Introduction
Since the inception of transcatheter aortic valve implantation (TAVI) approximately twenty years ago [
1], technological enhancements and procedural simplifications have allowed its routine use in current practice to treat symptomatic aortic valve stenosis [
2,
3] New-onset conduction disturbances (NOCD), including high-degree atrioventricular block (HAVB), permanent pacemaker implantation (PPI), and new-onset persistent left bundle branch block (NOP-LBBB), remain common complications following transcatheter aortic valve implantation (TAVI) and are associated with poorer long-term outcomes [
4‐
8]. As TAVI indications expand to younger and lower-risk populations, strategies to mitigate conduction disturbances have gained increasing importance.
Beyond patient-related and anatomical predictors, increased attention has recently been paid to per-procedural methods for reducing NOCD. The cusp-overlap (COL) technique has been described as an alternative to conventional three-cusps coplanar (TCC) projection, aiming to reduce parallax and improve visualization of the non-coronary cusp, thereby facilitating more predictable valve implantation depth (ID) [
9]. According to recent works, higher implantation of transcatheter heart valve (THV) decreases the rate of NOCD requiring PPI for balloon-expandable valve (BEV) [
10] and self-expandable valve (SEV) [
11‐
15]. The role of ID in NOCD is closely related to inter-individual anatomical variations like membranous septum length (MSL) [
16,
17] and the volumes and distribution of calcifications of the aortic valve complex [
18,
19], which were not systematically accounted for in recent observational studies. Most studies comparing the COL and TCC techniques involved a “historical” cohort of TCC, which also raises the problem of heterogeneous management of NOCD for which guidelines were only recently published [
20,
21]. We therefore hypothesized that COL does not exert a direct protective effect on NOCD per se, but rather may indirectly influence outcomes by facilitating higher implantation relative to patient-specific anatomy, quantified by the difference between MSL and ID (ΔMSID).
Accordingly, the present study aimed to compare COL and TCC techniques in a contemporary cohort, with a specific focus on the mechanistic role of implantation depth relative to membranous septum length.
Multidetector computed tomography
Enhanced-contrast electrocardiographically gated MDCTs were performed at end-systole (≈ 30% of the heart cycle) and acquired with collimation, as recommended for measuring aortic valve complex dimensions and peripheral vascular access [
25,
26]. Working projections, MSL, aortic annulus areas, and diameters were manually obtained and reported for each patient during postprocessing from the workstation AW Server 3.2 (General Electric Healthcare). Measurements were performed retrospectively by a single experienced operator. The aortic annulus was defined by the luminal contour within a virtual plane aligned with the basal attachment points of the three aortic valve cusps [
25]. For a standardized analysis of the membranous septum, the cursor in the perpendicular co-planar view was placed on the intersect of the non-coronary and the right coronary cusp. The coronal view was swept to find the lower part of the membranous septum [
25]. Its length was the Euclidean distance between its lower part and the aortic annulus plane. Intra-observer variability assessment was performed on 50 randomly selected cases (Fig.
1in the Supplemental Appendix).
Volumes of calcifications were obtained from a fully automated detection of landmarks within the aortic valve complex [
27], which was separated into three sections in the craniocaudal axis. The device landing zone (DLZ) was defined as the cylinder ranging from 3 mm above to 2 mm below the aortic annulus plane. The left ventricular outflow tract (LVOT) was defined as the 10 mm below the aortic annulus plane along the centerline. The inferior limit of the total leaflet (TL) sector was the plane of the annulus aortic plane. We used an arbitrary height of 12 mm for this sector. The upper leaflet (UL) section was limited below by the DLZ and has the same upper limit as the TL section. Each craniocaudal section was subdivided into three sub-sectors corresponding to the projections of the NCC, the LCC and the RCC [
18]. Overall, 17 sectors were individualized for the analysis (Fig.
