Criteria for Differentiating Left Bundle Branch Pacing and Left Ventricular Septal Pacing: A Systematic Review

Authors, Journal, Affiliations, Type, DOI

• Authors: Kailun Zhu; Linlin Li; Jianghai Liu; Dong Chang (corresponding); Qiang Li (corresponding)
• Journal: Frontiers in Cardiovascular Medicine 2022;9:1006966. Received 29 July 2022; accepted 13 September 2022; published 30 September 2022. Open access (CC BY)
• Affiliations: Department of Cardiology, Xiamen Cardiovascular Hospital of Xiamen University, School of Medicine, Xiamen University, Xiamen, China; School of Medicine, Xiamen University
• Type: Systematic review — PubMed, EMBASE, Web of Science and Cochrane Library searched to August 2022; 15 primary studies included (refs 3–17), all observational. No meta-analysis, no risk-of-bias appraisal, no registered protocol
• DOI: 10.3389/fcvm.2022.1006966
• Funding: Xiamen Municipal Bureau of Science and Technology (No. 3502Z20214ZD1165). The authors declare no conflict of interest
• Why it matters here: this is the only source in the wiki that assembles the entire LBBP-versus-LVSP discrimination literature into one table — every cut-off, every cohort, every sensitivity/specificity pair — and it is the origin of the "personalized criteria rather than a fixed Stim-LVAT" position that the LBBAP concept page now needs to carry

Overview

Between 60% and 90% of implants labelled "left bundle branch pacing" are confirmed to capture the left bundle at the time of implantation — which means up to one third are actually left ventricular septal pacing, capturing only working myocardium near where the Purkinje system exits. This systematic review asks how the two are told apart. Across 15 studies it finds that paced QRS duration does not separate them consistently (some studies show LBBP narrower, some show no difference, one shows LVSP narrower), while stimulus-to-LV activation time (Stim-LVAT) in V5/V6 is shorter in LBBP in every study that measured it — but with cut-offs that move from 70 to 101 ms depending on the population. The review therefore concludes that the output-dependent QRS transition remains the gold standard (while conceding it is often unobtainable), that an LBB potential combined with a short Stim-LVAT predicts capture better than either alone, and that patient-specific criteria — ΔStim-LVAT against His pacing, the V6–V1 interpeak interval, or the native IDT–TCT difference — are needed because a single fixed Stim-LVAT value cannot serve narrow-QRS, LBBB and heart-failure populations at once.

Keywords

left bundle branch pacing; left ventricular septal pacing; QRS complex; electrocardiogram; electrophysiology

Key Takeaways

Abstract

• Background: LBBP preserves LV electrical and mechanical synchrony by directly capturing the left bundle branch; ~60–90% of LBBP is confirmed to capture LBB at implantation, implying up to one third is LVSP; LBB capture is the discriminating variable
• Methods: 15 articles identified by database search to August 2022
• Results: paced QRS duration comparisons between LVSP and LBBP have not been uniformly concluded, but Stim-LVAT was shorter in LBBP than LVSP in all studies; Stim-LVAT determined LBB capture with sensitivity 76–95.2% and specificity 78.8–100%, varying by population
• Conclusion: the output-dependent QRS transition from NS-LBBP to S-LBBP or LVSP is direct evidence of LBB capture; LBB potential combined with short Stim-LVAT predicts capture better; personalized criteria rather than a fixed Stim-LVAT are necessary, especially in LBBB and heart failure

Introduction — why LVSP keeps being mislabelled LBBP

• The techniques are procedurally almost identical: both advance from the RV septum transseptally to the LV septum, and both can produce a relatively narrow paced QRS with the same dynamic morphology change from LBBB to RBBB pattern
• LVSP is the easier operation: it needs no confirmation of LBB capture, no recording of LBB or His potential, no precise fluoroscopic targeting and no extra pacing manoeuvres
• The consequence: 60–90% of LBBP is confirmed to capture the LBB, so "up to one-third of captures from the left septum are actually LVSP"
• Existing strategies report sensitivity 70–100% and specificity up to 100%, but on different populations and against different reference standards — the inconsistency the review sets out to map

