Patients with symptomatic sinus nodal dysfunction and atrioventricular block receiving conduction system pacing were prospectively enrolled in our hospital between July, 2020 and December, 2021. For comparative analysis of case-control study, gender- and comorbidity-matched patients receiving transvenous RVP for symptomatic bradyarrhythmia and gender- and comorbidity-matched control subjects without pacemaker between July, 2020 and December, 2021 were also enrolled in this study. The exclusion criteria included patients with: (1) any history of HF with left ventricular ejection fraction (LVEF) <50% due to severe valvular heart disease or cardiomyopathy, (2) permanent atrial fibrillation, (3) left bundle branch block, (4) end-stage renal disease, and (5) malignancy with expected lifespan <1 year.

The detailed procedures are provided in Supplementary Methods: Procedures of conduction system pacing.

Echocardiography was performed within 7 days and at 6-month follow-up after the pacemaker implantation. All echocardiographic measurements were obtained initially during sinus or intrinsic rhythm for patients with normal atrioventricular conduction at the time of echocardiography, and then obtained during the DDD pacing mode with an atrioventricular delay of 100–120 ms to ensure complete capture by pacing without fusion in patients with normal atrioventricular conduction, at 2-fold output of pacing threshold, and a pacing rate between 80 and 110 beats/min. Because of inconsistent pacing threshold of HBP group during follow-up, echocardiographic images were only acquired during non-selective HBP (NS-HBP). Transthoracic echocardiographic images were recorded using a Vivid E9 ultrasound system (GE Healthcare, Milwaukee, WI, US) with M5S 3.5 MHz transducers. All echocardiographic measurements were obtained for at least 3 consecutive beats for sinus and pacing rhythm, and 5 consecutive beats for atrial fibrillation rhythm, and were stored with digital loops for offline analysis with EchoPAC (Version 202, GE Healthcare, Chicago, IL, US).

Based on the recommendations for cardiac chamber quantification by echocardiography (16), left atrial diameters, and LV dimensions and volumes were measured using two-dimensional echocardiography, and LVEF was calculated by biplane Simpson's method from the apical 4-chamber and 2-chamber views. The mean E/e' ratio (the average value of septal E/e' and lateral E/e'), tricuspid annular plane systolic excursion (TAPSE), and S' wave of the RV free wall (FW) were also obtained.

The intraventricular and interventricular synchronies were evaluated by pulsed-wave Doppler imaging echocardiography, tissue Doppler (Ts) imaging and STE. The regional durations of time measured for the basal segments in the LVFW, RVFW and interventricular septum (IVS) were from the start of QRS to the peak velocity of S' wave of Ts in the LVFW (Ts-LVFW), RVFW (Ts-RVFW) and septum (Ts-IVS), respectively (17).

The LV GLS, peak strain dispersion (PSD), septal-to-posterior wall motion delay (SPWMD), and difference of Ts between LVFW and IVS (Ts-LVFW-IVS) were used as LV mechanical synchrony parameters (3, 4, 9, 10, 17, 18). The value of Ts-LVFW-IVS means the time of S' in the LVFW minus the time of S' in the septum. The LV GLS measured by STE at a frame rate of 50–80 frames/s, and the apical four-chamber (A4C) view, apical three-chamber (LAX) view, and apical two-chamber (A2C) view of STE were recorded as well as electrocardiography. Peak LV GLS was then performed using echocardiography software that divided the myocardium into 6 segments in each view, creating graphs of shortening over the cardiac cycle. A bull's-eye plot was created of peak LV GLS for each myocardial segment (Figure 1). That absolute values of GLS were used when comparison among different study groups. PSD is defined as the standard deviation of time to peak longitudinal systolic strain of LV segments (Figure 1). The difference of Ts between Ts-RVFW and Ts-IVS (Ts-RVFW-IVS) was used as RV intraventricular mechanical synchrony parameter (10). For interventricular synchrony, the value of LV pre-ejection period (LV PEP), interventricular mechanical delay (IVMD), and different of Ts between LVFW and RVFW (Ts-LVFW-RVFW) were calculated (10, 17). Pulsed-wave Doppler flow velocity signal was recorded from the LV outflow tract, and LV PEP were measured as the time intervals between the Q-wave on the surface ECG and the onset of Doppler flow. IVMD was defined as the time difference between LV PEP and RV PEP, which were the periods from the start of QRS wave to the start of pulsed-wave Doppler flow of LV and RV outflow tracts, respectively. The values of LV PEP >140 ms and IVMD >40 ms were both considered indicative of interventricular dyssynchrony (17, 19).

