Risk factors of pacing dependence and cardiac dysfunction in patients with permanent pacemaker implantation.

AIMS: Right ventricular pacing (RVP) dependence could impair left ventricular ejection fraction (LVEF). This study aimed to illuminate the relationship between RVP proportion and LVEF, as well as disclosing independent predictors of RVP dependence. METHODS AND
RESULTS: Patients indicated for permanent pacemaker implantation were included (2016-2020). The ventricular pacing lead was placed in right ventricular apex or septum. Pacing mode programming followed universal standard. Electrocardiographic, echocardiographic, and serological parameters were collected. RVP dependence was defined according to its influence on LVEF. This study was of case-control design. Included patients were matched by potentially confounding factors through propensity score matching. A total of 1183 patients were included, and the mean duration of follow-up was 24 months. Percentage of RVP < 80% hardly influenced LVEF; however, LVEF tended to decrease with higher RVP proportion. High degree/complete atrioventricular block (AVB) [odds ratio (OR) = 5.71, 95% confidence interval (CI): 3.66-8.85], atrial fibrillation (AF) (OR = 2.04, 95% CI: 1.47-2.82), percutaneous coronary intervention (PCI) (OR = 2.89, 95% CI: 1.24-6.76), maximum heart rate (HRmax ) < 110 b.p.m. (OR = 2.74, 95% CI: 1.58-4.76), QRS duration > 120 ms (OR = 2.46, 95% CI: 1.42-4.27), QTc interval > 470 ms (OR = 2.01, 95% CI: 1.33-3.05), and pulmonary artery systolic pressure (PASP) > 40 mmHg (OR = 1.93, 95% CI: 1.46-2.56) were proved to predict RVP dependence.
CONCLUSIONS: High RVP percentage (>80%) indicating RVP dependence significantly correlates with poor prognosis of cardiac function. High degree/complete AVB, AF, ischaemic aetiology, PCI history, HRmax  < 110 b.p.m., QRS duration > 120 ms, QTc interval > 470 ms, and PASP > 40 mmHg were verified as independent risk factors of RVP dependence.

Introduction

The appearance of the pacemaker fundamentally changed the way of bradycardia treatment. In addition, it has witnessed some revolutionary progression of technique in the area of pacemaker during recent years. For instance, leadless pacemaker is designed to integrate the pulse generator and the electrode as a whole with a mini size, which could solve the problems of lead‐related adverse events (e.g. lead infection, pocket infection, and tricuspid regurgitation). Other innovative techniques including bi‐ventricular based cardiac resynchronization therapy (CRT) and its derivative technique, multi‐site left ventricular pacing have been proved to be effective on heart failure (HF) treatment. After all, the mainstream selection for treatment of brady‐arrhythmia is still the classic right ventricular pacing (RVP) mode. Unfortunately, this well‐proved pacing technique has its own insufficiency. Traditional RVP mode could lead to elevated occurrence rate of asynchrony of ventricular motion, , ventricular remodelling, atrial fibrillation (AF), and even cardiac dysfunction. , , Clinical trials have confirmed that the aforementioned adverse effects positively correlated with RVP percentage, , , , , , , and incidence of HF significantly increase in patients with high level of RVP proportion. RVP dependence is defined as high proportion of ventricular pacing in total heart beats; however, it still lacks an acknowledged cut‐off point value of RVP dependence. Considering that the most important influence of RVP dependence is the compromised cardiac function, it is reasonable to set up the critical value of RVP dependence based on cardiac function. Besides, risk factors of RVP dependence remain unclear. Nowadays, the newly emergent physiological pacing modes such as His bundle pacing (HBP) and left bundle branch regional pacing (LBBP) draw a lot of attention, which may be potential to avoid the poor prognosis brought by long‐term RVP treatment. , , Nevertheless, it exists controversy about the specific characteristics of patients who could get benefits from the so‐called physiological pacing. It is reasonable to deduce that patients characterized with high probability of RVP dependence should have much more chance to benefit from physiological pacing. Therefore, to investigate predictors of RVP dependence would help to make strategy for choosing a certain type of pacemaker that is favourable to a specific patient. In addition, because high probability of RVP dependence might lead to premature battery depletion of leadless pacemaker for its relative small battery capacity, precise prediction of RVP could offer instruction for leadless pacemaker implantation. By no means, RVP is an irreplaceable technique, and RVP would maintain its predominant position for a long time, let alone its relatively low costs in comparison with novel pacing technique. Therefore, predictors should be deliberately found to screen patients with high probability of RVP dependence and select patients mostly benefiting more from physiological pacing. Accordingly, based on a retrospective cohort, the present study is designed to find out the specific value of pacing proportion, namely, RVP dependence, which could significantly influence cardiac function. Moreover, risk factors of RVP dependence are analysed as well to facilitate predicting prognosis of patients who received RVP treatment. Our study might assist EP physician to acquire a better understanding of RVP dependence and rational pacing scheme for patients. Furthermore, this study would help to customize a personalized scheme of pacemaker implantation, attaining the will of precision medicine somewhat. To our knowledge, similar study has been hardly seen before.

