| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Pediatric Imaging |
1 From the Clinic for Congenital Heart Disease (P. Beerbaum, H.K., H.E., P. Barth, J.H., H.M.) and Clinic for Thoracic and Cardiovascular Surgery (U.B.), Heart and Diabetes Center, North Rhine-Westfalia, Ruhr-University Bochum, Georgstrasse 11, D-32545 Bad Oeynhausen, Germany; and Philips Medical Systems, Best, the Netherlands (J.G.). Received June 28, 2002; revision requested August 28; final revision received December 10; accepted January 14, 2003. Supported in part by Philips Medical Systems, Best, the Netherlands. Address correspondence to P. Beerbaum (e-mail: pbeerbaum@hdz-nrw.de).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Sixty-five children (mean age, 5.4 years ± 2.7 [SD]) with ASD and inconclusive TTE results underwent phase-contrast MR imaging. Defect size and rim distances measured on MR imaging sections obtained in the ASD plane and from the defect to the venae cavae, aortic root, and atrioventricular valves were compared with transesophageal echocardiographic (TEE) findings (n = 30) during transcatheter closure or surgical measurements (n = 40) by using Bland-Altman analysis. Inflow MR angiography was compared with invasive cine angiocardiography for detection of associated venous anomalies.
RESULTS: For ASD size, mean differences were less than 1 mm between MR imaging and TEE measurements (with upper and lower limits of agreement between 2.3 and -3.3 mm) and were between 1.2 and -1.6 mm between MR imaging and surgical measurements (with upper and lower limits of agreement between 4.7 and -5.2 mm). Septal rim measurements at MR imaging agreed fairly well with TEE and surgical results. Septal length was overestimated at MR imaging versus TEE (mean difference, 3.0 mm; upper and lower limits of agreement, between 8.0 and -2.8 mm), but MR imaging septal length measurements agreed with surgical results. Rim distance to coronary sinus was difficult to assess. MR imaging enabled referral of 25 of 30 patients for successful transcatheter closure; five patients were found to have too large defects after balloon sizing. Multiple ASDs and/or associated vascular anomalies in 17 of 65 patients were clearly identified at MR imaging, compared with results of TEE, surgery, and cardiac catheterization.
CONCLUSION: In children with ASD and inconclusive TTE results, MR imaging can enable determination of defect size, rim distances to adjacent structures, and venous connections.
© RSNA, 2003
Index terms: Heart, abnormalities, 514.141 • Heart, interventional procedures, 514.1268 • Heart, MR, 58.121416, 58.12142 • Magnetic resonance (MR), in infants and children
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Alternatively, magnetic resonance (MR) imaging and MR angiography have the potential todepict the size and shape of an ASD (3,13), depict pulmonary and systemic venous anomalies (14) that are often associated with an ASD, and enable the quantification of left-to-right shunt in children (15,16). However, to our knowledge, use of MR imaging to determine the distance between septal rims and anatomic landmarks, which is important for the feasibility of closure device placement, has not been assessed, and, except for a single case report (17), no attempt to stratify patients for intervention or surgery with MR imaging has been reported. The sensitivity and specificity of MR angiography in the detection of pulmonary and systemic venous anomalies have not been evaluated prospectively in a larger pediatric population. The purpose of our study was to evaluate phase-contrast MR imaging for sizing of ASDs and inflow MR angiography for detection of associated venous anomalies in pediatric patients with inconclusive TTE results.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MR imaging and catheterization and TEE were well tolerated. Phase-contrast MR imaging sizing studies and MR angiography were usually completed within 40 minutes. The mean heart rate was 103 minutes ± 14 at MR imaging and 96 minutes ± 15 at catheterization. The mean time between MR imaging and surgery or transcatheter closure was 34 days ± 92. All patients were in sinus rhythm and had evidence of substantial left-to-right shunting. The study was approved by our institutional review committee, and informed written consent was obtained from parents or caretakers.
Study Design
Each patient underwent MR imaging for determination of (a) defect location, size, and distances to adjacent structures, and (b) pulmonary and systemic venous connections. Those children considered suitable were subsequently referred for transcatheter defect closure, and MR imaging results were compared with TEE measurements. In children who were considered unsuited for intervention on the basis of MR imaging data (see Results for selection criteria), MR imaging sizing results were compared with measurements obtained during surgical repair. Prior to surgery, according to hospital practice at the time, routine cardiac catheterization was performed to enable assessment of venous return, quantification of left-to-right shunting, and exclusion of pulmonary hypertension.