Global LV MW data were quantified using non-invasive LV pressure-strain loop estimations and the area within the LV pressure–strain loop represents MW (13, 14). Non-invasive blood pressure recordings to estimate LV pressure were measured by a sphygmomanometer to obtain the brachial cuff systolic blood pressure immediately prior to the echocardiographic study (14). The quantification of MW was conducted by EchoPAC (Version 202, GE Healthcare, Chicago, IL, US) according to the following steps. First, the duration of isovolumic and ejection phases was defined by valvular timing (the opening and closing time of mitral and aortic valves) according to pulse-wave Doppler imaging. Then, the LV pressure-strain loop was constructed automatically with the combination of LV strain and non-invasive blood pressure measurements adjusted by the duration of the isovolumic and ejection phases (Figure 2) (14). Global myocardial work index (GWI) was calculated as the average of MW from the three apical views using GLS pressure loop from mitral valve closure to mitral valve opening. Global constructive work (GCW) was defined as the MW performed by LV segmental shortening during systole or by LV segmental lengthening during isovolumic relaxation. Global wasted work (GWW) was defined as the MW performed by LV segmental lengthening during systole or MW performed by LV segmental shortening during isovolumic relaxation. Global myocardial work efficiency (GWE) was defined as the ratio of GCW to total work (GCW plus GWW), expressed as a percentage (Figure 2).

Intra- and inter-observer variabilities were calculated by intraclass correlation coefficient and the standard error of measurement. For intra-observer analysis of the LV synchrony data, 18 random measurements were reanalyzed, with a period of 2–4 weeks between the first and second analyses. Intraclass correlation coefficients for intra-observer agreement were 0.933 (P < 0.001) for GLS and 0.948 (P < 0.001) for GWI. Inter-observer variability was performed by a different cardiologist repeating measurements from the same images. For interobserver analysis of the LV synchrony data, 18 random measurements were analyzed. The intraclass correlation coefficients for interobserver agreement were 0.986 (P < 0.001) for GLS and 0.910 (P < 0.001) for GWI.

Continuous variables are expressed as mean ± standard deviation, and categorical variables were expressed as numbers and percentages. Continuous variables were compared using unpaired Student's t-test or the Kruskal–Wallis test in case of a non-normal distribution. Categorical variables were compared using Chi-square test or Fisher exact test as appropriate. Paired sample t-test or Wilcoxon test in case of a non-normal distribution was performed to assess differences of paired samples. Comparison of the continuous variables among groups was performed by one-way analysis of variance (ANOVA), with Bonferroni correction when significant events were detected. A two-sided P-value <0.05 was considered statistically significant. All statistical analyses were performed using the SPSS software (version 22.0; SPSS Inc., Chicago, IL, USA).

This study recruited 109 consecutive patients receiving transvenous permanent conduction system pacing. Thirty-nine patients were excluded, including 6 patients with history of HF due to severe valvular heart disease or cardiomyopathy, 7 patients with end-stage renal disease, 6 patients with permanent atrial fibrillation, 3 patients that refused to join the study and 17 patients with left ventricular septal pacing (LVSP) according to the reported criteria (20–22). Finally, 70 patients with physiological pacing, including 24 patients with NS-HBP and 46 patients with LBBP, were enrolled. For comparative analysis of case-control study, 16 patients with RVP and 10 subjects without pacemaker (control group) were also enrolled.

Table 1 lists the clinical characteristics of the study patients. The mean age of the entire study subjects was 74 ± 9 years and 51.0% of the study subjects were male. The NS-HBP group was older than the LBBP and control groups (Table 1). The NS-HBP and RVP groups had a higher prevalence of atrial fibrillation, and a higher prescription rate of antiarrhythmic and anticoagulants compared with the control and LBBP groups (Table 1). The control group had a higher prescription rate of angiotensin converting enzyme inhibitors/angiotensin receptor blockers compared with the other 3 groups (Table 1).

The LBBP group had a higher prevalence of atrioventricular block and pre-existing RBBB, and a higher burden of ventricular pacing during follow-up compared with the other 3 groups (Table 1). In the LBBP group, the stimulus to left ventricular activation time was 69 ± 8 ms, and the LBB potential to QRS was 23 ± 6 ms. The pacing QRS duration was wider in the RVP group compared with the LBBP and NS-HBP groups (145 ± 14 vs. 114 ± 11 vs. 116 ± 12 ms, respectively, P < 0.001) (Table 1). About pacing parameters, LBBP group had higher R wave amplitude at implant and follow-up, lower pacing threshold at follow-up, and higher pacing impedance at implant and follow-up compared with NS-HBP group (Figure 3).