2in the Supplemental Appendix). A threshold of 850 Hounsfield-Unity (HU) [
28] was used to quantify the volumes of calcifications, with a systematic review by an expert operator to detect any artifact that could lead to a misestimation. Each volume was expressed in mm
3 of calcifications and as a percentage of the total sector volume.
Discussion
In this contemporary cohort, the cusp-overlap technique was not independently associated with a reduction in HAVB or PPI at 30 days. Our findings therefore primarily provide confirmatory evidence that implantation depth relative to membranous septum length, rather than the angiographic projection itself, is a key determinant of conduction outcomes after TAVI. Nonetheless, COL significantly reduced ID, thus the COL technique should be interpreted as a facilitating imaging tool that may support anatomy-guided implantation strategies, rather than as a direct determinant of conduction outcomes. Importantly, the projection strategy itself was not independently associated with long-term outcomes after multivariable adjustment. In contrast, early conduction disturbances were strongly associated with subsequent mortality and heart failure hospitalization, underscoring that conduction injury rather than implantation projection is among the key drivers of late clinical events.
The COL uses right anterior oblique and caudal orientation to superimpose the RCC and the LCC, and display the NCC on the opposite side of the projection. This technique aims to reduce THV parallax and offer a more predictable ID. It is acknowledged that the occurrence of NOCD requiring PPI is directly linked to the MSL [
16,
17]. Indeed, the conduction pathways protrude in the lower part of the membranous septum [
30] and can be damaged by the deployment of the THV. Therefore, many observational studies on this subject demonstrated that COL may reduce PPI after TAVI, regardless of THV type [
10,
11,
13,
15]. In a study limited to the SAPIEN 3 BEV (Edwards Lifescience), Sammour et al. [
10] found that a systematic high deployment approach using the COL reduced both PPI (5.5% vs. 13.1%;
p < 0.01) and NOP-LBBB (5.3% versus 12.2%;
p < 0.01) at 30 days. Similarly, in a large cohort involving 995 propensity-matched pairs of patients who underwent TAVI with the self-expanding Evolut Platform (Medtronic), the rate of PPI at discharge was lower in COL group than in the conventional technique group (17.0 vs. 11.9%;
p = 0.001), yet more NOP-LBBB were observed in the COL group (27.4% vs. 22.0%;
p = 0.01), questioning the true independent effect of COL on the reduction of PPI from a pathophysiological standpoint. Most previous studies compared more contemporary patients treated with the COL technique to their “historical” TCC counterparts, which introduces significant biases that cannot be adequately adjusted for in statistical analyses. Indeed, increased operators’ awareness of the detrimental effect of post-TAVI PPI and the publication of experts’ consensus and international guidelines [
21] during the inclusion period of several of these studies likely influenced both the performance of TAVI procedures and immediate post-procedural care regarding NOCD [
11‐
13]. We did not demonstrate that using the COL technique per se leads to a significant reduction of HAVB or PPI at 30 days. Nevertheless, the COL technique was associated with a lower ID, resulting in less contact between the lower part of the membranous septum and the THV. Secondly, a lower ∆MSID-NCC significantly reduced PPI at 30 days. Therefore, COL may facilitate procedural conditions that favor higher implantation relative to the membranous septum. Furthermore, it should be highlighted that we were able to adequately adjust our analysis for major baseline (FD-AVB, RBBB), anatomical (bicuspid aortic valve, degree of calcification), and procedural (valve type, pre- and post-dilatation, year of the procedure) confounding factors to delineate the true independent effect of the COL technique [
31,
32]. Moreover, although we also observed an increasing adoption of the COL technique, this was not associated with a trend toward higher implantation over time. Finally, the inclusion period of the present study was on the shorter side of those of previous studies and more importantly was subsequent to the publication of the first major expert consensus document guiding the management of post-TAVI NOCD [
20].
