Objective Many patients undergoing transcatheter aortic valve implantation (TAVI) have a pre-existing, permanent pacemaker (PPM) or receive one as a consequence of the procedure. We hypothesised that chronic pacing may have adverse effects on TAVI outcomes.
Methods and results Four groups of patients undergoing TAVI in the Placement of Aortic Transcatheter Valves (PARTNER) trial and registries were compared: prior PPM (n=586), new PPM (n=173), no PPM (n=1612), and left bundle branch block (LBBB)/no PPM (n=160). At 1 year, prior PPM, new PPM and LBBB/no PPM had higher all-cause mortality than no PPM (27.4%, 26.3%, 27.7% and 20.0%, p<0.05), and prior PPM or new PPM had higher rehospitalisation or mortality/rehospitalisation (p<0.04). By Cox regression analysis, new PPM (HR 1.38, 1.00 to 1.89, p=0.05) and prior PPM (HR 1.31, 1.08 to 1.60, p=0.006) were independently associated with 1-year mortality. Surviving prior PPM, new PPM and LBBB/no PPM patients had lower LVEF at 1 year relative to no PPM (50.5%, 55.4%, 48.9% and 57.6%, p<0.01). Prior PPM had worsened recovery of LVEF after TAVI (Δ=10.0 prior vs 19.7% no PPM for baseline LVEF <35%, p<0.0001; Δ=4.1 prior vs 7.4% no PPM for baseline LVEF 35–50%, p=0.006). Paced ECGs displayed a high prevalence of RV pacing (>88%).
Conclusions In the PARTNER trial, prior PPM, along with new PPM and chronic LBBB patients, had worsened clinical and echocardiographic outcomes relative to no PPM patients, and the presence of a PPM was independently associated with 1-year mortality. Ventricular dyssynchrony due to chronic RV pacing may be mechanistically responsible for these findings.
Trial registration number (ClinicalTrials.gov NCT00530894).
Statistics from Altmetric.com
Transcatheter aortic valve implantation (TAVI) has emerged as a therapeutic option for patients with severe aortic stenosis who are inoperable or high-risk surgical candidates.1 ,2 The TAVI population typically includes elderly patients with significant comorbid cardiac disease, including many with pre-existing permanent pacemakers (PPM). In addition, the TAVI procedure itself can cause injury to the cardiac conduction system, resulting in conduction abnormalities that require PPM. The reported incidence of newly implanted PPM after TAVI ranges from approximately 6% with the balloon-expandable Edwards SAPIEN valve (Edwards Lifesciences, Irvine, California) to 25% with the self-expanding Medtronic CoreValve (Medtronic, Minneapolis, Minnesota).3–7 The rate of new PPM implant after surgical aortic valve replacement has been recently reported to be 5.8%.2
The most frequently employed configuration of permanent pacing involves a single RV lead, typically implanted in the apex. The haemodynamic and mechanical disadvantages of RV apical pacing are similar to those of spontaneous left bundle branch block (LBBB).8 The deleterious effects of RV pacing on clinical outcomes have been demonstrated in patients with depressed LVEF.9 Existing data also suggest adverse effects of RV pacing in patients with preserved or even normal LVEF.10 ,11 Furthermore, patients with new LBBB or PPM following TAVI have been shown to have impaired LVEF recovery.12–14
Given the abnormal ventricular substrate resulting from aortic stenosis and the demonstrated adverse effects of RV pacing in diverse patient groups, we hypothesised that chronic RV pacing may have negative clinical and functional effects on TAVI patients. We therefore compared clinical outcomes and LV function at 1 year among patients with a prior PPM, newly implanted PPM and no PPM with or without LBBB in the Placement of Aortic Transcatheter Valves (PARTNER) trial and associated registries.
Study population and design
The PARTNER trial design and primary results have been previously reported.1 ,2 Inoperable and high-risk surgical candidates underwent TAVI with a 23 or 26 mm, balloon-expandable, Edwards SAPIEN transcatheter heart valve (Edwards Lifesciences, Irvine, California, USA). The inclusion and exclusion criteria and follow-up evaluations were the same for the trial and registries, and all patients were evaluated by a national screening committee. A blinded events committee was provided data summaries from the study in order to adjudicate important clinical outcomes. ECG and transthoracic echocardiograms were obtained at baseline, hospital discharge or 7 days, 30 days, 6 months and 1 year and were interpreted by independent core laboratories.15
In the current analysis, clinical and echocardiographic outcomes were compared between patients with previously implanted PPM (prior PPM), PPM implanted within 30 days after TAVI (new PPM), patients without PPM but with chronic LBBB (LBBB/no PPM), and no PPM or LBBB (no PPM). The primary outcome measure was 1-year all-cause mortality, and secondary outcomes included rehospitalisation, the composite of mortality or rehospitalisation, cardiovascular mortality and LVEF at 1 year. Subgroup analyses were performed based on baseline LVEF.