Conventional echocardiographic findings at implant and at 6-month follow-up are summarized in Table 2. The left atrial size significantly increased during ventricular pacing at 6-month follow-up in the RVP group compared with the LBBP and NS-HBP groups (46 ± 15 vs. 39 ± 8 vs. 38 ± 8 mm, P = 0.031) (Table 2). The LV end-diastolic and end-systolic volumes during ventricular pacing did not differ among the RVP, LBBP and NS-HBP groups at implant and at 6-month follow-up. Likewise, LVEF during ventricular pacing did not differ among the RVP, LBBP and NS-HBP groups at implant and at 6-month follow-up. The mean E/e', TAPSE, and S' wave of RVFW during ventricular pacing also did not differ among the RVP, LBBP and NS-HBP groups at implant and at 6-month follow-up (Table 2).

During sinus or intrinsic rhythm, parameters of LV mechanical synchrony, in terms of SPWMD, PSD, LV GLS, and Ts-LVFW-IVS, did not differ among the 4 groups at implant and at 6-month follow-up. However, during ventricular pacing, RVP group had longer SPWMD compared with LBBP and NS-HBP groups (122 ± 39 vs. 78 ± 22 vs. 82 ± 15 ms, respectively, P < 0.001) at implant and (113 ± 35 vs. 72 ± 23 vs. 86 ± 24 ms, respectively, P < 0.001) at 6-month follow-up, longer PSD compared with LBBP and NS-HBP groups (63 ± 17 vs. 53 ± 15 vs. 47 ± 20 ms, respectively, P = 0.027) at implant and (65 ± 26 vs. 48 ± 14 vs. 42 ± 22 ms, respectively, P = 0.002) at 6-month follow-up, lower LV GLS, especially over segments of anteroseptal and septal walls, compared with LBBP and NS-HBP groups (16.6 ± 5.4 vs. 20.0 ± 2.8 vs. 18.5 ± 5.8%, respectively, P = 0.033) at 6-month follow-up, and longer Ts-LVFW-IVS compared with LBBP and NS-HBP groups (83 ± 55 vs. 16 ± 37 vs. 29 ± 39 ms, P < 0.001) at implant and (45 ± 50 vs. 18 ± 38 vs. 15 ± 30 ms, respectively, P = 0.042) at 6-month follow-up (Table 2).

During sinus or intrinsic rhythm and ventricular pacing, the parameter of RV mechanical synchrony, in terms of Ts-RVFW-IVS, did not differ among the RVP, LBBP and NS-HBP groups at implant and at 6-month follow-up (Table 2).

During sinus or intrinsic rhythm, parameters of interventricular mechanical synchrony, in terms of LV PEP, IVMD and Ts-LVFW-RVFW, did not differ among the 4 groups at implant and at 6-month follow-up. However, during ventricular pacing, RVP group had longer LV PEP (141 ± 16 vs. 104 ± 23 ms, P = 0.014) compared with LBBP group at 6-month follow-up, longer IVMD compared with the LBBP group (14 ± 13 vs. −6 ± 6 ms, P = 0.032) at implant and (21 ± 24 vs. −8 ± 11 ms, P = 0.002) at 6-month follow-up, and longer Ts-LVFW-RVFW compared with the LBBP group (58 ± 60 vs. 1 ± 46 ms, P = 0.001) at implant (Table 2).

There were no significant differences in all the parameters of MW (GWI, GCW, GWW and GWE) during sinus or intrinsic rhythm among the 4 groups at implant and at the 6-month follow-up (Table 3 and Figure 4). However, during ventricular pacing, LBBP group had higher GWI compared with RVP group (2,174 ± 529 vs. 1,689 ± 562 mmHg%, P = 0.023) at implant and (2,189 ± 527 vs. 1,493 ± 799 mmHg%, P = 0.002) at 6-month follow-up, and NS-HBP group had higher GWI compared to RVP group only at the 6-month follow-up (2,146 ± 703 vs. 1,493 ± 799 mmHg%, P = 0.010) (Table 3 and Figure 4). During ventricular pacing, NS-HBP group had higher GCW compared with RVP group (2,921 ± 771 vs. 2,203 ± 866 mmHg%, P = 0.009) at 6-month follow-up (Table 3 and Figure 4). During ventricular pacing, LBBP group had lower GWW compared with RVP and NS-HBP groups (211 ± 161 vs. 484 ± 281 vs. 406 ± 234 mmHg%, respectively, P < 0.001) at 6-month follow-up, and consequently, LBBP group had higher GWE compared with RVP and NS-HBP groups (89.5 ± 4.3 vs. 83.4 ± 7.2 vs. 85.0 ± 9.5%, respectively, P = 0.004) at implant and (91.4 ± 5.0 vs. 80.9 ± 8.3 vs. 86.4 ± 8.1%, respectively, P < 0.001) at 6-month follow-up (Table 3 and Figure 4).