Methods

Study population

Patients who were diagnosed with bradycardia and indicated for permanent pacemaker implantation were included in study cohort. Types of bradycardia involved in this study mainly included sick sinus syndrome and atrioventricular block (AVB). The inclusion criteria included the following: with age ≥ 18 years; definite diagnosis and indicated for permanent pacemaker implantation; ventricular lead being placed in either right ventricular apex or right ventricular outflow tract (septum); with left ventricular ejection fraction (LVEF) ≥ 40% at baseline; receiving pacemaker implantation for the first time; and with life expectancy ≥ 1 year. The exclusion criteria included the following: with age < 18 years; planning to receive CRT, HBP, or LBBP; planning to receive implantable cardioverter defibrillator; temporary pacemaker for perioperative transition; replacement of pacemaker for reason of battery depletion, device‐related infection, or lead dislocation, and so forth; and with diagnosis of malignant tumour.

Procedure of pacemaker implantation and pacing mode programming

Pulse generator was placed in the infraclavicular region, and the leads were inserted either via the cephalic vein or subclavian vein approach. Passive atrial leads and active ventricular leads were selected by the operator routinely except for unstable fixation. Atrial lead (if had) was placed in left auricular appendage, while right ventricular lead was placed in apex or septum (outflow tract of right chamber). Amplitude (sensing), threshold, and impedance were measured by the pacing analyser (Medtronic 5318) during the operation. Stimulation threshold was tested at a pulse duration of 0.4 ms. The patients were discharged 1–2 days after the procedure if there were no complications. Pacing mode programming followed the uniform programming scheme.

Electrocardiography

Standard 12‐lead electrocardiograms (ECGs) were obtained by using Philips ECG machine model TC20 at a speed of 25 mm/s within 12 h of STEMI onset and input into a computer for further analyses. The ECG was observed for at least 60 s by the ECG machine with a function of real‐time monitoring. The ECG in non‐ventricular pacing condition was selected for data collection. If the ECG showed continuous ventricular pacing, this ECG would be excluded. Average PR intervals, QRS durations, and QT intervals of 12‐leads were measured in random order by two independent cardiologists blinded to the study design. The PR interval was measured from the onset of P wave to that of QRS complex. The QRS duration was measured from the beginning of the earliest to the end of the last QRS deflection. The QT interval was regarded as the onset of QRS complex to the end of T wave and corrected by the Bazett formula (QTc). The sinus ECG was preferred if a patient had a paroxysmal AF history. In persistent AF, 3 continuous cardiac cycles were included, and mean values were obtained. All ECGs were double‐checked, and the final result was established by consensus of at least two cardiologists.

Echocardiography

Transthoracic echocardiographic echocardiograms (TTE) were obtained in all patients, on a Philips IE33 instrument (Philips, Netherlands) with a 2–3.5 MHz transducer (X3–1), following the American Society of Echocardiography (ASE) instructions, for LVEF, left ventricular end‐systolic diameter (LVESD), left ventricular end‐diastolic diameter (LVEDD), left atrial diameter (LAD), and pulmonary artery systolic pressure (PASP) measurements. Left atrial volume (LAV), left ventricular end‐systolic volume (LVESV), and left ventricular end‐diastolic volume (LVEDV) were calculated by corrected Teichholz's formula: V = 7.0/(2.4 + D) × D 3 (V = volume, D = diameter). LVEF was detected by the Simpson method. PASP was estimated from the tricuspid regurgitant jet velocity using the modified Bernoulli equation and adding the estimated right atrial pressure (RAP) on the basis of inferior vena cava (IVC) diameter and collapsibility according to the ASE guidelines for echocardiographic assessment of the right heart in adults: IVC diameter ≤ 2.1 cm that collapses >50% with a sniff suggests a normal RAP of 3 mmHg (range, 0–5 mmHg), whereas an IVC diameter > 2.1 cm that collapses <50% with a sniff suggests a high RAP of 15 mmHg (range, 10–20 mmHg). In indeterminate cases in which the IVC diameter and collapse do not fit this paradigm, an intermediate value of 8 mmHg (range, 5–10 mmHg) was used.