ECGs obtained at pre-specified follow-up time points in the PPM groups were analysed for pacing prevalence and morphology. Pacing morphology was defined as RV if the ECG pattern resembled typical LBBB, with a QRS complex wider than 140 ms, a predominantly negative complex in leads V1 and V2, and a positive complex in lead I. All other morphologies were categorised as ‘other’. Pacing prevalence was based on the presence of any pacing on a standard 12-lead ECG. LV systolic function was assessed by LVEF, as calculated by Simpson's biplane method.
Continuous variables are presented as means±SD, and categorical variables are presented as counts. Potential differences between study groups were assessed using the Wilcoxon rank-sum test, the χ2 test or Fisher's exact test as appropriate. Outcomes at 1 year were analysed with Kaplan–Meier (KM) estimates and compared between groups with the log-rank test. Propensity matching was performed for patients with a prior PPM compared with no PPM using a model derived from pertinent demographic, clinical and echocardiographic variables, and employed a 1:1 matching. Cox regression was performed to identify factors associated with mortality over 1 year of follow-up, with presence of a previous or new pacemaker retained in the model as a time-dependent variable regardless of its significance. Other candidate variables for the Cox model were baseline values selected on the basis of clinical importance or were required to have p<0.10 in univariate analysis. Clinical variables included age, sex, the Society of Thoracic Surgeons (STS) score, history of coronary artery disease (CAD) or coronary artery bypass grafting (CABG), diabetes mellitus, renal disease, liver disease, chronic obstructive pulmonary disease (COPD), smoking and anaemia. ECG and echocardiographic variables included LV mass index, LV end-diastolic dimension, LVEF or atrial fibrillation. A sensitivity analysis was performed with LBBB included. The presence of a prior or new pacemaker was also tested for interactions with all significant covariates in the final Cox model. A two-sided α value of <0.05 was required for statistical significance. All statistical analyses were performed using SAS software, V.9.2 (SAS Institute, Cary, North Carolina, USA).
Patient population and baseline characteristics
A total of 2531 patients underwent TAVI in the PARTNER randomised trial (n=519) and continued access registries (n=2012). Overall, PPM was present in 759 patients (29.7%), of which 586 patients had a PPM prior to TAVI (22.9%, figure 1). New PPM implantation was performed within 30 days after TAVI in 173 patients (6.8% of total and 8.8% of 1945 patients without pre-existing PPM) The majority of new PPM were implanted during the index hospitalisation (97.1%) and within 1 week of TAVI (86.1%). The most common indications for PPM were high-degree atrioventricular block (79%) or sick sinus syndrome (17.3%). Newly implanted devices were predominantly dual-chamber (75.7%) or single-chamber (19.7%) RV pacemakers. Only 29 additional patients (1.1%) received PPM beyond 30 days post-TAVI. For the patients with prior PPM, data on indication for implant and device type were not available. Of the 1772 patients without PPM, 160 had chronic LBBB.
Baseline patient clinical characteristics are shown in table 1. Overall, the population was elderly (mean age 84.5±7.2 years, 52.3% male) with significant medical comorbidities and high surgical risk (STS score 11.5±4.1 and logistic EuroSCORE 26.4±16.3). Compared with the other groups, prior PPM patients were more frequently male and had more CAD, prior CABG and prior percutaneous coronary interventions (p<0.0004), although not more myocardial infarction (MI) or congestive heart failure (CHF). The STS score was marginally higher in the prior PPM group (p=0.002). The no PPM group had more COPD compared with the prior PPM and LBBB/no PPM groups (p=0.002).
