Morphologic description of ventricular septal defect (VSD) is mandatory before performing the newly developed transcatheter closure procedure. Inaccurate estimation of defect size has been reported using conventional two-dimensional (2D) transthoracic echocardiography (TTE). The aim of this study was to assess VSD morphology and size using three-dimensional (3D) TTE compared with 2D TTE and surgery.
Forty-eight children aged 21.4 ± 29.3 months with isolated muscular ( n = 11 [22.9%]) and membranous ( n = 37 [77.1%]) VSDs were prospectively included. Three-dimensional images were acquired using full-volume single-beat mode. Minimal diameter, maximal diameter, and systolic and diastolic VSD areas were measured from 3D data sets using multiplanar reconstruction mode (QLAB 9). Maximal-to-minimal VSD diameter ratio was used to assess VSD geometry. Linear regression analysis and the Bland-Altman method were used to compare 3D measurements with 2D and surgical measurements in a subgroup of 15 patients who underwent surgical VSD closure.
VSD 3D diameters and areas were measured in all patients (100%; 95% CI, 92.6%-100%). Maximal diameter was lower on 2D TTE compared with 3D TTE (7.3 vs 11.3 mm, P < .0001). Mean bias was 4 mm, with 95% of values ranging from −1.76 to 9.75 mm. Correlation between 3D maximal diameter and surgical diameter was strong ( r 2 = 0.97, P < .0001), while correlation between maximal 2D diameter and surgical diameter was moderate ( r 2 = 0.63, P < .0001). VSDs had an oval shape when assessed by 3D TTE. Maximal-to-minimal diameter ratio assessed by 3D TTE was significantly higher in muscular VSDs compared with membranous VSDs (3.20 ± 1.51 vs 2.13 ± 1.28, respectively, P = .01). VSD area variation throughout the cardiac cycle was 32% and was higher in muscular compared with membranous VSDs (49% vs 26%, P = .0001).
Three-dimensional TTE allows better VSD morphologic and maximal diameter assessment compared with 2D TTE. VSD shape and its changes during the cardiac cycle can be visually and quantitatively displayed. Three-dimensional echocardiography may thus be particularly useful before and during percutaneous VSD closure.
Indications for surgical closure of congenital ventricular septal defects (VSDs) are based on pulmonary pressure and left ventricular volume overload. Two-dimensional (2D) transthoracic echocardiography (TTE) is usually sufficient to assess the anatomy and hemodynamics of a VSD and guide clinical management. Recently, percutaneous closure of muscular as well as membranous VSDs using different devices has emerged as an alternative to the surgical technique in selected cases. An accurate assessment of the size and morphology of a VSD is of crucial importance in these cases for device choice. Morphologic assessment of VSD, its diameters, and its dynamic variation throughout the cardiac cycle are difficult to achieve from sequential cuts of one plane by 2D TTE. We and others have reported that three-dimensional (3D) TTE is useful to describe the size, morphology, and dynamic morphology variation of intracardiac defects such as atrial septal defects in children. Thus, we hypothesized that 3D TTE would also be useful to assess VSD morphology and size in these population. Our aim was to compare the 3D measurement of VSDs with that by 2D TTE and with surgical findings in a pediatric population. We also aimed to describe and compare the shapes of membranous and muscular VSDs throughout the cardiac cycle using 3D TTE.
We performed a prospective single-center study including 48 unselected children with isolated membranous or muscular VSDs. Patients with multiple VSDs and patients with other congenital cardiovascular abnormalities were not included. The study was approved by our institutional review committee, and informed consent was obtained from each patient or his or her legal representative.
Two-dimensional TTE was used first to assess VSD size. Measurements were obtained from two orthogonal planes (long- and short-axis views) on the end-diastolic frames. The largest diameter of these two orthogonal diameters was considered the maximal 2D diameter and the smallest the minimal 2D diameter. A 3D full-volume single-beat data set was then acquired (iE33; Philips Medical Systems, Andover, MA) using X5-1 or X7-2 matrix probes (Philips Medical Systems). The data set was stored digitally and transferred to a workstation (QLAB 9; Philips Medical Systems) for offline analysis. All measurements from 3D data sets were independently performed by another operator, who was unaware of the results of 2D TTE.
