Recently, real-time 3-dimensional (3D) echocardiography (RT-

Transcription

1 CONGENTIAL HEART DISEASE Value of Real-Time 3-Dimensional Echocardiography Sectional Diagnosis in Complex Congenital Heart Disease Evaluated by Receiver Operating Characteristic Analysis Guo-zhen Chen, MD, Guo-ying Huang, MD, Zi-yu Tao, BS, Xiao-qin Liu, MD, and Qi-shan Lin, MD, Shanghai, China Objective: We sought to evaluate the value of real-time 3-dimensional (3D) echocardiography (RT-3DE) in the pathologic morphology of complex congenital heart disease (CHD) using receiver operating characteristic (ROC) analysis. Methods: Thirty patients with complex CHD were examined by 2-dimensional echocardiography (2DE) and RT-3DE. Their diagnoses of pathologic morphology were made by 3 echocardiographers blinded to clinical data. Their value was evaluated by ROC analysis using a 5-point categorical scale in major cardiovascular structures and spatial relationships among complex CHD. Results: Compared with surgical findings, there were 75.6% of the definitely correct in RT-3DE and 64.4% in 2DE. ROC curve for RT-3DE was located left-superior to that for 2DE, and the area under ROC curve for RT-3DE (A 0.96) was higher than that for 2DE (A 0.90). There was significant difference between them by the method of Z test (Z 2.64, P.0083). Conclusions: With the spatial 3D display of cardiovascular structural malformations, RT-3DE, coupled with conventional Doppler 2DE, may add information on complex CHD. Recently, real-time 3-dimensional (3D) echocardiography (RT- 3DE) proved to be able to display detailed spatial anatomic and pathologic information similar to surgeon s visualization, with online or instantaneous 3D imaging acquisition, anyplane sectional projections and rotations, and higher-quality dynamic 3D rendering. Thus, it could be expected to make an accurate noninvasive diagnosis for complex congenital heart disease (CHD). However, RT-3DE is an initial echocardiographic technique; so far there are rare reports of systematic objective evaluations on application value of RT-3DE in cardiac pathologic morphology of complex CHD. 1-4 Therefore, based on our previous studies, 5-10 the diagnostic effect of RT-3DE in awide range of complex CHD was objectively assessed by receiver operating characteristic (ROC) analysis, as compared with 2-dimensional (2D) echocardiography (2DE). 11,12 METHODS Patient Population From October 2004 to May 2006, 30 patients with complex CHD (19 boys and 11 girls; mean age years) who underwent operation were recruited to the study. The detailed information is showedintable1,includingthesex,age,mainsurgicaldiagnosis,and surgical procedure. Two-dimensional Echocardiography and RT-3DE Examination The patients with complex CHD were examined by 2DE using an echocardiographicsystem(sonos5500or7500,philips,saronna,italy) with common 2DE transducer, and recorded on videotape. Then the patientswereexaminedbyrt-3deusinganechocardiographicsystem (Sonos 7500, Philips) with its X4 matrix transducer in the subcostal, apical,andparasternalregions.alldataofrt-3dewerestoreddigitally onharddiskandcd-romforofflinepostprocessinganddataanalysis. RT-3DE Sectional Diagnosis Based on the previous studies, 5-10 pathologic morphology of complex CHD from the above-mentioned data sets of RT-3DE was reviewedanddiagnosedbythedesigneddiagnosticapproachofserial novelrt-3desectionalvolumetricviewscombinedwiththesequential segmental approach of Van Praagh et al Namely, detailed cardiovascular anatomic structures on different orientation, visual angle, and sectional view were displayed and observed by aseries of novel RT-3DE sectional volumetric views around the center of From the Pediatric Heart Center, Children s Hospital of Fudan University, Shanghai, China. This study was in part financially supported by China Postdoctoral Scienceatrioventricular and ventricular-arterial connective regions, which Foundation, the National Science Foundation (grant serial number ). included volumetric view of 4-chamber heart, volumetric view of Reprint requests: Guo-ying Huang, MD, Pediatric Heart Center, Children s interventricularseptumfromleftsideandfromrightside,volumetric Hospital of Fudan University, 183 Fenglin Road, Shanghai , China ( viewofinteratrialseptumfromleftsideandfromrightside,volumetric view of atrioventricular valves from top and from bottom, and gyhuang@shmu.edu.cn) /$34.00 volumetricviewofsemilunarvalvesfromtopandfrombottom.then Copyright 2008 by the American Society of Echocardiography. RT-3DE sectional diagnoses of complex CHD were made by analyzingthe3segmentsincludingatria,ventricles,andgreatarteries,andat doi: /j.echo