Clinical follow-up was available at 1 year (365±30 days) in 2528 patients (99.8%). Compared with no PPM, prior PPM, new PPM and LBBB/no PPM were all associated with significantly higher rates of mortality and combined mortality or rehospitalisation at 1 year of follow-up (table 2 and figure 2). As a sensitivity analysis, a landmark analysis beginning at day 30 was also performed, which showed similar trends (see online supplementary figure S1). Cardiovascular mortality was higher in the prior PPM and LBBB/no PPM groups, and rehospitalisation was higher for both the prior and new PPM groups, by both KM estimate and non-parametric competing risk analysis (figure 2 and online supplementary figure S2). When propensity matching was performed between prior PPM and no PPM patients (variables listed under table 2), prior PPM remained associated with all-cause mortality (p=0.01) and cardiovascular mortality (p=0.04). Stroke and atrial fibrillation (on EKG) were higher in no PPM relative to the prior PPM patients at 1 year. There were no significant differences in the New York Heart Association (NYHA) class III or IV CHF or incidence of MI at 1 year between the PPM and no PPM groups. When comparing the prior PPM and LBBB/no PPM groups, there were no significant differences in mortality or rehospitalisation.
Proportional hazards regression analysis for mortality
Cox regression analysis was performed to identify predictors of mortality, excluding patients with LBBB/no PPM (table 3). In this analysis, the presence of new or prior PPM were both significant independent predictors of mortality over 1 year of follow-up. Other significant predictors of mortality included male sex, renal disease, STS score, anaemia, liver disease, COPD and atrial fibrillation. When the LBBB/no PPM patients were added to the no PPM group in the analysis, new PPM and LBBB were not significant predictors of mortality, but prior PPM remained significant. The only covariate that had significant interaction with the presence of a pacemaker was COPD. When the interaction term was added to the Cox model, the order or the significance of the results were not affected.
Core laboratory analyses of echocardiograms were available at baseline in 98% of patients and at 1 year in 78% of surviving patients (77% of PPM patients and 79% of no PPM patients). LVEF at baseline and among surviving patients at 1 year is compared between groups in table 4. These results are also stratified by baseline LVEF (>50%, 35–50% and<35%). Overall, the prior PPM and LBBB/no PPM groups had significantly lower LVEF at baseline compared with no PPM. Compared with no PPM, the other three groups all had significantly lower LVEF at 1 year. When stratifying the groups by baseline LVEF, the LVEF for all four groups were similar prior to TAVI for each stratum of LVEF. However, at 1 year, prior PPM patients had significantly lower LVEF and decreased recovery of LVEF in each stratum. The recovery of LVEF was poorest among patients with baseline LV systolic dysfunction. New PPM and LBBB/no PPM patients with LVEF >50% had lower LVEF at 1 year than no PPM patients, and LBBB/no PPM patients with LVEF <35% had significantly poorer recovery of LVEF. When comparing the prior PPM and LBBB/no PPM groups to each other, there were no significant differences in the LVEF-stratified data.
ECG analysis was available for 99% of patients at baseline and among surviving patients, was available for 98% at discharge/7 days, 92% at 30 days, 86% at 6 months and 83% at 1 year (83% for both PPM and no PPM groups at 1 year). ECG data are summarised in table 5. Most no PPM patients had no intraventricular conduction disturbance. Pacing of any type was noted on 61% of ECGs among prior PPM patients and 51% among new PPM patients. An RV paced complex was present in 88% of paced ECGs in prior PPM patients and 95% of paced ECGs in new PPM patients.
The primary findings of this analysis of the PARTNER trial and registries are as follows: (1) patients with a new or prior PPM had significantly increased mortality, rehospitalisation and mortality or rehospitalisation at 1 year after TAVI relative to patients with no PPM. Furthermore, the presence of new or prior PPM were both independent predictors of mortality at 1 year. (2) Surviving new or prior PPM patients had lower LVEF at 1 year compared with no PPM patients, and prior PPM patients had poorer recovery of LVEF that was most pronounced in those with baseline depressed LV function. (3) Outcomes in the prior PPM patients closely mirrored those of the LBBB/no PPM group. (4) PPM patients had a high prevalence of RV pacing, while the majority of the no PPM group had no reported intraventricular conduction defect.