To avoid interobserver variability, all surgical procedures were performed and all surgical measurements were obtained by one surgeon (U.B.), and all TEE examinations were performed by one technician. Phase-contrast MR images obtained in 10 randomly selected patients were reanalyzed by two observers (P. Beerbaum, H.K.), who were blinded to their previous results and to each other’s results, so that we could determine interobserver variability. The TEE technician, the physician who performed the transcatheter intervention, and the cardiac surgeon (U.B.) were blinded to the specific phase-contrast MR imaging sizing data (beyond the interpretation of "suited for intervention" or "unsuited for intervention"). The MR imaging unit operator was blinded to any echocardiographic results.
General anesthesia was initiated during transcatheter defect closure and/or TEE, and intravenous sedation—with midazolam (Dormicum; Hoffman-La Roche, Grenzach-Wyhlen, Germany) and thiopental (Trapanal; BykGulden Lomberg Chemische Fabrik, Konstanz, Germany) as needed—was administered for MR imaging and diagnostic catheterization. Blood pressure, oxygen saturation, heart rate, and respiratory rate were monitored continuously.
MR Imaging Technique
All examinations were performed with a 1.5-T whole-body MR imaging unit (ACS-NT; Philips Medical Systems, Best, the Netherlands) with a maximum gradient performance of 23 mT/m amplitude and a slew rate of 105 mT/m/msec. A five-element cardiac phased-array coil was used for signal acquisition. Details of the MR imaging protocol are listed in Table 1. Turbo field-echo survey images in three orthogonal planes were obtained; this was followed by diastolic-gated, T1-weighted multisection spin-echo echo-planar imaging in which transverse sections were obtained from the diaphragm to the upper mediastinum for delineation of gross anatomy.
|
|
|
|
|
|
Pulmonary and systemic venous return was assessed with two-dimensional time-of-flight MR angiography with transverse-plane acquisitions. To enhance signal from thoracic veins, a trigger delay was chosen so that images were acquired during early diastole, when venous flow is maximal (15).
MR Image Analysis
Defect size was determined off-line on the basis of the reconstructed phase images (Fig 1) that displayed maximal shunt flow (3,15). The ASD diameter bar was then transferred to the corresponding magnitude image displaying the anatomic information, and the septal rims were measured toward the entrances of the SVC (Fig 2) and IVC (Fig 3) into the right atrium, to the aortic root and the opposite atrial wall (Fig 4), to the atrioventricular valves (mitral valve if compared with TEE findings and tricuspid valve if compared with surgical findings), and to the opposite atrial wall (Fig 5). Size measurements obtained from en face images were only accepted if the defect was truly located in plane, as suggested from a narrow rim of signal void in the magnitude images (13). The relationship of the ASD to the coronary sinus was estimated on the en face images when possible. Atrial septal lengths were directly assessed in the aortic-level and mitral-level planes. The craniocaudal septal length, however, could not be visualized on a single section and was therefore assessed by adding the distance between the rim and the SVC and the distance between the rim and the IVC to the craniocaudal measurement of defect diameter on the section on which the defect diameter was largest—either the section depicting the ASD-to-SVC length or the section depicting the ASD-to-IVC length.
Multiplanar reformations depicting pulmonary and systemic venous connections were reconstructed from the axial two-dimensional inflow MR angiographic data.
TEE Technique
A Sonos 4500/5500 (Hewlett-Packard, Palo Alto, Calif) TEE system with a biplanar pediatric 5.0-MHz phased-array transducer was used. Additional septal fenestrations were detected with color Doppler by using low pulse-repetition frequency, high frame rate, narrow color, and black-and-white angle. When the ASD was located in the upper portion of the atrial septum, septal rim distances to the aortic root and the opposite atrial wall were determined in the transverse plane, whereas when the defect was located in the lower part of the atrial septum, rim distances were measured to the mitral valve and the opposite atrial wall. In either projection, the maximum ASD diameter and the total septal length were also recorded. In all patients, the craniocaudal ASD diameter and rim distances to the entrances of the SVC and IVC were measured in the longitudinal plane, and the craniocaudal septal length was assessed by adding the craniocaudal measurement of defect diameter to the distance from the rim to the SVC and the distance from the rim to the IVC.
