OUP user menu

Electrocardiographic patterns and long-term clinical outcome in cardiac resynchronization therapy

Mads B. Kronborg , Jens C. Nielsen , Peter T. Mortensen
DOI: http://dx.doi.org/10.1093/europace/eup364 216-222 First published online: 14 November 2009

Abstract

Aims The present study aims to identify the predictive value of electrocardiographic (ECG) patterns on long-term clinical and echocardiographic outcome in patients treated with cardiac resynchronization therapy (CRT).

Methods and results Clinical information including a standard 12-lead ECG was collected from patient files in consecutive patients treated with CRT from 1997 to 2007. Symptomatic response was defined as improvement in New York Heart Association class (≥1) and echocardiographic response as improvement in left ventricular ejection fraction of ≥5% absolute. We included 659 patients [median age 66 years, 526 (80%) male]. There was a higher all-cause and cardiac mortality in patients with left bundle branch block (LBBB), prolonged PR interval, right-axis deviation combined with LBBB in the pre-implant ECG, and no QRS reduction after CRT. Patients with right bundle branch block and patients with an intermediate QRS duration (150–200 ms) had a higher chance of symptomatic improvement, and patients with normal PR interval and normal axis in LBBB had a higher chance of echocardiographic improvement.

Conclusion Cardiac resynchronization therapy does not change the predictive value of ECG patterns in heart failure patients with bundle branch block, where LBBB, a prolonged PR, and an abnormal axis in LBBB are signs of a more severe degree of myocardial disease, and therefore a worse outcome. Lack of electrical resynchronization defined as an unchanged or prolonged QRS duration is associated with higher all-cause and cardiac mortality in patients treated with CRT.

  • Cardiac resynchronization therapy
  • Biventricular pacing
  • Electrocardiogram
  • Bundle branch block
  • Heart failure
  • Mortality

Introduction

Cardiac resynchronization therapy (CRT) is a well-established treatment for patients with heart failure, low left ventricular ejection fraction (LVEF), and prolonged QRS duration.1,2 Randomized controlled trials have shown that CRT improves symptoms of heart failure, quality of life, exercise capacity, LVEF, and reduces all-cause mortality.37 Despite its convincing effect, the proportion of patients whose symptoms improve after CRT is only 50–70%.3,4,6 Most patients who have been enrolled in clinical trials of CRT have had left bundle branch block (LBBB). The impact of the baseline electrocardiogram (ECG) and changes in QRS duration on clinical and echocardiographic outcome has mostly been investigated in smaller studies with short follow-up. The present study aims to identify the predictive value of ECG patterns on long-term clinical and echocardiographic outcome in patients treated with CRT.

Methods

Patients

To evaluate the effect of ECG patterns in patients with CRT or CRT-D, we included consecutive patients from the Danish Pacemaker Register,8 treated at Aarhus University Hospital, Skejby, from 1 January 1997 to 31 December 2007. Patients undergoing an upgrade of conventional devices to CRT in the period were also included. Survival data were obtained for the entire cohort from the National Board of Health. The study was approved by the Danish Data Protection Agency and the National Board of Health, and conforms with the principles outlined in the Declaration of Helsinki.

Implantation

Various transvenous delivery systems, LV leads, and CRT devices with or without implantable cardioverter-defibrillator (ICD) from different companies were used. In the period from 1997 to 1999, LV pacing was achieved placing a Slimline® lead (Vitatron) in the coronary sinus transvenously using a coronary Amplatz catheter. A posterior/lateral and basal/mid-ventricular LV lead position was most often targeted; however, due to coronary sinus anatomy or technical difficulties, other stimulation sites were reached in some of the patients. Two patients received an LV epicardial lead via thoracotomy early in the study period. The VV interval was set between −20 and +20 ms, and the atrio-ventricular (AV) interval between 100–120 ms sensed and 130–150 ms paced unless patients were optimized by echocardiography after the implantation.9

Electrocardiograms

A standard 12-lead ECG was recorded with CardioLabIT GE during pacemaker implantation or with a traditional ECG machine during hospitalization before and after the implantation. In the majority of the patients, the ECG was interpreted at a speed of 100 mm/s, otherwise at a speed of 25 mm/s.

