A novel electrocardiographic predictor of clinical response to cardiac resynchronization therapy

Abstract

Aims

A wide QRS with left bundle branch block pattern is usually required for cardiac resynchronization therapy (CRT) in patients with dilated cardiomyopathy. However, ∼30% of patients do not benefit from CRT. We evaluated whether a detailed analysis of QRS complex can improve prediction of CRT success.

Methods and results

We studied 51 patients (67.3 + 9.5 years, 36 males) with classical indication to CRT. Twelve-lead electrocardiogram (ECG) (50 mm/s, 0.05 mV/mm) was obtained before and 3 months after CRT. The following ECG intervals were measured in leads V1 and V6: (i) total QRS duration; (ii) QRS onset–R wave peak; (iii) R wave peak–S wave peak (RS-V1 and RS-V6); (iv) S wave peak–QRS end; and (v) difference between QR in V6 and in V1. Patients were considered as responder when left ventricular ejection fraction (LVEF) increased by ≥5% and New York Heart Association class by ≥1 after 3 months of CRT. Of ECG intervals, only basal RS-V1 was longer in responders (n = 36) compared with non-responders (52.9 ± 11.8 vs. 44.0 ± 12.6 ms, P = 0.021). Among patients with RS-V1 ≥45 ms 83% responded to CRT vs. 33% of those with RS-V1 < 45 ms (P < 0.001). RS-V1 ≥ 45 ms was independently associated with response to CRT in multivariable analysis (odds ratio 9.8; P = 0.002). A reduction of RS-V1 ≥ 10 ms by CRT also significantly predicted clinical response. RS-V1 shortening correlated with improvement in LVEF (r = −0.45; P < 0.001) and in MS (r = 0.46; P < 0.001).

Conclusion

Our data point out that RS-V1 interval and its changes with CRT may help to identify patients who are most likely to benefit from CRT.

Introduction

Cardiac resynchronization therapy (CRT) has greatly improved clinical outcome in patients with heart failure, who also show a left bundle branch block (LBBB) pattern at the electrocardiogram (ECG).^1–3 However, despite the satisfactory results obtained with CRT, ∼30% of patients receiving this treatment do not achieve significant clinical benefits.^3,4

The reasons for CRT failure are multiple and include the presence of significant comorbidity, the difficulty to achieve an adequate left ventricular (LV) resynchronization, non-optimal device programming and recruitment of patients with highly advanced heart failure to respond to any treatment.^5–8 Evidence from clinical trials, however, suggests that CRT failure is mainly attributable to an unreliable assessment of the presence and degree of mechanical LV desynchronization.^5–8

Although the terms ‘desynchronization’ and ‘resynchronization’ refer to mechanical activation of the heart and are, in fact, usually assessed by analysing LV contraction by ultrasound methods,^6–8 CRT is an electrical therapy, which aims to achieve LV mechanical synchrony acting on electrical conduction abnormality. Electrical activation of the heart, indeed, underpins mechanical activation. Accordingly, the presence of ‘electrical desynchronization’ is mandatory for therapy success.

A QRS duration >120 ms is one of the required criteria for patient eligibility to CRT.⁹ However, several studies have shown that QRS duration does not significantly correlate with mechanical LV desynchronization^10–12; accordingly, it is not reliable enough to identify patients who will really benefit from CRT.^13,14

ECG signal in a precordial lead reaches its maximum amplitude when the vector of electrical current generated in the heart is maximally directed towards its exploring electrode, whereas it has maximum negativity when depolarization current runs towards the opposite direction. Accordingly, the interval between positive (R wave) and negative (S wave) peak of QRS might correlate with desynchronization of LV contraction of two opposite regions of the heart.

In this study, we evaluated whether a detailed analysis of QRS complex on surface ECG can be useful to identify the presence and the degree of mechanical LV desynchronization, and whether this method can predict the response to CRT more reliably than simple QRS duration.

Methods

Study population

We enrolled 51 patients fulfilling the following inclusion criteria: (i) heart failure due to ischaemic or idiopathic cardiomyopathy; (ii) New York Heart Association (NYHA) functional classes II–IV; (iii) LV ejection fraction (LVEF) ≤40%; (iv) QRS with an LBBB morphology and duration ≥120 ms.