2 Journal of the American Society of Echocardiography Chen et al 459 Volume 21 Number 5 Table 1 Clinical data of patients with complex congenital heart disease for receiver operating characteristic analysis No. Sex Age, y Surgical findings Surgical procedure 1 F 9.0 TOF and perimembranous VSD Intracardiac repair and enlarged RVOT and PA 2 F 3.8 Primum ASD and mitral cleft Surgical for primum ASD and mitral cleft repair 3 M 3.0 DORV (right anterior aorta) and doubly committed VSD and PS Bidirectional cavopulmonary shunt 4 M 0.7 Corrected TGA and secundum ASD and PS and perimembranous VSD repair and ASD closure and enlarged PA VSD 5 M 1.0 TOF and perimembranous VSD Intracardiac repair and enlarged RVOT and PA 6 F 0.7 Intracardiac TAPVD and complete AVSD and single atrium Surgical repair for complete AVSD and TAPVD 7 M 0.8 Single atrium and mitral cleft Surgical repair for single atrium and mitral cleft 8 M 9.0 Pulmonary atresia with VSD (aorta from RV) Right bidirectional Glenn shunt 9 M 0.1 Complete TGA (right anterior aorta) and perimembranous VSD Switch procedure and VSD repair 10 M 8.0 Ebstein s anomaly of tricuspid valve Plicated atrialized RV and tricuspid valvuloplasty 11 M 0.0 Complete TGA (right anterior aorta) and perimembranous VSD Switch procedure and VSD repair 12 M 1.0 TOF and perimembranous VSD and PFO Intracardiac repair and enlarged RVOT and PA 13 M 3.4 TOF and perimembranous VSD Intracardiac repair and enlarged RVOT and PA 14 M 7.0 Dextrocardia and pulmonary atresia and VSD and ventricular inversion Modified Blalock-Taussig shunt 15 M 0.0 Balanced single ventricle and TA Modified Blalock-Taussig shunt 16 F 0.6 TOF and VSD (superior to the crest supraventricularis) Intracardiac repair and enlarged RVOT and PA 17 M 0.6 Complete TGA (anterior aorta) and perimembranous VSD and PS Rastelli procedure 18 F 1.3 TOF and subaortic VSD and PFO Intracardiac repair and enlarged RVOT and PA 19 M 0.6 TOF and perimembranous VSD and PFO Intracardiac repair and enlarged RVOT and PA 20 F 1.3 TOF and VSD (inferior to the crest supraventricularis) Intracardiac repair and enlarged RVOT and PA 21 M 0.1 DORV and subpulmonary VSD PA banding procedure 22 M 0.6 TOF and perimembranous VSD and PFO Intracardiac repair and enlarged RVOT and PA 23 F 0.3 TOF and perimembranous VSD Intracardiac repair and enlarged RVOT and PA 24 M 0.1 Complete TGA (anterior aorta) with intact ventricular septum Switch procedure 25 M 0.1 Complete TGA (anterior aorta) and ASD Banding and Blalock-Taussig shunt then switch procedure 26 F 0.5 TOF and perimembranous VSD Intracardiac repair and enlarged RVOT and PA 27 F 0.0 Pulmonary atresia with intact ventricular septum Pulmonary valvulotomy and enlarged RVOT and PA 28 M 0.7 TOF and perimembranous VSD Intracardiac repair and enlarged RVOT and PA 29 F 4.0 TOF and perimembranous VSD Intracardiac repair and enlarged RVOT and PA 30 F 0.5 TA and secundum ASD and hypoplastic RV and inlet VSD Right bidirectional Glenn shunt ASD, Atrial septal defect; AVSD, atrioventricular septal defect; DORV, double outlet right ventricle; F, female; M, male; PA, pulmonary artery; PFO, patent foramen ovale; PS, pulmonary stenosis; RV, right ventricle; RVOT, right ventricular outlet; TA, tricuspid atresia; TAPVD, total anomalous pulmonary venous drainage; TGA, transposition of the great arteries; TOF, tetralogy of Fallot; VSD, ventricular septal defect. two connections including atrioventricular and ventricular-arterial connections. ROC Analysis The materials from 2DE recorded on videotape and RT-3DE raw data stored in CD-ROM were reviewed by 3 experienced echocardiographers blinded to the names and clinical details of patients with complex CHD. Using a series of novel RT-3DE volumetric views combined with the sequential segmental approach of Van Praagh et al, the 3 echocardiographers blinded to clinical date made their respective diagnoses of RT-3DE and 2DE in the patients with complex CHD, through observing atrioventricular connection, ventricular-arterial connection, location of septal defect, correlated position of great vessels, and others around the center of atrioventricular and ventricular-arterial connective regions. According to surgical findings as gold standard, all diagnostic results of RT-3DE and 2DE made by the 3 echocardiographers blinded to clinical date were compared and assessed by ROC analysis, using a 5-point categorical scale ranging from definitely correct to definitely wrong, including definitely correct, showing that all echocardiographic diagnoses were confirmed with surgical findings; correct, showing that main echocardiographic diagnoses were confirmed with surgical findings; partially correct, showing that partial echocardiographic diagnoses were not confirmed with surgical findings; wrong, showing that main echocardiographic diagnoses were not confirmed with surgical findings; and definitely wrong, showing that all echocardiographic diagnoses were not confirmed with surgical findings. 