Mortality after cardiac resynchronization therapy or right ventricular pacing in transcatheter aortic valve replacement recipients

This single-center trial was approved by the local ethics committee and conforms with the principles outlined in the Declaration of Helsinki. Data is used from a hospital registry – a single center registry to track outcomes in patients with severe aortic stenosis undergoing TAVR at the Heart and Diabetes Centre North Rhine-Westphalia, Bad Oeynhausen, Germany. Patient selection and the required procedural technique for TAVR were decided by a heart team. Consecutive patients undergoing TAVR from 2012 to 2022 were analyzed. Patients with pre-existing cardiac implantable electronic devices were excluded from further analysis.

The indication for post-TAVR PMI was determined according to current guidelines (2). PMI was performed if high grade AV-blockage or alternating bundle branch with conduction disturbance occurred after TAVR. New onset of sole left bundle branch or prolonged AV-interval was not seen as indication for PMI. Patients underwent PMI during the same hospital stay as TAVR. In patients with preserved sinus rhythm, a dual-chamber device was implanted. Ventricular leads were placed at the right ventricular apex or at the interventricular septum. The mode of pacing in patients with reduced LVEF was on discretion of the treating physicians and included RV pacing as well as CRT implantation. The main reasons for CRT implantation were the individual age and clinical status of the patient as well as the expected percentage of pacing and expected LVEF recovery after TAVR which was estimated by the treating physicians. The final decision whether to implant a CRT device or to perform RV pacing was based after individual decision in each patient at the discretion of the treating physicians.

For clinical endpoints, outcomes were evaluated at hospital discharge and follow up visits if available. Patients were seen in our outpatient clinic or by the referring cardiologist after 3–6 months and then once yearly. Visits included clinical inspection and patients´ history assessment, transthoracic echocardiography, ECG and pacemaker interrogation if necessary.

Five-year outcome data were used for analysis of outcome measures. The primary endpoint was all-cause mortality after 5 years. Secondary endpoints included QRS width after PMI and LVEF improvement in patients with LVEF ≤ 40%.

For analysis we divided these patients in 3 groups based on LVEF prior to TAVR because of two reasons. First, these thresholds are cut off values for diagnosis of Heart failure with reduced ejection fraction (HFrEF = LVEF ≤ 40%), with mildly reduced ejection fraction (HFmrEF = LVEF 41–49%) and preserved ejection fraction (HFpEF = LVEF ≥ 50%). Secondly, LVEF ≤ 40% is the cut-off value for European guideline recommendation of CRT in patients with heart failure in whom a high percentage of right ventricular pacing is expected [2].

Since not all patients with LVEF ≥ 50% are diagnosed with HFpEF, this group is referred to patients with “left ventricle with preserved ejection fraction” (LVpEF). Similarly, patients with a LVEF 41–49% are referred to “left ventricle with mildly reduced ejection fraction” (LVmrEF), and those with an LVEF ≤ 40% are referred to “left ventricle with reduced ejection fraction” (LVrEF).

A total of 4385 patients (mean age 81 ± 6 years, 53.1% female) without PMI before TAVR were analyzed. Baseline characteristics stratified for LVEF are displayed in Table 1. Beside other findings, LVrEF patients were more often male, had higher STS-Score and EuroScore II and were more symptomatic as LVmrEF and LVpEF patients as assessed by NYHA classification. Furthermore, LVrEF patients displayed more comorbidities including chronic kidney disease, peripheral artery disease and diabetes mellitus. Echocardiographic values can be found in Supplemental Table 1. LVrEF patients had lower mean aortic valve gradient (37.9 ± 15.4 mmHg vs. 41.4 ± 16.8 mmHg vs. 47.1 ± 22.1 mmHg; p < 0.001) prior TAVR and smaller effective orifice area after TAVR compared to LVmrEF and LVpEF patients (1.74 ± 0.47 cm² vs. 1.77 ± 0.44 cm² vs. 1.81 ± 0.46 cm²; p < 0.010). However, despite a higher rate of trans-apical access in LVrEF and LVmrEF patients, rates of major complications were not significantly different among these groups. Time on intensive-care-unit (2 days (IQR 1–4 days) vs. 1 days (IQR 1–3 days) vs. 1 day (IQR 1–3 days); p < 0.010) and hospital stay (13 days (IQR 8–18 days) vs. 11 days (IQR 8–15 days) vs. 10 days (IQR 8–14 days); p < 0.001) were significantly longer in LVrEF patients compared to LVmrEF and LVpEF patients (Table 2). Furthermore, in-hospital-mortality was significantly different among groups with LVpEF, LVmrEF and LVrEF (3.4% vs. 3.0% vs. 1.6%; p = 0.001). There was no difference in post-procedural aortic regurgitation/ paravalvular leakage between the three subgroups.

