Background
The aim of this study was to develop a mathematical model using two-dimensional echocardiographic parameters to estimate right ventricular end-diastolic volume (RVEDVi) in adults with repaired tetralogy of Fallot.
Methods
Linear regression equations were used to examine the relationship between two-dimensional echocardiographic and cardiac magnetic resonance (CMR) imaging measures of RVEDVi. Imaging studies in 101 adults were used to create and validate the model. The ability of the model to detect changes in CMR RVEDVi was tested in 57 adults with serial studies.
Results
The optimal model to quantitate CMR RVEDVi included two-dimensional echocardiographic right ventricular end-diastolic area measured in the apical four-chamber view, indexed to body surface area (AreaDi) (CMR RVEDVi = 11.5 + [7 × AreaDi]). The model reliably allowed the detection of stable and changing CMR RVEDVi (κ = 0.84 and κ = 0.82, respectively, P < .0001).
Conclusion
Quantitative assessment of right-ventricular volumes by echocardiography is feasible and may be used for serial follow-up in patients with contraindications for CMR.
The timing of pulmonary valve replacement in patients with repaired tetralogy of Fallot and residual severe pulmonary regurgitation remains a major challenge in the care of patients with congenital heart disease. Right ventricular (RV) volumes and function, as well as their changes over time, are among the most important factors influencing decision making. Published guidelines for adults with congenital heart disease recommend qualitative assessment of RV size and systolic function on two-dimensional echocardiographic (2DE) imaging in patients with repaired tetralogy of Fallot. However, because of the complex geometry of the right ventricle, the assessment of RV size on standard 2DE imaging is difficult. The currently proposed 2DE measurements have shown poor correlation with measurements of RV end-diastolic volume (RVEDVi) on cardiac magnetic resonance (CMR) imaging. Although CMR has become the reference method for assessment of RV volumes and systolic function, it is not portable and is expensive, time consuming, and unsuitable for patients with pacemakers or implantable defibrillators. Congenital patients in particular may also be excluded from CMR because of the presence of implanted metallic stents or coils, which frequently cause significant image artifacts. An accurate, reliable, and reproducible 2DE method to quantify RV dimensions and their changes over time would therefore be of clinical interest.
The objectives of our study were (1) to create a mathematical model using 2DE parameters to estimate CMR-derived RVEDVi; (2) to define cutoffs using the 2DE measurements to identify mild and severe RV dilatation; and (3) to assess the ability of the model to detect changes in RV volumes over time.
Methods
Study Subjects
We identified 101 consecutive adult subjects late after tetralogy of Fallot repair from our clinic database who underwent both CMR and 2DE imaging <6 months apart (mean time between CMR and 2DE imaging, 0.7 ± 2.3 months). Derivation of the model was based on echocardiograms from 67 randomly selected subjects from the study group, and the remaining 34 subjects were used as the validation group. In 57 patients (56%) who underwent serial 2DE and CMR studies (a median of 3.2 years [range, 1.6–6.2 years] after the initial study), the ability of the 2DE model to detect changes in RVEDVi over time was tested. The study was approved by the institutional ethical committee.
Echocardiography
All transthoracic echocardiographic studies at our institution are performed by experienced cardiac sonographers and read by experienced staff cardiologists. We have a predefined protocol for image acquisition, including parasternal short-axis and long-axis views as well as apical views. Right ventricular end-diastolic volumes are reported as normal, mildly, moderately, or severely enlarged, on the basis of visual estimation, supplemented by measurement of the basal RV dimension in the apical four-chamber view ( Figure 1 D, measurement 5).
![](https://i0.wp.com/thoracickey.com/wp-content/uploads/2018/06/gr1-442.jpg?w=960)
For the purposes of this study, all echocardiograms were reanalyzed on an Agfa IMPAX Cardiac Review Station (Agfa HealthCare Corporation, Westerly, RI). Measurements were made by a single experienced echocardiographer (M.G.) blinded to the CMR results. End-diastolic measurements of the right ventricle were performed in the parasternal long-axis view, at two levels of the parasternal short-axis view, and in the apical four-chamber view ( Figure 1 ). In contrast to the relatively uniform left ventricular contraction, the right ventricle contracts in a peristaltic pattern, and contractions propagate from the RV inflow tract to the RV outflow tract (RVOT). “True” end-diastole will therefore occur at slightly different time points within different regions of the right ventricle. By convention, end-diastole was defined as the time of closure of the tricuspid valve in the apical four-chamber view and as the largest visual diameter of the right ventricle in parasternal long-axis and short-axis views. Parasternal long-axis RV measurements were performed perpendicular to the basal interventricular septum ( Figure 1 A). Parasternal short-axis measurements of the RVOT were performed at the level of the aortic valve along a line from the center of the imaging probe to the center of the aortic valve ( Figure 1 B). Parasternal short-axis measurements of the RV body were performed at the level of the papillary muscles of the left ventricle ( Figures 1 C and 1 E). Measurements in the four-chamber view were performed on images in which the septum was aligned parallel to the central cursor line ( Figures 1 D and 1 F). The tracings of RV areas included all RV trabeculae into the blood pool, in concordance with measurements of the right ventricle on CMR at our institution ( Figures 1 E and 1 F). For all tracings, an average of two measurements was calculated.
