| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Ann Thorac Surg 2006;81:1436-1442
© 2006 The Society of Thoracic Surgeons
Cardiothoracic Centre, All India Institute of Medical Sciences, New Delhi, India
Accepted for publication August 29, 2005.
* Address correspondence to Dr Chowdhury, Department of CTVS, AIIMS, New Delhi-110029, India (Email: ujjwalchow{at}rediffmail.com).
| Pediatric cardiac surgery: To participate in The Annals of Thoracic Surgery CME Program, please visit http://cme.ctsnetjournals.org.
|
 |
Abstract |
|---|
 Top Abstract IntroductionPatients and MethodsResultsCommentAcknowledgmentsReferences |
|---|
METHODS: Two hundred and eighty survivors of repaired tetralogy of Fallot aged 42 months to 40 years (mean, 142.2 ± 85.3 months) underwent cardiac MRI, first-pass and gated radionuclide ventriculography (RNV) for the assessment of biventricular function, and PRF after 89.26 ± 42.40 months. The receiver operating characteristic curve analysis was done to quantify the diagnostic accuracy of MRI.
RESULTS: There was statistically significant agreement between MRI and RNV in evaluating right and left ventricular function. An MRI-derived right ventricular ejection fraction 47.2% or greater than normal was associated with a sensitivity of 92.3% and a specificity of 92.3%. An MRI-derived left ventricular ejection fraction 53.9% or greater than normal was associated with a sensitivity of 93.2% and a specificity of 93.3%. Area analysis indicated that 97.34% (standard error [SE] = 0.0118) and 98.56% (SE = 0.0052) of the time values of right and left ventricular ejection fraction were higher for patients with normal right and left ventricular functions, respectively, compared with abnormal. There was a strong agreement between velocity-encoded and stroke volume-derived PRF [(r = 0.886, p < 0.001; 
= 2.62 ± 1.12, p < 0.0001; r' = 0.121, p = 0.051; b = 0.96 (SE = 0.012); p< 0.0001; ICC = 0.98, p< 0.0001). Higher PRF was associated with increased indexed right ventricular dimensions and inversely correlated with biventricular ejection fractions.
CONCLUSIONS: The MRI-derived ejection fraction values predictably separate patients with normal ventricular function from abnormal. Velocity-encoded MRI can accurately quantitate PRF in tetralogy of Fallot.
Assessment of patients undergoing repair of tetralogy of Fallot (TOF) mandates an imaging technique that will provide adequate information on (1) residual anatomical problems (eg, ventricular septal defect, pulmonary stenosis, and right ventricular outflow tract [RVOT] aneurysm), (2) the amount of pulmonary regurgitation, and (3) biventricular size and function [1–3]. Two-dimensional echocardiography has limited value for accurate assessment of right ventricular size and function, and the severity of pulmonary regurgitation [3, 4]. Gated blood-pool scintigraphy is a superior alternative as it is devoid of geometric assumptions, but spatial resolution in children is poor, and the atrial blood pool cannot be clearly separated from the ventricular region [4]. The emergence of cardiovascular MRI has provided a technique to noninvasively assess cardiac anatomy and quantify ventricular mass, dimension, function, and pulmonary regurgitation [5–8].
This prospective study was performed to (1) evaluate the sensitivity, specificity, and predictive accuracy of MRI-derived left and right ventricular ejection fractions (RVEF, LVEF) as compared with the gold standard, radionuclide ventriculography (RNV), and (2) compare the PRF derived from phase-contrast velocity encoded (VEC)-MRI with that from ventricular stroke volumes by cine-MRI.
|
Patients and Methods |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentAcknowledgmentsReferences |
|---|
30 mm Hg), (3) supraventricular arrhythmias, and (4) significant ventricular arrhythmias (Lown grade > 2). Exclusion criteria included the presence of coil embolization-related artifacts on MRI (n = 18), and prolonged QRS duration (> 180 ms). A total of 600 patients underwent surgical correction of TOF between January 1992 and December 2004, and of the patients who fulfilled the selection criteria, 280 were selected randomly. Age at correction of the entire study group was 18 months to 35 years (mean, 52.85 ± 59.09 months). Age at reevaluation was 42 months to 40 years (mean, 142.18 ± 85.31 months). Descriptive characteristics of all patients including the details of operative procedures and the relevant hemodynamic data are summarized in Table 1. Postoperative evaluation consisted of clinical examination, electrocardiogram, and two-dimensional echocardiography every three months. All subjects were in normal sinus rhythm during the examination. These patients underwent RNV and MRI between January 2003 and December 2004 (closing interval). Patients were entered in the follow-up study protocol after informed consent had been obtained. There was no selection bias within the group because all patients were randomly selected. The examiners were blind to demographic, procedural, and hemodynamic data.
