The V6-V1 interpeak interval: a novel criterion for the diagnosis of left bundle branch capture

Abstract Aims We hypothesized that during left bundle branch (LBB) area pacing, the various possible combinations of direct capture/non-capture of the septal myocardium and the LBB result in distinct patterns of right and left ventricular activation. This could translate into different combinations of R-wave peak time (RWPT) in V1 and V6. Consequently, the V6-V1 interpeak interval could differentiate the three types of LBB area capture: non-selective (ns-)LBB, selective (s-)LBB, and left ventricular septal (LVS). Methods and results Patients with unquestionable evidence of LBB capture were included. The V6-V1 interpeak interval, V6RWPT, and V1RWPT were compared between different types of LBB area capture. A total of 468 patients from two centres were screened, with 124 patients (239 electrocardiograms) included in the analysis. Loss of LVS capture resulted in an increase in V1RWPT by ≥15 ms but did not impact V6RWPT. Loss of LBB capture resulted in an increase in V6RWPT by ≥15 ms but only minimally influenced V1RWPT. Consequently, the V6-V1 interval was longest during s-LBB capture (62.3 ± 21.4 ms), intermediate during ns-LBB capture (41.3 ± 14.0 ms), and shortest during LVS capture (26.5 ± 8.6 ms). The optimal value of the V6-V1 interval value for the differentiation between ns-LBB and LVS capture was 33 ms (area under the receiver operating characteristic curve of 84.7%). A specificity of 100% for the diagnosis of LBB capture was obtained with a cut-off value of >44 ms. Conclusion The V6-V1 interpeak interval is a promising novel criterion for the diagnosis of LBB area capture.


Introduction
Electrocardiographic diagnosis of conduction system capture remains one of the challenges of modern physiologic pacing. 1 This is because of the phenomenon of non-selective (ns) capture, that is simultaneous activation of the conduction system and the adjacent myocardium. While sensitive and specific electrocardiogram (ECG) criteria for His-bundle pacing were recently developed and validated, similar criteria for left bundle branch (LBB) pacing are lacking. 2 Three main types of capture are observed during LBB area pacing: non-selective (ns)-LBB, selective (s-)LBB, and left ventricular septal (LVS) myocardial capture. 3 The QRS and V 6 R-wave peak time (RWPT) fixed cut-off criteria proposed arbitrarily for differentiation between ns-LBB and LVS capture are widely used. [4][5][6] We recently demonstrated that a V 6 RWPT <75 ms is nearly 100% specific for LBB capture in patients with a narrow QRS or right bundle branch block (RBBB), as is V6 RWPT < _80 ms in case of left bundle branch block (LBBB)/intraventricular conduction delay/escape rhythm/asystole, but that there is a great deal of overlap in V 6 RWPT between ns-LBB and septal myocardial capture in all instances (with LBB capture possible even with values of >100ms). 7 These criteria are likely influenced not only by the LBB capture/non-capture but also by the degree of the initial latency and the velocity of intraventricular conduction. Furthermore, no QRS-based criteria for differentiation between ns-LBB and s-LBB capture exist. Therefore, a search for new ECG markers for the differentiation of the three main types of LBB area capture is justified.
We intended to develop a new QRS-based criterion that would omit the initial latency that is present after the pacing stimulus and would be measured from an individualized intra-QRS reference time point instead. We hypothesized that the various combinations of direct capture/non-capture of the interventricular septum and capture/ non-capture of the LBB result in several distinct patterns of the right ventricular (RV) and left ventricular (LV) activation ( Figure 1). This could translate into distinct timing of the intrinsicoid deflection in leads V 1 (a surrogate for RV activation delay) and V 6 (reflecting LV activation delay). Consequently, the V 6 -V 1 interpeak interval could differentiate the three types of LBB area capture ( Figure 2).

Methods
The study adhered to the Helsinki Declaration, all patients gave written informed consent for participation in this study and the Institutional Bioethical Committee approved the research protocol.

Aim
Our aim was to analyse the V 6 -V 1 interpeak interval as a potential new ECG criterion for the differentiation of various types of LBB area capture. In addition, the RWPT in V 1 and V 6 were analysed separately during different types of capture to formulate mechanistic hypotheses for the findings.