Search strategy and outcomes

• PubMed, EMBASE, Web of Science and Cochrane Library; terms combining "left bundle branch pacing" OR "left bundle branch area pacing" with "left ventricular septal pacing"
• Excluded: animal studies, abstracts, reviews, editorials, individual case reports
• 15 articles selected for detailed review (Table 1)

Physiology and practicality of LVSP

• The anatomical argument predates the technique by decades. In 1970 Durrer et al. mapped the total excitation of seven isolated normal human hearts with 870 intramural terminals. Within 5 ms of the LV epicardial/endocardial action potential, three LV endocardial areas activated synchronously — high anterior paraseptal wall below the mitral attachment, central LV septal endocardium, and posterior paraseptal about one third of the way from apex to base — with propagation across them over the next 5–10 ms and fusion by 15–20 ms. Pacing in these first-activated areas should therefore reproduce the intrinsic sequence
• Little et al. (1982) showed in nine open-chest dogs by echocardiography that pacing the left side of the interventricular septum produced the same septal excitation and motion sequence as intrinsic sinus rhythm
• Peschar et al. found by pressure–volume relations that LVSP best maintained normal LV pump function among sinus rhythm, LVSP, RV pacing and various epicardial sites — attributed to physiological electrical propagation
• Mafi-Rad et al. (2016) studied LVSP in sick sinus syndrome with normal cardiac structure: acute haemodynamics were comparable to atrial pacing and superior to RV apical and RV septal pacing, and paced QRS was shorter — 144 ± 20 ms LVSP vs 172 ± 33 ms RVAP vs 165 ± 17 ms RVSP (P = 0.02 and 0.004). At 6 months pacing parameters were stable with no lead-related complications
• In CRT-indication patients, LVSP has been reported as at least as good as biventricular pacing and potentially as His bundle pacing

Physiology and practicality of LBBP

• Anatomy: the LBB arises from the branching portion of the His bundle beneath the non-coronary/right-coronary cusp junction and spreads as a broad ribbon-like structure in the LV septal sub-endocardium. It has a slender left anterior fascicle to the anterior papillary muscle and a thick left posterior fascicle to the posterior papillary muscle; almost all LV septal fibres arise from the posterior fascicle and interlace into a network reaching the inferior third of the left septal endocardium
• Why LBBP is easier than HBP: the abundant interfascicular network means block of one fascicle need not prolong QRS much, and the ribbon-like structure presents a larger target
• Longitudinal dissociation gives LBBP its electrical rationale: the LBB and RBB are separated by an insulated sheath within the His bundle, so most bundle branch block may be intra-Hisian. Narula et al. normalised bundle branch block with distal HBP, shortening the prolonged HV by 20–35 ms; Upadhyay et al. mapped the LBBB LV septum and localised the block to the left-sided His bundle in 72%, with the LBB trunk accounting for the rest
• The founding clinical demonstration (Huang 2017): in dilated cardiomyopathy whose LBBB (QRS 180 ms) was not corrected by HBP, LBBP beyond the block at low output corrected it, and the residual RBBB was eliminated by adjusting AV delay — the low-output advantage over HBP's high-output requirement and loss of capture
• Subsequent work established LBBP for high-degree AV block, sick sinus syndrome and AV node ablation, and as CRT superior to conventional BVP in typical LBBB with low LVEF