The intrinsic and paced QRS duration was defined as the widest QRS complex on a standard 12-lead ECG. The intrinsic QRS duration was divided into three groups: <150, 150–200, and >200 ms. The difference between intrinsic and paced QRS duration was divided into two groups, QRS shortening (>0 ms) and QRS prolongation or unchanged QRS duration (≤0 ms). PR duration was measured in patients with intrinsic AV conduction and defined as the shortest duration between the start of the P-wave to the start of the QRS complex. A prolonged PR interval was defined as >200 ms. Left bundle branch block was defined as QRS ≥120 ms with an rS or QS morphologic type in V1 and a broad R-wave without a Q-wave in either lead I or V6. Right bundle branch block was defined as QRS ≥120 ms with an rSR or qR in lead V1 and a deep S-wave in leads I and V6. The QRS axis was divided into normal (−30° to +90°), left-axis deviation (LAD, <−30°), and right-axis deviation (RAD, >90°) in patients with intrinsic conduction.

Clinical data

We collected clinical information from the entire patient group retrospectively. We included the New York Heart Association (NYHA) classification and LVEF before implantation and at the first follow-up after the implantation, documented in patient files as a part of routine clinical care. The LVEF was measured using visual assessment by an experienced observer,10 and LV end-diastolic diameter (LVEDD) was measured using m-mode.11

An improvement in NYHA class ≥1 after implantation was defined as a clinical response to CRT, and an improvement in absolute LVEF ≥5% after implantation was defined as an echocardiographic response to CRT.

Statistics

Normally distributed data are presented as mean ± SD, otherwise as median and 25th–75th percentiles. Absolute frequency and percentages are reported for categorical data. The difference between groups was evaluated with t-tests for Gaussian variables and Wilcoxon's rank-sum tests for non-Gaussian variables. Pearson's χ2 test was used for all categorical data. Cox's regression analysis was used for all-cause and cardiac mortality. Logistic regression analysis was used for improvement in NYHA class and LVEF. Electrocardiographic measurements and patterns included in the analysis were QRS duration (>150, 150–200, and >200 ms), changes in QRS duration after implantation (QRS shortening >0 ms and QRS prolongation or unchanged QRS duration ≤0 ms), configuration [LBBB, RBBB, and right ventricular (RV) pacing], and PR duration (normal and prolonged). In patients with LBBB, we compared normal axis, LAD, and RAD. We adjusted for following variables known or suspected to be associated with the outcomes measured:12 age, gender, heart failure aetiology, NYHA class, diabetes, atrial fibrillation, ICD, and LVEF at baseline. Median LVEF at baseline was used to dichotomize the continuous variable. Hazard ratios and odds ratios with 95% confidence intervals (CI) are reported. Mortality rate was summarized by construction of Kaplan–Meier curves. All patients who underwent heart transplantation (HTX) were censored at the date of the surgery. All P-values are two-sided and nominal. A P-value of <0.05 was considered statistically significant. All statistical analyses were performed using STATA software (STATA for Windows, version 10.0).

Results

Patients

Seven hundred and six patients received CRT or CRT-D in the study period. In 659 patients (93%), useful ECGs were obtained. The median age at implantation was 66 (58–73) years, the median follow-up time for mortality was 2.5 (1.2–4) years, and the median time from implantation to clinical follow-up was 0.5 (0.2–1) years. Baseline data for all patients and patients with LBBB, RBBB, and RV pacing in the pre-implant ECG including clinical parameters, pharmacologic therapy, and echocardiographic measurements are shown in Table 1.