The measurement of membranous septum length (MSL) in our study was performed using pre-procedural CT in a standardized coronal reconstruction, consistent with the methodology originally described by Hamdan et al. and subsequently adopted in multiple studies evaluating ΔMSID as a predictor of conduction disturbances [
16]. In contrast to more recent approaches incorporating infra-annular reconstructions, percentage-based implantation metrics, or membranous septum area quantification [
33‐
35], our method intentionally focused on a reproducible linear measurement aligned with the non-coronary cusp nadir, facilitating integration with angiographic implantation depth assessment. Similarly, implantation depth was measured on final angiography at the non-coronary cusp, in line with prior reports, including Jilaihawi et al.
34, who emphasized the importance of implantation depth relative to the membranous septum in minimizing permanent pacemaker implantation after self-expanding TAVI. Although small absolute differences in implantation depth (≈ 1–2 mm) may appear modest, prior mechanistic studies suggest that such differences may be clinically meaningful given the close anatomical proximity of the conduction system to the virtual annular plane.
Sammour et al.
10 reported an ID of 3.2 ± 1.9 mm using the TCC technique vs. 1.5 ± 1.6 mm with the COL technique in a cohort of BEV. ID in our cohort was closer to Pascual and Mendiz’s works, encompassing only SEV (5.14 ± 2.6 mm vs. 4.2 ± 2.1 mm and 5.65 ± 3.48 mm vs. 3.43 ± 2.79 mm, respectively) [
11,
13]. Comparisons between studies remain difficult due to the different types of THV. Nonetheless, the small absolute differences in ID between the two techniques were comparable (approximately 1.6 mm). Nevertheless, these measurements remain subject to angiographic spatial resolution constraints (roughly 0.2 mm) and residual parallax, which should be considered when interpreting small between-group differences [
9]. Interestingly, Sammour et al.
10 reported a decreasing trend in ID in their population of TAVI recipients treated from April 2015 to December 2018. Of note, this trend was apparent before the implementation of the COL technique, and more pronounced in the TCC group (
p < 0.001 vs.
p = 0.052 in the COL group). This may reflect a paradigm shift that predates the use of COL, among operators, which aimed at a high implantation of THV to reduce NOCD and PPI.
In addition, very few studies reported the MSL, while it is an essential measurement in this context [
16,
18,
36]. Firstly, in patients with short MSL, systematic high implantation could be insufficient to avoid NOCD, all the more so if other risk factors are associated [
18,
28]. Secondly, high deployment might be useless or even detrimental to some patients with long MSL. Indeed, a recent study [
37] demonstrated that a target ID between 1 and 3 mm reduced NOCD, but exposed to sinus sequestration in case of redo-TAVI and unfavorable future coronary access. These data suggest that the strategy of implantation should tend to a compromise between the risk of NOCD and the possibility of subsequent coronary access using a patient-specific multimodal approach to limit long-term adverse events. The most important limitation to the use of this measurement in daily practice is probably the absence of a standardized method and inter-observer variability [
25].
The inclusion of multiple commercially available valve platforms reflects contemporary clinical practice and enhances the generalizability of our findings, as the primary objective of this study was to evaluate projection-based implantation strategies rather than device-specific performance. The observed association between Portico/Navitor implantation and conduction disturbances must be interpreted cautiously, as the patients included represent our initial institutional experience with this platform, corresponding to a limited number of procedures per operator. A learning-curve effect likely contributed to these findings. Importantly, larger contemporary series have reported permanent pacemaker implantation rates with the Navitor system comparable to other current-generation valves, suggesting that our findings should not be generalized beyond this early experience [
38].
Study limitations
This single-center retrospective study is subject to inherent limitations. First, operator-related selection bias in projection technique choice may play a significant role in our findings and cannot be adequately accounted for by statistical adjustment. Secondly, the measurements of the MSL and the ID were assessed retrospectively by a single expert operator thus not involving a central adjudication by a corelab. The intra-observer variability of ID measurement was not evaluated. Finally, given the modest absolute difference observed in the primary endpoint between groups, a risk of type II error cannot be ruled out. Indeed, a post-hoc power calculation demonstrated that more than 1300 patients would have been necessary in each group to demonstrate such a difference with an 80% power at the alpha level of 5%.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.