The electrocardiographic characteristics of septal flash in patients with left bundle branch block

Aims

In patients with systolic heart failure and left bundle branch block (LBBB), septal flash (SF) movement has been described by echocardiography. We evaluated the prevalence of SF in LBBB and non-LBBB patients and evaluated whether specific electrocardiographic (ECG) characteristics within LBBB are associated with the presence of SF on echocardiography.

Methods and results

One hundred and four patients with probable LBBB on standard 12-lead ECG were selected, 40 patients with non-LBBB served as controls. Left bundle branch block and non-LBBB were defined, according to the most recent guidelines. The presence of SF was assessed by echocardiography. Strict LBBB criteria were met in 93.3% of the patients. Septal flash was present in 45.2% of LBBB patients and was not present in non-LBBB patients. This was more prevalent in patients without anterior ischaemic cardiomyopathy (ICMP) compared with those with anterior ICMP (P = 0.008). The duration of QRS was longer in SF patients compared with that of non-SF patients (P < 0.05). The presence of a mid-QRS notching in more than two consecutive leads was a good predictor for the presence of SF (P = 0.01), and when combined with an absent R-wave in lead V1, the presence of SF is very likely (P = 0.001).

Conclusion

Our data show that SF is present in 45.2% of LBBB patients, whereas it was absent in patients with non-LBBB. Patients with SF fulfilled more LBBB criteria compared with LBBB patients without SF. Our findings raise the provocative question of whether the presence of SF identifies patients with ‘true LBBB’ and whether this echocardiographic finding might be considered as a selection parameter in cardiac resynchronization therapy.

Introduction

Left bundle branch block (LBBB) can be associated with a dyssynchronous contraction of the left ventricle (LV), and these haemodynamic changes negatively affect outcome in heart failure (HF).^1,2

The main purpose of cardiac resynchronization therapy (CRT) is to restore LBBB-induced dyssynchrony. Randomized trials have consistently shown significant improvements of morbidity and mortality in patients treated with CRT.³ Therefore, correct diagnosis of LBBB is crucial for selecting patients who most likely benefit from CRT. In 2009, the American Heart Association together with the American College of Cardiology Foundation and the Heart Rhythm Society published the recommendations for the standardization and interpretation of the electrocardiogram (ECG). In this document, several ECG criteria are proposed to identify LBBB.⁴

Echocardiographic manifestations of LBBB have been described for over 40 years. In experimental LBBB (ablation of the proximal left bundle branch in canine models)⁵ and in humans with LBBB, a typical septal motion called septal beaking or septal flash (SF) has been described by echocardiography.⁶ This myocardial dyssynchronous movement can be identified by echocardiography using visual ‘eyeballing’, anatomical M-mode, or strain imaging. In SF, the early septal contraction (right to left motion) occurs before aortic valve opening and is followed by late contraction of the lateral wall of the LV, which in turn causes a left to right motion of the septum. This dyssynchrony creates a highly inefficient LV pump function, as the brief septal contraction does not contribute to the ejection of blood.⁵ The SF movement is amenable for CRT therapy as recent studies have shown that the presence of SF is highly predictive of CRT response in HF patients with LBBB.^6,7 Nevertheless, this septal motion appears not to be always present in LBBB HF patients, and this might be one of the reasons of non-response in CRT patients, despite LBBB.⁶

We therefore examined the prevalence of SF in patients with typical LBBB, left anterior hemiblock (LAHB), left posterior hemiblock (LPHB), and right bundle branch block (RBBB). Furthermore, we investigated whether the presence of SF was associated with a specific ECG-pattern within LBBB.

Methods

Patient selection

The study patients were consecutively selected from a clinical, digitally stored ECG database at the Department of Cardiology and consisted of 125 randomly chosen subjects, with diagnosis of LBBB, judged by the treating physician. A different data set of 40 non-LBBB patients (LAHB, LPHB, or RBBB) was used as a control group, and another data set of 40 consecutive LBBB patients was included for confirmation.

Exclusion criteria were unavailability of echocardiography at the moment of LBBB diagnosis, continuous right ventricular pacing during echocardiography, or poor quality ECG. In the case of device therapy (pacing/CRT), ECGs were obtained from the episode prior to CRT implantation or, in the presence of a pacemaker, ECGs were recorded in non-pacing mode.

Patients were classified as ischaemic cardiomyopathy (ICMP) if they had a history of myocardial infarction and revascularization or showed angiographic evidence of significant coronary single-vessel disease.