After a mean follow up of 739 days (IQR 300 to 1465 days), Kaplan–Meier analysis revealed significantly different estimated survival rates in each subgroup with the lowest estimated survival after 1117 ± 30 days in LVrEF patients (37.0% vs. 43.5% vs. 55.1%; HR ≥ 50% vs. 41–49%: 0.68, CI 0.554–0.834, log rank p < 0.001; HR 41–49% vs. ≤ 40%: 0.769, CI 0.615–0.961, log rank p = 0.021) (Fig. 1). However, multivariate cox regression found age and baseline creatinine, but not LVEF as significant confounders for survival (Supplemental Table 2).

A total of 603 of 4385 patients (13.8% of all patients) underwent PMI after TAVR. There was no difference in patient survival for patients receiving PMI after TAVR as compared to patients without PMI in the overall cohort in uni- und multivariate analysis (48.6% vs. 48.5%; HR 0.977, CI 0.906–1.055; log rank p = 0.552) (Fig. 2, Supplemental Table 3).

Among patients receiving PMI, 105 were LVrEF patients (15.0%), 48 LVmrEF patients (15.7%) and 450 LVpEF patients (13.3%), p = 0.171). Of these, 19 LVrEF patients (18.1%), 1 LVmrEF patient (2.1%) and 1 LVpEF patient (0.2%) received a CRT system (p < 0.001).

In LVrEF patients, Kaplan–Meier analysis revealed that estimated all-cause mortality after a mean of 1382 ± 33 days of follow-up was significantly lower in CRT patients as compared to patients with RV pacing (21.1% vs. 48.8%; HR 0.48, CI 0.204–1.128; log rank p = 0.045) (Fig. 3) as well as compared to LVrEF patients not undergoing any device implantation (21.1% vs. 44.2%; HR 0.358, CI 0.133 – 0.962; log rank p = 0.033). After adjustment for clinical confounders, biventricular pacing at discharge remained a significant beneficial factor for 5-year patient survival in multivariate analysis (HR 0.3, CI 0.095–0.951, p = 0.041) (Table 3).

In LVrEF patients, the CRT-group had significantly lower LVEF values prior to TAVR as compared to patients receiving RV pacing (30 ± 6.3% vs. 34.4 ± 5.9%; p = 0.004). QRS widths prior TAVR (129 ± 21 ms vs. 126 ± 28 ms, p = 0.674) and QRS widths after TAVR but before PMI (153 ± 23 ms vs. 153 ± 23 ms, p = 0.305) were not significantly different between CRT and RV pacing patients with LVrEF. QRS width at discharge was significantly reduced in the CRT group (145 ± 20 ms vs. 166 ± 28 ms; p = 0.003) and ventricular pacing burden was higher (99 ± 1% vs 79 ± 34%; p = 0.001) as compared to patients receiving RV pacing (Table 4). The percentage of patients receiving an ICD between CRT patients and patients receiving RV pacing was not significantly different (15.8% vs. 5.8%; p = 0.16). After four months (mean follow up 133 days, IQR 103–394 days), LVEF improved significantly in patients receiving CRT pacing (30 ± 6.3% to 40% (31–50%); p < 0.001) as well as in patients receiving RV pacing (34.4 ± 5.9% to 40% (32–45%); p < 0.001). Although LVEF was comparable between patients receiving CRT and RV pacing (p = 0.442) after four months, there was a trend towards greater LVEF change in CRT patients (14.9 ± 13.4% vs. 6.1 ± 7.9%, p = 0.075) (Table 4).

Long-term survival has become an increasingly important study endpoint as more patients undergo TAVR at younger ages [17]. Today, ESC guidelines recommend TAVR for patients ≥ 75 years whereas current ACC guidelines recommend considering clinical factors if the patient age is ≥ 65 years [18, 19] and consequently, a relevant proportion of this population has a life expectancy of 5 years and more [20].

PMI did not influence overall 5-year mortality in our patient cohort and PMI rates after TAVR were not different among patients across LVEF groups. In our cohort, the overall PMI rate was 13.8% which is comparable to other trials [20]. Several studies found adverse outcome related to PMI after 1-year follow-up duration, but these findings were not confirmed in other studies [7, 8]. In heart failure patients, RV pacing is a known factor which is associated with increased risk of mortality and hospitalization [11].

