His bundle pacing, learning curve, procedure characteristics, safety, and feasibility: Insights from a large international observational study

Abstract Background His‐bundle pacing (HBP) provides physiological ventricular activation. Observational studies have demonstrated the techniques’ feasibility; however, data have come from a limited number of centers. Objectives We set out to explore the contemporary global practice in HBP focusing on the learning curve, procedural characteristics, and outcomes. Methods This is a retrospective, multicenter observational study of patients undergoing attempted HBP at seven centers. Pacing indication, fluoroscopy time, HBP thresholds, and lead reintervention and deactivation rates were recorded. Where centers had systematically recorded implant success rates from the outset, these were collated. Results A total of 529 patients underwent attempted HBP during the study period (2014‐19) with a mean follow‐up of 217 ± 303 days. Most implants were for bradycardia indications. In the three centers with the systematic collation of all attempts, the overall implant success rate was 81%, which improved to 87% after completion of 40 cases. All seven centers reported data on successful implants. The mean fluoroscopy time was 11.7 ± 12.0 minutes, the His‐bundle capture threshold at implant was 1.4 ± 0.9 V at 0.8 ± 0.3 ms, and it was 1.3 ± 1.2 V at 0.9 ± 0.2 ms at last device check. HBP lead reintervention or deactivation (for lead displacement or rise in threshold) occurred in 7.5% of successful implants. There was evidence of a learning curve: fluoroscopy time and HBP capture threshold reduced with greater experience, plateauing after approximately 30‐50 cases. Conclusion We found that it is feasible to establish a successful HBP program, using the currently available implantation tools. For physicians who are experienced at pacemaker implantation, the steepest part of the learning curve appears to be over the first 30‐50 cases.


| INTRODUCTION
Since the first report of successful permanent His bundle pacing (HBP) in 2000 by Deshmukh et al, 1 HBP has emerged as an alternative to right ventricular endocardial pacing for patients with a bradycardia indication for pacing. HBP produces rapid and physiological ventricular activation via the His-Purkinje conduction system, rather than the nonphysiological activation, which occurs with right ventricular endocardial pacing due to slow cellto-cell conduction of the electrical wavefront. HBP produces a narrow QRS duration and, by preserving physiological ventricular activation, may prevent the development of right ventricular pacing-induced cardiomyopathy in patients with a bradycardia indication for pacing. 2 Furthermore, HBP can often correct bundle branch block, resulting in a reduction in QRS duration and improvement in ventricular activation time. 3 The HBP pioneers encountered several challenges during the early years of HBP, such as low procedural success rates and high and rising thresholds. 1 However, with the arrival of specialized delivery equipment and a more suitable pacing lead, the published HBP success rates appear to have improved. [4][5][6] This published experience includes relatively small numbers of patients from a limited number of expert centers. As a result, there have been concerns about whether, with current technology, it is feasible for HBP to be adopted more widely.
In this study, we report the experience of seven international centers who have adopted HBP, with a particular focus on the learning curve for performing HBP, acute implant success rates as well as acute and longer-term HBP thresholds.

| Study design
This is a retrospective, multi-center observational study designed to evaluate the contemporary real-world practice of permanent HBP.
We focus on indications for pacing and patient characteristics, procedural characteristics, and device follow-up. The study population included all patients undergoing attempted permanent HBP, for any indication, at seven implanting centers (1 Germany, 1 Poland, 1 Switzerland, 3 United Kingdom, and 1 United States). Baseline patient demographics together with relevant clinical information (QRS duration, NYHA class, LV function, presence of IHD) were recorded. Patients identified as being recruited into blinded randomized controlled trials, including the HOPE-HF double-blind randomized crossover trial, 7 were excluded from this analysis.

| Procedural details
Procedural characteristics (indication, presence of pre-existing device, procedural success, lead/delivery equipment, lead-generator configuration, fluoroscopy time, HBP thresholds, and programmed stimulation energy) were recorded.