Given the similarity in the conduction abnormalities and outcomes between chronic RV pacing and LBBB, we hypothesise an adverse effect of RV pacing post-TAVI. The detrimental effects of RV pacing have been previously described in patients with LVEF <40% and implantable defibrillators.9 In the BLOCK HF trial, patients with LVEF up to 50% who had RV pacing for AV block were found to have higher mortality, a higher rate of heart failure admissions and worse LV echocardiographic indices compared with patients receiving biventricular pacing.10 In the Mode Selection Trial (MOST), patients with normal LVEF received PPM primarily for sinus node dysfunction, and the risk of heart failure hospitalisation or atrial fibrillation was directly correlated to the extent of RV pacing.11 Subsequent analysis of MOST demonstrated that RV pacing percentages as low as 40% were associated with a twofold increase in heart failure hospitalisations.16 A recent report similarly showed that patients who required RV apical pacing after AV node ablation had significant reduction of a previously normal LVEF, which could be predicted by early LV dyssynchrony.17
Data on the effects of conduction abnormalities or RV pacing after TAVI point to similar adverse effects. Houthuizen et al18 found an association between LBBB after TAVI and increased mortality, although this finding has not been universally replicated in other studies.12 ,19 ,20 Hoffmann et al14 reported that patients who developed LBBB or the need for permanent pacing after TAVI had increased LV dyssynchrony at 1 month and impaired recovery of LVEF. Similarly, Nazif et al12 demonstrated that patients who developed LBBB after TAVI had lower LVEF, and that the impaired recovery of LVEF was particularly severe among those with baseline depressed LVEF. A report from Urena et al13 showed that TAVI patients who received a new PPM had lower LVEF compared with non-paced patients over 22-month follow-up. The authors did not note an increase in mortality or rehospitalisation in the PPM patients. However, in this and a recent PARTNER analysis,21 patients with a new PPM after TAVI had higher rehospitalisation and mortality or rehospitalisation compared with no PPM patients. In the current analysis, which differs in the exclusion of chronic LBBB patients, new PPM is also associated with higher mortality. The new PPM patients also had significantly lower LVEF at 1 year compared with the no PPM patients.
In contrast to the prior studies that focused on new conduction abnormalities or pacemaker implantation after TAVI, we included the large group of patients with prior PPM, surmising that trends seen in studies of acquired conduction abnormalities may be more evident in chronically paced patients. Our analysis of the prior PPM patients showed both increased mortality and diminished recovery of LVEF at 1 year. It is notable that the recovery of LVEF after TAVI was most impaired in patients with lower LVEF at baseline, who constituted a greater proportion of the prior PPM patients. The magnitude of difference in recovery of LVEF between new and prior PPM patients may represent a continuum due to less duration of RV pacing or less frequent pacing in the former, as new PPM patients may recover intrinsic conduction after TAVI.22 ,23
The outcomes of the prior PPM group closely mirror the mortality, rehospitalisation and LVEF of the subset of no PPM patients with chronic LBBB, for which the conduction abnormality is similar to RV pacing and ventricular dyssynchrony is implicated. The worsened recovery of LVEF after TAVI at all baseline levels in the prior PPM group also argues for a mechanistic effect of pacing. The fact that the prior PPM group had higher cardiovascular mortality at 1 year compared with the new PPM group supports the possibility that the mortality associated with pacing was due to the duration of pacing and its long-term consequences rather than the need for pacing itself. The only clinical outcome that was improved for the prior PPM group was the presence of atrial fibrillation on ECG at 1 year and stroke, which could possibly be attributed to atrial pacing in patients with sinus node dysfunction.24 Alternatively, it is possible that there are unknown confounders or the need for chronic pacing may be associated with a sicker group of patients. However, after multivariable analysis and propensity score matching, the presence of a prior PPM remained significantly associated with mortality in this study.
The current mortality results are in contrast to the study by Buellesfeld et al,25 who included prior PPM and found no significant mortality differences compared with TAVI patients with new or no PPM However, that study was much smaller and there were only 48 patients with PPM prior to TAVI, for which the relative risk for mortality was the highest at 1 year among all three groups. Besides the significantly larger patient population, the current analysis also differs from previous studies in that the data were derived from a rigorously conducted trial with central adjudication of clinical events and core lab analysis of primary data.