Clinically unstable patients, i.e. those admitted for an acute clinical event (e.g. acute coronary syndrome or acute heart failure) or with a change in NYHA class and/or in diuretic dosage in the last month, were excluded from the study. Patients already implanted with a pacemaker were also excluded.

Study protocol

All patients underwent clinical evaluation, an amplified 12-lead surface ECG recording and a two-dimensional (2D) colour Doppler echocardiography at enrolment (before CRT) and after 3 months of CRT. Clinical, ECG, and 2D echo investigations were independently performed by expert cardiologists who were blinded to the results of the other tests.

Cardiac resynchronization therapy was performed at our centre following standard procedures,¹ and biventricular pacemakers were programmed by operators and clinical electrophysiologists who were unaware of our study protocol.

Patients who showed an improvement in functional capacity of at least one NYHA class along with an increase in LVEF ≥5% at 2D echocardiography at the 3-month assessment, compared with baseline, were considered as responders to CRT.

The study protocol was approved by the local ethics committee and complies with the Declaration of Helsinki. An informed consent was obtained from all patients.

Assessment of clinical status

Clinical status of patients at baseline and at 3-month follow-up was assessed by NYHA functional class and by the validated Minnesota Heart Failure Questionnaire.¹⁵

Amplified surface electrocardiogram

An amplified standard 12-lead amplified surface ECG recording, with a paper speed of 50 mm/s and an amplitude of 20 mm/mV, was obtained at baseline and at follow-up. The following time intervals were measured in leads V1 and V6 on three consecutive QRS complexes, and then averaged (Figure 1): (i) total QRS complex duration (QRS_tot); (ii) QRS onset–R wave peak interval (QR-V1 and QR-V6 intervals); (iii) R wave peak–S wave peak interval (RS-V1 and RS-V6 intervals); (iv) S wave peak–QRS end interval (SJ-V1 and SJ-V6 intervals); (v) difference between QR-V6 interval and QR-V1 interval (D_QR_V6-V1).

When R wave was absent in lead V1, QR-V1 interval was considered equal to 0 ms, and RS-V1 was measured from the onset of QRS. Similarly, when S wave was absent in lead V6, SJ-V6 interval has been considered equal to 0 ms.

Electrocardiogram leads V1 and V6 were chosen for measurements as, due to their position (anterior and postero-lateral, respectively), these leads can reflect time difference in a better manner between the anterior and the postero-lateral electrical activation of the heart.

Two-dimensional colour Doppler echocardiogram

Two-dimensional echocardiography was always performed by the same expert cardiologist, who was blind to ECG results and clinical status. Left ventricular end-diastolic volume (LVEDV) and LV end-systolic volume (LVESV) were calculated both in four- and two-chamber apical views using the modified Simpson method, and the two measurements were averaged.¹⁶ Left ventricular ejection fraction was then obtained with the formula: (LVEDV – LVESV)/LVEDV × 100.

Previous data indicated that intra-observer variability of LVEF measurement for our expert echocardiographer was excellent (coefficient of variation 2.8%).

Statistical analysis

Between-group comparisons were carried out by independent t-test. Analysis of variance with a repeated measure design was applied to compare variations in continuous variables at follow-up. In case of global statistical significance between-group and within-group multiple comparisons were carried out by unpaired and paired t-test, respectively, with correction by Bonferroni rule. Categorical variables were compared by Fisher's exact test. Correlations were assessed by Pearson test.

As RS-V1 interval was the strongest ECG variable associated with a favourable response to CRT (see below), we identified the best cut-off value of RS-V1 for predicting therapeutic response to CRT. For this aim increasing RS-V1 values (with 5 ms interval increases) were considered and the best cut-off was chosen as that showing the highest β coefficient value in logistic regression analyses. The independent association of RS-V1, as a dichotomized variable, with response to CRT was also assessed by multivariate logistic regression, in which variables with univariate P value ≤0.1 were included. Data are expressed as mean ± standard deviation or percentage. Statistical significance was fixed at P < 0.05. Statistical analyses were performed using the statistical software SPSS 17.0 (SPSS Italia, Inc.).