Serology

Routine blood test items of HF patients in admission were collected from our medical record system, including serum creatinine (Scr), N‐terminal prohormone of brain natriuretic peptide (NT‐proBNP), cardiac troponin‐T (cTnT), sodium, potassium, chloride, creatine kinase (MB), haemoglobin, albumin, alanine aminotransferase (ALT), and aspartate aminotransferase (AST). Furthermore, estimated glomerular filtration rate (eGFR) was calculated by Modification of Diet in Renal Disease (MDRD) formula and Chronic Kidney Disease Epidemiology Collaboration (CKD‐EPI) formula based on the value of Scr and adjusted by patient's age and sex. MDRD formula was described as follows: eGFR for male = 175 × (Scr × 0.0113)−1.154 × age−0.203 and eGFR for female = 175 × (Scr × 0.0113)−1.154 × age−0.203 × 0.742. Well, CKD‐EPI formula was more complicated because it was dependent on not only sex but also the range of Scr level, and the specific equations were depicted as follows: for male with Scr ≤ 80: eGFR = 141 × (Scr × 0.0113/0.9)−0.411 × 0.993age, for male with Scr > 80: eGRF = 141 × (Scr × 0.0113/0.9)−1.209 × 0.993age, for female with Scr ≤ 62: eGFR = 144 × (Scr × 0.0113/0.7)−0.329 × 0.993age, and for female with Scr > 62: eGFR = 144 × (Scr × 0.0113/0.7)−1.209 × 0.993age. As these formulae mentioned above, the unit of eGFR was mL/min/1.73 m2, the unit of Scr was μmol/L, and the unit of age was year.

Propensity score matching method

Patients with RVP dependence were matched by propensity scores to patients without RVP dependence with a greedy matching algorithm. First, all possible five decimal place matches were made, then four decimal places and so on down to one decimal place. Next, the five best matches for each RVP dependence patient were randomly selected with each non‐RVP dependence patient only selected once. The weighted average of each set of non‐RVP dependence patient was used to represent one non‐RVP dependence patient. Patients who could not be matched were excluded from the matched cohort analysis. Subjects were assigned their original enrolment until the end of follow‐up.

Statistics

To find out the incidence of different RVP percentage, the RVP percentage was categorized into five groups: >50%, >60%, >70%, >80%, and >90%, as an endpoint for cumulative event rate analysis. To investigate the influence of different RVP percentage on ECG parameters and Echo parameters, the RVP percentage was evenly divided into 10 groups: 0–10%, 10–20%, 20–30%, 30–40%, 40–50%, 50–60%, 60–70%, 70–80%, 80–90%, and 90–100%. The cardiac function most related RVP percentage was selected as the cut‐off point of RVP dependence. Patients were allocated to RVP dependence group and non‐RVP dependence group (match ratio 1:1) by propensity score matching method to eliminate the potential bias. Missing data were replenished by calculating the mean value. After multivariate logistic analysis, continuous variables with statistical significance were further processed by receiver operating characteristic (ROC) analysis to calculate the Youden index (sensitivity + specificity − 1) to identify the optimal cut‐off points.

Results

Distribution of different right ventricular pacing proportion in included patients and related changes of left ventricular ejection fraction

After screening and filtration, a total of 1183 patients were included (Figure ). Different percentage of RVP was proved to gradually increase for long‐term follow‐up. With mean value of 24 month follow‐up duration, nearly 80% patients finally attained RVP percentage > 50%, while almost half of patients have RVP percentage > 90% (Figure ). Among 10 groups with different RVP percentage, values of LVEF were comparable with each other at baseline (Supporting Information, Figure ). Comparison of the cumulative incidence of different RVP proportion in total of patients was shown in Figure . The highest occurrence rate was patients with RVP percentage > 50%.

Figure 1

Flow chart of patient filtration and further selection by propensity score matching. CRT, cardiac resynchronization therapy; ICD, implantable cardioverter defibrillator; LBB, left bundle branch; VPD, ventricular pacing condition.

Figure 2

Distribution of different RVP proportion in included patients and related changes of left ventricular ejection fraction. (A) The RVP percentage was evenly divided into 10 groups: 0–10%, 10–20%, 20–30%, 30–40%, 40–50%, 50–60%, 60–70%, 70–80%, 80–90%, and 90–100%. (B) The cumulative incidence curve showed the differential occurrence of different percentage of RVP (>50%, >60%, >70%, >80%, and >90%) as endpoint. RVP, right ventricular pacing.

