Transseptal Activation Delta: An Electrocardiographic Criterion for Left Bundle Branch Capture
Authors, Journal, Affiliations, Type, DOI
- Authors: Nerea Torres Gonzalez, Diego Valdivia Miranda, Julio Salvador Hernandez Afonso (corresponding author), Dacil Díaz Gómez, Emilio Darias Trujillo, Antonio Cabrera León, and Luis Alvarez Acosta
- Affiliations: Arrhythmia and Electrophysiology Unit, Department of Cardiology, Hospital Universitario Nuestra Señora de la Candelaria, Santa Cruz de Tenerife; Universidad de La Laguna; Research Unit of Primary Care, Canary Islands, Spain
- Journal: Heart Rhythm, volume 23, number 11, November 2026, pages e2934–e2941
- Type: Prospective, observational, single-centre diagnostic accuracy study with a separate prospective validation cohort
- DOI: 10.1016/j.hrthm.2026.05.033
- Funding and disclosure: No specific grant; the authors report no conflicts of interest
Overview
This study proposes an intraindividual ECG criterion for left bundle branch capture based on a within-patient comparison rather than an absolute interval: the transseptal activation delta (TAD), defined as the difference in V6 R-wave peak time between monopolar pacing from the lead's proximal ring electrode and from its distal tip. The reasoning is physical — ring pacing recruits septal myocardium only, whereas tip pacing may additionally recruit the left bundle branch, so a capture-dependent extra shortening in V6 activation should appear as a widening delta. In 142 pacing observations from 122 patients, a TAD ≥25 ms identified left bundle branch capture with 78.9% sensitivity, 95.7% specificity and 84.5% accuracy (AUC 0.88), and it added incremental discrimination beyond the established V6-RWPT and V6–V1 criteria. A 30-patient prospective validation cohort reproduced the performance. The main weakness is that capture itself was adjudicated by ECG and pacing manoeuvres rather than invasively, so the criterion is validated against a surrogate of the same kind it is trying to replace.
Keywords
Left bundle branch pacing, conduction system capture, transseptal activation delta, V6 R-wave peak time, V6–V1 interpeak interval, septal pacing, left ventricular septal pacing, physiological pacing
Key Takeaways
Rationale
- Confirming left bundle branch capture rather than septal myocardial pacing remains the central intraprocedural problem in left bundle branch area pacing
- Existing criteria — paced RBBB pattern in V1, abrupt stimulus-to-V6-RWPT shortening with lead advancement or output change, selective/nonselective transitions, the V6–V1 interpeak interval, and absolute V6-RWPT cut-offs — either depend on pacing manoeuvres or on baseline QRS characteristics that are not always available, and several have baseline-dependent cut-offs
- Direct markers such as a left bundle branch potential or a clean selective-capture transition are inconsistently observed, which is why indirect criteria are needed
- The insight is to use the patient as their own control across two pacing sites: ring pacing predominantly reflects septal myocardial conduction, tip pacing may recruit the left bundle branch, and the difference between the two should isolate the conduction-system contribution without reference to any population cut-off
- Under septal-only capture the delta is small because both electrodes use comparable pathways; under left bundle capture the tip shortens disproportionately
Study Design
- Prospective, observational, single-centre study, September 2024 to January 2026
- 126 consecutive patients underwent left bundle branch pacing; 4 were excluded because septal myocardial capture from the ring electrode could not be achieved, leaving 122 patients
- 20 patients contributed measurements both during septal myocardial pacing and after confirmed capture, yielding 142 pacing observations — 95 with left bundle branch capture and 47 with septal pacing
- Capture definition: paced QRS in V1 compatible with conduction-system engagement (QR, Qr, rSR′, rsR′) plus abrupt, reproducible QRS morphology change during output reduction or diagnostic manoeuvres indicating a transition from nonselective LBBP to selective capture or to LV septal pacing. Gradual changes and latency-related variation were not diagnostic. Septal pacing without dynamic evidence of recruitment was classified as non-LBB capture, subcategorised as LVSP or mid-septal/deep-septal myocardial pacing