Cardiac resynchronization therapy.
Where patients were noted to have multiple indications, only one was recorded-this was the one with the strongest pacing indication.
The pacing responses observed were recorded as selective His bundle capture, nonselective His bundle capture, and myocardialonly capture as defined using previously described standard criteria. 4 The criteria used by the different operators whether to accept a His lead position included (a) the His capture threshold, (b) the paced KEENE ET AL.

| 1985
QRS duration (particularly of relevance if intrinsic QRS is broad), and (c) R wave amplitude, (if the His lead is required for sensing). One center also imposed a time limit of 30 minutes.
Pulse width for HBP threshold testing was determined at the individual implanting centers.
The HBP lead delivery method was divided into three categories: the typically used approaches of the fixed curve C315 His sheath (Medtronic, MN) or the deflectable C304 delivery sheaths (Medtronic), and a third category allowing for hybrid approaches using different delivery sheaths. The type of lead implanted to deliver HBP was also recorded. Fluoroscopy time was stratified by both the number of leads implanted and the reported rhythm during implant.

| Clinical outcomes
Clinical outcomes (QRS duration, HBP thresholds, and programmed stimulation energy) were collected. These outcomes were reported at implant and at the latest recorded follow-up. HBP QRS duration was stratified by selective or nonselective capture. Nonselective capture of the His bundle results in the capture of both the local myocardium and the His bundle. The capture of the local myocardium with nonselective pacing produces a pseudo-delta-wave on the ECG, which prolongs the surface QRS duration. Programmed stimulation energy was analyzed as voltage and as a function of voltage and pulse width (V 2 t, V = voltage, t = pulse width).

| Statistical analysis
Descriptive statistics, including mean and standard deviation, were used, and for within-patient changes in parameters, paired t tests were performed. Differences between groups were assessed using analysis of variance. To assess the impact of experience on pacing parameters, regression models were constructed (ordinary leased squares or proportional odds as appropriate) with a restricted cubic spline with three knots for experience (measured as a number of cases performed by that center). Data analysis was performed using R version 3.3.

| Clinical indications
The majority of HBP attempts were performed in patients with intermittent or persistent high-degree atrioventricular block and 86.2% of procedures were de-novo device implantations; the remainder were upgrades of previously implanted devices.

| Procedural success rates
Overall, successful permanent HBP was achieved in 81% of the 322 patients implanted in the centers, where implant success rates were systematically recorded.
One center with 47 procedures over 3.5 years had a 91% success rate, another with 108 procedures over 1.7 years had an 81% success rate, and the third with 167 procedures over 4.3 years had a 78% success rate.
Patients who had a QRS duration less than 120 ms were more likely to be successfully implanted than patients with a broader QRS Pre-AV node ablation 6.6% Note: Values are mean ± SD or n (%) for all patients in whom permanent HBP was attempted. Where data are missing, n is provided for the patients in whom data are available. Abbreviations: HBP, His-bundle pacing; IVCD, interventricular conduction delay; NYHA, New York Heart Association; SD, standard deviation.
In 39 cases, the specific reason for HBP failure was not recorded. In the 22 cases where this information was available, the reasons for discontinuing the His pacing attempt were as follows: infra-Hisian block distal to the pacing site at high rates (n = 2), unmappable His signal with unsatisfactory pace-mapping (n = 8), lead instability/displacement (n = 3), failure to correct bundle branch block (n = 7), and His pacing threshold prohibitively high (n = 2). With experience, successful implantation rates were higher at 87% after 40 implant attempts.

| Devices, leads, and delivery equipment
In all patients, the SelectSecure 3830 69 cm lead was used to attempt HBP. In 89.0% of successful HBP cases, the C315 fixed curve sheath successfully delivered the lead. In 10.1% of implants, the deflectable C304 sheath was used and in 0.9%, a C315 sheath was passed through a modified coronary sinus catheter or a coronary sinus guide catheter alone was used to position the His lead.
A single-lead pacemaker with only a His bundle lead was implanted in 65 patients (14% of successful HBP implants). A dual-lead pacemaker was implanted in 182 patients (39%). In 36 of these patients, the His bundle lead was connected to the atrial port with a ventricular "backup" lead connected to the ventricular port. In 146 patients, the His bundle lead was connected to the ventricular port with a right atrial lead connected to the atrial port; in two such patients, a right ventricular sensing lead was also implanted and connected to the ventricular port along with a His bundle lead via a Y connector.
A total of 219 (47%) patients received triple lead devices. In 152 patients, a CRT-P device was used, and 67 patients received a CRT-D device. In almost all CRT devices, an atrial lead was connected to the atrial port, and the His lead and RV ("back-up") lead were connected to the ventricular ports; however, alternative configurations were used in a small minority of cases.
From the 399 patients who did not require a mandatory RV lead for defibrillation purposes, 47% were implanted with a "back-up" ventricular pacing lead in addition to the His lead. Back up lead utilization decreased with experience (P = .0018): this is shown in Figure 1.