11,12 Statistics In the atrioventricular and ventricular-arterial connective levels of complex CHD, all assessments of ROC analysis were categorically collected. Then, ROC curves of RT-3DE and 2DE were drawn, and their areas under ROC curves were calculated and analyzed by the method of Z test. Statistical significance was defined as a value of P less than.05. RESULTS The RT-3DE and 2DE materials of 30 patients with complex CHD were respectively diagnosed by the 3 echocardiographers blinded to clinical data. After their diagnostic results were compared and assessed by ROC analysis, 90 assessments of RT-3DE and 90 assessments of 2DE were collected respectively (Table 2). Compared with surgical findings, there were 64.4% (58/90) of the definitely correct in 2DE and 75.6% (68/90) in RT-3DE. With the spatial superiority of RT-3DE over 2DE, RT-3DE provided some additional cardiovascular

3 460 Chen et al Journal of the American Society of Echocardiography May 2008 Table 2 Categorical collection of receiver operating characteristic assessment for real-time 3-dimensional echocardiography and 2-dimensional echocardiography diagnosing the pathologic morphology of complex congenital heart disease ROC categorical scale Definitely correct Correct Partially correct Wrong Definitely wrong Total AUC SEM 2DE (Concordance) (Discordance) DE (Concordance) (Discordance) AUC, Area under ROC curve. information for evaluation of CHD including atrio-ventricular and ventricular-arterial continuity; details of septal defects such as size, shape, and location; and spatial appreciation of the defects related with great arteries. ROC Curve Based on calculation of sensitivity, specificity, and false-positive rate of RT-3DE and 2DE diagnosing complex CHD, ROC curves of RT-3DE and 2DE were respectively drawn with the sensitivity and falsepositive rate of different truncated dot (Figure 1). From Figure 1, ROC curve for RT-3DE was located left-superior to that for 2DE. Namely, ROC curve for RT-3DE was closer to the ideal discrimination function than that for 2DE. Area Under ROC Curve The area under ROC curve for RT-3DE (A 0.96) was higher than that for 2DE (A 0.90). There was significant difference (P.05) between them by the method of Z test (Z 2.64, two-sided P.0083). Thus, both ROC curves and the areas under ROC curve showed that the sensitivity and specificity of RT-3DE in the diagnosis of cardiovascular structural malformations of complex CHD were higher than those of 2DE (Figures 2 to 4). DISCUSSION Figure 1 Receiver operating characteristic curve for real-time 3-dimensional echocardiography (3DE) and 2-dimensional echocardiography (2DE) diagnosing pathologic morphology of complex congenital heart disease. Se, Sensitivity; 1-Sp, falsepositive rate. Color figure online. Although 2DE can provide an enormous amount of cardiac structural information, structural analysis requires that the observer be able to mentally integrate these planar data into a 3D construct of the heart by interrelating information from a series of adjacent views The advent of 3D echocardiography (3DE) eliminates the need for each observer to analyze the tomographic images and generate his or her concept of the heart. 19 In our previous studies of conventional 3DE, 5-8 3DE sectional views of cardiac septa from left side and from right side could display en face shape, size, and position of septal defect and its surroundings, and 3DE sectional views of cardiac valves from top and from bottom could display en face surface area and correlation of atrioventricular and semilunar valves. Compared with 2D image showing only a single cut, there are some clinical advantages of 3DE showing en face view of entire structure and the spatial relationship of cardiac abnormalities. The additional information of 3DE could modify therapeutic strategy. Especially, new RT-3DE acquiring online 3D imaging similar to standard 2DE progressively ameliorates some 