- Measurement: at the final lead position, V6-RWPT was measured from the pacing artifact to the peak of the R wave in V6 during monopolar pacing from the ring and from the tip at 5 V and 0.5 ms, sweep speed 100 mm/s. TAD = ring V6-RWPT − tip V6-RWPT
- Four lead–sheath systems were used: SelectSecure 3830/C315 HIS (Medtronic, 9 mm tip-to-ring), Solia S60/Selectra 3D (Biotronik, 10 mm), Ingevity Plus/SPSCC 2 (Boston Scientific, 10.7 mm) and Tendril 2088TC/CPS 3D Direct (Abbott, 10 mm)
- Mixed-effects logistic regression with a patient-level random intercept accounted for patients contributing more than one observation
- Incremental-value analysis: a base model containing the optimal V6-RWPT and V6–V1 criteria was compared with an extended model adding TAD ≥25 ms, by likelihood ratio test, Nagelkerke R² and DeLong test
- Prospective validation cohort: 30 consecutive patients with the same implantation and measurement protocol, with the 25 ms threshold prespecified and unmodified
Baseline Characteristics (n=122)
- Mean age 72 ± 9.8 years; 59% male
- Baseline QRS 139 ± 30.1 ms: narrow 27.9%, right bundle branch block 29.5%, left bundle branch block 33.6%, nonspecific intraventricular conduction delay or paced 9%
- Indications: atrioventricular block 67.2%, cardiac resynchronization therapy 18%, sick sinus syndrome 6.6%, atrial fibrillation with slow ventricular response 6.6%
- Leads implanted: Biotronik 57, Abbott 35, Medtronic 21, Boston Scientific 9
Primary Result — Diagnostic Performance of TAD
| Observation | Left bundle capture (n=95) | Septal pacing (n=47) | P |
|---|---|---|---|
| V6-RWPT at the tip — ms | 71.4 ± 12.9 | 90.5 ± 16.2 | <.001 |
| V6-RWPT at the ring — ms | 99.7 ± 15.2 | 104.6 ± 17.3 | .11 |
| TAD — ms | 28.5 ± 9.6 | 15.0 ± 7.5 | <.001 |
| Final paced QRS — ms | 116.4 ± 10.8 | 128.0 ± 16.4 | <.001 |
| V6–V1 interpeak — ms | 45.1 ± 12.5 | 28.9 ± 10.2 | <.001 |
- The structure of the result is the argument: tip pacing differed sharply between capture states, ring pacing did not (P=.11), so the delta isolates the tip's conduction-system recruitment
- ROC analysis: AUC 0.88 (95% CI 0.82–0.94; P<.001)
- TAD ≥25 ms, the Youden-derived cut-off: sensitivity 78.9%, specificity 95.7%, accuracy 84.5%, positive likelihood ratio 18.5, negative likelihood ratio 0.22
- Adjacent thresholds 2 ms either side performed only slightly differently, and 25 ms gave the highest Youden index overall and within baseline-QRS subgroups
Performance by Subgroup and Substrate
| Analysis | AUC (95% CI) | Notes |
|---|---|---|
| Overall (N=142) | 0.88 (0.82–0.94) | — |
| Narrow QRS or RBBB | 0.81 (0.71–0.92) | — |
| LBBB, NICD or paced QRS | 0.95 (0.89–1.00) | 100% specificity claimed in this subgroup |
| Patient-level, final lead position only (n=122) | 0.86 (0.78–0.94) | Confirms the result is not an artefact of multiple observations |
| Prominent S wave in V6 >20% of QRS (55 observations, 38.7%) | — | SN 82.6%, SP 85.7%, accuracy 83.6% |
| RS or rS morphology in V6 (39 observations, 27.5%) | — | SN 82.1%, SP 90.0%, accuracy 84.6% |
- Among 47 septal pacing observations, 32 were mid-septal/deep-septal and 15 LVSP; specificity was 96.9% and 93.3% respectively
- The authors' explanation for the gradient across substrates: in LBBB, proximal interruption of the left conduction system prolongs septal myocardial propagation, magnifying the tip's incremental shortening; in narrow QRS or RBBB the left bundle is intact and deep septal ring pacing may already recruit the Purkinje network early, compressing the delta even when capture is genuine
- No significant effect modification by LVEF (continuous P=.075; categorical P=.75 and P=.34), left ventricular end-diastolic diameter (P=.23) or lead manufacturer (all interaction P>.35)
Reproducibility
- Interobserver agreement on TAD in 30 randomly selected patients: intraclass correlation coefficient 0.93 (95% CI 0.86–0.97)
- Bland–Altman bias 1.13 ms, 95% limits of agreement −9.81 to 12.07 ms
Independent and Incremental Value
- In mixed-effects logistic regression, TAD was associated with capture with an odds ratio of 1.19 per 1 ms (95% CI 1.12–1.26; P<.001); baseline QRS duration and pattern were not independently associated, and adding them to a TAD-only model did not improve fit (P=.09)
- Against the established criteria: in a model containing optimal V6-RWPT and V6–V1, TAD ≥25 ms remained independently associated with capture with an odds ratio of 29.8 (6.1–145.6; P<.001) — the largest effect size among predictors
- Incremental fit: likelihood ratio Δχ² = 27.3 (P<.001); Nagelkerke R² 0.52 → 0.69; AUC 0.863 → 0.939 (DeLong P=.004)
