The present study demonstrates that PC-CMR-derived measurements of ASD diameter and rim sizes show strong positive correlation with the corresponding values assessed by conventional 2D TEE in adult patents with ostium secundum ASD. PC-CMR also appears to be superior to 2D-TEE for evaluating the preprocedural feasibility for device closure in these patients, as 2D-TEE significantly underestimates maximal ASD diameter and PI rim size and can occasionally miss an anomalously draining pulmonary vein. In this study, all the patients deemed to be suitable for device closure on CMR ultimately underwent a successful transcatheter closure with a device whose size correlated strongly with the defect size measured on CMR.
TEE is the most widely available and validated method for evaluating the feasibility for device closure in adult ASD patients and is recommended for all patients who are being planned for this procedure [8, 9]. It provides a thorough insight into the defect morphology including type, size, shape, number of defects, presence of adequate rims for device anchorage, and ruling out any associated anomalies that could potentially complicate or preclude device closure [5,6,7,8,9]. TEE also serves as an invaluable tool for intra-procedural guidance and allows safe and accurate device deployment by ensuring stable device position, ruling out residual shunts and impingement of important structures before device release [3, 4, 21]. However, TEE is a semi-invasive imaging modality causing discomfort to the patients and requires sedation or occasional administration of general anaesthesia in uncooperative patients. Furthermore, although a relatively safe procedure, serious complications including oesophageal laceration or perforation, upper GI bleeding, pharyngeal tears, cardiac arrhythmias, laryngospasm, and hypoxia have been described in 0.1–0.9%. Mortality is rare, occurring in 0.01–0.02% cases [22, 23].
2D TEE is the most widely used mode for assessing ASD anatomy before transcatheter closure. Given the two dimensional nature, it is fraught with inherent limitations of comprehensively defining the shape, eccentricity and multiplicity of the defects, especially in complex secundum defects [24]. With the advent of real-time 3D matrix array TEE transducers in recent years, many of these limitations have been circumvented. Real-time 3D TEE allows more comprehensive interrogation of atrial septal anatomy and dynamic en face visualisation of ASD throughout the cardiac cycle. However, limited availability and lack of widespread expertise are the major stumbling blocks in its routine use in our part of the world. Furthermore, the posteroinferior part of the atrial septum may be inadequately visualised due to artefactual dropout, small fenestrations may escape detection due to limited spatial resolution, and patient movement during image acquisition can result in malalignment and reconstruction artefacts [25, 26].
Besides the obvious advantages of three-dimensional imaging and non-invasive nature, CMR generally outperforms TEE in terms of larger field of view, unlimited tomographic planes acquisition, and lesser operator dependence [10, 11]. Conventional CMR sequences including spin-echo technique and cine gradient-echo technique are often inaccurate in depicting the size and margins of secundum ASD, owing to septal thinning adjacent to the defect and low interatrial pressure gradient across the defect [16]. PC-CMR imaging, on the other hand, has been demonstrated to be superior to these techniques in reliably defining the size and shape of ASDs as it more sensitively detects localised low velocity phase shifts across the defect and provides a 3D en face view of the defect based on flow related signal enhancement [15,16,17,18,19,20]. In a study of 30 adult patients, Holmvang et al. demonstrated that ASD dimensions measured on PC-CMR showed an excellent correlation with balloon sizing of the defect during catheterisation as well as template standards measured during surgery. Spin-echo imaging overestimated the defect diameter by 48% in the same study that was attributed to signal dropout in the fossa ovalis region due to septal thinning [16]. Beerbaum et al. in a study of 65 paediatric patients, demonstrated good agreement between PC-CMR and TEE-derived ASD size (mean difference < 1 mm). Among 30 patients who were scheduled for transcatheter closure based on PC-CMR findings, 5 patients were found to have unexpectedly large defects on stretched balloon sizing and hence referred for surgery [17]. Consistent with the previous studies, there was a strong correlation between TEE and CMR vis a vis ASD diameter in the present study (r = 0.81, p < 0.001). However, TEE significantly underestimated the defect size compared to PC-CMR (mean difference − 1.6 mm; 95% CI − 2.8, − 0.4 mm). Furthermore, 2 patients with defect diameter < 34 mm on TEE versus > 34 mm on CMR could not undergo successful transcatheter closure as the largest available device size (40 mm) did not suffice in completely occluding the defect while achieving a stable position before release. We did not use balloon sizing technique in this study as contemporary data suggest that it is no longer necessary and may occasionally lead to device oversizing [27, 28]. Our results also demonstrated that the final device size that successfully occluded the defect correlated more strongly with the defect dimensions on PC-CMR (r = 0.85, p < 0.001) as compared to TEE (r = 0.71, p = 0.02). These findings are in sync with those published by Thomson et al. and Durongpisitkul et al. which demonstrated better correlation between device size and ASD diameter on CMR in comparison with intracardiac echocardiography (ICE) and TEE, respectively. [15, 20] The ability to visualise the defect en face on PC-CMR provides an unequivocal advantage over conventional echocardiography in clearly outlining the defect especially when the location or shape is eccentric.
In so far as the assessment of various rims is concerned, we found a significant correlation between rim sizes measured on TEE and CMR (r = 0.68, p < 0.001 for AS rim; r = 0.64, p < 0.001 for PS rim; r = 0.54, p < 0.001 for AI rim; r = 0.56, p < 0.001 for PI rim). The mean difference of rim measurements was less than 1 mm for all rims except for PI rim. TEE significantly underestimated PI rim size compared to CMR (mean difference − 1.98; 95% CI − 3.68, − 0.28; p = 0.023). This was even after we used all the recommended TEE manoeuvres including retroflexion and withdrawal of the probe from stomach for adequate visualisation of PI rim. Among seven patients with insufficient PI rim on TEE, five had rim size > 5 mm on CMR and subsequently underwent a successful device closure. There are a few studies which have demonstrated that transcatheter closure, although difficult, may be safe and feasible in some patients with deficient PI rim on TEE [29, 30]. Given the facts that inadequate evaluation of the posteroinferior part of the atrial septum represents an inherent limitation of TEE, even with the latest 3D technology, and more than 70% patients with deficient PI rim on TEE actually had sufficient rim on CMR in our study, we can infer that CMR is superior to TEE in deciding the candidature for device closure based on PI rim size [13, 24,25,26]. In the present study, CMR picked up anomalous drainage of left upper pulmonary vein into innominate vein in one patient which was missed on TEE, hence excluding the patient from transcatheter closure. Previous studies have shown that CMR may identify additional cardiac or extracardiac anomalies in up to 20% patients that alter clinical management decisions [15, 17]. Meticulous echocardiographic evaluation and exclusion of patients with multiple defects or atrial septal aneurysms could have attributed to such lower proportion in the present study. Besides the lack of widespread availability, there are certain important limitations to CMR that need to be mentioned. Apart from prolonged scanning times and ability to hold breath, CMR has been shown to be inferior to TEE in detecting patent foramen ovale or small septal fenestrations [15, 31]. Furthermore, PC-CMR imaging does not reliably depict septal thickness because the method depends on flow imaging rather than on structural delineation of thin membranes. 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 PC-CMR imaging [17]. Finally, the utility of CMR is limited by the presence of metallic prostheses or claustrophobia in a small percentage of patients.