View this table:
Table 1

Baseline characteristics

All (n = 659)LBBB (n = 527)RBBB (n = 39)Paced (n = 93)
Patients characteristics
 Age66 (58–73)66 (59–73)66 (55–70)65 (58–73)
 Men/Women527 (80)/132 (20)411 (78)/116 (22)36 (92)/3 (8)*80 (86)/13 (14)
 IHD/non-IHD343 (52)/314 (48)272 (52)/253 (48)23 (59)/16 (41)48 (52)/45 (48)
 Diabetes104 (16)81 (15)5 (13)18 (19)
 Atrial fibrillation165 (25)115 (22)15 (38)*35 (38)*
 ICD225 (35)182 (35)24 (62)*19 (34)#
 QRS duration (ms)170 ± 49167 ± 51161 ± 24191 ± 38*#
NYHA
 I12 (2)8 (2)1 (3)3 (4)
 II94 (15)75 (15)9 (24)8 (9)
 III459 (71)365 (71)24 (65)67 (77)
 IV80 (12)62 (12)3 (8)9 (10)
Echocardiography
 LVEF (%)25 ± 824 ± 725 ± 727 ± 11
 LVEDD (mm)70 ± 1070 ± 1070 ± 1065 ± 9*#
Medicine (%)
 Beta-blockers487 (75)395 (76)27 (71)65 (73)
 ACE-inhibitors582 (90)477 (92)32 (84)73 (82)*
 Loop diuretics547 (84)445 (85)31 (82)71 (80)
 Aldosterone antagonists383 (59)323 (62)16 (42)*44 (50)#
 Digoxin254 (39)203 (39)12 (32)39 (44)
 Amiodarone129 (20)101 (24)9 (21)19 (20)
  • Baseline characteristics are presented as median and 25th–75th percentiles or mean ± SD for continuous variables and as absolute frequency and percentages for categorical variables. The differences between groups were evaluated with t-tests for Gaussian variables and Wilcoxon's rank-sum tests for non-Gaussian variables. Pearson's χ2 test was used for all categorical data.

  • *P < 0.05 vs. LBBB.

  • #P < 0.05 vs. RBBB.

In 527 (80%) patients, LBBB was the QRS configuration at baseline, 39 (6%) patients had RBBB, and 93 (14%) patients had RV pacing. The proportions of patients with female gender, CRT-D, and atrial fibrillation were higher and the proportion of patients taking aldosterone antagonists was lower in the group with RBBB vs. LBBB. In the group of patients with RV pacing, a higher proportion had atrial fibrillation, a lower proportion was taking ACE-inhibitors, the QRS duration was longer, and the LVEDD was lower when compared with patients with LBBB. In patients with RV pacing, less received CRT-D and a higher proportion was taking aldosterone antagonist, compared with patients with RBBB.

In 208 (47%) patients, the PR interval was prolonged. Patients with long PR were significantly older, higher proportions were men and taking loop diuretics, and lower proportions were taking digoxin, beta-blockers, and amiodarone compared with patients with a normal PR interval.

In 108 (17%) patients, the QRS duration was long (>200 ms), mean 214 ms (±22); in 375 (57%) patients, the QRS duration was intermediate (150–200 ms), mean 170 ms (±13); and in 170 (26%) patients, the QRS duration was short (<150 ms), mean 130 ms (±11).

Patients with intermediate QRS duration were significantly older, and compared with those with long QRS duration, there were more men and a higher proportion of patients with atrial fibrillation. There was a higher proportion of CRT-D and higher LVEF in patients with a short QRS. In the group with long QRS duration, more patients were taking digoxin and the patients had a higher LVEDD.

In patients with LBBB, 284 (56%) had normal axis, 207 (40%) had LAD, and 21 (4%) had RAD. Patients with RAD had a higher LVEF and those with LAD had a lower LVEF at baseline compared with those with normal axis. Patients with LAD also had a higher LVEDD compared with those with normal axis and a longer QRS duration compared with those with RAD.

In 634 (92%) patients, we obtained differences in QRS duration between the intrinsic and paced ECG. The mean change was a reduction of 19 (±32) ms, and 457 (72%) patients had a reduction of the QRS duration and 177 (28%) patients had no change or prolongation of the QRS duration. There were higher proportions of men and patients with atrial fibrillation and lower proportions of patients taking amiodarone and receiving an ICD in the group with no reduction of the duration QRS. These patients also had a shorter QRS duration at baseline.

Clinical outcome

In the study period, 25 (4%) patients underwent HTX and 215 (32%) patients died, in 156 (73%) of these patients, cardiac death was observed. The cumulative all-cause mortality was 14 (95% CI, 11–17) %, 24 (95% CI, 20–28)%, and 56 (95% CI, 47–66)% at 1, 2, and 5 years. Significant unadjusted predictors of all-cause and cardiac mortality were prolonged PR, LAD, and RAD in LBBB in the baseline ECG, and no QRS reduction in the ECG obtained after CRT. Adjusted for variables known or suspected to be associated with mortality, LBBB, prolonged PR, RAD in LBBB, and no QRS reduction after CRT were significant predictors of all-cause and cardiac mortality. The Kaplan–Meier estimates of all-cause mortality are shown in Figure 1.