All patients gave written informed consent. This retrospective study was approved by the Ethics Committee of the Ghent University Hospital.

Electrocardiographic parameters and analysis

Baseline standard supine 12-lead ECGs were recorded at a paper speed of 25 mm/s and a calibration of 10 mm/mV, with a standard General Electric Healthcare device type MACC 5500 or with a Schiller Cardiopulmonary Diagnostics AT 104 device. All measurements of PR interval, QRS width, and QTc duration were taken from the automated report of the ECG device (GE software version 2.37 or Schiller AT 104 version 2.13 or Schiller AT 104 version 2.51).

Left bundle branch block was defined according to the standard American Heart Association/American College of Cardiology Foundation/Heart Rhythm Society (AHA/ACCF/HRS) criteria:⁴ (i) QRS duration >120 ms in adults, with the presence of (ii) a broad and/or slurred R-wave in the lateral leads, (iii) deep S-wave in the anteroseptal leads, (iv) the absence of Q-waves in the leads V5, V6, and lead I; and (v) T-waves are usually opposite in direction to the QRS, positive concordance may be as well normal in LBBB, negative concordance (negative T-wave in leads with negative QRS) is not.

Echocardiography

Patients were imaged in left lateral decubitus with a commercially available system (GE Healthcare Ultrasound Vivid 7, Vingmed, Horten, Norway; GE Healthcare Ultrasound Vivid E9, Vingmed; Philips Ultrasound iE 33, Best, The Netherlands) in conventional parasternal and apical views (AP). Standard two-dimensional cine loops were recorded in all patients, and analysis was performed off-line using EchoPAC version 7.1.13 in the GE scanning system and Xcelera viewer R3 version 3.3.1 2013 in the Philips scanning system. The LV ejection fraction (EF) was judged as normal (≥55%), mildly reduced (45–54%), moderately reduced (30–44%), and severely reduced (<30%).

The echocardiographic examinations were performed during the pacing-off modus or prior to PM/CRT implantation.

Assessment of septal flash

To determine the presence and extent of SF, one independent echocardiography expert (F.T.), blinded to the ECGs, reviewed all echocardiography images. The presence of SF was defined as reported previously and assessed based on: (i) visual ‘eyeballing’ on parasternal short axis (PSSAX), parasternal long axis (PSLAX), or AP views; (ii) two-dimensional anatomic M-mode in the PSLAX or PSSAX, or using the off-line automated M-mode to allow adjustments/perpendicularity of the cursor; and (iii) using speckle tracking strain analysis (off-line, using EchoPAC version 7.1.13) in the AP views.^6,8,9 Septal flash was scored as absent, moderate, or manifest based on septal excursion amplitude. The degree of the inward SF excursion was assessed by eyeballing using the M-mode and by assessing the extent of the early negative peak strain.

As a control group, the presence of SF was evaluated in 40 non-LBBB patients (LAHB, LPHB, and RBBB). The diagnosis of LAHB, LPHB, and RBBB was based on the surface ECG, according to the AHA/ACCF/HRS criteria.⁴

For the second and separate validation set of 40 LBBB patients and to assess the interobserver variability of the echo parameters, two experienced echocardiographers (T.D.B. and F.T.) were blinded for the retrospective off-line analysis.

Interobserver variability

The interobserver variability for ECG characteristics was evaluated in 40 consecutive patients with LBBB by two independent observers (except for the QRSd and QTc as these parameters were measured automatically). The ECG analysis was performed by two cardiologists familiar with ECG reading and scoring (B.C. and J.D.P.). The interobserver variability for echocardiographic assessment of SF was evaluated in this same population of 40 LBBB patients by two independent echocardiography experts (T.D.B. and F.T.).

Statistical analysis

Statistical analysis was performed using SPSS software package Version 21 (IBM, Chicago, IL, USA). Continuous variables were presented as mean ± standard deviation (SD). Where appropriate, continuous variables were assessed using Student's t-test. To compare means of two variables, we used Student's t-test and the Mann–Whitney U-test. Categorical variables were expressed as total number (percentages) and compared between groups using the Fisher's exact test. The Kruskal–Wallis test was used for comparing categorical with continuous variables.

Multivariate analysis was used to test whether several combinations of ECG characteristics predicted SF. Binary logistic regression tests were used to assess whether some selected combinations of ECG characteristics predicted SF, independent of contributions of constituents of combinations and other selected ECG characteristics. Interobserver variability was assessed using intraclass correlation coefficients (ICCs).