|
Pulmonary regurgitation was assessed from both continuous-wave Doppler trace and color-flow mapping. Pulmonary regurgitation was classified as mild when the retrograde pressure drop was maintained throughout diastole, moderate when equilibration between pulmonary artery and right ventricular pressures occurred in late diastole, and severe when it met the baseline in middiastole or earlier [9].
Radionuclide Techniques: First-Pass Radionuclide Angiocardiography and Gated Radionuclide Ventriculography
Protocol. Each patient (n = 280) underwent first-pass radionuclide angiocardiography for RVEF followed by equilibrium gated RNV for LVEF, in the same sitting. The scan was done after in vivo red cell labeling with radioactive tracer technetium 99 m pertechnetate. For in vivo labeling, "cold" stannous ion (15 µg/kg body weight) was first injected intravenously. First-pass study was performed by administration of 0.2 to 0.3 millicurie (mCi)/kg body weight (10–15 mCi) (370–555 megabecquerel) of technetium 99 m pertechnetate (99 m TcO4) injected intravenously in the antecubital vein 20 minutes after the first step with the patient under the camera. The camera was positioned in the anterior view. With the energy discrimination centered on 140 keV, first-pass radionuclide angiographic data were acquired at 50 msecs per frame for a total of 1,000 frames and images acquired in a 64 x 64 matrix with a zoom factor 1.
After the first-pass study acquisition, RNV was performed. The acquisition was synchronized to the patients R-wave on the electrocardiogram (ECG) and the R-R interval was divided into 32 equal phases (gating windows). Images in each patient were acquired in three standard views: left anterior oblique (LAO) (or the best septal view to separate the right and left ventricular chambers; usually corresponds to LAO-40 degrees), anterior view, and left lateral view. Each view was acquired for 300 seconds (approx 500–600 K counts), collecting data from 300 to 400 cardiac cycles in a 64 x 64 matrix with a zoom factor of 2.67. At the end of the acquisition the counts were summed for all the cycles and the images obtained for each phase were then displayed in sequence, creating a cine-loop of the entire cardiac cycle.
The vendor-provided processing protocol was used to process the RNV data to derive a global time activity curve from forward sequential images to calculate the LVEF. The RVEF was calculated from first-pass data using Siemens ICON (Siemens, Erlangen, Germany)-provided "first-pass processing protocol."
Cardiovascular Magnetic Resonance Imaging
Scans were performed using a 1.5-Tesla Siemens Sonata system (Siemens Medical Solutions) in 298 patients on the same day the echocardiogram was obtained. Torso or cardiac-phased array coils were chosen according to body size. Scout images were obtained in a transverse, coronal, and sagittal planes using a standard, multislice spin echo sequence. The following imaging protocols were employed: (1) three-plane localizing images; (2) breath-hold ECG-triggered segmented k-space fast-spoiled gradient-recalled cine sequences in two-and-four chamber planes, followed by 12 contiguous short-axis slabs perpendicular to the long-axis of left and right ventricle (slice thickness 5 mm; interslice space 1–4 mm). We used the short-axis plane because it allows measurement of both right and left ventricular volumes in one sequence. Phase-contrast short-axis "CINE" images were used for ventricular volume-myocardial mass analysis. Velocity encoded phase-contrast sequence was used for direct measurement of forward and reverse flow in the main pulmonary artery. For selecting an axial section for analysis of pulmonary regurgitation, various angle manipulations were undertaken 1–2 mm superior and parallel to the pulmonary valve on a sagittal image before the exact, or closest to, circular cross-section of the main pulmonary artery was obtained.