Results

General findings

Overall, 51 consecutive patients undergoing CRT at our centre were enrolled in the study. The main clinical characteristics of patients are summarized in Table 1. Mean age of patients was 67.3 ± 9.5 years and 36 were males. Dilated cardiomyopathy was of ischaemic origin in 31 patients (61%) and idiopathic in 20 (39%). Data regarding basal functional capacity, ECG intervals, and echocardiographic parameters are summarized in Table 2.

Basal findings and response to cardiac resynchronization therapy

At 3-month follow-up, 36 patients (71%) were ‘responder’ and 15 patients (29%) were ‘non-responder’ to CRT. There was no significant difference in the main clinical variables at baseline between the two groups (Table 1).

Of the pre-CRT ECG parameters considered for analyses, total QRS duration was similar in the two groups (Table 2), whereas the RS-V1 interval was only significantly different between responders and non-responders, resulting longer in the former group of patients (P = 0.02). No difference in RS-V1 was found between patients with an ischaemic or a non-ischaemic origin of dilated cardiomyopathy (51.3 ± 13.9 vs. 48.8 ± 10.7 ms, P = 0.49).

A pre-CRT RS-V1 interval ≥45 ms discriminated in a better manner between responder and non-responder patients to CRT. Significant clinical response to therapy was observed in 86% of patients with RS-V1 ≥ 45 ms (n = 36), compared with only 33% of those with RS-V1 < 45 ms (n = 15; P < 0.001; Table 3). Patients with RS-V1 ≥ 45 ms showed greater improvement of LVEF, NYHA class, and Minnesota score at follow-up compared with those with RS-V1 < 45 ms (P<0.01 for all the variables; Table 3, Figure 2). The prevalence of a basal RS-V1 ≥ 45 ms was similar in patients with ischaemic and non-ischaemic dilated cardiomyopathy (P = 1.0, Table 3).

RS-V1 ≥ 45 ms was the only independent predictor of response to CRT at multivariate logistic regression analysis (odds ratio 9.8 (95% confidence interval 2.2–42.9); P = 0.002) (Table 4).

Change in electrocardiogram variables and response to cardiac resynchronization therapy

No patient died or was lost at 3-month follow-up. At 3 months, both total QRS duration and RS-V1 interval showed significant shortening compared with basal values in responder patients. Reduction of the time interval, however, appeared significantly greater for RS-V1 than for QRS. Only RS-V1 was indeed significantly shorter at follow-up in responders compared with non-responders (P = 0.03). Neither total QRS duration nor RS-V1 interval significantly narrowed at follow-up, compared with pre-CRT, in non-responder patients (Table 2).

A significant correlation was found between the variations from baseline to follow-up of RS-V1 and those of LVEF (r = −0.45, P = 0.001) and Minnesota score (r = 0.46, P = 0.001). In contrast, no significant correlation was found between changes in QRS duration and both changes in LVEF (r = −0.17, P = 0.22) and Minnesota score (r = 0.19, P = 0.18).

Finally, a reduction ≥10 ms at follow-up of RS-V1 interval, compared with baseline, seemed to predict clinical improvement with CRT in a better manner. Patients with a reduction ≥10 ms in RS-V1 interval following CRT showed, indeed, a greater improvement of both LVEF and of Minnesota score, compared with those with RS-V1 reduction <10 ms (P < 0.001 for both; Table 5). In contrast, there were no significant differences in variations of both LVEF and functional capacity when dividing patients according to several cut-off of changes in total QRS duration (e.g. 10, 15, 20, 25, and 30 ms).

Discussion

Our study shows that, in patients with LBBB, measurement of the interval between the peak of R wave and the peak of S wave in lead V1 (RS-V1 interval) can be helpful in identifying those who are most likely to have significant clinical improvement by CRT. Actually, a basal RS-V1 interval ≥45 ms was particularly predictive of favourable clinical response to treatment, even after adjustment for other confounding variables, suggesting that it can quite reliably indicate the presence of LV dyssynchrony. Accordingly, a reduction of RS-V1 by at least 10 ms by CRT predicted favourable clinical response to treatment, suggesting achievement of effective LV resynchronization. In particular, this last finding might indicate the need for aiming at an RS-V1 reduction as great as possible for optimization of biventricular stimulation.

