This study tested the feasibility, reliability, and reproducibility of a novel ABC classification for LAA sealing based on the pattern and grade of IDT as evaluated by both CCT and TEE.

The ABC classification was highly reproducible across imaging modalities and readers of differing levels of experience. TEE could reliably classify LAA sealing in agreement with CCT findings. Accordingly, type A patterns detected by TEE were a marker of sealing whereas type C was a marker of patency. Based on frequent transformation of type B patterns into A or C with time, type B patterns can most likely be regarded as small leaks or a transitional phase in device healing.

Cardiac imaging has contributed substantially to the advancement of LAAC [20,21,22].

Worldwide, image surveillance following LAAC is routinely performed using TEE and/or CCT to examine possible complications such as DRT and PDL as well as to monitor device sealing [23].

The diagnosis of PDL as a suboptimal or unwanted result after LAAC remains a significant challenge [12]. Variabilities in incidence rates and thresholds for clinical significance of leakage possibly stem from differences in definitions, imaging modalities (TEE vs. CCTA), and timing of evaluation [20,21,22, 24]. TEE remains the least expensive and most versatile imaging modality to support LAAC and is also the most frequently used modality utilized in most RCTs [25]. While offering numerous advantages, the patient compliance for follow-up due to its invasive nature. CCT, on the other hand, is a widely accepted non-invasive alternative to TEE for post-LAAC surveillance and displays superior sensitivity in detecting PDL [15, 17, 26]. Moreover, LAA patency without visible leaks might only be detected by CCT [15, 17, 25].

Despite all advancements in imaging techniques, a substantial discrepancy in leak quantification remains between TEE and CCT. Therefore, the quest for a classification of device sealing status following LAAC, valid across these two dominant imaging modalities, is desirable since a congruent communication between the two primary imaging modalities could help resolve some of the aforementioned remaining uncertainties in PDL evaluation.

Identifying patterns of IDT could facilitate the development of a unified classification system. These patterns may provide a surrogate marker for the process of device healing and even provide clues for the mechanisms of patency. Previous anecdotal studies have reported the feasibility of this approach, but verification of diagnostic accuracy with CCT as well as its reproducibility was lacking [18, 27, 28].

In this cohort, categorizing IDT into three major groups, we showed that TEE was very highly sensitive to detection of almost complete thrombosis of both the device and the LAA (Type A) using CT as the reference. Among patients with Type A, only 2.9% had concurrent CCT-detected PDL at 90 days, mostly of minor size, while no PDLs were found in this group on TEE. On the other hand, Type C patients were associated with PDL in 100% of cases irrespective of imaging modality. TEE showed less sensitivity in detecting Type C patients but with almost perfect specificity. Lower sensitivity might in part be due to innate limitations of the ultrasound itself. Whether using ultrasound-enhancing agents (UEA) might help increase the sensitivity remains to be systematically evaluated in future studies [23].

Considering the geometric characteristics of PDL, it becomes evident that a two-dimensional color-Doppler assessment of the jet may be inadequate for evaluating a crescent-shaped three-dimensional leakage. One recently published retrospective study has attempted to address the problems of 2D quantification of color jets by employing a 3D approach yielding promising results [29]. However inherent limitations of color 3D imaging still restrict its widespread clinical adoption and consistent real-world use.

Clinically significant PDL could also theoretically be caused by multiple gaps of smaller sizes, which further complicate the 2D evaluation of the jet. A non-Doppler-based evaluation of sealing using IDT might help avoid these pitfalls and enhance its clinical reproducibility.

We observed a tendency for larger PDL sizes in patients with Type C at 90 days compared with Type B1. This should be investigated further in a larger study.

The Type B2 represented a rare form of device patency in which failure of IDT was mainly proximal without visible leaks. This might represent a lack of complete device membrane endothelialization. This type was found only during early follow-up imaging (until 45 days) and might imply a transitional stage in the process of device thrombosis and surface healing.

Using a single antiplatelet discharge regimen, the observed DRT rate (4.6%) is aligned with previously observed frequencies [7, 30,31,32,33]. Any association between the ABC classification and DRT remains to be further studied.

The substantial inter-reader agreement (kappa range, 0.53–0.69) and intra-reader agreement (kappa range, 0.75–0.83 for TEE to predict the CT-based classification) also confirm the feasibility of this “common language” being implemented regardless of the imaging modality.

We have demonstrated that even with minimal training provided to a single novice reader, consistently high levels of both inter- and intra-reader agreement and reliability were observed across both groups. This underscores the easy applicability of the ABC Classification.