LBBP vs LVSP — the physiological trade-off

• LBBP conducts rapidly through the intrinsic His–Purkinje system, accelerating LV lateral wall depolarisation while comparatively delaying RV excitation — producing interventricular dyssynchrony
• LVSP captures only LV septal myocardium, so septal excitation spreads transversely; LV delayed excitation partly overlaps RV delayed excitation, producing more balanced interventricular synchrony that is nonetheless non-physiological
• Short-term effects on cardiac function appear similar in practice, and paced QRS duration has no consistent relationship across studies:
  • Shimeno: NS-LBBP vs LVSP 135 ± 7 vs 137 ± 9 ms (P > 0.05)
  • Curila: 104 (100, 108) vs 103 (100, 107) ms (P > 0.05); Peng: 105.3 ± 15.6 vs 109.2 ± 9.6 ms (P = 0.287)
  • Zhou: LBBP shorter — 104.26 ± 19.00 vs 118.09 ± 23.20 ms (P = 0.032) — but LVSP was faster to implant: 38.13 ± 11.52 vs 53.52 ± 14.39 min (P < 0.001)
  • Curila (second study): LBBP QRS shorter than LVSP when the LVSP lead was in close proximity to the LBB
• Stim-LVAT was shorter in LBBP in every study that measured it — the one exception being Heckman et al. (73 ± 15 vs 81 ± 13 ms, P = 0.138), which nonetheless found by QRS area on ECG and vectorcardiography a small but significant ventricular-synchrony advantage for LBBP
• The synchrony trade-off is measured directly:
  • Curila (ultra-high-frequency ECG): LBBP produces greater interventricular electrical dyssynchrony than LVSP while preserving physiological LV depolarisation
  • Jastrzębski (V6–V1 R-wave peak difference): LVSP has superior interventricular synchrony but longer Stim-LVAT and QRS
  • Chen (coronary sinus electrogram mapping, 27 LBBP vs 16 LVSP): intrinsic activation runs LV lateral → posterior (distal CS ahead of proximal). All LBBP patients retained the intrinsic sequence; all LVSP patients were non-physiological, with the earliest activation shifted from lateral to posterior wall
  • Qian (SPECT mechanical synchrony, 68 bradycardia patients): a constant Stim-LVAT <76 ms together with a recorded LBB potential identified patients with favourable LV mechanical synchrony

How to differentiate LBBP from LVSP

Stim-LVAT

• The criterion: Stim-LVAT measured at high output and at threshold remains shortest and constant, reported as 100% specific for LBB capture
• The transition logic: with decrementing output the lead moves from NS-LBBP to S-LBBP or to LVSP
  • NS → S-LBBP: Stim-LVAT stays shortest and constant if the lead is in the LBB trunk or branches, with an isoelectric interval between stimulus and ventricular electrogram
  • NS → LVSP (lead near but not in the LBB): Stim-LVAT abruptly lengthens by ~10 ms as LBB capture is lost and only septal myocardium remains
• ⚠ But only 41.2% of LBB-area pacing shows this output-dependent QRS transition (Shimeno) — the criterion is specific when present and unavailable in the majority

LBB potential and retrograde His potential

• An LBB potential should be recorded in all patients with a normal left conduction system, but its presence is NOT direct evidence of capture. It is found in ~68–98% of non-LBBB patients undergoing LBBP — and also in 12.5% of LVSP (Chen), where it is interpreted as a far-field potential downstream of the LBB; a higher output is then needed to conduct retrogradely and produce a His potential
• The correct interpretation: an LBB potential indicates only that the lead tip is close to the LBB area
• The stronger form: for non-LBBB patients, the sinus-rhythm His-to-LBB interval should equal the LBBP stimulus-to-retrograde-His interval. At <1.0 V/0.5 ms, a retrograde His and/or anterograde LBB potential recorded from an HBP lead or a transaortic multielectrode catheter is evidence of direct LBB capture — but not every LBBP patient yields a recordable potential

