Predictors of ventricular pacing burden after permanent pacemaker implantation following transcatheter aortic valve replacement

Abstract Background In the era of an expanding use of transcatheter aortic valve replacement (TAVR), conduction disturbances and the requirement for permanent pacemaker (PPM) implantation remains a clinical concern. Hypothesis Using a single‐center experience, we sought to identify predictors of ventricular pacing burden after TAVR in patients who required PPM implantation. Methods We conducted a retrospective study of 359 consecutive patients with symptomatic severe aortic valve stenosis who underwent TAVR at our institution between September 2013 and July 2019. Thirty patients (8.4%) required a PPM within 30 days after TAVR. Pre and post‐TAVR electrocardiograms, pre‐TAVR echocardiograms and computed tomography (CT), TAVR procedural details and post‐TAVR device interrogation records at 1, 3, and 6 months were reviewed. Results Mean percentage of ventricular pacing (VP%) at 1, 3, and 6 months was 58%, 59%, and 56% respectively. Using univariate logistic regression analysis, patients who had low VP% < 5% at 6 months were more likely to have a prosthesis/echocardiography‐derived left ventricular outflow tract (LVOT) diameter ratio < 1.3 (OR 7.00, P‐value .048), prosthesis/CT‐derived aortic annulus diameter ratio < 1.02 (OR 7.11, P‐value .047), post‐TAVR new‐onset LBBB (OR 16.80, P‐value .019), time to PPM implantation greater than 2 days post‐TAVR (OR 9.38, P‐value .026) and pre‐TAVR use of a beta blocker (OR 9.40, P‐value .026). Conclusions In patients who required a PPM implantation post‐TAVR, a lower TAVR prosthesis/LVOT or aortic annulus diameter ratio, post‐TAVR new‐onset LBBB and later time of PPM implantation showed a trend toward predicting a low VP% at 6 months.


| INTRODUCTION
Transcatheter aortic valve replacement (TAVR) is now established as a noninferior alternative to surgical aortic valve (AV) replacement in patients with symptomatic severe AV stenosis, not only in high-risk 1,2 and intermediate-risk patients, 3,4 but also in patients at low-surgical risk. 5,6 Despite the growing evidence of good clinical outcomes with TAVR, conduction disturbances and the requirement for permanent pacemaker (PPM) implantation remains a clinical concern. The incidence rate of PPM implantation within 30 days after TAVR is reported at 6.6% to 8.8% using balloon-expandable valves in the PARTNER 1, 2, and 3 trials, 3,5,7 and remains at a high rate of 17.4% using self-expandable valves in their latest low-risk trial. 6 These conduction disturbances are believed to be due to mechanical stress from the prosthetic valve on the conduction system in the atrioventricular node and interventricular conduction system, 8 and include variable degrees of atrioventricular block, bradycardia and left bundle branch block (LBBB). Recent studies have shown a trend in recovery of some of these conduction disturbances, resulting in a decrease in the percentage of patients who are PPM dependent with time. 9 Persistence of PPM dependency has been reported to be somewhere between 22% and 64% in recent single-center studies. [10][11][12][13][14][15] A better understanding of the natural course of these rhythm disturbances and identifying predictors of their recovery can be particularly useful in helping the clinical decision-making process when a PPM is being considered for certain disturbances where a strong consensus is not available. Recent institutional experiences have identified factors as a pre-TAVR right bundle branch block (RBBB), 10,11,14 higher prosthesis/LVOT diameter ratio, 14 and earlier time to PPM implantation within 1 to 2 days post-TAVR 10,13,14 as predictors of pacemaker dependency. In this study, we aimed to use our institutional experience to identify predictors of post-TAVR ventricular pacing burden with time, and investigate predictors of a low-ventricular pacing percent over a 6 month follow-up period that may suggest pacemaker independency.

| Study population
We conducted a single center retrospective cohort study of 359 con-

| Data collection
All data was obtained from electronic medical records. Pre-TAVR echocardiograms were reviewed for left ventricular (LV) function, LV outflow tract (LVOT) diameter and AV area. Computed tomography (CT) TAVR protocol images were reviewed for aortic annulus perimeter-derived diameter. TAVR procedural details were reviewed for valve prosthesis type and size, use of preimplant valvuloplasty, use of postimplant dilatation and AV hemodynamics. Prosthesis/echocardiography-derived LVOT diameter ratio and prosthesis/CT-derived aortic annulus diameter ratio were calculated. A supplementary figure (Supplementary Figure 1) demonstrates how these diameters were obtained. Pre and post-TAVR electrocardiograms (ECGs) were analyzed for heart rate, rhythm, PR interval, QRS duration, and the presence of any conduction disturbances. Indications for which PPMs where implanted were collected from clinical documentation and ECG interpretation by a cardiology electrophysiologist and categorized as: persistent complete or high-degree atrioventricular block (AVB), transient complete or high-degree AVB, pause(s) more than 3 seconds and symptomatic bradycardia (including sick sinus syndrome, atrial fibrillation with slow ventricular rate and prolonging PR interval).