| Fluoroscopy time
The mean fluoroscopy time across all procedures was 11.7 ± 12.0 minutes. Fluoroscopy times, stratified by the number of leads and procedural rhythm, are depicted in Table 2. Overall, fluoroscopy times were not significantly affected by the underlying rhythm during the implant procedure.
Thirty implants involved a coronary sinus (CS) lead and were excluded from fluoroscopy time analysis (Table 2) as variation in fluoroscopy use for CS lead implantation is not relevant to this analysis and could be responsible for differences in th overall Learning curve metrics A, shows the relationship between fluoroscopy time and center experience (P = .15). B, shows the relationship between pacing threshold and center experience (P = .04). C, shows the log odds of receiving a back-up pacing lead in the ventricle (P = .0018).
Overall successful implantation rate was 81% which increased to 87% after 40 cases. As number of cases per center increased, fluoroscopy time and His capture threshold decreased until plateauing after 30 and 50 cases fluoroscopy time. Fluoroscopy times tended to reduce with increasing His bundle pacing experience (P = .15) as shown in Figure 1 with a leveling off after 30-50 cases. In the remaining 249 patients, selective His bundle capture was not observed, the mean threshold of His bundle capture in this group was 1.4 ± 1.0 V at 0.9 ± 0.3 ms. In 64 of these patients, the myocardial capture threshold was reported and found to be lower than the nonselective threshold with a mean myocardial threshold of 0.9 ± 0.6 V at 0.9 ± 0.2 ms.

|
In patients with bundle branch block where either the left or right bundle was corrected by His bundle pacing (n = 27), the mean threshold for bundle recruitment was 2.1 ± 1.0 V at 0.9 ± 0.2 ms and the mean threshold of His bundle capture (without necessarily recruiting both bundles) was 1.6 ± 0.8 V at 0.9 ± 0.2 ms. Figure 2 depicts the withinpatient change in thresholds for the different pacing responses seen.
The relationship between the threshold of His bundle capture and center experience is shown in Figure 1. As experience increased, the threshold of His bundle capture reduced, before plateauing (P = .04), again demonstrating a learning curve.
The mean programmed His lead pacing output across all cases at implant was 4.2 ± 1 at 0.9 ± 0.2 ms (data available in 322 cases) with a mean safety margin of 2.9 ± 1.2 V above the threshold of His bundle capture.  Table 3.
Of the 366 patients with follow up threshold data available, 31 In 17 patients, no follow up capture threshold data was reported because they were not receiving His pacing. In three patients, this was because there was only myocardial capture present (n = 3) (mean threshold 0.5 V at 1ms). In 14 patients, His lead had been deactivated. Reasons for deactivation were a rise in threshold (n = 6) and displacement and loss of capture (n = 7). One patient experienced exercise-induced symptoms and 2:1 infra-Hisian block was observed during an exercise test. Therefore, His lead was deactivated and RV only pacing was performed.

| His bundle lead complications
In addition to the 17 patients without follow-up threshold data reported due to His lead issues, eight additional patients had undergone lead revisions either due to displacement (n = 5) or unsatisfactory threshold rise (n = 3). In further 10 patients, the His pacing lead had been deactivated due to displacement and loss of capture (n = 3) or due to unsatisfactory threshold rise (n = 7). Therefore, overall 7.5% of initially successful His lead implants experienced issues requiring either a lead revision or lead deactivation.

| ECG responses
Mean preimplant QRS duration across all attempted patients was 118 ± 33 and 116 ± 31 ms in those successfully implanted. Across the group as a whole, there was no significant difference in paced QRS duration, which was 115 ± 24 ms (P = .5).
In patients with an intrinsic QRS duration <120 ms, the mean within-patient change from intrinsic QRS to selective His bundle capture QRS was +8 ± 18ms (n = 125) (P < .0001) and the mean within-patient change from intrinsic QRS to nonselective His bundle capture QRS was +17 ± 23 ms (n = 130) reflecting the presence of a pseudo-delta-wave on the ECG (P < .0001).
In patients with intrinsic QRSd > 120 ms, the mean within-patient change in QRS duration from intrinsic QRS to selective His bundle capture QRS was −21 ± 29 ms (n = 60) and the mean within-patient change in QRS duration from intrinsic QRS to nonselective His bundle capture QRS was −27 ± 32 ms (n = 102) (P < .0001) (this narrowing was despite the presence of a pseudo-delta wave). Figure 3 shows the summary of ECG changes, and full details are provided in Table 4.