Figure 1

Kaplan–Meier estimates of all-cause mortality in (A) patients with LBBB, RBBB, and RV pacing before implantation (RVP); (B) patients with a normal and prolonged intrinsic PR duration (>200 ms); (C) patients with short (>150 ms), intermediate (150–200 ms), and long QRS duration (>200 ms); (D) patients with LBBB before implantation combined with a normal axis (Normal), LAD, and RAD; (E) patients with QRS prolongation or unchanged QRS duration (ΔQRS ≤0 ms) and QRS shortening (ΔQRS >0 ms) during CRT.

In 550 patients, we obtained data on NYHA class at baseline and follow-up. Significant unadjusted predictors of improvement ≥1 NYHA class were a QRS duration between 150 and 200 ms. Adjusted for variables known or suspected to be associated with symptomatic improvement RBBB and a QRS duration between 150 and 200 ms were significant predictors of symptomatic improvement.

In 346 patients, we obtained data on LVEF at baseline and at later follow-up. Significant unadjusted predictors of improvement in absolute LVEF ≥5% were baseline QRS duration between 150 and 200 ms, normal PR interval, normal axis vs. RAD in LBBB, and reduction in QRS duration after the implantation. Adjusted for variables known or suspected to be associated with echocardiographic improvement, normal PR, and normal axis vs. RAD in LBBB were significant. Electrocardiographic patterns as predictors of all-cause mortality, cardiac mortality, symptomatic response, and echocardiographic response are listed in Table 2.

View this table:
Table 2

Clinical outcome

UnadjustedAdjusted
All-cause mortality (n = 659)HR95% CIP-valueHR95% CIP-value
 LBBB11
 RBBB0.590.29–1.200.150.410.19–0.910.03
 Paced0.660.42–1.030.070.470.28–0.79<0.01
 QRS < 150 ms11
 QRS 150–200 ms1.020.73–1.430.910.820.57–1.180.29
 QRS > 200 ms1.040.69–1.580.851.280.81–2.010.30
 Prolonged PR1.761.24–2.49<0.011.671.14–2.43<0.01
 LAD in LBBB1.481.09–2.020.011.310.95–1.840.10
 RAD in LBBB2.301.19–4.470.014.302.17–8.51<0.01
 QRS reduction0.600.45–0.81<0.010.600.42–0.85<0.01
Cardiac mortality (n = 659)HR95% CIP-valueHR95% CIP-value
 LBBB11
 RBBB0.410.15–1.110.080.380.12–0.950.04
 Paced0.650.39–1.090.100.460.30–0.970.05
 QRS <150 ms11
 QRS 150–200 ms1.060.71–1.570.790.870.57–1.330.52
 QRS >200 ms1.040.64–1.710.871.210.71–2.090.47
 Prolonged PR1.891.26–2.86<0.011.701.08–2.660.02
 LAD in LBBB1.511.06–2.170.021.450.98–2.150.06
 RAD in LBBB2.531.20–5.310.014.562.08–10.0<0.01
 QRS reduction0.580.41–0.82<0.010.600.40–0.900.01
NYHA class improvement ≥1 (n = 550)OR95% CIP-valueOR95% CIP-value
 LBBB11
 RBBB1.540.75–3.160.242.581.13–5.930.03
 Paced1.110.69–1.810.671.230.69–2.200.48
 QRS < 150 ms11
 QRS 150–200 ms1.651.09–2.480.021.701.07–2.700.02
 QRS > 200 ms1.360.79–2.320.271.520.82–2.840.19
 Prolonged PR (n = 373)0.920.61–1.380.681.120.70–5.340.22
 LAD in LBBB (n = 474)1.020.69–1.510.931.120.73–1.730.61
 RAD in LBBB0.520.19–1.460.220.470.16–1.400.17
 QRS reduction1.500.99–2.20.0581.20.73–2.00.46
Improvement in LVEF ≥5% (n = 346)OR95% CIP-valueOR95% CIP-value
 LBBB11
 RBBB1.010.43–2.370.981.360.56–3.360.50
 Paced0.960.55–1.690.891.200.61–2.330.60
 QRS <150 ms11
 QRS 150–200 ms1.811.08–3.030.031.670.94–2.950.08
 QRS >200 ms1.370.71–2.600.361.340.65–2.750.43
 Prolonged PR0.500.29–0.850.010.480.27–0.860.01
 LAD in LBBB0.970.58–1.600.900.850.47–1.470.56
 RAD in LBBB0.090.01–0.760.030.110.01–0.970.046
 QRS reduction1.921.18–3.13<0.011.790.97–3.310.062
  • Hazard ratios (HR) and odds ratios (OR) with 95% confidence interval (CI) for all-cause mortality, cardiac mortality, clinical response, and echocardiographic response. Left bundle branch block (LBBB), right bundle branch block (RBBB), and right ventricular pacing before implantation (paced), left-axis deviation (LAD), and right-axis deviation (RAD). All analyses were adjusted for age, gender, heart failure aetiology, NYHA class, diabetes, atrial fibrillation, ICD, and LVEF at baseline.