3DE limitations such as time-consuming, cumbersome, and movement artifacts from conventional 3DE with respiratory and electrocardiographic gating and, thus, would become the powerful imaging tool of noninvasive 3D diagnosis in complex CHD. 1-4,9,10 However, ROC analysis originally used in the assessment of radar images was introduced to an evaluation of comparing some static images of transesophageal 3DE to those of 2DE by Belohlavek et al 11 in 1994, and firstly confirmed the feasible and effective diagnostic value of 3DE assessed by ROC analysis, but most of the estimations of RT-3DE diagnostic effect in previous reports were easily subjective as a result of individual images, specific patients, and private opinions of echocardiographers. Currently, there is still a lack of objective evaluating of studies of RT-3DE diagnostic value on a wide range of CHD in children. 1-4,20-23 Therefore, ROC analysis, which was recognized as the objectively better half-quantitive method to assess two or more imaging techniques, was used in the study. 11,12 Compared with surgical findings as gold standard, the effect of RT-3DE and 2DE diagnosing atrioventricular and ventricular-arterial abnormal connections and other structural malformations in complex CHD assessments were blindly assessed by ROC analysis. From our findings, both ROC curve for RT-3DE and that for 2DE were located left-superior, and their areas under ROC curves were also large. The results showed that the two imaging techniques had the better value in the diagnosis of complex CHD. However, because the assessment of anatomic details and relationships in CHD are very important, all of which may influence management options, decision

4 Journal of the American Society of Echocardiography Volume 21 Number 5 Chen et al 461 Figure 2 Real-time 3-dimensional echocardiography (RT-3DE) images of complex congenital heart disease were compared with those of 2-dimensional echocardiography (2DE). (A), Volumetric view of 4-chamber heart of partial atrioventricular septal defect (ASD) displayed by RT-3DE. (B), 2DE was used as control of A. (C), Volumetric view of interventricular septum from right side of Ebstein s anomaly displayed by RT-3DE. (D), 2DE was used as control of C. AO, Aorta; LA, left atrium; LV, left ventricle; RA, right atrium; RV, right ventricle. Color figure online. Figure 3 Real-time 3-dimensional echocardiography (RT-3DE) images of complex congenital heart disease were compared with those of 2-dimensional echocardiography (2DE). (A), Volumetric view of cardiac valves from top of pulmonary atresia with ventricular septal defect displayed by RT-3DE. (B), 2DE was used as control of A. (C), Volumetric view of atrioventricular valves from bottom of tetralogy of Fallot displayed by RT-3DE. (D), 2DE was used as control of C. AO, Aorta; LA, left atrium; LV, left ventricle; MV, mitral valves; PA, pulmonary artery; RA, right atrium; RV, right ventricle; TV, tricuspid valves. Color figure online.

5 462 Chen et al Journal of the American Society of Echocardiography May 2008 Figure 4 Real-time 3-dimensional echocardiography (RT-3DE) images of complex congenital heart disease were compared with those of 2-dimensional echocardiography (2DE). (A), Volumetric view of long axis of overriding aorta (AO) displayed by RT-3DE. (B), 2DE was used as control of A.(C), Volumetric view of interventricular septum (IVS) from right side of complete transposition of great arteries displayed by RT-3DE. (D), 2DE was used as control of C. LA, Left atrium; LV, left ventricle; RV, right ventricle. Color figure online. making, surgical operations, and percutaneous interventional procedures, in the study the detail observation of intracardiac structures and their spatial correlations was stressed and regarded as an assessment standard of ROC analysis, including atrioventricular connection, ventricular-arterial connection, location of septal defect, correlated position of great vessels, and others around the center of atrioventricular and ventricular-arterial connective regions. From our findings, with the advantage of dynamic 3D display of cardiovascular structures and morphologies, ROC curve for RT-3DE was superior to that for 2DE, and its area (0.96) was also higher than 2DE (0.90). Namely, ROC curve for RT-3DE was closer to the ideal discrimination function than that for 2DE. Conclusions With the spatial 3D display of cardiovascular structural malformations, RT-3DE, coupled with conventional Doppler 2DE, may add information on complex