The value of the criteria — cut-offs, and why they do not transport

• The V6–V1 interval has a physiological basis, not just a statistical one. Transition NS → S-LBBP prolongs stimulus-to-late-R in V1 (120.7 ± 16.7 → 138.5 ± 21.5 ms) with no change in Stim-LVAT (77.2 ± 13.6 vs 76.6 ± 14.1 ms, P = 0.36); transition NS → LVSP raises Stim-LVAT by ≥15 ms (78.4 ± 10.8 → 98.4 ± 13.9) while prolonging V1 R-wave peak time only 6.2 ± 6.3 ms. The interval is therefore longest in S-LBBP (62.3 ± 21.4), intermediate in NS-LBBP (41.3 ± 14.0) and shortest in LVSP (26.5 ± 8.6 ms)
• Why fixed cut-offs fail: Stim-LVAT is prolonged by large ventricular size and by LBBB in heart failure, so the same anatomical capture produces different numbers in different patients. The review's answer is within-patient comparison — against His pacing (ΔStim-LVAT), against the native V6 intrinsicoid deflection and transseptal time (IDT − TCT), or against the patient's own LBB potential
• Vijayaraman's demonstration of the principle: in LBBB, Stim-LVAT was 91.7 ± 8.4 ms during HBP, 75.2 ± 8.8 NS-LBBP, 76.9 ± 8.6 S-LBBP and 90.4 ± 9.1 LVSP (P < 0.001) — capture shortens it by ~15 ms relative to His pacing, and the Δ carries the diagnostic information
• An unexplained outlier the review does not account for: Zhou et al. reported Stim-LVAT of 48.70 ± 13.67 ms (LBBP) and 58.70 ± 13.67 ms (LVSP) — "significantly shorter than those in any other studies". A measurement convention difference (QRS onset rather than stimulus) is the likely explanation, but the review simply notes the discrepancy

Conclusion

• The transition from NS-LBBP to S-LBBP or LVSP is the gold standard for confirming LBB capture — but it may be unobtainable because LBB and adjacent myocardium can share a similar capture threshold
• LBB potential combined with a short Stim-LVAT predicts capture better than either alone
• Personalized criteria (ΔStim-LVAT, ΔStim-LVAT%, comparison of paced and native LVAT, or IDT − TCT) are required, particularly in LBBB and heart failure
• The review's flowchart separates patients without LBBB from patients with HF and/or LBBB, assigning different cut-offs to each — the structural expression of its argument that no single value applies to all

Limitations of the document

Authors' own stated limitations

• None are stated. The article has no limitations section, no risk-of-bias assessment and no discussion of the strength of the pooled evidence

Additional limitations evident in the review

• It is a systematic review in search strategy only. There is no PRISMA flow diagram, no registered protocol, no formal quality appraisal, no meta-analysis and no confidence intervals. The abstract's "sensitivity 76–95.2% and specificity 78.8–100%" is a range across heterogeneous studies with different reference standards, not a pooled estimate — and the review does not say so
• ⚠ The same 468-patient cohort appears twice, as two "studies". Jastrzębski refs 6 (Heart Rhythm 2021) and 9 (Europace 2022) both list n = 468 and 124 (26%) with LBB capture, and Table 1 gives them different values for the same measurements — S-LBBP QRS 175.8 ± 26.5 vs 144.5 ± 24.4 ms, and NS-LBBP Stim-LVAT 74.7 ± 12.0 vs 78.4 ± 10.8 ms. So the table's "15 studies" includes one centre's registry counted twice, and the two entries are internally inconsistent about the same patients
• "Stim-LVAT was shorter in LBBP than LVSP in all studies" is contradicted by the review's own table. Heckman et al. report 73 ± 15 vs 81 ± 13 ms, P = 0.138 — not significant — which the body text acknowledges ("only one study found no significant difference") while the abstract and results summary do not
• Table 1 has missing cells that weaken it as a reference. Wu et al.'s patient number is recorded as "None" although the text and the primary paper give 30; several rows record Stim-LVAT or QRS as "None" when the cited study reports the value elsewhere
• The reference standard differs between included studies and is never reconciled. Some define capture by QRS transition, some by LBB potential, some by Stim-LVAT shortening, some by direct electrophysiological recording. A criterion's sensitivity against a surrogate reference standard is not the same quantity as its sensitivity against direct evidence, so the ranges quoted are not commensurable
• The "gold standard" label is in tension with the review's own data. The conclusion calls QRS transition the gold standard while reporting that it is demonstrable in as few as 41.2% of cases — and the review offers no protocol for the majority in whom it is absent, which is precisely the population in which a discrimination criterion is needed
• No clinical outcomes. The review never asks whether correctly distinguishing LBBP from LVSP changes pacing parameters, remodelling, heart failure events or mortality — so the practical value of the entire diagnostic exercise is assumed rather than demonstrated
• Selection and reporting constraints: English-language databases only, no grey literature, publication cut-off August 2022, and a literature dominated by a small number of high-volume groups (Wenzhou, Maastricht, Ningbo, Xiamen). The included studies are all observational, mostly single-centre
• The cut-offs were derived post hoc in the primary studies and are presented here as if validated; none has prospective external validation in an unselected population. The review's own conclusion — use personalized criteria — follows from this, but it does not propose a decision rule an operator can apply when the transition is absent and no LBB potential is recordable
• The LVSP evidence base is largely preclinical and small. Durrer (7 hearts), Little (9 dogs), Peschar (animal pressure–volume), Mafi-Rad (small SSS cohort) establish plausibility, not clinical equivalence; the claim that LVSP is "at least as good as BVP and potentially HBP" in CRT rests on a single small study
• Rating rationale: rated high. This is the most complete synthesis of the LBBP/LVSP discrimination literature available, its table is directly usable, and its central argument — that fixed Stim-LVAT cut-offs do not transport across populations — is well supported by the spread it documents. It is not rated very high because it is a secondary source with no quality appraisal, no meta-analysis, no confidence intervals, a duplicated cohort, an internal contradiction about the "all studies" claim, and no outcome data