Prospective Validation (n=30)
- 19 of 30 (63.3%) had confirmed capture; mean age 76.9 ± 7.7 years; 66.7% male; 60% narrow QRS or RBBB; 80% preserved LVEF
- TAD ≥25 ms: sensitivity 73.7% (56.3–94.3), specificity 90.9% (58.7–99.8), accuracy 80% (61.4–92.3), PPV 93.3% (68.1–99.8), NPV 66.7% (38.4–88.2)
- Continuous TAD: AUC 0.912 (0.790–1.000)
- TAD was again higher with capture: 28.5 ± 6.9 vs 17.4 ± 5.6 ms (P<.001)
Interpretation
- The authors present TAD as a confirmatory marker with high specificity and high positive predictive value, not as an exclusion test — a value below 25 ms "may not reliably exclude" capture and should be read alongside other criteria
- Its selling point is that it is intraindividual, so it needs no population cut-off and performs consistently across baseline QRS patterns — the limitation that has repeatedly undermined absolute timing criteria
- Its practical limit is equally intraindividual: it requires the ability to pace from both ring and tip, which confines it to implantation and, the authors concede, "may not be readily reproducible during routine postimplant device follow-up with most available generators"
- They position it for intraoperative decisions when baseline-dependent cut-offs or ECG transitions are inconclusive
Limitations of the document
- Single centre, with wide confidence intervals in the 30-patient validation cohort; the authors describe the validation as supportive rather than confirmatory
- The reference standard is itself electrocardiographic. Capture was adjudicated using V1 morphology plus dynamic pacing manoeuvres and transitions, not by invasive His–Purkinje recording. The criterion is therefore validated against a surrogate, and the one study with a direct electrophysiological reference standard is not used here
- No blinding. Capture adjudication and ECG measurement were performed by the implanting operator during the same procedure, so measurement and classification are not independent
- Paired ring-and-tip recordings were available in only 20 of 122 patients; the remaining observations are unpaired, and the patient-level sensitivity analysis exists to compensate but does not restore the pairing
- Selection by measurement feasibility. Four of 126 consecutive patients were excluded because ring pacing did not capture septal myocardium, so the criterion is unavailable in a small but real subset
- Septal substrate was not characterised. Septal scarring, septal thickness, lead depth and lead orientation — each of which would change ring-pacing conduction time and therefore TAD — were not systematically assessed; a more longitudinal lead trajectory could allow both electrodes to engage the conduction system and compress the delta
- Multiple subgroup and interaction analyses without multiplicity control, in a single-centre cohort; the claim of stable performance across ventricular function and lead type rests on non-significant interaction terms in an underpowered analysis
- The 100% specificity claim in the LBBB/NICD/paced subgroup is not accompanied by confidence intervals and rests on a subgroup of the 142 observations
Key Concepts Mentioned
- concepts/LBBAP-Capture-Criteria — where this criterion sits against the established ones, and the page that owns the comparison table
- concepts/Left-Bundle-Branch-Area-Pacing — the parent procedure
- concepts/LBBAP-Implantation-Technique — the intraprocedural context in which a ring-versus-tip measurement is even possible
- concepts/ECG-Conduction-Disturbances — the baseline QRS patterns that determine which criteria are interpretable
- concepts/Electrical-Resynchronization-Metrics — activation time as a resynchronisation measure
- concepts/LBBB-Criteria — identifying the substrate before implantation
Key Entities Mentioned
- Medtronic, Biotronik, Boston Scientific and Abbott — four lead–sheath systems used; their spacing ranges from 9 to 10.7 mm tip-to-ring, which the interaction analysis found did not modify performance. No entity pages exist and none is warranted on the strength of an implant registry
Wiki Pages Updated
wiki/sources/tad-lbbap-hrs-2026.md— created (this page)wiki/concepts/LBBAP-Capture-Criteria.md— new TAD subsection with derivation and validation data; master comparison table row; two contradictions addedwiki/concepts/Left-Bundle-Branch-Area-Pacing.md— capture-criteria summary table row addedwiki/sourceindex.md— index rowwiki/wikiindex.md— LBBAP-Capture-Criteria row revisedlog.md— ingest entry