Cardiac Catheterization and Intervention
All 65 children underwent diagnostic cardiac catheterization as a routine clinical procedure (per hospital protocol at that time); 30 of these children underwent attempted transcatheter defect closure with Amplatzer Septal Occluder (AGA Medical, Golden Valley, Minn) systems (2) on the basis of the phase-contrast MR imaging sizing results.
Measurements during Surgery
Surgery was performed through a midline sternotomy by using induced ventricular fibrillation and intermittent cross-clamping of the aorta. An Overholt clamp (Nobby Medical Instruments, Los Angeles, Calif) was used to assess long-axis (craniocaudal) and short-axis defect size and rim distances to the entrances of the SVC and IVC, the aortic root, the tricuspid valve (no accurate measurement of the rim to the mitral valve was possible with an approach through the right atrium), and the coronary sinus. The craniocaudal septal length was assessed by adding the craniocaudal measurement of defect diameter to the distance of the rim to the SVC and the distance of the rim to the IVC. The distance between the tips of the Overholt clamp was measured with a ruler. No repeated measurements were obtained so that measuring could be performed in less than 1 minute of myocardial ischemia time.
Statistical Analysis
All comparisons between MR imaging and TEE and/or surgical results were performed by one author (H.K.) by using Bland-Altman analysis (18) to evaluate the agreement between sizing measurements at phase-contrast MR imaging, TEE, and/or surgery. Data were log transformed when differences and means were linearly related (18). Interobserver variability for phase-contrast MR imaging measurements of all defect and rim sizes were determined by using two-factor analysis of variance (SAS, version 8.0; SAS Institute, Heidelberg, Germany). P < .05 was considered to indicate a statistically significant difference.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
ASD Sizing: Phase-Contrast MR Imaging versus TEE Results
As outlined in detail in Table 2, measurements of defect sizes at phase-contrast MR imaging had excellent agreement with measurements at TEE for aortic level, mitral valve level, and craniocaudal orientation (Bland-Altman analysis). Estimation of precision of the limits of agreement (upper limit of agreement = mean difference + 2 SDs; lower limit of agreement = mean difference - 2 SDs) was based on calculation of 95% CIs (18). In the aortic-level, mitral-level, and craniocaudal projections, there was a difference of less than 1.0 mm in measurements between the methods, and upper and lower limits of agreement were between 2.3 and -3.3 mm. TEE was superior to phase-contrast MR imaging in the detection of minor additional septal fenestrations smaller than 1 mm.
|
Atrial septal length measurements at phase-contrast MR imaging agreed with those at TEE for aortic level septal length but were overestimated at phase-contrast MR imaging for craniocaudal and mitral valve level septal lengths (mean difference, 3.0 mm), with some scatter (upper limits of agreement, between 2.8 and 8.0 mm; lower limits, between -1.8 and -2.8 mm) (Table 2).
ASD Sizing: Phase-Contrast MR Imaging versus Surgical Measurements
Surgical measurements of defect size and rim dimensions were available in 40 children with ASD. Twenty-nine of these children had simple ASD, four had associated PAPVR, one had multiple peripheral pulmonary arterial stenoses, one had net of Chiari (ie, a large, perforated remnant of the Thebesian valve), and five had one to two small ASDs in addition to the major defect being evaluated. Craniocaudal and short-axis (transaortic) defect size measurements at phase-contrast MR imaging agreed well with the corresponding diameters assessed intraoperatively in all 40 children (mean differences in these projections, between 1.2 and -1.6 mm) (Table 3). Phase-contrast MR imaging measurements of septal rim distances to the tricuspid valve, SVC, and IVC had fair agreement with surgical results, but phase-contrast MR imaging measurements of rim distances to the aorta had excellent agreement with surgical results (Table 3). In 21 patients in whom phase-contrast MR imaging measurements of rim distance to the coronary sinus were possible, these measurements showed good agreement with surgical results. Phase-contrast MR imaging measurements of craniocaudal atrial septal length agreed moderately well with surgical results in 40 children, in contrast to the observed disagreement between phase-contrast MR imaging and TEE results for this measurement (Tables 2 and 3).
