Atrial fibrillation (AF) is a major cardiac arrhythmia and is known to increase the risk of mortality (1) including an increased risk of stroke (2, 3). In patients with AF, the left atrial appendage (LAA) was identified as a source of thrombus associated with a higher risk of causing stroke (4) with oral anticoagulation representing an effective therapy to reduce the risk of stroke (5). In patients with contraindications to oral anticoagulation (6) closure of the LAA (LAAc) represents an efficient therapy option (7). However, the high variability of LAA shapes and dimensions (8) challenges optimal occluder selection and final implantation (9, 10), indicating the need for accurate preprocedural assessment of the LAA.

The gold standard imaging modality for LAA assessment for LAAc is 2D transesophageal echocardiography (TEE), also recommended in conjunction with x-ray fluoroscopy (XR) (11). The selection of device size is based on the maximum diameter obtained from the LAA diameters assessed at the anticipated landing zone of the occluder (12). However, despite the high spatial resolution of both imaging modalities, underestimation of the maximal diameter derived from 2D TEE and XR has been reported, which was attributed to the often oval shape of the LAA and the non-optimal choice of view geometry selected in 2D TEE or projective XR for measurement (13–15).

In general there is a high demand for preprocdural assessment of the LAA geometry for accurate procedure planning. The use of 3D TEE has been shown to be more accurate than those based on 2D TEE (14, 16), although the maximum diameter is still underestimated compared to measurements based on 3D computer angiography (CTA) (17, 18). CTA has been shown to be an appropriate imaging tool for LAAc planning (19). In case of 3D imaging data, device size selection based on perimeter and area measurement derived landing zone diameters have been reported to be more reliable than simply using the maximum diameter (16, 20). In addition to the correct assessment of the diameter, the optimal XR angulations for periprocedural imaging during implantation could be retrieved from the CTA data (21). Even though preprocedural assessment of the LAA based on CTA has the potential to improve the outcome and efficiency of the intervention (22, 23), its application is limited in patients with reduced kidney performance and due to its intrinsic risk of ionizing radiation is under safety/benefit debate for general application in LAAc (24). Here, cardiac magnetic resonance imaging (CMR) may gain interest especially considering the additional excellent soft-tissue contrast, enabling thrombus detection in the LAA (25, 26). However, where there is general consensus that 3D CT is appropriate for LAAc planning, the respective role of 3D CMR for LAAc planning needs further investigation (11).

As further approaches for facilitating improved preprocedural device selection printed 3D models either derived from CTA (27, 28) or CMR data (29) as well as an interactive modelling tool for LAAc planning (30) have been reported.

The objective of this pilot study is to investigate the application of non-contrast-enhanced 3D CMR in preprocedural planning of LAAc. For initial accuracy assessment of CMR, the landing zone diameters were quantified retrospectively at the same LAA location as used for the XR-derived periprocedural measurements. For estimation of the value of CMR for LAAc planning, landing zones were determined prospectively based on CMR and the resulting diameters and optimal C-arm angulations compared to periprocedural data.

13 patients (85% male, 76 ± 9 years) with atrial fibrillation (paroxysmal, persistent, long-persistent, permanent) not suited for oral anticoagulation therapy were enrolled in this pilot study [n = 10 Watchman FLX™ (Boston Scientific, Massachusetts, USA); n = 2 LAmbre™ (Lifetech Scientific, Shenzhen, China); n = 1 Amplatzer™Amulet™ (Abbott, Illinois, USA)]. The procedure was performed according to the current clinical recommendations. For device size selection, the LAA of the patients was measured periprocedurally with 2D TEE (EPIQ CVxi, Philips Medical Systems, Best, The Netherlands) at view angles ∼45°, ∼90°, and ∼135°. To approve the TEE measurements, contrast-enhanced XR (Allura Clarity, Philips Medical Systems, Best, The Netherlands) were additionally routinely acquired in right anterior oblique and caudal angulation as recommended (31). If according to the objective impression of the interventional cardiologist the LAA was not well displayed in the XR projections acquired in recommended angulation, additional XR angulations were acquired and analyzed. The maximum diameter measurement of the identified landing zone was considered for the selection of the device size. All patients underwent an CMR examination the day before the actual intervention. The evaluation was conducted in accordance with the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the local ethical committee. Written informed consent was obtained from all individual participants included in the study (drks.de DRKS00015649).