Changes of echocardiographic parameters at different time‐point during follow‐up

In comparison with baseline, percentage of RVP < 80% hardly influenced LVEF; however, LVEF would decrease if RVP dependence took place. Compared with baseline, LVEF of patients with 60% RVP was hardly changed after long‐term follow‐up (63.1 ± 5.9 vs. 63.9 ± 6.7, P = 0.733). Seventy per cent RVP tended to dampen cardiac function (57.9 ± 9.1 vs. 62.3 ± 10.8, P = 0.15); however, 80% RVP was proved to significantly influence LVEF (55.7 ± 10.8 vs. 64 ± 10.4, P = 0.005), indicating an optimal cut‐off point of RVP dependence (Figure and ). Changes of other echocardiographic parameters (LAD, LVEDD, LVESD, LAV, LVEDV, and LVESV) from baseline to the endpoint of follow‐up were analysed. No matter RVP dependence happened or not, LAD and LAV both increased indicating enlarged left atrial size (Supporting Information, Figure ). Parameters of cardiac diastolic function (LVEDD and LVEDV) and cardiac systolic function (LVESD and LVESV) were changed if RVP proportion was more than 80%, indicating enlarged left ventricular size (Supporting Information, Figures and ). In addition, higher value of PASP was detected in patients with RVP proportion > 80% (Supporting Information, Figure ). From the follow‐up data, both cardiac diastolic dysfunction and cardiac systolic dysfunction were influenced by RVP dependence. Accordingly, RVP proportion > 80% (80%~) was defined as cut‐off point of RVP dependence for its significant influence on cardiac function.

Figure 3

The trend of LVEF change in patients with different range of right ventricular pacing proportion. (A) Changes of LVEF at different time‐point during follow‐up. (B) Trends of LVEF in VPD group and non‐VPD group (cut‐off point of pacing proportion as 80%). LVEF, left ventricular ejection fraction; VPD, ventricular pacing condition.

Changes of electrocardiographic parameters at different time‐point during follow‐up

QRS duration and QTc interval showed different changes of tendency in different ranges of RVP percentage (Figures and ). QRS duration showed no difference between baseline and different time‐point of follow‐up with RVP percentage < 80% (Figure ). In Figure , RVP dependence group, QRS duration significantly increased during long‐term follow‐up (98.0 ± 26.0 ms vs. 147.5 ± 37.7 ms, P < 0.01). Similarly, at each time‐point, QTc interval was not influenced if RVP percentage was <80% (Figure ). However, QTc interval strikingly increased if RVP dependence took place (357.4.0 ± 73.6 ms vs. 410.0 ± 95.0 ms, P < 0.01) (Figure ).

Figure 4

The trend of QRS duration change in patients with different range of right ventricular pacing proportion. (A) Changes of QRS duration at different time‐point during follow‐up. (B) Trends of QRS duration in VPD group and non‐VPD group (cut‐off point of pacing proportion as 80%). VPD, ventricular pacing condition.

Figure 5

The trend of QTc interval change in patients with different range of right ventricular pacing proportion. (A) Changes of QTc interval at different time‐point during follow‐up. (B) Trends of QTc interval in VPD group and non‐VPD group (cut‐off point of pacing proportion as 80%). VPD, ventricular pacing condition.

Propensity score matched cohort and demographic data of included patients at baseline

In original patient cohort at baseline, the mean value instead of distribution of age showed no difference between groups (69.2 ± 19.8 vs. 68.7 ± 11, P = 0.49). Male proportion in RVP dependence group was larger than non‐RVP dependence group (58.4% vs. 44.45%, P < 0.0001). Other demographic characteristics such as medication, concomitant of heart disease/hypertension/diabetes mellitus, AF, attribute of pacemaker, and description of ECG or Echo factors were shown in Tables and . To eliminate the potential confounding factors, the patient whose RVP proportion was more than 80% was allocated to RVP dependence group, matched with one patient without RVP dependence (match ratio 1:1) in non‐RVP dependence (Figure ). Finally, the following variables were included in the logistic regression model for calculation of propensity score matching: 1024 patients matched with sex, age, hypertension, diabetes, right ventricular lead location (apex/septum), and chamber (single/dual‐chamber). After propensity score matching method, demographic factors were finally comparable between groups (Tables and ). All the serological markers at baseline showed no difference between groups (Supporting Information, Figure ).