Postprocessing
Left and right ventricular end-diastolic and end-systolic volumes, mass, and ejection fractions were analyzed as described by Lorenz [7]. Quantification of flow rates and calculations of pulmonary regurgitation were performed as described by Powell and colleagues [8]. The PRF was measured by VEC-MRI, as well as short-axis cine-MRI. Regurgitant fraction was defined as reverse flow divided by forward flow as measured by VEC-MRI. Using breathhold segmented k-space cine MRI in the short-axis plane, the right and left ventricular end-diastolic and end-systolic volumes were measured. Stroke volume (SV) was defined as (end-diastolic volume-end-systolic volume), and PRF was defined as (RVSV-LVSV)/RVSV.
Statistical Methods and Analysis
Statistical analysis was done using Intercooled STATA 8.0 software (Stata Corp, College Station, TX). Interval-related data were expressed as the mean ± standard deviation and the categorical variables were expressed as percentages. Noncategorical variables were analyzed using the paired t test.
The agreement between two methods of measurement of ejection fraction (MRI versus radionuclide) and pulmonary regurgitant fraction (VEC-MRI vs cine-MRI) was assessed using five statistics: Pearsons' correlation coefficient (r), mean value of the difference between the two sets of observations (), correlation between averages and difference between observations of the two variables (r'), regression coefficient (b), and intraclass correlation coefficient (
). Of the five statistics, if r 
1, 0, r' 0, b 1, and 1, then there is a good agreement between both clinical measurements.
The receiver operating characteristic curve (ROC) analysis was done to determine the cutoff of the MRI-derived EF values, which will predictably separate the normal from the abnormal taking RNV as the gold standard. The correlation between PRF and biventricular function was assessed using Pearson's correlation coefficient (r). To quantify the diagnostic accuracy of MRI study, the area under the ROC curve was statistically analyzed. The p value of 0.05 or less was considered statistically significant.
|
Results |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentAcknowledgmentsReferences |
|---|
0.5) in 215, and depressed (ejection fraction 
0.5) in 65 patients. One hundred and twelve patients (40%) had mild and 120 (42.9%) had moderate pulmonary regurgitation. Twelve patients (4.3%) had mild tricuspid regurgitation. There were no residual anatomical problems on any patients.
MRI Parameters
Image quality was diagnostic in 280 patients (18 were excluded due to the presence of coil embolization-related artifacts). Both intracardiac and large vessel anatomy could be depicted clearly on all patients with ECG-gated spin-echo images.
Analysis of Agreement
(1) Validation of MRI-Derived RVEF and LVEF (Table 2 (a)). There was a significant correlation between MRI-derived right and left ventricular ejection fraction (r = 0.58, p < 0.05, Fig 1). The mean value of difference between the two sets of observations (ie, MRI and RNV) depicted a systematic higher value by MRI. For MRI-derived RVEF and LVEF the value was 2.77 ± 1.16 and 3.44 ± 1.48, respectively. We found a good reliability for MRI-derived ejection fraction values, though it was overestimated when compared with RNV ( = 0.9881, p < 0.0001).
|
|
|
|
(3) Agreement Analysis of PRF (Table 2(b), Fig 1): (a) VEC-MRI vs Short-Axis Cine-MRI. There was a strong agreement between measurements of PRF derived from VEC-MRI and from stroke volume-assessment of short-axis cine-MRI ( = 0.9835, p < 0.001). Stroke volume-derived RF significantly overestimated PRF (mean, 25.24 ± 3.82%) in patients with mild tricuspid regurgitation (n = 12). The PRFs were greater in the patients in whom a transannular patch had been used for reconstruction of the RVOT (n = 232) compared with patients with an intact pulmonary annulus (n = 48).
Agreement Analysis of PRF: (b) Correlation Between PRF and Biventricular Function. Higher PRF was associated with increased indexed right ventricular dimensions [end-diastolic volume (r = 0.816, p = 0.005) and end-systolic volume (r = 0.891, p = 0.004)]. The VEC-derived PRF inversely correlated with the MRI-derived RVEF (r = -0.67, p < 0.05) and LVEF (r = -0.48, p < 0.05) (Fig 1).
|
Comment |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentAcknowledgmentsReferences |
|---|
Ventricular Function After TOF Repair
Assessment of biventricular function necessitates a reliable and reproducible technique. The complex crescentic shape of the right ventricle and its relationship with the infundibulum does not permit the use of two-dimensional echo algorithms based on linear measurements and geometric assumptions [3, 4].