Population
Consecutive patients who received LBB area pacing device for bradycardia and/or heart failure indications were screened. The LBB implantation procedure was described by us and others in detail elsewhere. [3][4][5][6] To develop LBB capture criteria, we included only patients with direct evidence of LBB capture obtained during dynamic ECG manoeuvres-this served as a diagnostic 'gold standard' in the current study. Such a situation was considered to take place when the paced QRS morphology in lead V 1 was of the QR/ rSR' type and at least one of the three types of dynamic QRS morphology change below was observed during the procedure ( Figure 1): (1) Transition from ns-LBB to s-LBB capture during decrease in pacing output. (2) Transition from ns-LBB to LVS capture during decrease in pacing output. (3) Transition from ns-LBB to s-LBB capture during programmed stimulation and/or burst/incremental pacing. 6 (4) Diagnostic pacing was performed only in unipolar pacing mode to avoid the confounding phenomenon of anodal capture. 1

QRS measurements
Implantation procedures were recorded on the digital electrophysiological system (LabsystemPRO, Boston Scientific, USA). To ensure high precision, the measurements were performed using all 12 surface ECG leads recorded simultaneously, digital calipers, fast sweep speed (200 mm/s), and appropriate signal augmentation. At least three QRS complexes were measured and the values were averaged. In each studied patient, every available paced QRS type (s-LBB, ns-LBB and LVS) and native QRS were measured. The following QRS characteristics were obtained: (1) Global QRS duration, measured from the pacing stimulus or QRS onset (in case of intrinsic rhythm or s-LBB capture) to the final QRS component in any of the 12 ECG leads. (2) V 1 RWPT, measured from the pacing stimulus to the peak of the dominant R wave in lead V 1 . (3) V 6 RWPT, measured from the pacing stimulus to the peak of R wave in lead V 6 . (4) V 6 -V 1 interpeak interval, measured from the R-wave peak in lead V 6 to the R-wave peak in lead V 1 during simultaneous recording of all 12 ECG leads (Figures 1 and 2).
Measurements from the QRS onset were used for the native QRS and s-LBB paced QRS.

Statistical analysis
Continuous variables are presented as means and standard deviations. The distribution of the QRS, V 6 -V 1 interpeak interval and RWPT in lead V 1 and V 6 was estimated by the kernel method. Categorical variables are presented as percentages. Between-group differences were assessed using the Fisher's exact test for 2 Â 2 tables, Student's t-test, or analysis of variance, as appropriate. The performance of binary decision rules was described using sensitivity (SN) and specificity (SP). The performance of the QRS duration and V 6 RWPT in discriminating between ns-HB and RV pacing was assessed using the receiver operating characteristic (ROC) curve. Bland-Altman statistics were then used to assess the inter-observer agreements of QRS duration interval measurements. Statistical analyses were performed using 'R' software (The R Foundation). P-values <0.05 were considered statistically significant.

