Impact of Right Ventricular Pacing Sites on Right Ventricular Function: Systematic Review and Meta-analysis

There are various indications for pacemakers, with sinus node dysfunction and high-grade atrioventricular (AV) block being among the most common reasons for implantation.^1,2 Most pacemakers rely on an endocardial pacing lead that is typically positioned at the right ventricular (RV) apex. However, evidence suggests that this conventional RV apical pacing can adversely affect cardiac structure and left ventricular (LV) function. Numerous studies have shown that RV pacing, especially from the RV apex, results in altered regional perfusion, mechanical dyssynchrony and adverse ventricular remodelling, all contributing to LV dysfunction over time.^3–13 Growing evidence suggests that alternative RV sites may reduce LV dyssynchrony compared with conventional RV apical pacing.^14–16

However, the upgrade from RV pacing to CRT has shown the greatest benefit in pacing-induced cardiomyopathy by reversing remodelling, improving mitral valve function and overall LV haemodynamics.^14–19 Although there has been more focus on the effects of RV pacing on LV function, few studies have evaluated the effects of RV pacing on RV function.

Most research on the effects of RV pacing on RV function is concentrated on tricuspid regurgitation (TR).^20–24 Other parameters of RV function, such as RV ejection fraction (RVEF), pulmonary artery systolic pressure (sPAP) and RV global longitudinal strain, have not been reviewed. We performed a systematic review of randomised controlled trials and observational studies to evaluate the effects of RV pacing on RV function.

Methods

This systematic review and meta-analysis were exempt from the requirement for local institutional review board approval and were compliant with the Health Insurance Portability and Accountability Act – US legislation. The PRISMA checklist was used while conducting this study.

Study Design

A systematic search of the PubMed, SCOPUS and Embase databases was performed, covering the period from database inception to 10 March 2024. Relevant articles were identified using MeSH terms, associated keywords and Boolean operators. Various keyword combinations of ‘right ventricular pacing’, ‘right ventricular function’ and ‘cardiac pacing’ were used as search terms (Supplementary Material 1 ). In addition, for a more comprehensive search, we manually searched the reference lists of related articles.

The inclusion criteria for study selection were studies using RV pacing strategies and those reporting data on RV function parameters. Studies that were duplicates, missing RV function data, involved patients aged <18 years and were either computer simulation studies, animal studies, case reports or conference abstracts were excluded.

Data Extraction

Two independent authors (SD, BA) screened the titles and abstracts to determine which articles to include for data extraction. Any inconsistencies were resolved by consensus. In the second phase, full text appraisal was completed by the same two authors.

Quality Assessment

The quality of each article was assessed using the modified Quality Assessment of Diagnostic Accuracy Studies (QUADAS) criteria.²⁵ The QUADAS-2 risk-of-bias domains (patient selection, index test, reference test and flow and timing), as well as the applicability domains, were rated as low risk, high risk, or unclear risk.

Right Ventricular Function

The primary outcome for this study was to examine the effect of RV pacing on RV function parameters. In studies where relevant data were accessible, the pacing site, pacing mode, pacing percentage and baseline patient characteristics were assessed. Patients with any indication for permanent pacemaker implantation were included. RV pacing sites included the RV apex, RV septum, leadless and left bundle branch area pacing (LBBAP). Data on the non-volumetric assessment of RV systolic function and haemodynamic assessment of RV and pulmonary circulation were extracted using the following parameters: 2D and 3D RVEF, tricuspid annular plane systolic excursion (TAPSE), RV volume (RV basal, mid-cavity or longitudinal), tissue Doppler-derived tricuspid lateral annular systolic velocity (S′), sPAP, 2D fractional area change (FAC), RV index of myocardial performance (RIMP), RV derivative pressure/time (dP/dt) and TR. Reference standards were adapted from the 2010 American Society of Echocardiography guidelines for the echocardiographic assessment of the right heart in adults and the 2017 American Society of Echocardiography recommendations for the non-invasive evaluation of native valvular regurgitation (Supplementary Table 1 ).^25–27

