Incidence and Predictors of Right Ventricular Pacing-Induced Cardiomyopathy

Authors, Journal, Affiliations, Type, DOI

• Authors: Shaan Khurshid, Andrew E. Epstein, Ralph J. Verdino, David Lin, Lee R. Goldberg, Francis E. Marchlinski, David S. Frankel
• Journal: Heart Rhythm 2014;11(9):1619–1625
• Affiliations: Cardiovascular Division (Electrophysiology Section) and Heart Failure Section, Hospital of the University of Pennsylvania, Philadelphia
• Type: Retrospective single-centre cohort study
• DOI: 10.1016/j.hrthm.2014.05.040

Overview

This study sought to determine how often right ventricular (RV) pacing produces a cardiomyopathy and which patients are at risk. Of 1,750 consecutive pacemaker implants performed between 2003 and 2012, 277 met the criteria of a normal baseline LVEF, a non-ICD/non-biventricular pacemaker, at least 20% RV pacing and a repeat echocardiogram after at least one year. After excluding 20 patients with an alternative cause of LV dysfunction, 50 of 257 (19.5%) had pacing-induced cardiomyopathy (PICM), with mean LVEF falling from 62.1% to 36.2% over a mean 3.3 years. Male sex (HR 2.15) and wider native QRS duration (HR 1.03 per ms; ≥115 ms 90% specific) were independent predictors, and PICM occurred in 13% of patients pacing only 20–39% of the time — below the conventional 40% threshold. The authors conclude that PICM may be more common than previously reported and that high-risk patients warrant closer surveillance or upfront biventricular pacing.

Keywords

Right ventricular pacing; Pacing-induced cardiomyopathy; Heart failure; Pacing

Key Takeaways

Background

• RV pacing can cause progressive LV systolic dysfunction and heart failure, presumably through electrical and mechanical dyssynchrony; this is termed pacing-induced cardiomyopathy (PICM).
• Only a subset of paced patients develop PICM, and predictors were undefined. Identifying them would allow high-risk patients to receive biventricular pacing upfront, avoiding LV dysfunction and reoperation — but because most patients do not develop PICM, and biventricular pacing is costlier with more complications, it should be reserved for those most likely to benefit.

Methods

Study population

• Consecutive patients undergoing pacemaker implantation at the Hospital of the University of Pennsylvania, January 2003 – December 2012.
• Inclusion: baseline echocardiogram within 6 months showing LVEF ≥50%; single-chamber ventricular or dual-chamber pacemaker; ≥20% RV pacing; repeat echocardiogram ≥1 year after implantation.
• Exclusions: pulse generator change; ICD; single-chamber atrial or biventricular pacemaker.

Measurements

• Baseline demographics, comorbidities and medications were extracted from the electronic record; LV end-diastolic and end-systolic diameters and LVEF were measured on Penn echocardiograms using a standardised, validated protocol.
• Native QRS duration, heart rate and bundle branch block were measured from the ECG closest to implantation, using conducted complexes and standard AHA/ACC/HRS criteria; paced QRS was measured on the ECG immediately after implantation.
• RV lead position was classified from the post-implant chest radiograph: a horizontal line was drawn from the heel of the ventricular lead at the bottom of the right atrium across the RV; below = apical, above = septal.
• Ventricular pacing percentage came from the first interrogation at least 2 weeks after implantation; when unavailable, it was averaged from all ECGs in the first 6 weeks.

Definition of PICM

• PICM = a ≥10% decrease in LVEF with resultant LVEF <50%.
• Alternative causes were actively excluded: myocardial infarction, ischaemia on stress testing, severe valvular disease, atrial arrhythmia with rapid ventricular response, frequent (>20%) ventricular premature depolarisations, and uncontrolled hypertension (≥50% of readings >160/100 mmHg, the Framingham threshold at which lifetime HF risk doubles). Patients with any alternative explanation were excluded.
• For survival analysis, PICM was dated to the first echocardiogram documenting the decline; patients were then followed for nadir LVEF, biventricular upgrade and post-upgrade LVEF. Patients without PICM were censored at their last echocardiogram.

Statistical analysis

• Student t test and Pearson χ² for baseline comparisons; Kaplan–Meier for survival free of PICM; Cox proportional hazards for predictors, with univariate screening followed by multivariate modelling; ROC curve for native QRS duration.