Population
A total of 468 patients with LBB area pacing from two centres were screened, out of which 124 cases with confirmed diagnosis of LBB capture were included in the further analysis. The remaining cases What's new?
• The V 6 -V 1 interpeak interval >33 ms is a novel criterion that differentiates non-selective left bundle branch (ns-LBB) capture from left ventricular septal (LVS) myocardial capture; it is accurate, easy to measure and reproducible.
• The novel concept to assess the R-wave peak time (RWPT) in V 6 using an intra-QRS reference time point (that is R-wave peak in V 1 rather than the pacing stimulus) circumvents the pacing latency or intraventricular conduction delay related to V 6 RWPT prolongation.
• Loss of LVS myocardial capture results in an increase in V 1 RWPT by > _15 ms but does not impact V 6 RWPT. This indicates that the predominant pathway of right ventricular activation during ns-LBB pacing is direct transseptal route rather than retrograde right bundle branch activation.
• Loss of LBB capture results in an increase in V 6 RWPT by > _15 ms with a specificity of 100%, and results in minimal change in V 1 RWPT.
The V6-V1 interpeak interval criterion were excluded as the LBB capture was confirmed only with some non-direct arbitrary criteria (LBB potential, V 6 RWPT, QRS duration, etc.). The basic clinical and procedure-related characteristics of the included patients are presented in Tables 1 and 2. A total of 239 ECG tracings were analysed: 124 with ns-LBB capture, 69 with transition to s-LBB capture, and 46 with transition to LVS capture.
Categorization of various types of capture during LBB area pacing based on V 6 -V 1 interpeak interval The V 6 -V 1 interpeak interval was longest during s-LBB capture (62.3 ± 21.4 ms), intermediate during ns-LBB capture (41.3 ± 14.0 ms), and shortest during LVS capture (26.5 ± 8.6 ms). The ROC curve for the differential diagnosis of ns-LBB and LV septal capture is presented in Figure 3, and the distribution of V 6 -V 1 interpeak interval values for different LBB area capture types is presented in Figure 4. The diagnostically optimal V 6 -V 1 interpeak interval value for the differentiation of ns-LBB and LVS pacing was 33 ms (SN and SP of 71.8% and 90.0%, respectively). A SP of 100% for the diagnosis of LBB capture was obtained with a cut-off value >44 ms, albeit at the cost of low SN ( Figure 4).

V 1 RWPT behaviour during transition from ns-LBB capture to other types of LBB area capture
The distribution of V 1 RWPT values for different LBB area capture types is presented in Figure 4.
An arbitrary criterion of increase in V 1 RWPT > _15 ms was fulfilled during transition from ns-LBB to s-LBB and from ns-LBB to LVS Transition from non-selective LBB capture to selective LBB capture results in delay of right ventricular (RV) activation due to loss of RV depolarization via direct septal myocardial activation. RV activation proceeds via transseptal conduction from left septal fascicles. Electrocardiographic markers of delayed RV activation are V 1 R-wave peak time prolongation and increase in V 6 -V 1 interval. (B) Transition from non-selective LBB capture to septal does not influence right ventricular activation but delays left ventricular activation due to loss of direct LBB capture. capture in 41/69 (59.4%) and 2/46 (4.3%) patients, respectively. Consequently, the SN and SP of this criterion for the diagnosis of transition to selective LBB capture were 59.4% and 95.6%, respectively. V 6 RWPT behaviour during transition from ns-LBB capture to other types of LBB area capture The distribution of V 6 RWPT values for different LBB area capture types is presented in Figure 4.
An increase in V 6 RWPT > _15 ms was observed in 38/46 (82.6%) cases during transition from ns-LBB to LVS capture and in none during transition to s-LBB capture. Consequently, an arbitrary criterion of increase in V 6 RWPT > _15 ms for the diagnosis of transition to LVS capture had an SN and an SP of 82.6% and 100%, respectively.