In contrast to RS-V1, both basal total QRS duration and its change following CRT were of limited value in predicting CRT effect.

Cardiac resynchronization therapy is able to improve cardiac performance by reducing the degree of electrical desynchronization of LV activation, which is the basis for mechanical LV desynchronization. At present, the indication to CRT is based on the presence of a QRS duration ≥120 ms with an LBBB pattern.⁸ However, a favourable response to CRT is absent in ∼30% of patients,^3,4,9 as also confirmed in our study, suggesting that total QRS duration is not an optimal parameter for selection of patients who can benefit from CRT.

A large QRS, indeed, does not necessarily reflect LV desynchronization, despite slow electrical LV activation.^10,12 This has stimulated the research of other parameters, mainly based on echocardiographic methods,^6–8 to identify patients with real LV desynchronization, who could therefore actually benefit from CRT.

In this study, we have evaluated the possibility that detailed analysis of LV electrical activation (i.e. of QRS) might improve identification of patients with LV desynchronization and, therefore, with significant probability to benefit from CRT. Our data suggest that a simple ECG interval, the RS interval in lead V1 (RS-V1) can be particularly helpful to this aim.

This interval is likely helpful because R wave peak in lead V1 indicates when LV electrical activation mainly concerns the anterior wall, whereas S wave peak indicates when LV electrical activation mainly concerns the postero-lateral wall. Accordingly, the interval between the two peaks can indicate difference in time between two opposite regions of the heart and therefore reflect the grade of LV desynchronization.

Accordingly, our study shows that a larger RS-V1 interval is present in patients who will respond to CRT in terms of improvement of both clinical status and LV function, suggesting that this specific ECG parameter can actually be sufficiently reliable in identifying LV desynchronization that can be significantly improved by CRT. Specifically, patients who had a basal RS in V1 lead ≥45 ms showed a positive outcome after CRT in 86% of cases, yielding a sensitivity value that seems better than that obtained by selecting patients on the basis of a QRS >120 ms, according to current scientific evidence.

This is confirmed also by the fact that, in responder patients, RS-V1 shows a more striking reduction after biventricular stimulation compared with total QRS duration. It is worth noting that an RS-V1 reduction of 10 ms or more identified a group of patients, in our study, who showed an increase of both LVEF and functional capacity significantly greater than those of patients with RS-V1 reduction <10 ms, and a significant correlation existed between reduction of RS-V1 after CRT and improvement of both LVEF and functional capacity.

In contrast, the total duration of QRS did not differ between responders and non-responders and no quantitative change in QRS duration during CRT significantly predicted clinical response to CRT, confirming the limited value of total QRS duration for selection of patients who could reliably benefit from CRT.

It should be noted that we used an arbitrary 5% increase in LVEF as a criterion to define a patient as responder to CRT, as such a change seemed quite valuable and beyond intra-observer variability in LVEF measurements. Moreover, the definition of responder in our study was not only based on LVEF but also required valuable clinical improvement, as indicated by a reduction of NYHA class.

Limitations of the study

Some limitations of our study should be acknowledged. First, our data have been obtained in an observational study including only a small number of patients. Therefore, they need to be confirmed in randomized controlled studies with larger populations of patients. More importantly, future studies should also clarify whether programming of biventricular pacing driven by RS-V1 interval reduction can improve clinical response to CRT and whether this variable is independent from any other variable already known to predict CRT effectiveness.

Secondly, it should be acknowledged that, as we included in the study only patients with LBBB, these results may not apply to patients with other QRS morphologies.

Thirdly, although a larger RS-V1 probably indicates a higher degree of LV desynchronization, this was not demonstrated in our study; accordingly, the relation of RS-V1 with mechanical desynchronization needs assessment in future studies.

Finally, it should be noted that a low proportion of our patients were taking ACE-inhibitors and anti-aldosterone drugs as compared with those enrolled in large clinical studies of similar patients. The reasons for that are unclear, as we were not directly involved in clinical care of patients; however, there were no statistical differences between responders and non-responders with regard to drug therapy.

Conflict of interest: none declared.

References