P-values of less than 0.05 were considered statistically significant.

Results

Of 125 patients, 14 patients were excluded because of unavailable echo images. Another four patients had continuous right ventricular or biventricular pacing during echocardiography and were excluded. In two patients, ECG tracings were inappropriate and one patient refused to participate. Of the included patients, 66.3% were males and 42.3% had ICMP. The mean age at the time of ECG was 70.4 years (± 12.7 years). Of all patients, 10 (9.6%) had atrial fibrillation. Overall, 31.7% of the patients had a pacing device at the time of ECG recording: 8 (24.2%) CRT (P or D) device and 25 (75.8%) pacemaker or implantable cardioverter defibrillator (Table 1).

One hundred and fifty patients were initially considered as negative controls, but 40 patients were finally selected on the basis of the presence of true isolated LAHB (n = 14), LPHB (n = 6), or RBBB (n = 20).

Left bundle branch block and septal flash

According to the criteria of AHA/ACCF/HRS guidelines, 97 (93.3%) of the 104 patients were identified as having LBBB. Seven patients were initially misclassified as LBBB, but none of these patients revealed SF on echocardiography. Of all 104 patients, SF was detected in 47 (45.2%) patients. By eyeballing, 97.9% of the SF patients were detected; anatomic M-mode and speckle tracking strain analyses identified 25 (53.2%) and 33 (70.2%) patients with SF, respectively (Table 2). The ICC showed a good agreement between the three echocardiographic methods to detect SF (Cronbach's alpha coefficient = 0.94).

There were no significant gender or age differences in the LBBB population between subpopulations, as shown in Table 1. Ischaemic cardiomyopathy was less likely to be present in patients with SF and LBBB [15 (31.9%)], whereas ICMP was present in 26 (52%) patients without SF and LBBB (P = 0.064) (Table 3). When categorizing for location of ischaemia, 42 out of 47 (89.4%) SF patients did not have ischaemia in the anterior region (no left main or left anterior descending artery infarction or stenosis). Compared with the SF-negative patients (50), there were 34% with anterior ischaemia or stenosis (P = 0.008).

Septal flash and right bundle branch block, left posterior hemiblock, or left anterior hemiblock

No SF was detected, with either detection method (eyeballing, M-mode, strain analysis), in 40 patients with true, isolated LAHB, LPHB, or RBBB. Eyeballing and M-mode could be assessed in all 40 patients, whereas strain analysis was feasible in 82.5% of the patients.

Septal flash and QRS morphology in left bundle branch block-electrocardiography

The SF group had significantly longer QRS duration¹⁰ with a mean of 149 ms (±12 ms), compared with 142 ms (±15 ms) in patients without SF (P = 0.031) (Table 3). There was no difference between the SF-positive and SF-negative groups, considering broad slurred R-wave (P = 0.495) (Table 4). Absent R-wave^3,4,10,13,14 (or R-wave <1 mm for a scale of 10 mm/mV) in lead V1 was categorized as the presence of QS wave in V1. For the SF population, 36 (76.6%) had a QS in lead V1 (P = 0.056). Notching in R-wave^10–12 in leads I and aVL was present in 45 (95.7%) SF patients (P = 0.093), and a notch in leads V5 and V6 was present in 41 (87.2%) SF patients and absent in 16 (32%) non-SF patients (P = 0.03). In all LBBBs, Q-waves in leads V5, V6, and I are absent.^4,10 A Q-wave in aVL did not differ significantly between SF and no SF (P = 0.093). Finally, considering T-wave concordance,⁴ there was no correlation between T-wave inversions and presence/absence of SF (P = 0.181). All considered ECG characteristics are listed in Table 4. Of all single measurements, the only significant ECG characteristic between SF and non-SF is the presence of notch in the R-wave of leads V5 and V6 (P = 0.03). For some of the other ECG characteristics, only a trend but no statistical significance was reached.

Combining electrocardiographic characteristics

When combining different ECG patterns in LBBB, for instance, notched R-wave in more than two leads, there is a highly significant difference between the SF group and the group without SF (Table 4). Furthermore, combining QRS notching in all lateral leads and inferior leads, SF presence is very likely (P = 0.01). When combining the morphology of V1 (QS or rS) and the presence of notching within two subsequent leads (anterolateral V5, V6 or high lateral I, aVL, or inferior II, III, aVF), the likelihood of SF presence is high (P = 0.008, 0.011, and 0.014, respectively) (Table 4).