CMR
CMR protocols and technical acquisition parameters used at our institution have been previously published. Clinical studies were analyzed and interpreted by four experienced CMR readers. Measurements of RVEDVi are obtained on axial views with manual tracing of endocardial contours. All readers follow our institutional convention of including papillary muscles and trabeculae in the blood pool. Volumes were indexed (normalized) for body surface area. The upper limits of normal for RV volumes as measured by steady-state free precession imaging sequences, similar to the methods used in this study, have been defined in healthy volunteers and range from 100 to 114 mL/m 2 . The upper limits of RVEDVi at which remodeling to normal values after pulmonary valve replacement is not likely to occur vary among studies from 150 to 170 mL/m 2 . Thus, an RVEDVi > 150 to 170 mL/m 2 is an important determinant of the timing of surgery at many institutions. For comparison of agreement of classes of RV dilatation between echocardiography and CMR, we defined normal RVEDVi as <110 mL/m 2 , mildly dilated as 110 to 139 mL/m 2 , moderately dilated as 140 to 169 mL/m 2 , and severely dilated as ≥170 mL/m 2 . Interobserver variability for RVEDVi was tested by recontouring right ventricles on 20 CMR studies.
Statistical Analysis
Statistical analysis was performed using SPSS version 17.0 (SPSS, Inc., Chicago, IL). Descriptive data for continuous variables are presented as means or medians and dichotomous variables as percentages.
In a derivation sample of 67 randomly selected subjects, correlations between 2DE measurements and CMR RVEDVi were examined. Pearsons correlation coefficients were used to examine the relationship between CMR RVEDVi and (1) absolute 2DE measurements; (2) 2DE measurements indexed to body surface area; and (3) measurements indexed to left ventricular dimension on basal four-chamber and parasternal long-axis views. On the basis of the RV measures that correlated most closely with CMR RVEDVi, linear regression models were created. The RVOT dimension was also included in one of the models because it was felt to be clinically relevant. The traced areas on the four-chamber view, the traced area on the short-axis view, and the dimension of the RVOT were included in the models alone and in combination. The final model was chosen on the basis of its accuracy to classify categories of RV dilatation. The accuracy of the 2DE model was tested in the validation group ( n = 34).
To have a simple 2DE screening tool to identify severe and normal or mild RV dilatation, optimal cutoffs for the 2DE RV end-diastolic areas were defined for the detection of CMR RVEDVi ≥ 170 mL/m 2 (severe) and ≤ 125 mL/m 2 (normal or mild). Receiver-operating characteristic curve analysis was used to define optimal cutoffs for identifying mild or normal and severe RV dilatation. Echocardiographic cutoff values were based on obtaining high sensitivity for severe and high specificity for mild RV dilatation. Kappa statistics were used to test agreement between classes of RV dilatation on the basis of the visual assessment of RV size by echocardiography, RVEDVi on the basis of the mathematical model, and actual CMR measurements. The accuracy of the threshold for normal RV dimensions was additionally validated in 50 patients with structurally normal hearts.
We also tested the ability of the model to detect changes in RVEDVi over time. Differences in CMR RVEDVi between the baseline and follow-up studies within a range of ±15% were defined as stable RVEDVi (on the basis of estimated interobserver variability for the measurement of RVEDVi on CMR), and a difference of >25% was defined as a significant change. The Bland-Altman method was used to test agreement between changes in RVEDVi as measured by CMR and by the 2DE model.
In a random sample of 20 subjects, we measured the intraobserver and interobserver variability for the 2DE measurements. Interobserver variability was tested by using prespecified echocardiographic loops and by leaving the choice of echocardiographic loops for measurement to the discretion of the observer. The Bland-Altman method was used to test intraobserver and interobserver variability.
Results
The mean age of the study cohort was 33 ± 12 years. Sixty-one percent of subjects had severe pulmonary regurgitation, and the mean RVEDVi at baseline as measured by CMR was 158 ± 51 mL/m 2 . Fifty-seven subjects (56%) had serial 2DE and CMR studies. A summary of baseline characteristics is given in Table 1 . There were no significant differences in baseline characteristics between the derivation and the validation group. Acoustic window quality on 2DE was judged good in 6%, fair in 86%, and poor in 8%. No studies were excluded because of inadequate acoustic windows.