In-hospital and long-term mortality was higher for LVrEF patients as compared to patients with preserved or only mildly reduced LVEF. However, LVEF did not remain as a significant risk factor for patient mortality after multivariate cox regression analysis. This might indicate that LVEF serves as a surrogate factor but not as a true predictor of patient mortality after TAVR, especially considering that aortic stenosis might have contributed to impaired LVEF which potentially resolves after TAVR.

Individual risk for hospitalization and death after TAVR may rely on LVEF improvement and right ventricular pacing burden. Tsushima et al. found RV-pacing ≥ 30% was associated with higher rates of heart failure and death in pacemaker recipients after TAVR [21]. However, the patients in this study had preserved LVEF and did not undergo CRT implantation [21]. Predictors for high pacing burden and failure of LVEF improvement after TAVR would help to decide whether a patient needs CRT or may be treated with RV pacing alone. Additionally, there are no valid predictors for high RV pacing burden after TAVR, although the rate of recovery of AV block after TAVR is relatively low [22]. There are only limited data available on the role of CRT implantation after TAVR. According to current guidelines CRT is indicated in patients with reduced LVEF and no differentiation between patients with or without TAVR is made. In patients with LVrEF undergoing TAVR, LV function may recover. In our study, patients receiving RV pacing as well as patients receiving CRT pacing displayed significantly higher LVEF four months after TAVR. Therefore, a substantial fraction of patients with LVEF ≤ 40% at baseline had no longer an indication for CRT implantation. However, the additional effect of CRT as heart failure therapy is neglected in these patients.

It is unknown if the results of studies including patients without TAVR, like the BLOCK-HF study [11], can be transferred to this particular patient collective. In a study by Ananwattanasuk et al. electrical dyssynchrony either associated with a high burden of RV pacing or left bundle branch block was associated with increased mortality after TAVR [23].

Our study found reduced all-cause mortality associated with CRT in LVrEF patients as compared to RV pacing and highlights the potential role of biventricular pacing in these patients. It remains unclear whether patients with preserved or mildly reduced LVEF benefit from CRT-device implantation. In our cohort, only 2 of 498 patients with PMI (0.4%) and LVEF > 40% underwent CRT implantation after TAVR because current heart failure and pacing guidelines were respected for these patients. It is also not clear whether patients with reduced LVEF and a low percentage of RV pacing, either with or without procedure-induced left bundle branch block, would benefit from CRT. In our study, patients receiving CRT had a high rate of biventricular pacing which was associated with the indication-specific pacemaker programming. Due to the retrospective design of our study no definite conclusions can be drawn. Further analysis in larger scale randomized studies are needed to define the role of CRT or conduction system pacing in TAVR recipients.

We found that CRT recipients had better survival as compared to patients with reduced LV function who received RV pacing as well as to patients who did not receive any pacing modality. Reasons for this are unknown. We performed a multivariate analysis and found only one other factor (renal function) besides CRT which was associated with different patient survival. A relatively high amount of patients with left bundle branch block who did not receive CRT implantation directly after TAVR (due to missing high-degree AV block) may play a role. Additionally, patients who received a CRT device may benefit from concomitant ICD therapy. These observations need further investigation in future studies on larger patient cohorts.

QRS width in the CRT group was 145 ± 20 ms which might have been optimized during follow up. However, it has been shown that QRS width < 150 ms during biventricular pacing is associated with reduced risk of heart failure and death [22, 23].

There are several limitations with this study. At first, we could not distinguish between cardiovascular and non-cardiovascular deaths. Furthermore, there are no information regarding the hospitalization rate which serves as a relevant endpoint in most other trials. Additionally, the number of patients with LVrEF and PMI was relatively small compared to other trials.

This is a retrospective study and patients were not randomized. We do not have any further information regarding the cause of death. The number of patients who received PPI and CRT was relatively small. Albeit no statistically different amount of patients who received ICD therapy among patients with RV pacing and CRT (5 vs. 15%), this numerical difference may have played a role with regards to patient survival. Patients were treated by CRT or RV pacing after individual decision for each patient based on the preference of treating physicians and patients taking potential LV recovery as well as patient comorbidities into account. Multivariate analysis was performed to eliminate potential confounders. Nevertheless, there might have been several clinical reasons that have contributed to the decision for or against pacemaker or CRT. Adjustment for these reasons is difficult and might not be part of multivariate analyses.