Multivariate analysis showed no statistical significance when using several combined ECG characteristics to predict SF.

Interobserver variability

The interobserver variability for assessing the ECG characteristics revealed an ICC of 0.82 ± 0.12. The interobserver variability of echocardiographic assessment of SF revealed an ICC of 0.79 (eyeballing), 0.79 (M-mode), and 0.83 (speckle tracking strain analysis).

Second validation set

In Supplementary material online, Tables S1 and S2, a second independent cohort of 40 consecutive and randomly chosen LBBB subjects is shown. The characteristics of the patients are shown in Supplementary material online, Table S1. There is no difference with regard to New York Heart Association (NYHA) class, EF, age, presence of myocardial infarction, and so on between SF and non-SF patients.

In Supplementary material online, Table S2, there is a confirmatory trend for the association of SF with QRS duration (P = 0.002), presence of QS wave in V1 (P = 0.029), notching in leads I and aVL (P = 0.088), and notching in leads V5–V6 (P = 0.052).

Discussion

In the European and American guidelines on CRT therapy, critical predictors of response are included in the selection of HF patients that may benefit from CRT, such as QRS duration, QRS morphology (LBBB), reduced EF, and variable degrees of NYHA classification.^3,13 The main purpose of CRT is to restore LBBB-induced dyssynchrony, and randomized trials have consistently shown significant reductions of morbidity and mortality in patients treated with CRT.³ On the contrary, HF patients with non-LBBB (RBBB or other intraventricular conduction pathology) do not appear to benefit from CRT to the same extent, do not benefit at all, and CRT may be harmful in some.¹³ The major reason for this observation in non-LBBB HF patients probably relates to the fact that conduction disorders other than LBBB do not induce the typical electrical dyssynchronous SF that is corrected by CRT (‘You cannot fix what is not broken’). Furthermore, although LBBB is a major predictor of CRT response, the diagnostic criteria of LBBB vary considerably among the different studies, which probably explain part of the non-response in CRT patients. The heterogeneous LBBB typing by the studies pleads for an international consensus on the correct diagnosis of LBBB (Table 5)^{3,4,10,13–15} In contrast, because of previous controversial studies,¹⁶ current guidelines do not recommend echocardiographic measures of dyssynchrony to select HF patients for CRT treatment.¹³ Recently, however, Doltra et al.⁶ have shown that in LBBB patients, the presence of SF on echocardiography strongly predicts response (reverse remodelling) in CRT HF patients.

In this report, we investigated the prevalence of SF in LBBB patients and whether SF may be associated with specific ECG morphologies within LBBB. We used rigorous criteria to define ‘true’ LBBB in our cohort. In this regard, in MADIT, a limited number of criteria were used to define LBBB in CRT candidates. For instance, QS in lead V1 was considered an LBBB criterion in MADIT, whereas QRS notching was not (Table 5).³ However, we found a strong correlation between the combined presence of QS in V1 and QRS notching in LBBB and SF patients. Therefore, MADIT-selected LBBB patients might display less SF and hence lower response rates in terms of reverse remodelling, as suggested by Doltra et al.⁶

Septal flash patients in our cohort met all LBBB criteria as defined by the AHA/ACCF/HRS,⁴ and the combined ECG characteristics of LBBB better identified SF patients when compared with the non-SF group. Importantly, SF was not detected in LAHB, LPHB, or RBBB. This raises the provocative question of whether SF might be a major criterion for ‘true LBBB’ or even ‘redefines’ true LBBB or identifies a particular ‘subset’ of LBBB patients. Current criteria that define LBBB on the surface ECG are not sensitive enough to characterize either the location or the extent of specific ventricular delays. Interestingly, electrophysiological (EP) studies have been performed in patients with SF, and these data showed a long transseptal activation time (due to slow muscle-to-muscle conduction within the septum) and functional lines of block in the LV.⁹ These EP characteristics were also described by Auricchio et al.,^9,17 but no echocardiographic data of dyssynchrony (no SF mentioned) were reported in that study. Importantly, in many other patients with LBBB, no such EP characteristics were present and often, several early breakthrough sites in the septum were revealed and were associated with much shorter transseptal activation times.¹⁷ Therefore, it appears that not all LBBBs are created equally, and it is likely that the heterogeneous EP findings in LBBB may relate to the variable anatomy of the left bundle,¹⁸ the site, and/or the extent of conduction block in the left bundle.^10,11,17 Moreover, myocardial substrate modification (e.g. infarction of the septum or left lateral wall) may also impact the presence and extent of SF and hence its CRT response.¹⁷