Results

Study population and incidence

• Of 1,750 patients undergoing implantation, 277 (15.8%) met inclusion criteria.
• 207/277 (74.7%) remained with preserved LVEF; 70 (25.3%) developed a cardiomyopathy, of whom 20 were excluded for an alternative cause (myocardial infarction 8, ischaemia 3, severe valvular disease 6, uncontrolled arrhythmia 2, uncontrolled hypertension 1) — leaving 50 patients with PICM (19.5%).
• Incidence by pacing burden: 20–39% 13.0%; 40–59% 16.7%; 60–79% 26.1%; 80–100% 19.8% — P = .7 across groups.
• Mean follow-up was 3.3 years (range 0.08–9.4).

Course of PICM

• Mean LVEF fell from 62.1% to 36.2%; 18/50 (36%) also developed new or worsened mitral regurgitation.
• The shortest time to PICM was 1 month and the longest 8.4 years.
• 14/50 (28.0%) underwent upgrade to a biventricular pacemaker or defibrillator; LVEF improved in all but 2, from a mean 32.1% before to 44.4% after upgrade.

Baseline comparison

• Patients who developed PICM were more often men (72.0% vs 49.8%, P = .005), had a lower baseline LVEF (62.1 ± 8.7% vs 65.3 ± 9.6%, P = .03) and a wider native QRS (107.2 ± 20.8 vs 96.0 ± 15.7 ms, P < .001).
• No significant differences in age, body mass index, coronary disease, atrial fibrillation, hypertension, diabetes, post-surgical heart block, β-blocker or ACE-inhibitor/ARB use, LV dimensions, heart rate, LBBB, RBBB, single-chamber device, apical lead position (70.0% vs 80.2%, P = .1), paced QRS (171.6 ± 17.0 vs 168.1 ± 17.8 ms, P = .2) or pacing percentage (86.9 ± 21.9% vs 84.2 ± 24.4%, P = .5).

Predictors

• Univariate: male sex, lower baseline LVEF and longer native QRS predicted shorter time to PICM.
• Multivariate: male sex HR 2.15 (95% CI 1.14–4.02; P = .02) and native QRS HR 1.03 per 1 ms (95% CI 1.01–1.05; P < .001) remained independent; baseline LVEF lost significance (HR 0.97, 95% CI 0.95–1.00; P = .09).
• ROC analysis: native QRS ≥100 ms gave the best balance (sensitivity 63%, specificity 78%); ≥115 ms was 90% specific for PICM.
• Pacing percentage did not predict PICM (HR 1.00, 95% CI 0.99–1.02; P = .7), and neither did apical lead position (HR 0.62, 95% CI 0.34–1.13; P = .1) or paced QRS (HR 1.01, 95% CI 0.99–1.03; P = .3).

Discussion

• PICM developed in ~20% of patients over a decade, and the authors argue the risk begins well below the accepted 40% pacing threshold, because 13% of patients pacing 20–40% developed PICM.
• Their incidence is higher than Yu et al.'s 9% (1-year follow-up) and consistent with Zhang et al.'s 26% — but Zhang did not exclude alternative aetiologies and diagnosed heart failure clinically rather than by LVEF.
• PICM can present as early as 1 month and as late as 9 years, so risk persists with continued exposure.
• In MOST, pacing >40% tripled heart-failure hospitalisation and incident AF versus <40%, which entrenched 40% as the threshold; the authors counter that MOST's risk also began at 20%, and draw the analogy with >20% VPD burden.
• The absence of a pacing-percentage effect may reflect inadequate representation of low burdens (both groups averaged >80%) or a threshold model — once the ≥20% insult occurs, predisposing factors determine whether cardiomyopathy develops, as the authors previously found for VPD burden.
• Native QRS is interpreted as a marker of substrate vulnerability, consistent with wider QRS predicting death in paced patients, with the non-significant interaction favouring biventricular pacing above 110 ms in an RCT, and with wider VPD QRS predicting VPD-induced cardiomyopathy.
• Male susceptibility parallels other non-ischaemic cardiomyopathies (stress-induced, hypertrophic, myocarditis, dilated), but the mechanism is unknown.
• Lead position was not associated with PICM, consistent with recent randomised comparisons of septal versus apical pacing.