In international guidelines, CRT is only indicated in patients with NYHA class III and IV. If we excluded patients with NYHA class I and II from the present analysis, the same variables were significant predictors of long-term clinical outcome.

Since ICD indications have changed during the study period, we also analysed the mortality data on patients with CRT without an ICD, and the same variables were significant predictors for mortality except a prolonged PR interval which became borderline significant.

Discussion

This study demonstrates the predictive value of different ECG patterns on all-cause mortality, cardiac mortality, symptomatic response, and echocardiographic response in a large single-centre cohort of consecutive patients treated with CRT.

Patients

Cardiac resynchronization therapy is indicated in patients with heart failure, NYHA class III and IV, reduced LVEF, and broad QRS. Prophylactically or as a part of other clinical trials in this study period, CRT devices were also implanted in a small group of primarily younger patients in NYHA class I and II. We also included patients with atrial fibrillation and previous devices, which were excluded from most clinical trials in this period. Other baseline characteristics such as age, gender, heart failure aetiology, QRS duration, and LVEF were similar to those observed in other comparable studies. There were some baseline differences between different groups in this study that may have influenced on the outcomes measured, but all analyses were adjusted for these variables. Left ventricular lead position was not included in the analysis, based on a previous study on the same cohort which showed no difference in the outcomes measured between different left ventricular lead positions.13

Clinical outcome

Right bundle branch block

Our study showed lower mortality and symptomatic response in a larger proportion of patients with RBBB compared with LBBB. Previous randomized controlled trials mainly included patients with LBBB, and the proportion of patients with RBBB varies from 7 to 13%; therefore, the potential benefit of CRT in this population is uncertain. In an experimental model, the degree of dyssynchrony was less in RBBB compared with LBBB14 and one clinical study showed that the 12-lead ECG is less reliable in the characterization of ventricular activation.15 In contrast, Fantoni et al.16 showed that the LV conduction delay and activation pattern were similar in patients with RBBB and LBBB. In a study by Garrigue et al.,17 the majority of 12 patients with RBBB treated with CRT improved symptoms and echocardiographic measurements. The degree of dyssynchrony at baseline was larger in the patients who improved during CRT. In subgroup analysis of patients with RBBB included in larger trials, diverging effect of CRT has been found.4,18,19 One recent large single-centre prospective study with a similar size and follow-up period compared with our study showed a worse outcome in patients with RBBB.20 This was also found in all patients included in the CARE-HF trial, but in this trial, patients with RBBB assigned to the control arm also showed a worse outcome.21

The results in our study may be due to some degree of intra- or interventricular dyssynchrony and therefore a potential benefit of CRT in our patients with RBBB. Furthermore, patients with heart failure and RBBB have a better prognosis compared with those with LBBB also without CRT.22 The benefit of CRT in patients with RBBB is still uncertain and need further investigation.