Ancillary analyses
Relationship between the V 6 -V 1 interpeak interval and V 6 RWPT The V 6 -V 1 interpeak interval was not related to the paced V 6 RWPT. The distribution of the V 6 RWPT values was similar in patients with diagnostic and non-diagnostic values of the V 6 -V 1 interpeak interval (see Figure 5). Consequently, the V 6 -V 1 interpeak interval was able to correctly re-classify 44/69 (63.8%) misdiagnosed cases (categorized as false negative by a V 6 RWPT value >75 ms). This V 6 RWPT value was found as indicative of lack of LBB capture with SP of 98.6% and SN of 41%. 7 Use of both criteria together, i.e. a combined LBB capture criterion (either V 6 RWPT <75 ms or V 6 -V 1 > _33 ms) had SP of 94.2% and SN of 78.2%.
Analysis of cases where transition from ns-to s-LBB capture did not result in prolongation of V 1 RWPT In 14/69 (20.3%) patients, transition to s-LBB capture did not result in an increase in V 1 RWPT (difference < _ 10 ms). In 4 of these 14 patients (28.7%), LBB potential to QRS interval was > _30 ms. In contrast, only Figure 2 Three types of capture during left bundle branch area pacing (LBB-P) observed in the same patient during pacemaker implantation; leads V 1 and V 6 are recorded simultaneously. During the transition from non-selective LBB-P to selective LBB-P, there is an increase in V 1 R-wave peak time (V 1 RWPT) but not in V 6 RWPT, resulting in a wider V 6 -V 1 interpeak distance. In contrast, during the transition from non-selective LBB-P to left ventricular septal (LVS) myocardial capture there is an increase in V 6 RWPT but not in V 1 RWPT, resulting in a shorter V 6 -V 1 interpeak distance.
Impact of native QRS type on V 1 RWPT, V 6 RWPT, and V 6 -V 1 interpeak interval during non-selective LBB capture The presence of a diseased conduction system (LBBB/NIVCD/ RBBBþfascicular block/asystole/ventricular escape rhythm) was related to longer V 6 RWPT and longer V 1 RWPT (P < 0.001), while the presence of RBBB did not influence ns-LBB paced V 1 RWPT or V 6 RWPT. Importantly, the V 1 RWPT (measured from QRS onset) during intrinsic rhythm with RBBB was the same as during s-LBB pacing (106 ± 11.7 ms vs. 105.0 ± 14.4 ms, respectively, P = 0.60). In contrast to V 1 RWPT and V 6 RWPT, the V 6 -V 1 interpeak interval was not influenced by the native QRS type (P = 0.47). The above relationships are presented in Figure 6.
Stimulus to QRS latency Some degree of latency (i.e. an interval between the pacing spike and QRS onset) was observed in all cases. Substantial variability in latency was observed both during ns-LBB and LVS capture and with significant correlation between latency and V 6 RWPT in both capture types (Supplementary material online, Figures S2 and S3). In contrast, V6-V1 interval did not correlated with the latency interval.  interpeak interval for the diagnosis of non-selective left bundle branch capture vs. left ventricular septal capture; the V 6 -V 1 interpeak interval value of 32.5 ms was found as an optimal cut-off point with a sensitivity of 71.8% and a specificity of 90.0%. AUC, area under the curve.

Interobserver agreement
Interobserver variation in measurement of V 6 RWPT and V 6 -V 1 interpeak interval was assessed by comparing the first observer measurements with 224 consecutive measurements by a second independent observer. For V 6 RWPT, the mean difference [95% confidence interval (CI)] was 0.47 ms (À0.24; 1.17) and the mean limits of agreement were -5.55 and 6.48 ms. For V 6 -V 1 interpeak interval, the mean difference (95% CI) was 0.97 ms (À0.5; 2.44) and the mean limits of agreement were -11.45 and 13.40 ms. The Bland-Altman plots for measurements by the observers are shown in Supplementary material online, Figure S1. Both observers agreed on the presence of LBB capture in 88% of cases.

Discussion
The major finding of this study is that a straightforward electrocardiographic parameter, the V 6 -V 1 interpeak interval, can differentiate between LBB capture and LVS myocardial capture. A V 6 -V 1 interval >33 ms showed an SN of 71.8% and an SP of 90% for confirming ns-LBB capture, while the value >44 ms was 100% specific. Importantly, this criterion was often diagnostic in cases where the V 6 RWPT interval provided false negative results. An additional finding was that an increase in V 1 RWPT and V 6 RWPT of approximately 20 ms each was observed with loss of LVS myocardial capture and loss of LBB capture, respectively (indicating delayed activation of the RV and of the LV, respectively). This sheds light on the physiology of depolarization during LBB area pacing and served to formulate novel criteria for the diagnosis of transition from ns-LBB capture to s-LBB or LVS capture on the basis of simultaneous assessment of sudden increase in V 1 RWPT and V 6 RWPT.