Statistical Analysis

Prior to conducting the meta-analysis, a descriptive analysis was performed using Microsoft Excel to summarise the characteristics of the included studies and to assess their individual contributions to the overall dataset. Meta-analyses were conducted to pool the mean differences of six RV function parameters (RVEF, TAPSE, RIMP, S′, FAC, sPAP) across studies before and after RV pacing. Although TR and RV volume measurements were a part of the initial data extraction, their values contained varying non-uniform measurements and, therefore, a meta-analysis was not plausible. In addition, pacing burden was excluded from the meta-analysis due to missing and non-uniform measurements (i.e. average pacing, median pacing percentage). Studies were included in the meta-analysis and subsequent meta-regression if they reported mean and SD values for RV parameters before and after treatment, sample sizes, treatment duration (months) and site of pacing. The site of pacing was categorised as follows: 1 = apex, 2 = septum, 3 = leadless, and 4 = LBBAP. The mean difference between the RV parameter after and the RV parameter before treatment was calculated for each study. SEs of the mean differences were computed using the following formula, assuming a correlation coefficient (r) of 0.5 between pre- and post-treatment measurements:

A logit transformation was applied to stabilise the variances of the event proportions and analyses were performed using the metaprop function from the R meta package. A random-effects model was used with the restricted maximum likelihood method to pool the mean differences and account for variability across studies.

Forest plots were generated using the RevMan5 layout to visualise individual study estimates and overall results. Heterogeneity before and after the meta-regression was assessed using Cochrane’s Q test and I² statistic (including H² to represent heterogeneity as a percentage), with I² values >50% indicating substantial heterogeneity. All analyses were conducted using R version 4.52 (2024; R Foundation for Statistical Computing) and the meta package. All statistical analyses were conducted by two authors (MH, SD).

Results

Study Characteristics and Search Results

The preliminary literature search generated 969 studies. After applying the eligibility criteria, 18 studies met inclusion criteria.^28-45 The PRISMA flowchart in Figure 1 summarises the literature search, with the baseline characteristics of the studies summarised in Supplementary Table 2. The included studies were published between 2006 and 2024, with a total of 1,220 patients, 42% (n=513) of whom were female. As indicated in Supplementary Table 3, the most common comorbidities were hypertension (51.3%; n=626), followed by type 2 diabetes (24.4%; n=298) and coronary artery disease (14.9%; n=182). Most of the studies (66.7%; n=12) were prospective non-randomised studies. In 14 studies, the population consisted of patients with a standard indication for permanent pacemaker. Follow-up periods ranged from 3 days to 76.8 months.^{28–32,34,35–39,42–45} Over this time, eight studies reported on pacing percentage, with 68.3% (n=462) having a pacing percentage over 40% and 31.7% (n=214) having a pacing percentage ≤40%.^{28,29,31,34,36,38,39,41} Among the 18 studies, 3D echocardiography was used in (16.7%) and 2D echocardiography was used in 16 (88.9%). In the five studies that reported baseline use of antiarrhythmic medications, 21.0% (n=34) of patients were on antiarrhythmic therapy compared with 79.0% (n=128) not on antiarrhythmic therapy.^{28,29,32,35,44} In the 16 studies, that reported baseline LV ejection fraction (LVEF), 74.6% (n=910) of patients had a baseline LVEF >50% and 17.6% (n=215) had a baseline LVEF ≤50%.^{28–33,35,36,38,39,41–45}

Supplementary Table 4 summarises the effect of RV pacing on RV function parameters.^28–45 In 14 of the 18 studies there was at least one parameter of RV function showing a statistically significant improvement or worsening (p<0.05).^{27–32,35–37,39,41–44} The other four studies showed neither a statistically significant improvement nor a statistically significant worsening of an RV function parameter.^34,35,39,41 Of the 14 studies showing significant changes in at least one RV function parameter, 12 had at least one significantly worsened parameter and two studies had at least one improved parameter. The most frequent RV function parameters that showed statistically significant worsening were TAPSE, TR and RIMP, each present in four of 12 studies; however, these were not reported in every study and are likely under-represented. In the eight studies that reported an average or median pacing percentage, three had a statistically significant worsening RV parameter in patients with a pacing percentage over 40% and one had a statistically significant improvement in an RV parameter.^{28,29,31,34,36,38,39,41}

Meta-analysis of Right Ventricular Function

In all 12 studies comprising 701 participants, were included in the meta-analysis assessing the pooled difference in RV function parameters after RV pacing.^{28-30,32,34,37,38,40,41,43-45} Subsequent meta-regression investigated whether the length of treatment and site of pacing explained variability in effect sizes across studies.

Right Ventricular Ejection Fraction

The estimated mean difference in RVEF following RV pacing was 2.28% (95% CI [−1.2818, 5.8385]; p=0.2097; Figure 2A). The heterogeneity between studies was high (I²=91.51%; H²=11.78; Q=54.0471; p<0.0001), indicating that 91.51% of the total variability was due to heterogeneity between studies, not within-study sampling error. Neither length of treatment (coefficient=0.1362; p=0.2082) nor site of pacing (coefficient=3.3500; p=0.1083) showed a statistically significant association with RVEF. Length of treatment and site of pacing accounted for 15.49% of the heterogeneity (pseudo R²=15.49%). Residual heterogeneity remained high (I²=87.89%), indicating the covariates did not fully explain the variability (Supplementary Table 5). The high residual heterogeneity suggests that other unexamined factors are contributing to variability.