| Follow-up
Patients were retrospectively followed up for a duration of 6 months.
Device interrogation records from outpatient visits were reviewed at 1, 3, and 6 months after PPM implantation. All the PPMs were programmed in the DDD mode at 60/min or the VVI mode in cases of persistent atrial fibrillation. The percentage of ventricular pacing (VP %) at each visit was documented, and ventricular pacing burden was categorized as low (VP% less than 5%), intermediate (VP% between 5% and 95%) and high (VP% more than 95%). These cutoffs were previously used and interpreted as VP% less than 5% suggesting pacemaker independency and VP% more than 95% suggesting absolute pacemaker dependency. 10

| RESULTS
A total of 30 out of 359 patients who underwent TAVR (8.4%) required a PPM within 30 days, with a median time to PPM implantation of 2 days post-TAVR (interquartile range 1 to 6 days). All but one patient were males (97%), and the mean age was 76 ± 8 years. The majority of valves were Edwards Sapien (original, XT or 3; 77%). Indications for PPM implantation were persistent high-degree AVB (n = 13, 43%), transient high-degree AVB (n = 6, 20%), pause more than 3 s (n = 4, 13%) and symptomatic bradycardia (n = 7, 23%). Mean VP% at 1, 3, and 6 months was 58%, 59%, and 56%, respectively, with no significant change over the 6-month follow-up period. None of the patients was lost to follow-up, but four died within the follow-up period. The final cohort for comparison of patients according to VP% at 6 months included 26 patients, categorized into low VP% <5% (n = 7), intermediate VP% 5% to 95% (n = 10) and high VP% >95% (n = 9). Table 1 summarizes the preoperative baseline characteristics, medication use, ECG details and echocardiographic findings of patients in the three groups. All variables were tested for differences between the groups, and prosthesis/LVOT diameter ratio was the only variable meeting statistical significance (P-value .021, with mean values of 1.21, 1.37, and 1.39 in the low, intermediate and high-VP% groups, respectively). Prosthesis/aortic annulus diameter ratio showed a similar trend but the difference between groups did not reach statistical significance (mean values of 1.03, 1.08, and 1.12, respectively). Table 2  Using univariate logistic regression analysis (see Table 3), patients who had low VP% <5% over a 6-months follow-up period were more likely to have a prosthesis/LVOT diameter ratio < 1.3 (OR 7.00, Pvalue .048), prosthesis/aortic annulus diameter ratio < 1.02 (OR 7.11, P-value .047), post-TAVR new-onset LBBB (OR 16.80, P-value .019), time to PPM implantation greater than 2 days post-TAVR (OR 9.38, Pvalue .026), and pre-TAVR use of a beta blocker (OR 9.40, P-value .026). The cutoff values used for continuous variables were based on ROC curves. An ROC curve was plotted for discriminative ability of prosthesis/LVOT diameter ratio to predict a VP% > 5%, giving an area under curve (AUC) of 0.86 and identifying a highest C-statistic of 1.3 ( Figure 2). Similarly, a value of 1.02 was identified for prosthesis/aortic annulus diameter ratio (AUC 0.74). A multivariable regression analysis model of all significant variables on univariate analysis did not reveal any independent predictors of a low-VP% <5% at 6 months, likely limited by the small sample size.
On the other hand, the only predictor of a high-VP >95% at 6 months was a PPM indication of persistent complete or high-degree AVB (OR 8.40, P-value .027). In addition, the following variables were not found in any of the patients in the low VP% group at 6 months: preoperative RBBB with bifasicular or trifasicular block, use of selfexpanding prosthetic valve, use of larger 34 mm valve and preimplantation valvuloplasty (see Tables 1 and 2). While these variables are associated with a low odds of having a low-VP% at 6 months, they were unable to be analyzed by logistic regression.

| DISCUSSION
In this study, we used a single institutional experience to investigate predictors of ventricular pacing burden after TAVR in patients who required PPM implantation. Our findings showed: (i) the incidence of TAVR-related PPM implantation in our institution was 8.4%, (ii) mean VP% over a 6-month period was 56-59%, (iii) a lower median VP% occurred in patients with post-TAVR new-onset LBBB at 1, 3 and 6 months, (iv) predictors of low VP% <5% suggestive of PPM independency over a 6-month follow-up period included prosthesis/LVOT diameter ratio < 1.3, prosthesis/aortic annulus diameter ratio < 1.02, post-TAVR new-onset LBBB, time to PPM implantation greater than 2 days post-TAVR and pre-TAVR use of a beta blocker, (v) on multivariable analysis, none of these variables independently predicted a low-VP%, which may be due to small sample size and model overfitting rather than lack of significance, (vi) on the other hand, a PPM indication of persistent complete or high-degree AVB predicted a high-VP% > 95% suggestive of complete PPM dependency over a 6-month follow-up period.
The incidence rate of PPM implantation within 30 days after TAVR in our institution (8.4%) is comparable to that reported by the PARTNER 1, 2, and 3 trials that used balloon-expandable valves (6.6%-8.8%). 3,5,7 It is lower however than a lot of recently reported single institutional experiences that used both self-expanding and balloon expandable valves (incidence rates between 11% and 28%). 10,11,13,14,16,17 These interinstitutional differences may be partly explained by our higher use of balloon-expandable valves (77%) which are generally associated with a lower risk of PPM implantation due to their higher level of implantation, 18 as well as our use of newer generation self-expanding Evolut valves rather than first generation Cor-eValves which have a lower risk of PPM implantation. 19 However despite these factors, there is still a high variability between institutional experiences with rates of PPM implantation after TAVR 19 which in the atrioventricular node and interventricular conduction system. 8,16 They can be permanent or temporary as a result of postprocedural local inflammation, edema or ischemia that recover with time. 9 In our institution, 27% of PPM patients had a low VP% <5% at 6 months and there was no significant difference in the mean VP% at The significance of pre-TAVR RBBB is especially important, as it is the most persistently reported predictor of PPM requirement in the literature, 7,9,17 as well as often being reported as a predictor of PPM dependency. 10 F I G U R E 2 Receiver Operating Curve (ROC) for discriminative ability of TAVR prosthesis/LVOT diameter ratio to predict a percent of ventricular pacing >5% at 6 months after TAVR. LVOT, left ventricular outflow tract; TAVR, transcatheter aortic valve replacement