Table 1A

Baseline demographic characteristics of study patients (categorical variables)

Categorical variables	Unmatched					Propensity score matched
	VPD		Non‐VPD		P value	VPD		Non‐VPD		P value
	N	%	N	%		N	%	N	%
	534	45.1	649	54.9		512	50	512	50
Age (years)					<0.0001					<0.0001
<30	12	2.2	4	0.6		12	2.3	4	0.8
30–49	30	5.6	23	3.5		30	5.9	23	4.5
50–69	215	40.3	338	52.1		209	40.8	338	66
>70	277	51.9	284	43.8		261	51	147	28.7
Sex					<0.0001					0.414
Male	312	58.4	290	44.7		290	56.6	276	53.9
Female	222	41.6	359	55.3		222	43.4	236	46.1
Medication
ACEI/ARB	274	51.3	327	50.4	0.77	260	50.8	247	48.2	0.453
CCB	312	58.6	357	55.1	0.238	296	58	305	59.7	0.611
Antiplatelet	233	43.7	325	50.1	0.03	225	44	241	47.1	0.347
Anticoagulant	73	13.7	149	23	<0.0001	65	12.7	103	20.1	0.002
Aetiology					0.614					0.308
None	415	77.7	499	76.9		402	78.5	386	75.4
Ischaemic disease	48	9	69	10.6		42	8.2	56	10.9
Non‐ischaemic disease	71	13.3	81	12.5		68	13.3	70	13.7
High degree/complete AVB	151	28.3	39	6	<0.0001	145	28.3	30	5.9	<0.0001
LBBB/RBBB					0.02					0.001
None	427	80	575	88.6		412	80.5	456	89.1
LBBB	24	4.5	17	2.6		77	15	41	8
RBBB	83	15.5	57	8.8		23	4.5	15	2.9
RV lead location					0.025					0.659
Apex	398	74.5	520	80.1		388	75.8	395	77.1
Septum	136	25.5	129	19.9		124	24.2	117	22.9
Chambers					0.029					1
Single‐chamber	231	43.3	284	43.8		215	42	215	42
Dual‐chamber	303	56.7	365	56.2		297	58	297	58
Atrial fibrillation	157	29.4	142	21.9	0.004	147	28.7	103	20.1	0.002
Heart valve surgery/PCI					0.001					0.002
None	475	89	572	88.2		455	88.9	449	87.7
Heart valve surgery	33	6.2	19	2.9		33	6.4	17	3.3
PCI	26	4.9	58	8.9		24	4.7	46	9
HTN	292	54.7	317	48.8	0.047	270	52.7	240	46.9	0.07
DM	93	17.4	95	14.6	0.202	78	15.2	74	14.5	0.792

ACEI, angiotensin‐converting enzyme inhibitor; ARB, angiotensin receptor blocker; AVB, atrioventricular block; CCB, calcium channel blocker; DM, diabetes mellitus; HTN, hypertension; LBBB, left bundle branch block; PCI, percutaneous coronary intervention; RBBB, right bundle branch block; RV, right ventricular; VPD, ventricular pacing condition.

Table 1B

Baseline demographic characteristics of study patients (continuous variables)

Continuous variables	Unmatched					Propensity score matched
	VPD		Non‐VPD		P value	VPD		Non‐VPD		P value
	Mean	SD	Mean	SD		Mean	SD	Mean	SD
Age (years)	69.2	13.8	68.7	11	0.49	68.9	13.8	65.6	10.1	<0.0001
Duration since diagnosis (months)	3.2	5.2	4.3	6.5	0.001	3.1	5.1	4.2	5.9	0.003
Follow‐up duration (months)	20.7	2.562	21.4	2.3	0.8308	20.6	3.1	21.5	2.8	0.698
Total heart beats by Holter	81 546.9	6680.8	83 654.6	36 119.2	0.184	81 556.9	6760.2	81 942.3	8162	0.411
Mean HR by Holter	54.7	6.5	57.1	9.9	<0.0001	54.7	6.3	56.6	9	<0.0001
HRmax by Holter (b.p.m.)	101.1	12.6	106.4	18.6	<0.0001	101.1	12.7	105.6	17.1	<0.0001
HRmin by Holter (b.p.m.)	35.2	4.5	36	5.3	0.005	35.3	4.5	35.9	5.4	0.041
The longest RR interval (ms)	5	1.4	6	25	0.327	5	1.3	6.4	28.1	0.278
Numbers of RR interval > 2 s	1866	3418	1304	1275	<0.0001	1810.9	3201.2	1346.4	1336.7	0.003
Baseline HR (b.p.m.)	58	12.4	63.3	14	<0.0001	57.9	12.3	63	13.3	<0.0001
Baseline QRS duration (ms)	108.6	20	101.8	14.7	<0.0001	108.3	19.8	102.3	14.7	<0.0001
Baseline QT interval (ms)	454.1	52.4	431.5	40.4	<0.0001	454.6	52.6	432.5	40.2	<0.0001
Baseline QTc interval (ms)	490.8	99.5	449	76	<0.0001	491.5	99.1	450.6	75.3	<0.0001
Baseline LVEF (%)	63.6	6.7	63.7	6.9	0.833	63.5	6.7	63.4	7.3	0.836
Baseline AORD (mm)	33.3	3	32.8	2.9	0.003	33.3	3.1	32.9	3	0.075
Baseline LAD (mm)	42	5.5	41.9	6.2	0.591	42	5.6	41.7	6.4	0.531
Baseline LVEDD (mm)	48.7	5	47.7	5.6	0.003	48.6	5	48.2	5.9	0.26
Baseline LVESD (mm)	32.2	13.9	31.1	5.6	0.084	32.2	14.2	31.4	6	0.278
Baseline PASP (mmHg)	38.6	8.5	37.1	6.9	0.001	38.5	8.4	36.6	7.1	0.0001