Tricuspid Annular Plane Systolic Excursion

The overall mean difference in TAPSE (after versus before treatment) was estimated as −0.0444 cm (95% CI [−0.1976, 0.1089]; p=0.5705; Figure 2B). The heterogeneity between studies was high (I²=93.56%; H²=15.53; Q=113.4468; p<0.0001), indicating that 93.56% of the total variability in the effect sizes was due to differences between studies rather than sampling error. The meta-regression indicated that the covariates reduced the unexplained heterogeneity (pseudo R²=39.05%). Residual heterogeneity was still high (I²=88.18%), indicating that additional factors likely contributed to the variability. Length of treatment did not show a statistically significant effect on TAPSE outcomes (coefficient=0.0056; p=0.5304); however, the effect of site of pacing (coefficient=0.1073; p=0.0198) was statistically significant. Specifically, apical pacing site had the greatest effect on worsening TAPSE, followed by leadless and then LBBAP (septal pacing was not included due to inadequate numbers; Supplementary Table 5 ).

Right Ventricular Index of Myocardial Performance

The overall estimated mean difference in RIMP was −0.0714 (95% CI [−0.2888, 0.1459]; p=0.5195; Figure 2C). The heterogeneity between studies was high (I²=97.16%; H²=35.23; Q=29.6509; p<0.0001) indicating that 97.16% of the variability in effect sizes is due to heterogeneity between studies, not sampling error. The meta-regression indicated that the covariates explained 81.17% of the heterogeneity (R²=81.17%), significantly reducing residual variability. Despite this, residual heterogeneity remained moderate (I²=73.83%), indicating other variables may also contribute to variability. Length of treatment was a significant predictor of the mean difference in RIMP (coefficient=0.0380; p=0.0106), suggesting that longer treatment durations are associated with greater reductions in RIMP (Supplementary Table 5). Site of pacing was not used in the meta-regression for RIMP due to a lack of data.

Tissue Doppler-derived Tricuspid Lateral Annular Systolic Velocity

The estimated mean difference in S′ was −1.3522 cm/s (95% CI [−2.3647, −0.3396]; p=0.0089), suggesting that, on average, S′ significantly decreased with RV pacing (Figure 2D). The heterogeneity between studies was high (I²=89.09%; H²=9.17; Q=47.0475; p<0.0001), indicating that 89.09% of the variability in effect sizes is due to differences between studies rather than random sampling error. In the meta-regression of covariates, 58.51% of the heterogeneity (R²=58.51%) was accounted for by the moderators, with residual heterogeneity remaining moderate (I²=76.65%). The length of treatment (coefficient=0.0273; p=0.2209) did not significantly affect S′ outcomes; however, the results for site of pacing (coefficient=1.0250; p=0.0124) indicated that different pacing sites may have a significant impact on S′ (Supplementary Table 5). Specifically, pacing from the apex was associated with the greatest impact on S′, followed by septal pacing, leadless pacing and LBBAP having the least impact compared to others.

Percentage Fractional Area Change

The overall estimated mean difference in FAC% was −0.1208 (95% CI [−3.6630, 3.4215]; p=0.9467; Figure 2E). Heterogeneity was high (I²=94.95%; H²=19.80; Q=81.3802; p<0.0001), with 94.95% of the variability in effect sizes due to differences between studies rather than random sampling error. The meta-regression with length of treatment and site of pacing as covariates reduced the heterogeneity, explaining 67.87% of the variability (R²=67.87%). Residual heterogeneity remained high (I²=83.03%), indicating that additional unmeasured factors may contribute to the variability. Length of treatment (coefficient=−0.0678; p=0.6765) did not significantly affect FAC% outcomes; however, FAC% outcomes varied significantly depending on the pacing site (coefficient=3.8720; p=0.0014; Supplementary Material Table 5). Specifically, septal pacing was associated with a greater reduction in FAC% than was LBBAP (no apical and leadless data in the cohort).