|
Interobserver Variability at Phase-Contrast MR Imaging
In 10 randomly selected patients, phase-contrast MR imaging measurements of all three ASD diameters (aortic level, mitral level, and craniocaudal) and all rim distances to adjacent structures were independently reevaluated by two observers. A P value of .659 (nonsignificant) was determined by using two-factor analysis of variance. From this, it can be concluded that there was no systematic deviation in measurements at phase-contrast MR imaging between the two independent observers.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
In this context, the results of our evaluation of 65 children with ASD and inconclusive TTE results demonstrate the reliability of cardiovascular MR imaging versus TEE or surgery and invasive cineangiocardiography for noninvasive defect sizing and delineation of associated vascular anomalies. In all 35 children in our study who were considered unsuited for intervention at phase-contrast MR imaging, the surgeon confirmed this unsuitability on the basis of the criteria mentioned in the Results section. MR imaging results enabled us to correctly schedule 25 (83%) of 30 children with inconclusive TTE results for transcatheter ASD closure. However, we incorrectly scheduled five of 30 children for transcatheter intervention on the basis of phase-contrast MR imaging results; the ASDs in these children were found to be too large for transcatheter closure at stretched-balloon sizing. This may be explained by the fact that phase-contrast MR imaging does not reliably depict septal thickness because the method depends on flow imaging rather than on structural delineation of thin membranes (13). Therefore, "floppy" septa, which are often found to require a large closure device after balloon-sizing in the catheter laboratory, can sometimes be missed at phase-contrast MR imaging (22).
ASD Sizing at Phase-Contrast MR Imaging: Methodological Issues
Holmvang and colleagues first used phase-contrast MR imaging to delineate ASD diameter and shape from a section position that matched the plane of the defect (13). The ASD was displayed en face according to the width of the transseptal flow stream at its base. With use of this flow imaging technique in adults, former limitations of spin-echo or flow-compensated gradient-echo MR imaging (23) were overcome. Spatial resolution was limited, however, and no measurements of rim distances to adjacent structures were provided in the report of Holmvang et al (13). Use of this method in children in the present study revealed difficulties in consistently matching the section to the ASD plane. This may have been due to (a) the use of thinner sections (5–6 mm) to evaluate bulging septa in the present study versus the 8–10 mm sections used in the study of Holmvang et al (13), and/or (b) the relatively increased bulk cardiac and respiratory motion in children. Furthermore, adjacent structures such as the systemic and/or pulmonary veins, aortic root, and atrioventricular valves may not always be visualized on a single section.
To avoid possible measurement errors resulting from a slightly up- or downstream section position relative to the atrial septal plane (13), we decided to add four single phase-contrast MR imaging sections, prescribed from the initially obtained en face projection to the adjacent structures, to determine ASD diameters and quantify rim distances. Detecting the maximum shunt flow is mandatory for aligning the section with the maximal ASD diameter (13) because there is considerable movement of the defect and a variation in size throughout the cardiac cycle (3,4,21). We used a five-element cardiac phased-array coil to optimize signal-to-noise ratio. This allowed for the high in-plane resolution of 1.2 x 1.2 mm that was required for precise delineation of adjacent structures and detection of smaller additional defects on en face phase images.
Inflow MR Angiography for Associated Vascular Anomalies
Both sensitivity and specificity were 100% for inflow MR angiography in the prospective delineation of systemic and pulmonary venous anomalies—including area of drainage and relative vessel size—in our population when inflow MR angiographic results were compared with cardiac catheterization results. We preferred two-dimensional inflow MR angiography to three-dimensional (3D) contrast material–enhanced MR angiography because of the relative robustness of two-dimensional inflow MR angiography to respiratory motion artifacts that may degrade 3D contrast-enhanced MR angiograms in freely breathing sedated children. In this setting, the use of 3D contrast-enhanced MR angiography may not be advantageous and requires a contrast agent.
Unlike with 3D contrast-enhanced MR angiography, with the proposed inflow MR angiographic sequence (designed to maximize venous blood signal), the atrial septum is displayed well enough that the precise topographic relation of a pulmonary venous ostium to the atrial septum is demonstrated. This may theoretically be important in children with an atrial-level PAPVR, although we did not observe this finding in our population. In children who can hold their breath, however, 3D contrast-enhanced MR angiography is faster, depicts more of the pulmonary and arterial vasculature, and allows one to assess vessel stenoses (14), which may be difficult to assess with inflow MR angiography owing to signal voids from turbulent blood flow. With inflow MR angiography, a smaller vessel with an in-plane course may be subject to saturation effects—a disadvantage not encountered with 3D contrast-enhanced MR angiography—although this was not a practical problem in our population.