Multislice gradient-echo MR imaging is accurate and independent of ventricular geometry, and therefore is well-suited for the quantitation of both left and right ventricular volumes at frequent intervals during the cardiac cycle [5–8]. This is particularly important in the follow-up of TOF patients with abnormal right ventricular loading conditions in which the shape of the left ventricle is known to get altered as well [10].
In this study there was significant agreement between MRI and RNV in evaluating right and left ventricular function (Table 2[a], 3, Figs 2A and 2B). An MRI-derived RVEF 47.2% or greater as normal was associated with a sensitivity of 92.3% and a specificity of 92.3%. An MRI-derived LVEF 53.9% or greater as normal was associated with a sensitivity of 93.2% and a specificity of 93.3%. Area analysis under the ROC curve indicated that 97.34% (SE = 0.0118) and 98.56% (SE = 0.0052) of the time the values of the RVEF and LVEF were higher for patients with normal right and left ventricular functions, respectively, compared with abnormal.
The MRI studies of left ventricular function in repaired TOF are limited and controversial. Some studies have indicated impaired left ventricular function, whereas others have reported normal left ventricular function [5–8, 10]. A possible explanation for this discrepancy might be that the groups differed with regard to the amount of pulmonary regurgitation. In our study higher PRF was associated with increased indexed right ventricular dimensions and inversely correlated with biventricular ejection fractions (Fig 1).
The significant correlation between RVEF and LVEF in this study population (r = 0.58; p < 0.05) indicates the importance of ventriculo-ventricular interaction in patients with repaired TOF that may ultimately be responsible for the late functional deterioration (Fig 1). Although such interaction has previously been demonstrated, the mechanism that links right ventricular dysfunction to a decrease in left ventricular function remains incompletely understood. The recent implementation of a single imaging session by using tagged MRI makes this technique attractive for assessment of both global and regional left ventricular function [10].
Validation of Pulmonary Regurgitation
Pulmonary regurgitation is perhaps the single most important determinant of late outcome for patients with repaired TOF. Several studies suggest that it predisposes to exercise intolerance, ventricular dysfunction, ventricular arrhythmias, and sudden cardiac death [1–6]. Although timing of pulmonary valve replacement is subject to discussion, several investigators are currently considering pulmonary valve replacement in postoperative patients with severe pulmonary regurgitation with or without symptoms [2, 11, 12]. This approach attempts to improve clinical status, preserve ventricular function, modify the risk of arrhythmias, and perhaps decrease sudden death [2, 11, 12].
It is therefore mandatory to obtain precise quantitation of pulmonary regurgitation and assess its influence on biventricular function. Attempts have been made to assess the severity of pulmonary regurgitation by Doppler echocardiography, ventriculography, pressure-volume loops, and video-densitometry [1–4, 9]. There was reasonable agreement between echocardiography and MRI in patients in whom Doppler estimated mild pulmonary regurgitation. In the presence of Doppler-estimated moderate or severe pulmonary regurgitation, reliable assessment of the severity requires further study [4, 9]. Furthermore, the pulmonary flow velocities measured by Doppler are merely a reflection of pressure difference between pulmonary artery and right ventricle and not necessarily an accurate representation of the severity of pulmonary regurgitation.
Sechtem and colleagues [13] have used MR measured stroke volume differences to calculate aortic and mitral regurgitation. A similar approach was used in the current study by comparing PRF measured by velocity-encoded MRI with the difference between right-sided and left-sided stroke volume. In the absence of shunt lesions or regurgitation through the other valves (eg, tricuspid regurgitation), pulmonary regurgitation should equal the difference. The agreement between velocity mapping measurements of PRF and stroke volume-derived-PRF as obtained with multisection cine-MRI was excellent (Table 2). Because stroke volume-derived regurgitation is a measure of regurgitant volume of both valves on one side of the heart, these measurements overestimated PRF in patients with combined pulmonary and tricuspid regurgitation.