Pulmonary Artery Systolic Pressure

The mean change in sPAP after RV pacing was 3.7282 mm Hg (95% CI [−6.0651, 13.5216]; p=0.4556; Figure 2F). Heterogeneity across studies was high (I²=98.02%; H²=50.60; Q=309.6241; p<0.0001), confirming large variability between studies. The meta-regression explained 84.38% of the heterogeneity (R²=84.38%), significantly reducing unexplained variability. Residual heterogeneity remained high (I²=84.01%), suggesting additional unmeasured factors were contributing to the remaining variability. Length of treatment (coefficient=0.0910; p=0.9160) did not significantly affect sPAP in RV pacing; however, site of pacing (coefficient=−7.9763; p=0.0006) considerably affected sPAP (Supplementary Table 5 ). Specifically, apical pacing is associated with the greatest increase in sPAP, whereas LBBAP is associated with the least impact on worsening sPAP.

Temporal Changes in Right Ventricular Function

The analysis demonstrated variability in RV functional changes over time (Figure 3). An increase in RVEF (+2.28%) was observed over a mean follow-up of 21.6 months, whereas sPAP increased by 3.73 mmHg over 6.5 months. In contrast, TAPSE (−0.04 cm at 13.5 months), RIMP (−0.07 at 7 months), S′ (−1.35 cm/s at 24.2 months) and FAC (−0.12% at 9.8 months) showed minimal or negative changes over time.

Risk of Bias Assessment

Risk of bias was evaluated using QUADAS-2 by two authors (SD, BA), independently and in duplicate (Supplementary Table 6). Most studies were rated as low risk and showed no major applicability concerns. The study of Huang et al. was identified as high risk in the ‘Flow and Timing’ domain, suggesting potential issues with the study’s methodology or reporting of follow-up procedures.³⁰ In addition, the study of Nunes et al. had high applicability concerns in the ‘Index Test’ domain, which may limit generalisability.³⁴

Discussion

This study suggests that RV pacing generally worsens RV function, with more pronounced effects when the pacing lead is in the RV apex and the pacing percentage is high. However, certain pacing strategies, such as LBBAP, can improve specific parameters of RV function. This indicates that optimising the pacing site and percentage could mitigate adverse effects on RV function. This study also highlighted the possible deleterious effect of RV pacing on TR. Heterogeneity in the data did not allow for proper meta-analysis; however, other studies point to a similar finding.^3–8 The heterogeneity in the present meta-analysis was consistently high across all echocardiographic parameters, suggesting that additional unmeasured factors, such as variations in patient populations, treatment protocols or follow-up duration, likely contributed to the observed heterogeneity. These findings underscore the importance of a nuanced approach to RV pacing, considering the underlying disease mechanisms and individual patient profiles to harness potential benefits and minimise risks.

This study shows that the effect of RV pacing is not uniform and varies based on the pacing location and the percentage of pacing. Despite their roles in causing LV dysfunction, it remains unclear why these factors affect RV function. It is possible the same factors that affect LV dysfunction are responsible for the decline in RV function. In addition, the LV dysfunction incurred by RV pacing may, in turn, lead to upstream deleterious effects, such as pulmonary hypertension and reduced RV ejection fraction.

Pacing at the RV apex has been associated with more pronounced adverse effects than pacing at other sites.⁴⁶ Specifically, chronic RV apical pacing has been associated with increased rates of pacemaker-induced cardiomyopathy.^5,6,13,47 Therefore, other pacing strategies have been pursued and have shown mixed results. In the PROTECT-PACE study, LVEF was significantly reduced regardless of RV apex pacing or high septal region pacing.⁴⁸ In a small study by Fruelund et al., 30% of patients developed pacemaker-induced cardiomyopathy over a median follow-up of 3.1 years regardless of RV septal or non-septal lead position.⁴⁹

Alternatively, conduction system pacing has recently evolved as a more physiological method to prevent ventricular dyssynchrony associated with RV pacing. His bundle pacing (HBP) has been shown to lead to a narrower paced QRS duration and marked improvement in LVEF in patients with pacemaker-induced cardiomyopathy.⁵⁰ When comparing HBP to biventricular pacing, Gardas et al. found that LV reverse remodelling was more significant with HBP than biventricular pacing in patients with pacemaker-induced cardiomyopathy.⁵¹

In addition to HBP, LBBAP has been shown to provide improvements in LV function. Moreover, the benefits seen with LV improvement may explain the improvements in RV function. Tian et al. demonstrated a beneficial effect of LBBAP on RV function in patients with RVEF below 45%, finding a significant improvement in RVEF, RV volume, RV strain and TAPSE.⁴⁰ In the study of Bednarek et al., RV systolic function, assessed by RV free wall strain, improved during LBBAP at 21 months.⁴¹ These findings suggest that optimising the pacing site and minimising the pacing percentage could mitigate the negative effects on RV function. Tailoring pacing strategies to individual patient characteristics could therefore play a crucial role in preserving RV function.