All patients | Derivation group | Validation group | |
---|---|---|---|
Variable | ( n = 101) | ( n = 67) | ( n = 34) |
Age (years) | 33 ± 12 | 33 ± 12 | 34 ± 13 |
Men (%) | 61 (60) | 40 (60) | 21 (62) |
Body surface area (m 2 ) | 1.8 (1.1–2.4) | 1.8 (1.1–2.4) | 1.8 (1.3–2.2) |
Age at corrective surgery (years) | 5.0 (0.5–36.0) | 5.0 (0.5–35.0) | 4.6 (1.4–36.0) |
Palliation prior to corrective surgery | 49 (49%) | 31 (46%) | 18 (53%) |
Prior pulmonary valve replacement | 24 (24%) | 14 (21%) | 10 (29%) |
QRS duration (ms) | 159 ± 27 | 159 ± 26 | 158 ± 28 |
Severe PR | 62 (61%) | 42 (63%) | 20 (59%) |
CMR baseline data | |||
RVEDVi (mL/m 2 ) | 158 ± 51 | 159 ± 51 | 157 ± 51 |
RVEDVi ≥ 170 mL/m 2 | 40 (40) | 27 (40) | 13 (38) |
RVEDVi ≤ 110 mL/m 2 | 19 (19) | 12 (18) | 7 (21) |
RVEF (%) | 41 ± 8 | 41 ± 8 | 41 ± 8 |
LVEF (%) | 53 ± 9 | 53 ± 10 | 52 ± 8 |
There was poor agreement between echocardiographic visual estimation of the degree of RV dilatation and the degree of RV dilatation on CMR (κ = 0.105, P = .046). Visual estimates of RV size systematically underestimated RV volumes in patients with moderate or severe RV dilation (by one class in 45% and by two classes in 18%) and systematically overestimated RV volumes in patients with normal or mildly dilated right ventricles (by one class in 44% and by two classes in 13%).
Derivation of the Model
As shown in Table 2 , the optimal correlation between 2DE measurements and CMR RVEDVi involved end-diastolic RV area in the apical four-chamber view (AreaD). When measurements were indexed for body surface area (AreaDi), correlation improved ( r 2 = 0.70 vs r 2 = 0.58). The correlation did not improve when AreaD was indexed to left ventricular dimensions (measurements 2 and 8) on either the parasternal long-axis or the four-chamber view ( r 2 = 0.53 and 0.48 vs r 2 = 0.58). On the basis of these correlations, linear regression was used to determine the relationship between 2DE RV area in the apical four-chamber view and CMR RVEDVi, resulting in the following model: RVEDVi = 11.5 + (7 × AreaDi) .
Absolute measurements | Measurements indexed to body surface area | |||
---|---|---|---|---|
RV measurement | r | r 2 | r | r 2 |
Parasternal long-axis view (measurement 1) | 0.684 | 0.468 | 0.691 | 0.477 |
RVOT (measurement 3) | 0.518 | 0.268 | 0.519 | 0.269 |
Short axis at the level of left ventricular papillary muscles (measurement 4) | 0.653 | 0.426 | 0.640 | 0.409 |
Basal diameter in the apical four-chamber view (measurement 5) | 0.503 | 0.253 | 0.525 | 0.275 |
Midventricular diameter in the apical four-chamber view (measurement 6) | 0.713 | 0.509 | 0.713 | 0.509 |
Length in the apical four-chamber view (measurement 7) | 0.369 | 0.136 | 0.353 | 0.124 |
Traced area in the short-axis view (measurement 9) | 0.650 | 0.423 | 0.727 | 0.529 |
Traced area in the apical four-chamber view (measurement 10) | 0.763 | 0.582 | 0.835 | 0.697 |
Combination of RVOT and traced area on the four-chamber view (measurements 3 and 10) | 0.787 | 0.619 | 0.863 | 0.744 |
Combination of RVOT and traced areas on the four-chamber and short-axis views (measurements 3, 9, and 10) | 0.805 | 0.648 | 0.898 | 0.806 |
![](https://freepngimg.com/download/social_media/63059-media-icons-telegram-twitter-blog-computer-social.png)
Stay updated, free articles. Join our Telegram channel
![](https://clinicalpub.com/wp-content/uploads/2023/09/256.png)
Full access? Get Clinical Tree
![](https://videdental.com/wp-content/uploads/2023/09/appstore.png)
![](https://videdental.com/wp-content/uploads/2023/09/google-play.png)
![](https://clinicalpub.com/wp-content/uploads/2023/09/banner1.png)