Physiological background of the V 6 -V 1 interpeak interval
During transition from ns-LBB to s-LBB capture, the activation of the RV is delayed (resulting in longer V 1 RWPT) because the direct capture of the interventricular septum is lost, while the activation of the lateral wall of the LV remains unchanged. This leads to an increase in the V 6 -V 1 interpeak interval. In contrast, transition from ns-LBB to LVS capture has no impact on RV activation as, in both situations, RV activation still proceeds from the same pacing site in the interventricular septum. However, LV activation is delayed due to loss of LBB capture, resulting in longer V 6 RWPT. Consequently, during LVS myocardial capture, the V 6 -V 1 interpeak interval decreases (Figures 1 and 2).
These phenomena are behind the observation that during LVS myocardial capture, despite lack of direct engagement of the His-Purkinje system, the paced QRS is relatively narrow and without pronounced features of either the right or left bundle branch block pattern. This is because there are two oppositely directed activation wavefronts, one in the RV and the other in the LV, resulting in shortened total ventricular activation time and a substantial cancellation effect in the lead V 1 (lack of pronounced, broad R'). This activation results in a unique QRS pattern in the precordial leads. Since both the RV free wall and LV free wall activations are delayed, the time to intrinsicoid deflection in both V 1 and V 6 is delayed. Consequently, the R-wave peaks in V 1 and V 6 occur at more or less the same time. Such a unique QRS pattern stands in contrast to the classic RBBB and The V6-V1 interpeak interval criterion LBBB patterns, where there is a substantial delay between the R-wave peaks in leads V 1 and V 6 .

Clinical translation
The arbitrary V 6 RWPT criterion, popularly used for LBB capture diagnosis, suffers from some limitations. Firstly, apart from our very recent study, there are no validation studies and data regarding its SN and SP for LBB capture diagnosis. 7 Secondly, while it is reasonable to believe that the currently used V 6 RWPT cut-off points of 75-80 ms are very specific, they are likely not sensitive enough to diagnose LBB capture in many cases. This is because, despite LBB capture, the V 6 RWPT might be prolonged due to latency, slower propagation via a diseased His-Purkinje conduction system, substantial LV dilatation, or a combination of these factors. The V 6 -V 1 interpeak interval is likely less influenced by these limitations. If there is substantial initial latency or slow conduction through the myocardium, it will affect, to a similar degree, the timing of the activation of the RV and LV. Consequently, the R-wave peak will be delayed in both V 1 and V 6 , and the V 6 -V 1 interpeak interval will not be much affected. In other words, the R-wave peak in V 1 may provide a better reference time point than the pacing stimulus to assess the timing of the R-wave peak in V 6 . Our results show that while V 6 RWPT and V 1 RWPT were strongly influenced by latency and the degree of the disease of the conduction system, as revealed by the type of the native QRS, the V 6 -V 1 interpeak interval was not ( Figure 6, Supplementary material online, Figures S2 and S3).
Our study supports this, showing that in patients with V 6 -V 1 interpeak indicating ns-LBB capture, the V 6 RWPT indicated myocardial capture only in about half the cases. The distribution of V 6 RWPT values in patients with long and short V 6 -V 1 interpeak intervals was not very different-this suggests that use of a combined criterion (V6RWPT <75 ms or V6-V1 > _33 ms) might be diagnostically optimal.
Diagnosis of the type of QRS transition during LBB area pacing Sudden prolongation of the V 6 RWPT is a recognized marker of transition from ns-LBB to LVS capture. 3 However, no data support any specific cut-off value, therefore it is not known how much V 6 RWPT should prolong to be diagnostic of loss of LBB capture. Our data suggest that prolongation by even 15 ms is 100% specific; still, in some patients, loss of LBB capture results in even smaller V 6 RWPT prolongation.
Diagnosis of selective capture during His-bundle pacing is straightforward due to the presence of an evident isoelectric interval after the pacing stimulus. However, during LBB pacing, it is often not possible to rely upon the initial isoelectric interval as a diagnostic feature. This is because, firstly, it is often very short and might be obscured by the post-pacing stimulus artefact, and, secondly, an initial latency interval is also present during LVS myocardial pacing. The criterion of sudden V 1 RWPT prolongation > _15 ms proposed by the current study might be helpful in such situations, as it is very specific for this type of transition during LBB area pacing, especially considering that other QRS morphological features potentially useful for such diagnosis (increase in width of R wave in V 1 , increase in amplitude of S wave in leads I and V 6 , drop in amplitude of R wave in leads I, II, and V 6 ) remain to be validated.