Right ventricular pacing

The clinical effect of upgrading a conventional pacemaker to a CRT system has been showed in two small randomized studies23,24 and in a number of prospective non-randomized studies.2529 The present study shows a similar echocardiographic and clinical response, but a lower mortality in patients with RV pacing at implantation compared with those with LBBB. This is partially in accordance with previous studies showing no difference in clinical outcome or mortality between patients with de novo CRT and those undergoing upgrade from conventional devices.20,30,31

The aetiology and severity of myocardial disease is probably different in patients with LBBB and those with heart failure and a conventional pacemaker indication in whom heart failure was occasionally diagnosed or maybe was induced by RV pacing. The results in our study may be due to a less severe degree of disease and therefore a pre-implant prognosis in the latter group. Cardiac resynchronization therapy in patients with heart failure that developed during RV pacing has a good effect on clinical outcome, and these patients should be upgraded when they are diagnosed.

QRS duration

Prolonged QRS duration is related to disease severity and increased mortality in heart failure patients.32 Although a longer QRS duration increases the likelihood of mechanical dyssynchrony, it has failed as a predictor of response to CRT.33 The present study shows no difference in mortality and echocardiographic response between a short, intermediate, or long QRS duration, but an increased symptomatic response in patients with a QRS between 150 and 200 ms, which may be due to the fact that the number of patients with mechanical dyssynchrony is larger in this group compared with those with a shorter QRS duration (<150 ms), and that patients with an extreme long QRS duration (>200 ms) may have more irreversible myocardial damage.

PR interval

In a general population, the predictive value of a prolonged PR interval is low. Two previous studies have found no correlation between PR duration and response to CRT.34,35 In contrast, our study showed a higher mortality and a decreased echocardiographic response in patients with prolonged PR interval. This was also found in patients in the CARE-HF study,21 and it might be explained by more severe myocardial disease before initiation of CRT in this group.

Different axis in left bundle branch block

Patients with LBBB and an abnormal axis deviation have more severe clinical manifestations; LAD is associated with a more severe conduction system disease and RAD suggests biventricular enlargement. In a previous study by Lecoq et al.,34 there were no differences in the electrical axis between responders and non-responders; similar to these findings, Reynolds et al.35 showed that an abnormal axis did not predict response. The present study showed an increased mortality and a lower chance of echocardiographic response in patients with RAD.

Reduction in QRS duration

Previous studies have shown that non-responders have a less degree of QRS reduction after CRT.21,34,36 Our study shows similar findings that a lack of electrical resynchronization, defined as a prolonged or unchanged QRS duration after CRT, is associated with a higher all-cause and cardiac mortality.

Limitations

This study presents the typical limitations of similar retrospective studies. The patients are more heterogeneous, there were different clinical and echocardiographic observers during the study period, and there was no blinding or control group. Furthermore, our study did not include other parameters usually considered of interest in modern CRT, especially intraventricular dyssynchrony. Electrocardiograms, echocardiographic measurements, and symptomatic evaluation were missing in a minority of the patients. Some of the subgroups in our study were small, including patients with RBBB, and the results should therefore be interpreted with caution. However, to our knowledge, this is the largest study evaluating the effects of different ECG patterns on long-term clinical outcome in a single-centre cohort of consecutive patients treated with CRT.

Conclusion

In a large consecutive cohort of CRT patients, LBBB, prolonged PR interval, RAD in LBBB in the pre-implant ECG, and no QRS reduction after CRT are associated with an increased all-cause and cardiac mortality. Right bundle branch block and intermediate QRS duration (150–200 ms) were associated with a better symptomatic response, and normal vs. prolonged PR interval and normal axis vs. RAD in LBBB were associated with a higher echocardiographic response. In general, the predictive value of baseline ECG patterns is similar in heart failure patients undergoing CRT and in a general population of heart failure patients, where LBBB, a prolonged PR, and an abnormal axis in LBBB are signs of a more severe degree of myocardial disease, and therefore a worse outcome. Lack of electrical resynchronization defined as an unchanged or prolonged QRS duration is associated with higher all-cause and cardiac mortality in patients treated with CRT.

Conflict of interest: none declared.

Funding

This work was supported by the Danish Heart Foundation (07-4-B695-A1464-22378).

References

View Abstract