| Implant success rates
The overall success rate of attempted HBP was 81%. After the first 40 implants at a center, the success rate was higher at 87%.
Other present-day reports of HBP success rates are consistent with this, ranging from 80% to 92%. 4,8,9 Some hospitals did not have comprehensive success rate data and it is conceivable that their rates may have been lower. Nevertheless, these rates compare favorably to the pioneering reports of HBP, which did not benefit from specialized delivery systems, having success rates as low as 66%. 1 The current 87% success rates that can be expected by patients undergoing implantation at experienced centers are the result of both improvements in lead delivery approach and also the growth of experience by operators at the implanting centers. Further increases in collective operator experience and advancements in delivery tools may bring HBP success rates up to the level now seen in biventricular pacing. In fact, since biventricular pacing was the previous such pacing revolution, it is notable that after a decade of experience major RCTs were reporting success rates (83% in PAVE 10 and 82% in RD-CHF 11 ) similar to the HBP success rates in this study. Importantly, HBP leads and delivery equipment can also be used for conventional atrial or ventricular pacing and even distal conduction system pacing where the left bundle itself is directly stimulated, offering flexibility when HBP capture cannot be achieved. 12

| His pacing safety
We found that 7.5% of His pacing leads required either a repeat intervention or were deactivated. The most common reasons for this were lead displacement and rise in pacing threshold. This is similar to the 6.7% rate reported in a prior long term follow-up study of His bundle pacing 5 and is comparable to the approximately 7% LV lead revision rate seen in 15 222 patients included in the studies of biventricular pacing. 13 Indeed, given the decades more of technical developments for RV leads, it is remarkable that the rate of RV lead repeat interventions is still even now as high as 3%. 14 While these rates for His pacing are not yet down to the level of RV pacing, they are close enough for a trial to be warranted to determine whether F I G U R E 3 Summary of QRS duration changes for all patients receiving His bundle pacing and then stratified according to intrinsic QRS duration either less than or more than 120 ms. (Error bars are standard error of the mean)  However, there are also potential disadvantages of using a backup lead. First, implanting more leads is associated with an increased risk of infection. 16  thresholds. This process continued until a plateau formed after around 40 cases (Figure 1). This provides reassurance that His bundle pacing is a feasible alternative to RV pacing.

| Choice of lead and delivery kit
Even though the His bundle is small, it has regions along with it with different pacing parameters. Potential explanations for our finding of improvements in His pacing thresholds with increasing experience may result as with increasing experience, operators become increasingly confident that they will be able to reposition the lead back on the His bundle should they elect to reposition it if the threshold in the first position is higher than desired. With greater experience, operators may also be able to more frequently deliver the lead perpendicular to the septum with better resultant pacing parameters.

| ECG capture responses
Patients with QRS > 120 ms showed clear shortening in QRS duration with HBP, with a mean reduction of 26 ms in this study. Patients with QRS < 120 ms showed small but statistically significant increases in QRS duration. This effect was particularly small in those with selective capture.
Even though one may, at first, expect nonselective capture to broaden QRS duration, if the intrinsic QRS is broad, it can be narrowed. Across our patients with nonselective capture, there was essentially no change in the mean QRS with pacing. This arose from a significant QRS widening in those with an intrinsic QRS < 120 ms and a significant QRS narrowing in those with an intrinsic QRS > 120 ms.

| Limitations
This is an observational, registry data set reliant on physicians reporting of data not acquired prospectively. Data on R wave sensing were not available from all centers and therefore we were unable to report on sensing in this study. Although most centers were not recruiting into randomized control trials, at least three were: this may have an influence on the data derived for learning curves.
Only three centers systematically collected implant success rate data. It is possible that implant success rates may vary depending on implant volume. Our study did not address this question which will need to be addressed in a prospective study.
There was no ECG core lab for confirming His capture; therefore the presence and type of His bundle capture were verified by the implanting physician according to the accepted criteria published in the His pacing consensus document. 17 The method for measuring QRS duration was determined by the local protocol within each institution. Although this was standardized within an institution, it was not standardized across the whole study.
As this was a retrospective study, there was not a strict protocol regarding the His pacing threshold and R wave amplitude accepted.
Individual operators made this decision on clinical grounds.

| CONCLUSION
This large, contemporary, and global experience of HBP shows that current practice in HBP has acceptable procedural success rates, acceptably low fluoroscopy time, low and overall reliable His bundle capture thresholds. The current deactivation or reintervention rate is not prohibitive and is similar to that of LV leads.
The learning curve for achieving His bundle pacing appears to plateau after around 40 cases with a progressive reduction in fluoroscopy time and His bundle capture threshold.