Limitations
A potential confounder in the above hypothesized RV/LV activation patterns and V 6 -V 1 interval behaviour during LBB area pacing is the retrograde activation of the right bundle branch. 8 As the majority of paced patients (64.4% in our study) had RBBB, LBBB or complete atrio-ventricular block, i.e. conditions that likely precluded retrograde conduction via the right bundle branch, our finding may not be applicable to patients with preserved conduction. Nevertheless, the following data support the concept that retrograde activation contributes little to the QRS during ns-LBB capture: (i) patients with narrow QRS complex had minimal difference in V 1 RWPT compared to those with RBBB or LBBB (only 3 ms); (ii) in patients with narrow QRS, transition from ns-LBB capture to selective capture also resulted in significant V 1 RWPT prolongation and obvious broadening of the R' in V 1 ; (iii) during intrinsic rhythm with RBBB, the V 1 RWPT was the same as during s-LBB capture and shortened with ns-LBB pacing (i.e. with additional septal myocardial activation).
These data suggest that the dominant way of RV activation during ns-LBB pacing is intraseptal depolarization from the LBB area pacing lead as proposed in Figure 1. The supero-basal part of the interventricular septum is quite thin, and the pacing helix is partially intraseptal and not completely on the LV side. At the same time, the pathway to the RV via retrograde activation of the conduction system is quite long. Contribution to RV depolarization of retrograde conduction to the right bundle branch probably increases with very proximal LBB lead, close to the His-bundle bifurcation. We have observed that in such cases, characterized by LBB potential to QRS interval of 30-35 ms, transition to s-LBB often results in only minimal V 1 RWPT prolongation.
It is possible that in some cases of LVS capture, there might be nearly instantaneous secondary activation of conduction tissue. Such Figure 5 The distribution of V 6 R-wave peak time (RWPT) values during non-selective left bundle branch capture in patients with V 6 -V 1 interpeak interval > _33 ms (blue line) and V 6 -V 1 interpeak interval <33 ms (red line). In approximately half of the patients with V 6 -V 1 interpeak interval indicating left bundle branch capture (i.e. > _33 ms, blue line), the V 6 RWPT was >75 ms (shaded area), indicating left ventricular septal myocardial capture and leading to misdiagnosis. cases might not be differentiated by the V6-V1 criterion from ns-LBB capture, but from a clinical point of view, they may be considered to be equivalent.
The measurements were performed on an EP recording system at 200 mm/s with digital calipers, and other measurement techniques (e.g. using a compass and 25 mm/s printouts) may be less accurate and borderline cases might be difficult to categorize. However, this is a common limitation to all ECG based duration criteria, which are nevertheless commonly used.

Conclusions
The V 6 -V 1 interpeak interval was found to be valuable for confirming LBB capture and distinguishing it from pure LVS myocardial capture. The diagnostic performance of a V 6 -V 1 interpeak delay >33 ms extends beyond the currently used criterion of V 6 RWPT >75-80 ms.
The impact of ns-LBB, s-LBB, and LVS capture on RWPT in V 1 and V 6 has provided insight into the physiology of LBB pacing and enabled the development of novel criteria for the differentiation of the three main types of LBB area capture.

Supplementary material
Supplementary material is available at Europace online.
Conflict of interest: P.V. has received research and fellowship support as well as speaker and consultant fees from Medtronic. He has also received consultant fees from Abbott, Biotronik, Eaglepoint LLC, and Boston Scientific. M.J. has received consultant fees from Medtronic. K.C. has received consultant fees from Medtronic. P.M. has received consultant fees from Medtronic.

Data availability
The data that support the findings of this study are available from the corresponding author upon reasonable request. [ms] Figure 6 Impact of native QRS type on V 1 R-wave peak time (V 1 RWPT), V 6 RWPT, and V 6 -V 1 interpeak interval in patients with narrow QRS, right bundle branch block (RBBB), and more diseased conduction system, including RBBB with anterior or posterior fascicular block (FB), left bundle branch block (LBBB), non-specific intraventricular conduction disturbance (NIVCD), or ventricular escape rhythm/asystole. In patients with more severely diseased conduction system, the V 1 RWPT and V 6 RWPT were prolonged to a similar degree; consequently, their difference, the V 6 -V 1 interval, remained stable, that is it was not affected by the type of native QRS. Measurements were done during non-selective left bundle branch capture.