Study Limitations
First, no conclusions from our data are applicable to patients with arrhythmia or irregular breathing patterns, because sizing results may be degraded by blurring from motion artifacts. Second, there is no validated in vitro or in vivo standard of reference for ASD measurements at phase-contrast MR imaging. Although surgery was performed by using induced ventricular fibrillation rather than cardioplegic solution to maintain the cardiac tone, the different cardiac preload makes both measurement conditions different to some extent (3,12). Due to the limited projections available with biplanar TEE, it was not always possible to collect all desired measurements in each patient. Moreover, phase-contrast MR imaging section orientation and TEE imaging planes may have been slightly different; this probably explains some of the scatter and the disagreement observed between phase-contrast MR imaging and TEE measurements of atrial septal lengths.
Third, imaging time is still relatively long and sedation is required in younger children. With the advent of parallel imaging techniques (24), however, a substantial decrease in MR imaging time can be expected. Ultrafast steady-state free-precessing gradient-echo MR imaging techniques have the potential to enable delineation of thin septal structures and ASD size, but these techniques are presently limited to use in patients who are able to hold their breath and are therefore unsuited for imaging sedated free-breathing children. Fourth, it was not our intention in this feasibility study to provide cost-effectiveness information on the use of this relatively expensive technique in patients with ASD. Costs may be saved, however, if the performance of unnecessary TEE and/or interventional procedures with general anesthesia is avoided through improved selection of children with ASDs and inconclusive TTE results.
In conclusion, cardiovascular MR imaging enables the noninvasive determination of ASD size, morphology, and spatial relationships and precisely depicts pulmonary and systemic venous connections, thereby assisting in the planning of transcatheter defect closure procedures in children with inconclusive TTE results.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Author contributions: Guarantors of integrity of entire study, P. Beerbaum, H.M.; study concepts, P. Beerbaum, H.E., H.M., J.G.; study design, P. Beerbaum, H.E., H.K., U.B.; literature research, P. Beerbaum, H.K., H.E., P. Barth, J.H.; clinical studies, H.E., U.B.; data acquisition, H.E., U.B., P. Barth, J.H.; data analysis/interpretation, H.K., P. Beerbaum, J.G., P. Barth, J.H.; statistical analysis, H.K.; manuscript preparation, P. Beerbaum; manuscript definition of intellectual content, all authors; manuscript editing, P. Beerbaum; manuscript revision/review and final version approval, all authors.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  K DEBL, B DJAVIDANI, S BUCHNER, N HEINICKE, F POSCHENRIEDER, S FEUERBACH, G RIEGGER, and A LUCHNER Quantification of left-to-right shunting in adult congenital heart disease: phase-contrast cine MRI compared with invasive oximetry Br. J. Radiol., May 1, 2009; 82(977): 386 - 391. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. E.J. Thomson, A. L. Crowley, J. F. Heitner, P. J. Cawley, J. W. Weinsaft, H. W. Kim, M. Parker, R. M. Judd, J. K. Harrison, and R. J. Kim Direct En Face Imaging of Secundum Atrial Septal Defects by Velocity-Encoded Cardiovascular Magnetic Resonance in Patients Evaluated for Possible Transcatheter Closure Circ Cardiovasc Imaging, July 1, 2008; 1(1): 31 - 40. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. E. Williamson, J. Kirsch, P. A. Araoz, W. B. Edmister, D. D. Borgeson, J. F. Glockner, and J. F. Breen ECG-Gated Cardiac CT Angiography Using 64-MDCT for Detection of Patent Foramen Ovale Am. J. Roentgenol., April 1, 2008; 190(4): 929 - 933. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. S. Sorensen, H. Korperich, G. F. Greil, J. Eichhorn, P. Barth, H. Meyer, E. M. Pedersen, and P. Beerbaum Operator-Independent Isotropic Three-Dimensional Magnetic Resonance Imaging for Morphology in Congenital Heart Disease: A Validation Study Circulation, July 13, 2004; 110(2): 163 - 169. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