Study limitations (as stated)

• Inclusion bias is possible and acknowledged: because a follow-up echocardiogram was required, patients who developed heart-failure symptoms may have been more likely to be imaged and therefore included, which could overestimate the true incidence; it would not be expected to affect the risk-factor findings.
• Residual misclassification is possible despite exclusion of alternative causes; the improvement in LVEF after upgrade in all but 2 patients supports the diagnosis.

Conclusion

• PICM may be more common than previously reported. Men with wider native QRS, particularly >115 ms, are at increased risk and may be best treated with biventricular pacing upfront — though the effectiveness and cost of that strategy need prospective study — and at minimum warrant closer clinical and echocardiographic follow-up. Risk likely begins below the traditional 40% pacing burden.

Limitations of the document

• Retrospective, single-centre, observational. Of 1,750 screened patients only 277 (15.8%) met criteria; the remaining 1,473 were excluded for missing baseline echocardiography, low LVEF, device type, <20% pacing or absent follow-up imaging. The study cohort is therefore a selected subset of the paced population.
• Inclusion bias is the principal threat to the incidence estimate, and the authors say so: follow-up echocardiography was clinically driven, so patients who became symptomatic were preferentially included. The 19.5% figure is likely an upper bound; the risk-factor associations are less affected.
• No control group and no randomised comparison. The 14 upgrades were not randomised; the improvement after upgrade (32.1% → 44.4%) is supportive but susceptible to regression to the mean and selection.
• The PICM definition is composite and arbitrary (≥10% LVEF fall with resultant LVEF <50%). LVEF measurement variability alone can approach 10 points, and no core laboratory or blinding is reported.
• Small event count: 50 events constrain the multivariable model, and the male-sex confidence interval is wide (1.14–4.02).
• Native QRS was measured manually without reported blinding, and patients with bundle branch block were excluded from the QRS analysis — so the ≥115 ms rule applies only to non-BBB patients.
• Pacing percentage was a short-term proxy: the first interrogation at ≥2 weeks, or averaged ECGs over 6 weeks, was used for chronic exposure.
• Lead position was classified from a two-dimensional chest radiograph, not CT or fluoroscopy.
• Follow-up is uneven (0.08–9.4 years) and censoring at the last echocardiogram may underestimate late events.
• No pre-implant LVEF trajectory: a single baseline echocardiogram cannot exclude subclinical decline already in progress.
• Two internal inconsistencies: the male-sex hazard ratio is reported in the abstract as 2.15 (95% CI 1.17–3.94; P = .01) but in the results and Table 2 as 2.15 (95% CI 1.14–4.02; P = .02); and the latest onset of PICM is 8.4 years in the results but "9 years" in the discussion.
• The study predates conduction-system pacing; there is no comparison with His bundle or LBBAP strategies.

Key Concepts Mentioned

• concepts/Pacing-Induced-Cardiomyopathy — the entity this paper defines, quantifies and risk-stratifies
• concepts/Permanent-Pacing-Indications — pacing-mode selection when ventricular pacing is anticipated
• concepts/PVC-Induced-Cardiomyopathy — a parallel dyssynchrony-mediated cardiomyopathy, with QRS width as a shared risk marker
• entities/Heart-Failure — the clinical consequence
• entities/CRT — biventricular upgrade as treatment

Key Entities Mentioned

• Hospital of the University of Pennsylvania — single study centre
• Mode Selection (MOST) Trial — source of the conventional 40% pacing threshold
• Framingham Heart Study — source of the blood-pressure threshold used to define uncontrolled hypertension
• Yu et al.; Zhang et al. — prior PICM incidence estimates (9% and 26%)

Wiki Pages Updated

• Created: wiki/sources/pacing-induced-cmp-hrs-2014.md
• Created: wiki/concepts/Pacing-Induced-Cardiomyopathy.md
• Updated: wiki/entities/CRT.md
• Updated: wiki/concepts/Permanent-Pacing-Indications.md
• Updated: wiki/concepts/PVC-Induced-Cardiomyopathy.md
• Updated: wiki/entities/Heart-Failure.md
• Updated: wiki/sourceindex.md
• Updated: wiki/wikiindex.md
• Updated: Arrhythmia-index.md
• Updated: log.md