AORD, aortic root diameter; HR, heart rate; LAD, left atrial diameter; LVEDD, left ventricular end‐diastolic diameter; LVEF, left ventricular ejection fraction; LVESD, left ventricular end‐systolic diameter; PASP, pulmonary artery systolic pressure; VPD, ventricular pacing condition.

Univariate analysis

In accordance with conventional viewpoint, high degree AVB or complete (third‐degree) AVB was predominant in RVP dependence group (28.3% vs. 6%, P < 0.0001), indicating high probability of RVP dependence occurrence [odds ratio (OR) = 6.35, 95% confidence interval (CI): 4.19–9.62]. Left bundle branch block (LBBB) rather than right bundle branch block (RBBB) was also related to RVP dependence (OR = 2.08, 95% CI: 1.39–3.11). Patients with AF tended to have RVP dependence (OR = 1.6, 95% CI: 1.2–2.13). Non‐ischaemic heart disease did not influence occurrence of RVP dependence (OR = 1.07, 95% CI: 0.75–1.54). However, ischaemic aetiology of heart disease (OR = 1.30, 95% CI: 1.07–2.18) and heart‐related operation including percutaneous coronary intervention (PCI) (OR = 3.72, 95% CI: 1.73–8.00) and heart surgery (OR = 1.94, 95% CI: 1.17–3.24) were related to RVP dependence. The duration after diagnosis of bradycardia was related to RVP dependence (OR = 1.04, 95% CI: 1.01–1.06). Baseline ECG parameters including mean heart rate (HR), HRmax, HRmin, numbers of RR interval > 2 s, QRS duration, and QTc interval were significantly related to RVP dependence. In baseline Echo parameters, only PASP was relevant to RVP dependence (Table ).

Table 2

Univariate analysis of candidate risk factors for right ventricular pacing dependence prediction

Factors	OR	95% CI		P value
		Lower	Upper
High degree/complete AVB	6.35	4.19	9.62	<0.0001
LBBB/RBBB				0.001
LBBB	2.08	1.39	3.11	<0.0001
RBBB	1.23	0.58	2.60	0.598
Atrial fibrillation	1.60	1.20	2.13	0.001
Aetiology of heart disease				0.31
Ischaemic disease	1.30	1.07	2.18	0.033
Non‐ischaemic disease	1.07	0.75	1.54	0.706
Heart valve surgery/PCI				0.003
Heart valve surgery	1.94	1.17	3.24	0.011
PCI	3.72	1.73	8.00	0.001
Duration after diagnosis	1.04	1.01	1.06	0.004
Mean HR	1.03	1.02	1.05	<0.0001
HRmax	1.02	1.01	1.03	<0.0001
HRmin	1.03	1.00	1.05	0.043
The longest RR interval	1.01	0.98	1.04	0.539
Numbers of RR interval > 2 s	1.00	1.00	1.00	0.008
QRS duration	1.02	1.01	1.03	<0.0001
QTc interval	1.01	1.00	1.01	<0.0001
Baseline LVEF	1.00	0.99	1.02	0.836
Baseline PASP	1.03	1.02	1.05	<0.0001

AVB, atrioventricular block; CI, confidence interval; HR, heart rate; LBBB, left bundle branch block; LVEF, left ventricular ejection fraction; OR, odds ratio; PASP, pulmonary artery systolic pressure; PCI, percutaneous coronary intervention; RBBB, right bundle branch block.

Receiver operating characteristic analysis

After univariate analysis, factors with statistical significance were further processed by ROC analysis to make them transform from continuous variables to categorical variables. Youden's index (sensitivity + specificity − 1) was calculated to identify the optimal cut‐off points. Area under curve (AUC) of HRmax, QRS duration, QTc interval, and PASP were 0.455, 0.574, 0.595, and 0.575, respectively. The optimal cut‐off points were determined as HRmax < 110 b.p.m., QRS duration > 120 ms, QTc interval > 470 ms, and PASP > 40 mmHg (Supporting Information, Figure ).