This study has several limitations. No data were available on the type of devices that were implanted prior to TAVI. The criteria for RV pacing morphology excluded the common morphologies for biventricular or LV pacing,26 ,27 but there can be overlap in pacing morphologies for different sites. Based on these criteria, one can deduce that the majority of devices were RV pacing systems and not biventricular devices, and based on the study population and implant trends in the newly implanted devices, were likely pacemakers and not defibrillators. In addition, the presence of defibrillators or cardiac resynchronisation devices would tend to mitigate mortality or LV dysfunction differences in the prior PPM group. Second, without PPM interrogation data, the exact amount of RV pacing cannot be determined, although estimates from reviewed ECGs suggest that the amount of pacing was substantial. Lastly, because the majority of clinical events occurred in outpatients, we cannot exclude the possibility that more ischaemic events occurred in the prior PPM group, which had a slightly higher frequency of CAD.
In the PARTNER trial experience, the presence of a new or prior PPM in patients who underwent TAVI was associated with significantly higher rates of adverse clinical events, including mortality, at 1 year. PPM patients had lower LVEF at 1 year, and prior PPM patients had poorer recovery of LVEF after TAVI. It is possible that ventricular dyssynchrony from RV pacing may explain these findings. Whether the use of cardiac resynchronisation devices will benefit TAVI patients requiring permanent pacing or prove to be cost effective in this elderly patient population cannot be concluded from these data. Further studies will be needed to confirm these results and to establish the optimal pacing configuration after TAVI in order to maximise the benefits of the procedure.
What is already known on this subject?
New conduction abnormalities or pacemaker implantation after transcatheter aortic valve implantation (TAVI) have been associated with adverse outcomes. We hypothesised that chronic pacing from pacemakers implanted before TAVI may also be associated with negative events.
What might this study add?
We analysed data from the Placement of Aortic Transcatheter Valve (PARTNER) trial and registries and found that at 1 year patients with prior pacemakers, newly implanted pacemakers or with chronic left bundle branch block all had higher all-cause mortality compared with patients without pacemakers (27.4%, 26.3%, 27.7% vs 20.0%, p<0.05) as well as lower LVEF among surviving patients at 1 year (50.5%, 55.4%, 48.9% vs 57.6%, p<0.01). The presence of a pacemaker was an independent predictor of mortality after TAVI (HR 1.31, 1.08 to 1.60, p=0.006, prior pacemaker; HR 1.38, 1.00 to 1.89, p=0.05, new pacemaker).
How might this impact on clinical practice?
These data suggest an adverse effect of RV pacing and indicate a consideration for alternative pacing modalities such as cardiac resynchronisation therapy in patients requiring pacing after TAVI.
The authors would like to thank Girma M. Ayele, PhD, for statistical assistance with this study.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
- Data supplement 1 - Online figures
Contributors JMD and TMN were responsible for drafting this manuscript and are willing to serve as guarantors for the accuracy of the data. Analyses were conducted by KX, with additional statistical support provided by Girma M. Ayele. PLH and PSD contributed to the collection of data. All authors contributed to the design of this substudy and offered critical revisions throughout the process of writing and revising the manuscript.
Funding The PARTNER Trial was funded by Edwards Lifesciences and the protocol was developed collaboratively by the Sponsor and the Steering Committee. The Sponsor had no role in the analysis or interpretation of this substudy or the drafting or decision to publish this manuscript.
Competing interests TMN: Consultant, Edwards Lifesciences. PSD: institutional grant support, Edwards Lifesciences. HCH: grant support, Boston Scientific Corporation, Edwards Lifesciences, Medtronic, St. Jude Medical; Consultant, Edwards Lifesciences. WYS: Consultant, MicroInterventional Devices. WFF: grant support, St. Jude Medical. EMT: travel reimbursements from Edwards Lifesciences related to participation as an unpaid member of the PARTNER Trial Executive Committee. ADP: Consultant, Edwards Lifesciences. RM: grant support, Edwards Lifesciences and St. Jude Medical; Consultant, Abbott Vascular, Cordis, and Medtronic; Equity, Entourage Medical. MW: Consultant, Edwards Lifesciences. RTH: grant Support, Philips Healthcare; Consultant/Honoraria: Edwards Lifescience, St. Jude Medical. CRS: travel reimbursements from Edwards Lifesciences related to participation as an unpaid member of the PARTNER Trial Executive Committee. MBL: travel reimbursements from Edwards Lifesciences related to participation as an unpaid member of the PARTNER Trial Executive Committee. SKK: Consultant, Edwards Lifesciences; Scientific Advisory Board, Thubrikar Aortic Valve.
Patient consent Obtained.
Ethics approval Institutional review boards at each site approved the study.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement The study protocol is available upon request to the corresponding author.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.