CMR and LAAc were successful in all patients. Two cases (Watchman FLX™) were excluded from the XR-CMR comparison due to the non-availability of suited XR data caused by inadequate XR angulations and final device selection solely based on TEE. In two of the remaining 11 cases initially selected devices were underestimated and required subsequent reselection during the ongoing procedure.

The potential advantages of preprocedural planning for LAAc like improved device size selection and angulation prediction have already been reported for CTA (17, 43). By utilizing a non-contrast-enhanced CMR protocol for preprocedural planning, radiation and contrast agent dose might further be reduced in comparison to CTA. This small pilot study showed the possibility of 3D imaging of the LAA of patients with atrial fibrillation using CMR for preprocedural planning.

Dimension assessment of the LAA based on the CMR at the identical landing zone as the XR-derived measurements showed high congruency indicating potential support of CMR for preprocedural planning. The significant differences in maximal diameters between CMR and XR are likely caused by the limited spatial resolution in CMR and non-optimal XR angulation as also previously reported in the context of preprocedural planning for LAAc based on CTA (13, 23, 44). This assumption was supported by the correlation of the ovality of the LAA with the deviation between the two modalities. Calculation of the CMR derived diameter at similar angulation direction as the XR clearly improved the agreement between the modalities.

From the current data, using the derived diameters for device selection appears feasible, as already motivated for CTA-based measurements (20, 45) and 3D TEE (16). Alternatively, a new device selection chart might be established for measurements based on 3D imaging as the currently used sizing charts are intended for TEE-based measurements and are intrinsically considering the associated underestimation (20). Even though underestimation with TEE has been considered in the device sizing chart, a more accurate LAA size measurement with CT has reported to lead to a lower number of re-device selection (46). LAA dimensions of CMR-derived landing zones overestimated the dimensions derived from TEE significantly, which is in concordance with previous studies (14, 47).

Previous studies (26, 28), as well as this study, compare periprocedural TEE and fluoroscopy measurements with measurements based on preprocedural data. Differences in observed diameters may be explained by the moment of imaging rather than the imaging modality. Patients undergoing TEE during the procedure generally have been fasting for a while, resulting in a more hypovolemic state. This could lead to underestimation of LAA diameter in XR and TEE imaging. This is in line with the current study, showing larger LAA diameters in CMR compared to XR and TEE. Ideally, imaging modalities should be compared at the same moment to account for this possible bias.

A general limitation in CMR rises from the rather low spatial resolution, which did not allow for clear identification of the trabecular structures in the LAA, and hence a clear landmark for exact determination of the landing zone maybe missing. The difficult landing zone identification in CMR may explain the somehow larger deviations of the CMR-derived measurements for device size prediction as compared to the CMR validation at identical landing zones.

However, even with the potentially non-optimal landing zone derived from CMR successful angulation prediction could be performed. The predicted angulation had high congruency with the final angulation clinically chosen showing the high potential of preprocedural CMR-based angulation determination in saving radiation, contrast agent, and interventional time.

A further limitation for CMR in LAAc procedure planning may rise from the rather long scan time, high related costs, and necessary exclusion of patients with non-MR compatible pacemaker. As such, CMR is unlikely to completely replace CT as preprocedural routine imaging modality. Nonetheless, for patients not suitable for CT, e.g., with impaired renal function, CMR has the great potential to become an alternative non-invasive contrast agent-free imaging modality for providing the information required for preprocedural planning. For assessment of the cost effectiveness of this approach larger blinded studies including the assessment of procedure time, procedural success and long-term patient outcome need to be performed.

This pilot study demonstrates the high potential of non-invasive, contrast agent-, and radiation-free CMR imaging in planning for LAAc. Based on the 3D CMR, the shape of the LAA can be evaluated and the landing zone can be determined for dimension assessment and angulation prediction. However, device size selection based on CMR-derived measurements maybe limited by the current sizing charts. Due to the limited accuracy in the maximal diameter the usage of derived diameters seems to be recommended and investigated in future studies.