Multivariate analysis

Finally, all factors (high degree/complete AVB, LBBB/RBBB, AF, aetiology of heart disease, heart valve surgery/PCI, HRmax < 110 b.p.m., QRS duration > 120 ms, QTc interval > 470 ms, and PASP > 40 mmHg) passed univariate analysis and transformed into categorical variables were further analysed by multivariate logistic regression. Ultimately, eight factors (high degree/complete AVB, AF, ischaemic aetiology, PCI history, HRmax < 110 b.p.m., QRS duration > 120 ms, QTc interval > 470 ms, and PASP > 40 mmHg) were verified as independent risk factors of RVP dependence (Figure ). Among them, high degree/complete AVB (OR = 5.71, 95% CI: 3.66–8.85), AF (OR = 2.04, 95% CI: 1.47–2.82), PCI (OR = 2.89, 95% CI: 1.24–6.76), HRmax < 110 b.p.m. (OR = 2.74, 95% CI: 1.58–4.76), QRS duration > 120 ms (OR = 2.46, 95% CI: 1.42–4.27), QTc interval > 470 ms (OR = 2.01, 95% CI: 1.33–3.05), and PASP > 40 mmHg (OR = 1.93, 95% CI: 1.46–2.56) were proved to be potential to predict RVP dependence. Nonetheless, LBBB or RBBB, heart valve surgery history, or non‐ischaemic aetiology of heart disease failed to predict RVP dependence independently (Supporting Information, Table ).

Figure 6

The forest plot of multivariate analysis. AVB, atrioventricular block; CI, confidence interval; LBBB, left bundle branch block; PASP, pulmonary artery systolic pressure; PCI, percutaneous coronary intervention; RBBB, right bundle branch block.

Discussion

This study is characterized with large sample size and retrospective design, providing a distinctive perspective of RVP dependence on permanent pacemaker. In the present study concentrating on pacing dependence, RVP percentage > 80% is proved to significantly influence cardiac function, indicating adverse RVP dependence taking place, within a long‐term follow‐up (mean duration: 24 months). ECG parameters and Echo parameters (high degree/complete AVB, AF, ischaemic aetiology, PCI history, HRmax < 110 b.p.m., QRS duration > 120 ms, QTc interval > 470 ms, and PASP > 40 mmHg) could effectively predict RVP dependence. Our study could provide novel perspective of pacing dependence, and conclusion originated from this study would help to identify patients with high risk of RVP dependence.

Pacemaker is one of the most important life‐saving cardiac implantable electrical devices for patients who are suffering from life‐threatening bradycardia. However, its operating principle actually goes against the physiological condition of electrocardio‐conduction, and this underlies the development of ventricular dyssynchrony and cardiac dysfunction. , , From this study, the percentage of RVP > 80% is verified to significantly correlate with diminished LVEF, enlarged left ventricular diameter, and increased left ventricular volume. These evidences suppose that high proportion of RVP induce ventricular remodelling and cardiac dysfunction. Long‐term RVP could result in diminished LVEF. , Besides, it was reported that long‐term RVP contributed to ventricular enlargement. , However, the specific value of RVP proportion that was related to dampened cardiac function was scarcely seen before.

Herein, baseline QRS duration and QTc interval are both proved to effectively predict RVP dependence. QRS duration reflecting ventricular depolarization has been reported to be negatively related to cardiac function. In our study, QRS duration presented an eminent trend of growth in RVP dependence group. The prolonged QRS duration indicates the electronic remodelling, and this is probably responsible for the ventricular pacing dependence. QTc interval reflecting the process of ventricular depolarization and repolarization could predict cardiac function decline. , Significant difference of QTc interval at baseline was shown between RVP dependence group and non‐RVP dependence group; however, QTc interval shrank during follow‐up. Our results showed that RVP could decrease QTc interval compared with baseline, indicating RVP might decrease repolarization dispersion in both RVP dependence group and non‐RVP dependence group. This seemed to be a contradiction. However, patients at baseline had lower level of HR and more ventricular escape beats, which could profoundly increase QT interval. On the contrary, RVP could increase HR and therefore decrease QT interval. In addition, to our surprise, HRmax rather than HRmin could predict RVP dependence. To some extent, HRmax might represent the HR reserve, and compromised HRmax value should be related to injured HR reserve, resulting in tendency to RVP dependence. It was reported that HR reserve could partly represent the function of cardiac conduction system, especially the sinus node function, , and this offered indirect evidence to support this perspective. Nevertheless, ROC analysis disclosed that HRmax failed to accurately predict RVP dependence (Figure ). In included echocardiographic parameters, only PASP significantly correlated with RVP dependence. It was reported that RVP could lead to pulmonary artery hypertension, indicating close relationship between RVP and pulmonary artery pressure. Patients with higher PASP at baseline might be more susceptible to develop into pulmonary artery hypertension after long term and high proportion of RVP. Therefore, traditional RVP pacemaker might not be suitable for patients with high value of PASP (>40 mmHg) and high probability of RVP dependence at baseline.

