In this multicenter study, using three-dimensional cardiac CT image-based measurements to predict the size of LAAO devices, the perimeter-derived diameter of the landing zone was found to be the most accurate predictor of device size. The perimeter-derived diameter was advantageous when the cross-sectional shape of the LAA was eccentric (EI > 0.2). Additionally, oversizing based on the TEE sizing chart was unnecessary due to the superior spatial resolution of CT imaging. To the best of our knowledge, this is the first clinical study to compare the accuracy of device size estimations obtained with different CT parameters using the size of the successfully implanted device as reference.
LAAO planning with CT measurements
TEE is a conventional pre-procedural method for evaluating the LAA. It is used to measure the width of the ostium and landing zone in multiple planes; the maximal diameter of the landing zone is used to determine the ACP or Amulet device size.[8] However, CT images have higher spatial resolution than TEE images and can be reconstructed into three-dimensional images, thereby providing the operator with a better understanding of the morphology of the LAA and its surrounding structures.[11,13,15] Therefore, CT is being used increasingly for the pre-procedural assessment of other percutaneous procedures, including transcatheter aortic valve replacement.[19] Recent studies have reported that CT provides better accuracy for LAAO planning than TEE.[13,15] However, as it is unclear which parameter should be used for sizing, no standardized protocol or recommendations have been established for using CT measurements to plan LAAO. Therefore, our study aimed to establish a practical method of successfully using CT-based parameters to predict LAAO device size.
Perimeter-derived diameter
The ACP and Amulet devices both have eight different sizes in 2–3 mm increments. In this study, the mean difference of the perimeter-derived diameter and actual device width was significantly smaller than 2 mm (-0.8 ± 2.4 mm), indicating a minimal error and excellent match. Previous studies have suggested that the perimeter is the most dependable parameter for evaluating the LAA ostium. Wang et al. compared the different parameters during different cardiac phases and found that the perimeter-derived diameter had minimal changes (1–2 mm) and was reliable for reproducing the ostium.[20] However, this study compared the parameters measured using two-dimensional oblique and three-dimensional measurement methods, focusing on the reproducibility of the LAA ostium rather than the accuracy of the sizing. More recently, Jia et al. compared the parameters measured using three-dimensional printed models and found a good correlation between the perimeter of the LAA orifice and LAmbreTM device size.[21] However, the ability of 3D printing to reflect actual cardiac anatomy may be limited, as the volume within the chambers changes throughout the cardiac cycle. Our study used successfully implanted devices as the reference to compare predicted device sizes based on CT measurements.
Underlying mechanism of superior accuracy in sizing with perimeter-derived diameter
In the current study, the minimal, average, and area-derived diameters underestimated the size of the LAAO device. Significant undersizing may lead to complications, including device malpositioning, embolization, or peri-device leakage. A comparison of the parameters in this study is shown in Supplement Figure 5. The average diameter was calculated as the arithmetic mean of the minimal and maximal diameters . When the minimal diameter was significantly smaller than the maximal diameter, or when the shape was more eccentric, the average diameter was relatively a small value. The area-derived diameter of an ellipse can be calculated as the geometric mean of the minimal and maximal diameters . Inequality of the two mean values indicates that the geometric mean is always less than or equal to the arithmetic mean, leading to an underestimation of size when using area-derived diameters.
The EI was identified as the important factor when comparing the results of device size for each parameter. The cross-sectional shape of the LAA ostium and landing zone is typically elliptical or irregular, while the occluding device is circular.[22-24] This difference may lead to a discrepancy between the predicted sizes and actual device sizes. EI can be used to determine the shape of the LAA ostium and landing zone, as the shape is more circular when the EI is approximately 0. When the occluding devices are inserted, the shape of the landing zone deforms to adapt to the device.[20] This adaptation does not lead to significant changes in the LAA parameters in patients with more circular EI. However, in patients with more eccentric landing zones, the adaptation significantly changes the minimal, maximal, average, and area-derived diameters, while the perimeter-derived diameter does not change significantly. In our study, the discrepancy between the predicted device size was greater when the EI was >0.2. Almost half of the patients in this study (48.4%) had an EI >0.2, indicating that using diameters other than the perimeter-derived diameter may lead to a mismatch in the device selection.
The maximal diameter measured in this study was similar to the actual implanted device size and not significantly different from the perimeter-derived diameter. The LAA is a relatively distensible structure within the heart, serving as a volume reservoir during the systolic phase.[25] This anatomical characteristic may allow for the maximal diameter to remain as an important measurement in LAAO, along with the perimeter-derived diameter. However, the insertion of grossly large devices may lead to malpositioning of the device and post-procedural complications, including device embolization, peri-device leakage, thrombus formation, and cardiac tamponade.[18,26] Additionally, in more eccentric cases with EI >0.2, the maximal diameter showed significant error compared with the perimeter-derived diameter, leading to overestimation of device size. Therefore, in patients with highly eccentric LAA ostium shapes, the perimeter-derived diameter may be the most accurate parameter for device size selection.
Unnecessity of oversizing with CT-based measurements
Our study also assessed the need for oversizing when planning for LAAO using CT. Sizing charts provided by the device manufacturer were used to determine the ideal device size correlating to the obtained measurements.[6] These charts typically suggest oversized device disc diameters due to an underestimation of the dimensions when two-dimensional TEE is used.[27] Three-dimensional TEE has improved the accuracy for the assessment of the true LAA orifice compared with two-dimensional TEE;[28] however, the measurements were smaller than those obtained using cardiac CT.[27]
When the oversizing method was used with CT-based measurements, each parameter was significantly mismatched with the actual device size. Oversizing improved the accuracy of the predicted device size when the minimal diameter was used from a mean error of -4.8 ± 3.3 mm to -1.8 ± 3.5 mm. However, the minimal diameter underestimates the lobe size irrespective of oversizing, and this improvement is clinically irrelevant as the minimal diameter is not used independently for the sizing of LAAO devices. All other parameters significantly overestimated the device size when oversizing was used. Therefore, oversizing may be unnecessary when CT images are used for pre-procedural LAAO planning.
Limitations
This study has a few limitations. First, only patients who underwent LAAO with no complications or device size mismatching were included, and our results were not directly compared with those obtained using the conventional TEE method. However, we used the actual implanted devices as reference to compare the device sizes predicted using different parameters. Second, this was a retrospective study and may have been influenced by unobserved confounders and selection or referral biases. Thus, the clinical feasibility and usefulness of this sizing method, such as the improved success rate, decreased procedure time, and decreased number of attempts, must be verified in future prospective studies. Lastly, this study only included patients implanted with ACP or Amulet devices, which led to the exclusion of a significant number of patients with Watchman devices. This exclusion criterion was based on the different sizing techniques used for Watchman devices. A method for using CT-based images for the pre-procedural planning of the implantation of Watchman devices is necessary.