Key Concepts Mentioned

• concepts/Left-Bundle-Branch-Area-Pacing — the technique and its capture-confirmation algorithm; this review is the map of that literature
• concepts/Conduction-System-Pacing — the umbrella category
• Stim-LVAT — stimulus artefact to LV activation time in V5/V6; the review's central discriminating measurement
• LVSP (left ventricular septal pacing) — myocardial-only capture on the left side of the septum; the "look-alike" that 10–40% of LBBP cohorts actually represent
• Selective vs non-selective LBBP — S-LBBP has no local myocardial capture and an isoelectric stimulus–QRS interval; NS-LBBP captures both
• LBB potential — near-field conduction-system signal indicating proximity, not capture
• Retrograde His potential / anterograde left conduction system potential — direct electrophysiological evidence of capture
• ΔStim-LVAT and ΔStim-LVAT% — the within-patient comparison against His bundle pacing that the review recommends over fixed values
• V6–V1 interpeak interval — RV-to-LV activation difference; longest in S-LBBP, shortest in LVSP
• IDT and TCT — native V6 intrinsicoid deflection time and transseptal conduction time; their difference is the personalised LBBB comparator
• Longitudinal dissociation — the intra-His separation of LBB and RBB that allows distal pacing to bypass block
• Interventricular dyssynchrony — the cost of LBBP's rapid LV activation, and LVSP's apparent advantage

Key Entities Mentioned

• concepts/Left-Bundle-Branch-Area-Pacing — primary concept page
• Xiamen Cardiovascular Hospital of Xiamen University — the investigating and corresponding institution
• Durrer, Little, Peschar, Mafi-Rad — the physiological lineage establishing LVSP's plausibility
• Narula; El-Sherif; Upadhyay — longitudinal dissociation and the localisation of LBBB block
• Huang, Wu, Chen, Su, Vijayaraman, Jastrzębski, Curila, Shimeno, Qian, Zhou — the primary investigators whose cohorts supply Table 1
• Frontiers Media / Frontiers in Cardiovascular Medicine — publisher

Wiki Pages Updated

• Created: wiki/sources/lbbp-vs-lvsp-criteria-fcm-2022.md
• Updated: wiki/concepts/Left-Bundle-Branch-Area-Pacing.md
• Updated: wiki/concepts/Conduction-System-Pacing.md
• Updated: wiki/sourceindex.md
• Updated: wiki/wikiindex.md