The 3D data set analysis was performed using multiplanar reconstruction mode. Each of the three axes was moved to obtain a plane along the ventricular septum including the whole VSD from the en face view ( Figure 1 , Video 1 ; available at www.onlinejase.com ). The two orthogonal diameters of the VSD were measured from this view on the end-diastolic frames. The largest diameter was considered the maximal 3D diameter and the smallest the minimal 3D diameter. The ratio of maximal to minimal 2D and 3D VSD diameters was calculated. Systolic and diastolic VSD areas were also obtained by delineating the outline of the VSD on the end-systolic and end-diastolic frames, respectively. The variation of the VSD surface area throughout the cardiac cycle was calculated as (diastolic VSD area − systolic VSD area)/diastolic VSD area and expressed as a percentage. All measurements were obtained from the right-sided surface of the VSD.
In 15 children (31.3%; 95% CI, 18.7%-46.3%) who underwent surgical VSD closure, the defect was examined by the surgeon from the right ventricular side through a standard right atrial opening. The maximal diameter of the VSD was measured directly by the surgeon.
Quantitative variables are expressed as mean ± SD. Body surface area was calculated according to the Mosteller formula. Comparisons between measurements were performed using paired t tests or Wilcoxon signed rank test. Paired t tests were used when variables were normally distributed and after the homogeneity of variance was checked. The Shapiro-Wilk test was used to test the normality of distribution. The Levene test was used to assess the homogeneity of variance. When these conditions were not satisfied, a nonparametric Wilcoxon test was used. Correlations between normally distributed variables were assessed using the Pearson test. Otherwise, a Spearman correlation coefficient was estimated.
A linear regression plot analysis expressing 3D diameters according to 2D diameters was performed for the whole population and among subgroups according to the position of the defect. The same method was applied to display 2D and 3D diameters according to surgical findings. The Bland-Altman method was used to further explore agreement between the two techniques. P values < .05 were considered to indicate statistical significant. Statistical analysis was performed using Stata version 8 (StataCorp LP, College Station, TX).
Forty-eight patients with isolated VSDs were included. VSDs were membranous in 37 patients (77.1%) and muscular in 11 (22.9%). Surgical VSD closure was performed in 15 patients (31.3%). Characteristics of the whole study population and of subgroups according to the position of the defect are reported in Table 1 . The mean age of the study population was 21.4 ± 29.3 months (range, 1-123 months). There was a trend toward younger children in the muscular VSD group.
|All ( n = 48)||Muscular VSD ( n = 11 [22.9%])||Membranous VSD ( n = 37 [77.1%])||P|
|Age (mo)||21.4 ± 29.3||10.4 ± 18.9||24.7 ± 31.3||.06|
|Weight (kg)||9.3 ± 7.3||6.8 ± 4.8||10.0 ± 7.8||.07|
|Height (cm)||74.3 ± 24.1||65.0 ± 18.8||77.1 ± 25.0||.10|
|BSA (m 2 )||0.4 ± 0.2||0.3 ± 0.2||0.4 ± 0.2||.07|
|Surgical VSD closure||n = 15 (31.3%)||n = 4 (36.4%)||n = 10 (27.0%)|
VSD Measurements on 2D and 3D TTE
Acquisition of a 3D volume data set and measurements were feasible in all children (100%; 95% CI, 92.6%-100%). The mean obtained 3D image resolution was 30 MHz (range, 22-35 MHz).