Atrial fibrillation tends to be a comorbidity along with abnormal cardiac conduction and bradycardia, and AF patients with long RR interval or slow ventricular rate response are indicated for pacemaker implantation. , However, AF is related to HF, , and AF patients with pacemaker implantation would be more susceptible to compromised cardiac function. Therefore, AF patients might be more vulnerable to adverse effects brought by RVP dependence. Investigation of RVP dependence in AF patients could help to identify someone who is not suitable for traditional RVP treatment in future.

Heart valve surgery might lead to irreversible injury of cardiac conduction system; however, this study suggested heart surgery did not bring additional risk of RVP dependence. Instead, patients with previous PCI history tended to develop into RVP dependence. This should be related to that the blood supply of cardiac conduction system largely depended on coronary artery, especially right coronary artery. Unfortunately, investigation about the relationship between specific branch of coronary artery lesion and RVP dependence was yet to be done, and further study should focus on this issue.

In recent years, renovation of cardiac implantable electronic device brought leadless pacemaker, subcutaneous implantable cardioverter defibrillator, HBP/LBBP, and CRT into clinical application. The progression of pacing technique is so fascinating; however, the effect of cost–benefit should be considered from the perspective of society economy. Screening patients who are not suitable for traditional RVP would improve prognosis of patients' cardiac function, decrease unnecessary medical expenses, and optimize the selection of pacemaker type for a certain patient. On the one hand, a patient is not suitable for leadless pacemaker implantation if he is predicted to have RVP dependence in future, for fear of premature pacemaker battery depletion. On the other hand, traditional RVP pacemaker might not be indicated for a patient who is very probable to develop into RVP dependence and cardiac dysfunction. Accordingly, physiological pacemaker (HBP/LBBP or CRT) should be alternatively chosen for these patients who are vulnerable to RVP dependence and its related cardiac dysfunction. Currently, RVP dependence of permanent pacemaker is yet to be deeply investigated. Lacking of definite instruction for RVP dependence diagnosis, to recommend a specific type of pacemaker to a certain patient which is somewhat fitting with the conception of personalized medicine cannot be attained. Our study providing complementary evidence of RVP dependence could help to better understand the pathogenesis, determine the prognosis, and provide the theoretical basis for individualized pacemaker treatment.

Limitation

Instead of a prospective study cohort included in this study, the specific study design is a kind of retrospective types (nested case–control study). As a result, a thoroughly prospective study concentrated on RVP dependence is highly needed in future. Risk factors attained from this study should be further testified by a validation cohort. Investigation about the relationship between specific branch of coronary artery lesion and RVP dependence was yet to be done, which was worthy of further study.

Conclusions

High RVP percentage (>80%) indicating RVP dependence significantly correlates with poor prognosis of cardiac function. High degree/complete AVB, AF, ischaemic aetiology, PCI history, HRmax < 110 b.p.m., QRS duration > 120 ms, QTc interval > 470 ms, and PASP > 40 mmHg were verified as independent risk factors of RVP dependence. Importantly, patients' ECG and Echo parameters at baseline could efficiently predict the occurrence of RVP dependence.

Conflict of interest

The authors declare that they have no conflict of interest.

Funding

This study was supported by Science and Technology Commission of Shanghai Municipality (Grant Number: Shanghai Sailing Program, 20YF1406500 and Grant Number: 201409004200), National Natural Science Foundation of China (Grant Number: 82100370 and Grant Number: 82170386), and Shanghai Clinical Research Center for Interventional Medicine (Grant Number: 19MC1910300).

Table S1. Baseline serological characteristics of study patients.

Table S2. Multivariate analysis.

Figure S1. The baseline value of LVEF showed no significant difference among patients with different range of RVP proportion.

Figure S2. LAD and LAV changes after long‐term RVP.

Figure S3. LVEDD and LVEDV changes after long‐term RVP.

Figure S4. LVESD and LVESV changes after long‐term RVP.

Figure S5. PASP changes after long‐term RVP.

Figure S6. Transformation of factors from continuous variables to categorical variables by ROC analysis.

Click here for additional data file.