOWLEDGE: In patients treated with polymeric BVS, coronary CTA can be used to assess for scaffold patency and luminal dimensions.
TRANSLATIONAL OUTLOOK: This study demonstrates that coronary CTA is accurate for the evaluation poly-l-lactic acid–based BVS. Futures studies are needed to establish the accuracy of coronary CTA in patients with scaffolds made from nonpolymeric materials.
For supplemental tables and figures, please see the online version of this paper.
Dr. Fajadet has received educational grants from Abbott, Boston Scientific, Medtronic, and Terumo. Drs. Dudek, Piek, Serruys, and Onuma are members of the International Advisory Board for Abbott Vascular. Dr. Chevalier is a consultant for Abbott Vascular. Dr. Windecker has received research contracts from Boston Scientific, Biotronik, Bracco, and Terumo. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
H. Vernon Anderson, MD, served as the Guest Editor for this paper.
- Abbreviations and Acronyms
- area under the curve
- bioresorbable vascular scaffold
- coronary artery disease
- computed tomography angiography
- intravascular ultrasound
- minimal lumen area
- minimal lumen diameter
- quantitative coronary angiography
- Received February 17, 2017.
- Revision received April 11, 2017.
- Accepted April 18, 2017.
- 2017 American College of Cardiology Foundation
- Motoyama S.,
- Sarai M.,
- Harigaya H.,
- et al.
- Williams M.C.,
- Hunter A.,
- Shah A.S.,
- et al.
- Yan R.T.,
- Miller J.M.,
- Rochitte C.E.,
- et al.
- Rief M.,
- Zimmermann E.,
- Stenzel F.,
- et al.
- Onuma Y.,
- Dudek D.,
- Thuesen L.,
- et al.
- Serruys P.W.,
- Chevalier B.,
- Sotomi Y.,
- et al.
- Diletti R.,
- Serruys P.W.,
- Farooq V.,
- et al.
- Onuma Y.,
- Serruys P.W.
- Abbara S.,
- Blanke P.,
- Maroules C.D.,
- et al.
- Serruys P.W.,
- Chevalier B.,
- Dudek D.,
- et al.
- Jang J.S.,
- Shin H.C.,
- Bae J.S.,
- et al.
- Altman D.G.,
- Bland J.M.
- Herrington D.M.,
- Siebes M.,
- Sokol D.K.,
- Siu C.O.,
- Walford G.D.
- Bossuyt P.M.,
- Reitsma J.B.,
- Bruns D.E.,
- et al.
- Budoff M.J.,
- Nakazato R.,
- Mancini G.B.,
- et al.
- Rochitte C.E.,
- George R.T.,
- Chen M.Y.,
- et al.
- Kubo T.,
- Akasaka T.,
- Shite J.,
- et al.
- Sotomi Y.,
- Onuma Y.,
- Suwannasom P.,
- et al.
- Collet C.,
- Sotomi Y.,
- Cavalcante R.,
- et al.
- Voros S.,
- Rinehart S.,
- Qian Z.,
- et al.
- Doh J.H.,
- Koo B.K.,
- Nam C.W.,
- et al.