Long-term RVP can cause LV systolic dysfunction, pacing-induced cardiomyopathy and HF (1–4, 23). The mechanism of adverse effects is due to asynchronous electrical activation of LV caused by RVP-induced left bundle branch block pattern (3, 4, 23). Our recent study revealed that delta increment in pacing QRS duration, similar to electrical dyssynchrony, increased the risk of cardiovascular mortality in patients with pre-existing bundle branch block (24). Nahlawi et al. reported that RVP caused deterioration of LVEF after one-week continuous pacing, that persisted even after cessation of pacing 24 h later (25).

The left atrial size significantly increased during ventricular pacing at 6-month follow-up in the RVP group compared with the LBBP and NS-HBP groups (Table 2). Consequently, RVP increased the risks of atrial fibrillation, HF hospitalization and stroke (2).

Interventricular and intraventricular mechanical dyssynchronies play a crucial role on ventricular pump function (3, 4, 17, 23). RVP can induce significant left intraventricular mechanical dyssynchrony, especially between the anteroseptal wall (with early activation) and the posterolateral wall (with late activation) (4, 9, 23). Recently, several studies reported that NS-HBP with high output (mean threshold 1.8 ± 1.2 V at 0.42 ms) and LBBP could prevent LV mechanical dyssynchrony, but NS-HBP with low output (mean threshold 1.0 ± 0.4 V at 0.42 ms) could not (9, 26). In this study, LBBP had better left intraventricular mechanical synchrony at implant and 6-month follow-up and better LV GLS, particularly over the anterior wall (LAX view) and septum (A4C view), at 6-month follow-up compared with RVP (Table 2). Moreover, LBBP provided stable threshold at long-term follow-ups compared with NS-HBP (Figure 3).

Interventricular dyssynchrony indicates a delay in activation between LV and RV, resulting in a lack of coordinated contraction between LV and RV. In HF patients, interventricular dyssynchrony was associated with higher risk of cardiac events, independent of the QRS duration and LVEF (18). In this study, LBBP had less interventricular mechanical dyssynchrony during ventricular pacing compared with RVP. An intra-patient-controlled study (10) reported that LBBP resulted in modest delay in RV activation and less interventricular synchrony measured by IVMD and Ts-LVFW-RVFW compared with NS-HBP in patients with bradycardia. However, in this study, LBBP, including 15 patients (32.6%) with pre-existing right bundle branch block, did not differ in interventricular synchrony compared to NS-HBP (Tables 1, 2).

Global LV MW is a novel method to estimate LV systolic function, derived from computed LV pressure-strain loops incorporating both brachial cuff systolic blood pressure recordings and STE data (13, 14). Previous studies have demonstrated that there is a good correlation between LVEF and GWE in the general population, and GWE can be applied to assess the outcomes in patients with cardiac diseases (14, 15, 27, 28). Moreover, lower resting values of LV GLS and GWE in HF patients with preserved LVEF suggest an early subclinical myocardial damage, seemly associated with lower exercise capacity, greater pulmonary congestion, and blunted LV contractile reserve during effort (28). Furthermore, in the Geisinger-Rush Conduction System Pacing Registry, Sharma et al. (29) reported that LBBP patients with a ventricular pacing burden >20% had a lower risk of all-cause mortality, HF hospitalization, or upgrade to biventricular pacing compared to RVP. In an animal study, RVP caused significant redistribution of work away from the septum toward the LV lateral wall and significant redistribution of perfusion toward the LV anterolateral regions and a 30%–40% decreased myocardial efficiency (stroke work/myocardial oxygen consumption), whereas LV apical and septal pacing (via transventricular septal approach) did not significantly change myocardial efficiency (30). In this study, we first reported that during atrial synchronized ventricular pacing, the LBBP group had better GWE compared with RVP and NS-HBP groups (Table 3 and Figure 4). Further studies are warranted to assess that the better GWE achieved with LBBP compared with RVP in this study translates into better long-term clinical outcomes in different patient population with high burden ventricular pacing.