Regarding the site of conduction block in the left bundle, radiofrequency ablation of the proximal part of the left bundle in dogs results in LBBB, with typical characteristics of SF. Eventually, these LBBB-induced dog hearts develop LV dysfunction, which was restored with CRT.⁵ It has been suggested that a similar pathophysiology of LBBB-induced cardiomyopathy may also occur in humans. These patients (mostly with NICMP) are characterized by a so-called CRT super-response, i.e. complete recovery of EF and reverse remodelling. In our cohort, a considerable number of LBBB patients did not have ICMP or NICMP and had normal EF. It is possible that these LBBB patients with SF may have worse cardiac outcome compared with the LBBB patients without SF, but this remains to be explored in large longitudinal follow-up studies.

On the basis of our findings and aforementioned reports, SF associated with typical LBBB is most likely caused by proximal block of the left bundle branch in humans (with the prerequisite of relatively intact myocardium, particularly the lateral and septal walls).⁹ We therefore hypothesize that SF might be highly prevalent in patients undergoing a transcatheter aortic valve implantation (TAVI), who develop (proximal) LBBB following the procedure, but this remains to be explored. Interestingly, this pure proximal LBBB in TAVI patients may also explain (as an independent risk factor) the worse outcome in these patients because of unfavourable LV haemodynamics associated with SF.¹⁹

The initial R-wave of ≥1 mm in lead V1 was suggested to be a sign of persistent left to right ventricular septum activation and considered as an incomplete LBBB in an EP study.²⁰ However, a septal scar might also have an initial R-wave in the right precordial deflections from unopposed RV free wall activation.²¹ Our data showed a trend of association between absent R-wave in V1 (classified as QS in V1) and the presence of an SF. This fits with the hypothesis that in proximal LBBB, activation of the septum occurs via the right bundle and activates the LV endocardium in a myocyte-to-myocyte activation, as mentioned earlier. However, theoretically, the presence of R-wave in V1 in patients with SF may also reflect fast RBBB conduction and RV activation, which is unopposed by the slower septal depolarization.

In the recent guidelines, LBBB morphology and QRS duration are the two major criteria to select patients for CRT. Randomized trials showed that patients with LBBB and QRS duration of >150 ms have the best CRT response.^6,13 In line with these findings and considering the current guidelines, we found a correlation between QRS duration and the presence of SF: patients with manifest SF had significantly longer QRS durations. Again, this is in line with recent data from Doltra et al., showing best CRT response in patients with SF compared with those with non-SF. One-third of patients fail to respond to CRT, which indicates that current established patient selection criteria might be suboptimal. Considering new criteria to define LBBB, several groups claim for the inclusion of mid-QRS notching or slurring to correctly diagnose true LBBB.^10,11 Mid-QRS notch in the lateral leads (I, aVL, and V6) has been shown to be a good predictor of CRT response.¹² In line with these findings, we found a significant correlation between mid-QRS notch in leads V5–V6 and the presence of SF. When considering more than two subsequent leads with a mid-QRS notch (leads I, II, III, aVF, aVL, V5, or V6), there is a highly significant correlation with SF. Risum et al.¹¹ concluded that the presence of a mid-QRS notch is necessary to distinguish true LBBB from LV hypertrophy, LV dilatation, and incomplete LBBB. Thus, these findings argue for a better and uniform manner to redefine (true) LBBB.

Recently, a scoring method has been proposed for selecting HF patients for CRT treatment.⁷ This system is based on both clinical and ECG-baseline characteristics together with SF on echocardiography. It provides a better predictive power than clinical and ECG characteristics alone. Left bundle branch block and SF are the predominant factors in this scoring system, again pointing to the fact that LBBB patients with SF movement are the best CRT targets.⁷

Conclusions

In conclusion, based on recent clinical and experimental data and the present study, we suggest to include SF as an easy, fast, and specific echocardiograhic marker of ‘true LBBB’ that might be helpful in the prediction of CRT response. However, because of the heterogeneity and the complex (dynamic) nature of the electromechanical myocardial substrate, and because HF patients may improve with CRT despite the absence of SF,⁶ no holy grail exists to detect potential CRT responders. Also, SF has not been evaluated prospectively in randomized trials.

Supplementary material

Supplementary material is available at Europace online.

Conflict of interest: none declared.

References

Author notes