TR is commonly seen after implantation of permanent pacemakers and implantable cardioverter defibrillators, and is associated with poor survival. Specifically, transtricuspid leads (not implanted in the conduction system) are found to be associated with a much higher risk of TR than conduction system pacing, CRT and leadless pacing. In a recent systematic review and meta-analysis, Yuyun et al. found an almost fivefold increase in post–device implantation TR in transtricuspid leads.⁵² In contrast, the authors found that conduction system pacing, CRT and leadless pacing did not significantly affect the risk of TR.⁵² Nevertheless, pacing strategies outside of transtricuspid leads are not without consequences. For instance, leadless pacemakers are a newer alternative that has shown a lower prevalence of TR, but there are few studies that provide robust evidence for this. Our study examined two of these studies and found mixed results. In the study of Salaun et al., a small number of patients were followed for 2 months after leadless pacemaker implantation without any significant difference in TR.⁴⁴ In a longer study with a 12-month follow-up, Beurskens et al. reported that 43% of patients with a leadless pacemaker had worsening TR compared with baseline (p<0.001).⁴⁵

Lead-related TR can result from various mechanical interferences, such as the lead impeding leaflet movement or coaptation, adherence or entanglement with the tricuspid valve or subvalvular apparatus and damage to the tricuspid valve due to perforation or laceration of leaflets, papillary muscles or chordae tendineae. In addition, asynchrony caused by abnormal RV activation can lead to the right atrium contracting against a closed tricuspid valve. Similarly, the causes of worsening TR after leadless pacemaker implantation are not fully understood, but suggested mechanisms include valve damage during implantation, mechanical effects of the device on subvalvular structures and pacing-induced RV dyssynchrony. Both transtricuspid RV leads and leadless pacemakers can cause TR through mechanical interference and damage to the tricuspid valve structures, as well as through pacing-induced RV dyssynchrony.^53-56

Not only do leads affect TR, but the presence of TR is associated with poor survival. In a systematic review and meta-analysis by Alnamat et al., worsening TR increased mortality by 140%.⁵⁶ Similarly, Zhang et al. found all-cause mortality to be higher among patients with TR deterioration 1 year after pacemaker implantation (HR 1.598; 95% CI [1.275–2.002]; p<0.01).⁵⁴ Although transtricuspid leads significantly elevate the risk of TR and associated poor survival, alternative pacing strategies, such as leadless pacemakers, show a lower prevalence of TR but still pose potential risks, necessitating further robust studies to confirm their long-term safety and efficacy.

In this study, we found a small, non-significant increase in RVEF after RV pacing in the meta-analysis. Furthermore, neither pacing site nor duration of treatment had a statistically significant association with RVEF. Although it is difficult to draw conclusions from this analysis, other studies with different populations have shown improved parameters with RV patients. For instance, there may be a benefit in RV pacing during RV infarction, cardiogenic shock and decompensated pulmonary hypertension. In a small case series, Love et al. showed that restoration of AV synchrony with atrial and AV sequential pacing resulted in a highly significant (p≤0.001) increase in systolic blood pressure, cardiac output and stroke volume.⁵⁷ In another study, Abraham et al. found that in two patients with clinical evidence of RV infarction, atrial and AV sequential pacing resulted in immediate and sustained improvement in systolic blood pressure and clinical indices of perfusion.⁵⁸ Although these two studies are small, the findings may show benefit in settings of acute decompensation of the RV by enhancing RV function and, in turn, cardiac output.

Limitations

Our study has several limitations. First, the number of studies that met our inclusion criteria was limited. Second, the methods were heterogeneous in the patient selection and follow-up. In addition, the majority were single-arm studies with missing or heterogeneous data, therefore leading to heterogeneous outcome data in the meta-analysis and meta-regression. Third, many studies examined RV function as a secondary endpoint, resulting in missing patient characteristics, such as pacing percentage and certain RV function parameters. Finally, most of the studies reviewed had small sample sizes, indicating the need for future research with larger populations to provide more robust evidence.

Conclusion

The primary finding of our study is that RV apical pacing tends to worsen RV function. Assessing RV function is crucial because worsening TR increases mortality, RV dysfunction exacerbates pulmonary hypertension and improving RV function during acute RV decompensation enhances LV haemodynamics. Future studies should focus on evaluating both LV and RV functions to better understand the mechanisms of RV dysfunction associated with RV pacing.

Click here to view Supplementary Material.