CHD. REFERENCES 1. von Bardeleben RS, Kuhl HP, Mohr-Kahaly S, Franke A. Second-generation real-time three-dimensional echocardiography. Finally on its way into clinical cardiology? Z Kardiol 2004;93:IV Chan KL, Liu X, Ascah KJ, Beauchesne LM, Burwash IG. Comparison of real-time 3-dimensional echocardiography with conventional 2-dimensional echocardiography in the assessment of structural heart disease. J Am Soc Echocardiogr 2004;17: Nanda NC, Miller AP. Real time three-dimensional echocardiography: specific indications and incremental value over traditional echocardiography. J Cardiol 2006;48: Hung J, Lang R, Flachskampf F, Shernan SK, McCulloch ML, Adams DB, et al, American Society of Echocardiography. Three-dimensional echocardiography: a review of the current status and future directions. J Am Soc Echocardiogr 2007;20: Chen GZ, Sun K, Chen SB, Zhang YQ, Jiang H, Huang MR, et al. Study on transthoracic three-dimensional echocardiographic diagnosis of complex congenital heart malformations in children [in Chinese]. Chin J Ultrasonogr 2002;11: Chen GZ, Sun K, Chen SB, Zhang YQ, Jiang H. Study on transthoracic three-dimensional echocardiographic diagnosis of conotruncal defects in children [in Chinese]. Chin J Med Imaging Technol 2002;18: Chen GZ, Sun K, Chen SB, Zhang YQ, Jiang H. Transthoracic threedimensional echocardiographic diagnosis of double outlet right ventricle [in Chinese]. Chin J Ultrasound Med 2003;19: Chen GZ, Sun K, Huang GY, Chen SB, Zhang YQ. Patho-morphologic diagnosis of conotruncal defects by using transthoracic three-dimensional echocardiography [in Chinese]. Chin J Pediatr 2005;43: Chen GZ, Huang GY, Sun K, Liang XC, Chen SB, Chen S. Study of real-time three-dimensional echocardiographic diagnostic method of pathological morphology in complex congenital heart disease [in Chinese]. Chin J Ultrasound Med 2006;22: Chen GZ, Huang GY, Liang XC, Ma XJ, Chen WD, Tao ZY, et al. Methodological study on real-time three-dimensional echocardiography and its application in the diagnosis of complex congenital heart disease. Chin Med J 2006;119: Belohlavek M, Foley DA, Seward JB, Greenleaf JF. Diagnostic performance of two-dimensional versus three-dimensional transesophageal echocardiographic images of selected pathologies evaluated by receiver operating characterstic analysis. Echocardiography 1994;11: Hanley JA, Mcneil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982;143:29-36.

6 Journal of the American Society of Echocardiography Chen et al 463 Volume 21 Number Van Praagh R, Durnin RE, Jockin H, Wagner HR, Korns M, Garabedian H, et al. Anatomically corrected malposition of the great arteries (S, D, L). Circulation 1975;51: Van Praagh R. The segmental approach clarified. Cardiovasc Intervent Radiol 1984;7: Van Praagh R, Santini F, Geva T. Segmental situs in congenital heart disease: a fundamental concept. G Ital Cardiol 1990;20: Saraclar M, Cil E, Ozkutlu S. Echocardiography for the diagnosis of congenital cardiac anomalies with multiple lesions. Pediatr Cardiol 1996;17: Sanders SP, Bierman FZ, Williams RG. Conotruncal malformations: diagnosis in infancy using subxiphoid 2-dimensional echocardiography. Am J Cardiol 1982;50: Tometzki AJ, Suda K, Kohl T, Kovalchin JP, Silverman NH. Accuracy of prenatal echocardiographic diagnosis and prognosis of fetuses with conotruncal anomalies. J Am Coll Cardiol 1999;33: Bartel T, Müiller S, Geibel A. Preoperative assessment of cor triatriatum in an adult by dynamic three-dimensional echocardiography was more informative than transesophageal echocardiography or magnetic resonance imaging. Br Heart J 1994;72: van den Bosch AE, van Dijk VF, McGhie JS, Bogers AJ, Roos-Hesselink JW, Simoons ML, et al. Real-time transthoracic three-dimensional echocardiography provides additional information of left-sided AV valve morphology after AVSD repair. Int J Cardiol 2006;106: Rawlins DB, Austin C, Simpson JM. Live three-dimensional paediatric intraoperative epicardial echocardiography as a guide to surgical repair of atrioventricular valves. Cardiol Young 2006;16: Seliem MA, Fedec A, Cohen MS, Ewing S, Farrell PE Jr, Rychik J, et al. Real-time 3-dimensional echocardiographic imaging of congenital heart disease using matrix-array technology: freehand real-time scanning adds instant morphologic details not well delineated by conventional 2-dimensional imaging. J Am Soc Echocardiogr 2006;19: Houck RC, Cooke JE, Gill EA. Live 3D echocardiography: a replacement for traditional 2D echocardiography? Am J Roentgenol 2006; 187: