Population studied

We examined 62 consecutive children referred for closure of an atrial septal defect using transthoracic three-dimensional echocardiography. There were 22 males and 40 females, the mean age being 8.5 ± 5 years, with a range from 2 to 18 years. The mean weight was 30 ± 15 kg, with a range from 10 to 85 kg. All patients had significant left-to-right atrial shunting with right ventricular overload as assessed by transthoracic cross-sectional echocardiography. The mean delay between the transthoracic three-dimensional echocardiographic study and the planned transcatheter or surgical closure of the defect was 4 ± 1 months, with a range from 1 to 6 months.

Acquisition of transthoracic three-dimensional echocardiographic reconstructions

Transthoracic echocardiography was performed with 3,5 MHz transducer connected to Acuson Sequoia C256 platform (Acuson, Montain View, CA, USA). The cross-sectional images were acquired with a small rotational device adapted for use in children (TomTec, Munich, Germany). Electrocardiographic gating by measurements of thoracic impedance ensured optimal temporal registration of the images, which were acquired from the subcostal position in patients weighing less than 20 kg, or from the precordial position in the heavier patients. The software obtains 36 sequential cross-sections for every 5° of sweep, each during a complete cardiac cycle, and stores them in the memory of the computer. We performed an average of 2 acquisitions for each patient, the one with minimal motion artefact being chosen for three-dimensional echocardiographic measurements. The mean time required for acquisition was 45 seconds, with a range from 30 to 90 seconds, depending on the heart rate of the patient.

Analysis of the three-dimensional images

From the conical data volume, we reconstructed specific cut planes to image en face the atrial septal defect and its rims as seen from the right atrium. We measured the maximal diameter of the defect in each patient when the septal deficiency was maximal in size during the cardiac cycle.¹¹ We then measured the extent of the rims surrounding the septal deficiency, taking the antero-inferior rim as the distance from the tricuspid valve, the postero-superior rim as the distance from the superior caval vein, and the postero-inferior rim as the distance from the inferior caval vein. The mean time required to acquire the three-dimensional echocardiographic reconstructions and make the measurements was 15 minutes per patient. The criterions used to select patients for transcatheter closure were adapted to the geometry of the device to be used for closure.¹⁴ In all patients, we used the Amplatzer atrial septal occluder, which is a circular double disk frame with a conjoint waist made of windings of nitinol. We required rims of greater than 4 mm to be present all around the septal deficiency, apart from the antero-superior rim, before proceeding to attempt interventional closure. Patients with rims of no more than 4 mm as measured from the three-dimensional echocardiographic reconstructions were referred for surgical closure (Fig. 1).

Figure 1. Transthoracic three-dimensional echocardiographic views from the right atrium. The left panel shows the centrally located deficiency within the atrial septum (arrow). The large rims, greater than 4 mm, all round the defect allowed transcatheter closure with the Amplatzer septal occluder. The right panel shows a large atrial septal defect (arrow). The restricted rim towards the tricuspid valve, less than 4 mm, precluded interventional closure. TV: tricuspid valve; ICV: inferior caval vein; SCV: superior caval vein.

Transcatheter closure

We measured the stretched diameter of the defects by profiling a balloon (PTA-OS NuMed, Inc) in left anterior oblique view, inflating the balloon with diluted contrast until a waist appears.¹⁵ The size of the Amplatzer septal occluder was then adjusted to the stretched diameter of the defect. After placement of the defect, residual shunting was assessed by cross-sectional transesophageal echocardiography.

Surgical closure

The atrial septal defect was exposed through a right atrial incision. The extent of the rims deemed deficient by three-dimensional echocardiography was measured intraoperatively on an empty flaccid heart, and the defect was closed using a pericardial patch. Absence of residual shunting was confirmed by cross-sectional transthoracic echocardiography.

Statistical analysis

Data are expressed as mean values with standard deviations. Blinded observers made the transthoracic three-dimensional echocardiographic measurements. Comparisons between the transcatheter and three-dimensional echocardiography methods for measuring the size of the septal deficiency were done by linear regression analysis. The agreement between methods was evaluated using the approach of Bland and Altman.¹⁶ Interobserver and intraobserver variability for measurement of maximal diameter by transthoracic three-dimensional echocardiography was determined in all studies. Measurements for the purposes of comparison were made on separate three-dimensional echocardiographic reconstructions.

Transcatheter closure

We referred 42 patients for transcatheter closure. The comparisons between the measurements made by transthoracic three-dimensional echocardiography and those using the sizing balloon are shown in Table 1. The correlation between the measurements was good (y = 3.45 − 0.73x; r = 0.78; p < 0.0001), with the mean difference between the average values for the maximal diameter obtained by the two methods being 2.4 ± 2.8 mm (Fig. 2). Of the 42 patients, 40 (95%) underwent successful transcatheter closure with no residual shunt. The mean size of the Amplatzer septal occluder was 22 ± 4 mm, with a range from 15 to 30 mm. In two patients with floppy septums, the stretched diameters as measured with the balloon were half as great again as predicted by the three-dimensional echocardiographic reconstructions. In these patients, therefore, the transcatheter procedure was abandoned, and patients referred for surgical closure.

Table 1. Patients selected by transthoracic three-dimensional echocardiography for transcatheter closure.

Figure 2. Comparisons of measurements made with transthoracic three-dimensional echocardiography (TT3DE) compared to the stretched diameter assessed using the balloon (BSD). The left panel shows the correlations between the measurements, while the right panel shows the differences of each pair of measurements as assessed using the method of Bland and Altman.

Surgical closure

We referred 20 patients directly for surgical closure. In these, the mean maximal diameter of the defect measured by transthoracic three-dimensional echocardiography was 23.8 ± 5.2 mm, with a range from 15 to 35 mm. From the transthoracic three-dimensional echocardiographic views, the antero-inferior rim was found to be deficient in 5 patients, the postero-superior rim in 6 patients, and the postero-inferior rim in 9 patients. We then found complete agreement on these assessments when comparing the three-dimensional echocardiographic and the intraoperative findings (Fig. 3). In the 2 patients in whom we had abandoned the transcatheter procedure, the surgeon found a floppy septum with multiple holes. All patients underwent uncomplicated surgical closure with no residual shunt.

Figure 3. Transthoracic three-dimensional echocardiographic and surgical views from the right atrium. The left panel shows a large round atrial septal defect (arrow). The size of the rim separating the defect from the inferior caval vein, measured at 3 mm, indicated surgical closure. The right panel shows the view obtained during the surgical correction, confirming the deficient postero-inferior rim. RPV: right pulmonary vein; ICV: inferior caval vein; SCV: superior caval vein.

The rims surrounding the septal deficiency

Should the rims be insufficient, then the device is prone to migration.^1–3 The minimal size of these rims depends on the geometry of the device. The umbrellas of the Amplatzer septal occluder extend from the waist for 7 mm all around the defect.¹⁴ Since the cross-sectional echocardiographer approaches the atrial septum from multiple orthogonal planes, mental three-dimensional constructs are needed to comprehend the relation between the defect and the surrounding structures. Transthoracic three-dimensional echocardiography allows surface imaging of the atrial septum. The extent of the septal deficiency, and the size of the rims throughout the circumference of the defect, is simultaneously and directly imaged from the three-dimensional echocardiographic views. Based on the transthoracic three-dimensional echocardiographic measurements of the rims, we excluded almost one-third of the patients from attempted transcatheter closure. The operative findings in these patients endorsed completely the three-dimensional echocardiographic data. The agreement on the localisation of the deficient rims was complete, while the mean difference between the echocardiographic and surgical measurements was less than 1 mm. It was the rim towards the inferior caval vein that was deficient in nearly half the cases. As did Magni et al.,¹⁷ we encountered difficulties in visualising the entire length of the postero-inferior rim using transesophageal three-dimensional echocardiography¹⁷. In contrast, transthoracic three-dimensional echocardiography depicted very well the inferior rims, as shown in Figures 4 and 5. Because of the geometric profile of the Amplazter septal occluder, we chose to ignore the extent of the antero-superior rim, which charts the distance from the aorta.^1–3

Figure 4. Transthoracic three-dimensional echocardiographic views from the right atrium. The left panel shows an oval septal deficiency (arrow). The rim from the tricuspid valve is well seen. The right panel shows the same patient after transcatheter closure. The Amplatzer septal occluder has a concave shape in the right atrial cavity, and does not obstruct flow through the inferior caval vein. ICV: inferior caval vein.

Figure 5. Transthoracic three-dimensional echocardiographic views from the right atrium. There is a large and round deficiency within the oval fossa. The narrowness of the rim separating the defect from the tricuspid valve, less than 3 mm, contra-indicated transcatheter closure. The inferior caval vein, and the Eustachian valve floating in the right atrial cavity, are well seen.

The maximal diameter of the septal deficiency

Three-dimensional echocardiography mimics the view of the beating heart as seen through an atriotomy, and allows description of the septal deficiency throughout the cardiac cycle.¹⁸ One striking finding was the great variability in the shape of the atrial septal defect, this observation emphasising how such alterations in the shape of the defect might alter the accuracy of calculations made using cross-sectional echocardiography. In those defects that are circular, cross-sectional images certainly cut the defect in its maximal diameter, but this is no longer true in those defects having complex shapes. Thus, in patients with an elongated oval-shaped defect, cross-sectional echocardiography could underestimate significantly the maximal diameter. Moreover, we found that the surface area of the septal deficiency changed significantly during the cardiac cycle, being maximal in late ventricular systole, and minimal in late left ventricular diastole.¹⁹ Such variation in surface area through the cardiac cycle increases the difficulty of measuring the maximal diameter by selecting the appropriate cross-sectional frame. It is the sizing balloon, nonetheless, which is still used most frequently to determine the extent of the septal deficiency.^{1–3, 15} The relation between the stretched diameter measured with the balloon and the maximal diameter of the septal deficiency is controversial. We found a mean difference of 2 mm between our three-dimensional measurements and the diameters determined using the balloon. This explained the slight oversize of the Amplatzer septal occluder used compared to the maximal diameters predicted from the three-dimensional reconstructions, these being 22 and 19.2 mm, respectively. In particular, we found that in our patients with floppy septum, the three-dimensional reconstructions define different properties of the defect compared to balloon sizing. The three-dimensional echocardiographic views demonstrate the major axis, incorporating information about the shape of the atrial septal defect. Balloon sizing provides information about the degree of physical stretch, a feature that cannot be predicted by echocardiography.

Limitations

Our measurements were obtained only from reconstructions that recreated the right atrial views obtained by the surgeon. While transesophageal three-dimensional echocardiography offers reconstructions from both sides of the atrial septum, our transthoracic three-dimensional approach did not allow good visualisation of the atrial septal defect from the left atrium. This is explained by the smaller cavity of the left compared to the dilated right atrium, and the greater distance from the probe, permitting poorer resolution for the left atrial reconstructions. The absence of obstruction to pulmonary venous flow, nonetheless, was assessed by cross-sectional transesophageal echocardiography after deployment of the device. The resolution of the transthoracic three-dimensional reconstructions was insufficient in most of patients to visualise the Amplatzer septal occluder. Three-dimensional color Doppler reconstructions may help to further define small defects or residual shunting.^{20, 21} The ability to reconstruct directly from the digital images, without transformation into video signals, would also improve the quality of the three-dimensional images, which is still very dependent on the gain settings.^{22, 23}

In conclusion, we have shown that transthoracic three-dimensional echocardiography is accurate in measuring the maximal diameter of defects within the oval fossa, and the extent of the rims surrounding them. This non-invasive technique should prove invaluable in selecting children for interventional closure. The use of the sizing balloon during closure provides necessary additional information, such as the resistance of the septum. Sizing and placement of the device should take in account both the three-dimensional echocardiographic and transcatheter findings.