Previous investigations that used quantitative methods to assess ventricular function late after TOF repair have demonstrated that the degree of pulmonary regurgitation is closely associated with the degree of right ventricular dilatation, but the data regarding the effect of pulmonary regurgitation on right ventricular systolic function are inconsistent [1–9]. Recently, Davlouros and colleagues [14] demonstrated that the combination of chronic pulmonary regurgitation and the presence of an aneurysmal or akinetic RVOT are the predisposing factors for right ventricular failure in adult patients after TOF repair.
Although the present study was not designed to determine the optimal timing of pulmonary valve replacement, it has provided validation of the velocity mapping technique for accurate quantification of pulmonary regurgitation after TOF repair. It has clearly demonstrated that a higher PRF is associated with increased right ventricular dimensions and is inversely correlated with biventricular function.
The findings of this study indicate that repair using a transannular patch (n = 232) was associated with a higher PRF, which is in accordance with previous studies [1–3]. Nevertheless, preventing severe pulmonary regurgitation or its hemodynamic effects by preserving the pulmonary annulus or use of appropriate technical modifications may reduce the need for reoperations, while research for an ideal prosthetic valve for RVOT reconstruction continues.
Clinical Implications
Pulmonary valve replacement has been performed increasingly in recent years after repair of TOF. However, there is no consensus about candidate selection, optimal timing of surgical intervention, and the long-term benefits of restoring pulmonary valve function. Appropriate timing of pulmonary valve replacement involves balancing the need to intervene early to prevent the development of irreversible right ventricular dysfunction with the understanding that all biologic valves have a limited life expectancy.
The present study demonstrates that a sizeable number of patients with moderate pulmonary regurgitation have right ventricular dilatation and dysfunction but are asymptomatic. They did not have associated lesions such as RVOT obstruction, residual ventricular septal defect, distal pulmonary artery stenosis, or RVOT aneurysm. The optimal management algorithm for this subgroup is not well-defined in the literature. Our study was performed in patients within 1 to 12 years after surgery unlike most published series, which have patients after long follow-up ranging from 15 to 36 years. It demonstrates that the sequelae of pulmonary regurgitation on the right ventricle manifests relatively early in some patients. At the other end of the spectrum are asymptomatic patients with severe pulmonary regurgitation and normal right ventricular function. Whether these patients merit pulmonary valve replacement or close follow-up is open to question. What is certain is that the advent of cardiac MRI provides the best tool to serially follow these patients to find answers to these unresolved questions.
Limitations
In this study, echocardiography was routinely used in the follow-up of patients. However, since it provided only semiquantitative estimates of pulmonary regurgitation and right ventricular function, a statistical comparison with MRI/RNV was not undertaken. The MRI cannot be applied to all clinical situations.
Magnetic resonance imaging is limited in patients with pacemakers and after coil embolization. Since MRI data acquisition is triggered to the R-wave of the ECG and the images are built up over several hundred cardiac cycles, it does not provide real time images. Inability to obtain beat-to-beat information with the current MRI techniques has hampered evaluation of respiratory influences on the amount of pulmonary regurgitation. The image quality is known to reduce in the presence of arrhythmias. Other limitations include the time-consuming nature of analysis of a multisection, multiphase data set. The coarse trabeculations of the right ventricular endocardium may hamper drawing of the right ventricular endocardial border. Automated contour detection programs are in the process of development. Studies like ours, with a single point observation, cannot comment upon the progression of pulmonary regurgitation and ventricular dysfunction.
Conclusions
The MRI-derived ejection fraction values predictably separate patients with normal biventricular function from the abnormal and testifies significant agreement with RNV. Velocity-encoded cine-MRI is an accurate method for quantification of pulmonary regurgitation after repair of TOF.
Magnetic resonance imaging can be performed serially with a high degree of reproducibility, and may be used for late postoperative assessment obviating the need for frequent catheterization. Echocardiography compliments MRI by providing a noninvasive estimate of right ventricular pressure and pulmonary regurgitation. Prospective and serial follow-up of patients with MRI after repair of TOF will determine the natural history of pulmonary regurgitation and may provide guidelines for appropriate timing of pulmonary valve replacement.
|
Acknowledgments |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentAcknowledgmentsReferences |
|---|
|
References |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentAcknowledgmentsReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