VSD measurements using 2D and 3D TTE in the whole study population and among subgroups according to the position of the defect are reported in Table 2 . Maximal and minimal VSD diameters were not significantly different between the two subgroups.
|All ( n = 48)||Muscular VSD ( n = 11)||Membranous VSD ( n = 37)||P|
|Maximal diameter (mm)||7.3 ± 2.4||6.0 ± 2.5||7.7 ± 2.3||.06|
|Minimal diameter (mm)||5.1 ± 2.0||4.2 ± 2.2||5.4 ± 1.9||.05|
|Maximal-to-minimal diameter ratio||1.5 ± 0.4||1.5 ± 0.3||1.5 ± 0.5||.48|
|Maximal diameter (mm)||11.3 ± 3.3||12.1 ± 3.0||11.1 ± 3.4||.20|
|Minimal diameter (mm)||5.6 ± 2.2||4.5 ± 2.1||5.9 ± 2.1||.07|
|Maximal-to-minimal diameter ratio||2.3 ± 1.4||3.2 ± 1.5||2.1 ± 1.3||.01|
|dVSDA (cm 2 )||0.6 ± 0.3||0.5 ± 0.4||0.6 ± 0.3||.51|
|sVSDA (cm 2 )||0.4 ± 0.3||0.3 ± 0.2||0.4 ± 0.3||.05|
|Surface area variation (%)||32 ± 15||49 ± 11||26 ± 12||<.001|
Correlation between maximal 3D and 2D diameters was moderate (Spearman correlation coefficient = 0.51, P < .001). Correlation between minimal 3D and 2D diameters was good (Spearman correlation coefficient = 0.75, P < .0001). Linear regression plots expressing 3D diameters according to 2D diameters are displayed in Figure 2 for the whole population and among subgroups according to the position of the defect. Minimal 3D and 2D diameters were well correlated ( r 2 = 0.77, P < .0001). Correlation between 3D and 2D maximal diameters was significant, but the strength of the correlation was low ( r 2 = 0.29, P < .0001). Correlation between 3D and 2D maximal diameters was better in membranous versus muscular VSDs ( r 2 = 0.46 vs r 2 = 0.11). Correlations between 3D and 2D minimal diameters remained very good in membranous and muscular VSDs ( r 2 = 0.72 and r 2 = 0.90).
Maximal diameter assessed by 2D TTE was lower than by 3D TTE ( P < .0001). Bland-Altman analysis confirmed a mean bias of 4 mm. The magnitude of intertechnique differences observed for individual patients was quite high; 95% of values ranged from −1.76 to 9.75 mm. Mean bias was higher in muscular compared with membranous VSDs (6.02 vs 3.39 mm, respectively) ( Figure 3 ).
Although VSD minimal diameter was significantly lower by 2D TTE compared with 3D TTE ( P = .001), the difference was less pronounced. Mean bias was low, with a mean difference of 0.44 mm ( Figure 3 ). The magnitude of intertechnique differences observed for individual patients was quite low; 95% of values ranged from −1.80 to 2.69 mm. Trends were similar with regard to membranous and muscular subgroups.
VSD Measurements by Echocardiography Compared with Surgical Findings
VSD maximal diameters obtained by 2D and 3D TTE in 15 patients who underwent surgical repair were compared with surgical findings (summarized in Table 3 ). Surgical and 3D diameters were not significantly different from and higher than 2D diameters. Correlation between maximal 3D and surgical diameters was excellent (Spearman correlation coefficient = 0.98, P < .0001). Correlation between maximal 2D and surgical diameters was good (Spearman correlation coefficient = 0.73, P < .01). Linear regression analysis of surgical diameter measurements confirmed an excellent correlation with maximal 3D diameter ( r 2 = 0.97, P < .0001). Bland-Altman analysis confirmed the lack of any significant bias. The mean difference between surgical and 3D measurements was 0.25 mm, with 95% of values ranging from 1.35 to 0.85 mm. The correlation between surgical diameter and 2D maximal diameter was lower ( r 2 = 0.63, P < .0001). The mean maximal 2D diameter was significantly lower than mean surgical measurements ( P < .001). The mean bias by Bland-Altman analysis was 3.71 mm, with 95% of values ranging from −7.85 to 0.43 mm ( Figure 4 ).