Predictors of recovery of left ventricular dysfunction after ablation of frequent ventricular premature depolarizations

Predictors of recovery of left ventricular dysfunction after ablation of frequent ventricular premature depolarizations Marc W. Deyell, MD, MSc, Kyoung-Min Park, MD, Yuchi Han, MD, MSc, David S. Frankel, MD, Sanjay Dixit, MD, FHRS, Joshua M. Cooper, MD, Mathew D. Hutchinson, MD, FHRS, David Lin, MD, Fermin Garcia, MD, Rupa Bala, MD, Michael P. Riley, MD, PhD, Edward Gerstenfeld, MD, FHRS, David J. Callans, MD, FHRS, Francis E. Marchlinski, MD, FHRS From the Electrophysiology Section, Hospital of the University of Pennsylvania, Philadelphia, Pennsylvania. BACKGROUND Frequent ventricular premature depolarizations (VPDs) can cause reversible left ventricular (LV) dysfunction. However, not all patients normalize their LV function after VPD elimination. OBJECTIVE To evaluate predictors of recovery of LV function following the elimination of frequent VPDs. METHODS We identified patients with ⱖ10% VPDs/24 h and an LV ejection fraction of ⬍50% who underwent successful ablation between 2007 and 2011. Subjects were classified as having reversible (ⱖ10% increase to a final LV ejection fraction of ⱖ50%) or irreversible (ⱕ10% increase or final LV ejection fraction ⬍50%) LV dysfunction on the basis of echocardiographic follow-up. A reference group with ⱖ10% VPDs but normal LV function was identified. RESULTS One hundred fourteen patients with ⱖ10% VPDs were identified; 66 had preserved and 48 had impaired LV function. Over a median follow-up of 10.6 months, 24 of 48 were classified as reversible and 13 of 48 as irreversible and 11 of 44 were excluded. There was a gradient of VPD QRS duration between the control, reversible, and

Introduction The concept that frequent ventricular ectopy may produce a potentially reversible form of left ventricular (LV) dysfunction has generated increasing interest since it was first reported in 1998.1 Coupled with this, electrophysiologists now have access to improved electrocardiogram localization criteria and mapping technology that has led to greater success at elimination of ventricular premature depolarizations (VPDs) with ablation.2,3 However, much of the natural history and pathophysiology of VPD-associated ventricular dysfunction remains unknown. An increased burden of VPDs has been associated with LV dysfunction in prior studies of patients referred for ablation.4 – 8 This is supported by a longitudinal study that This study was funded by the F. Harlan Batrus Research Fund and the Susan and Murray Bloom Research Fund. Address for reprint requests and correspondence: Dr Francis E. Marchlinski, MD, Electrophysiology Section, Hospital of the University of Pennsylvania, 9 Founder’s Pavilion, 3400 Spruce St, Philadelphia, PA 19104. E-mail address: francis. [email protected].

irreversible groups (mean VPD QRS 135, 158, and 173 ms, respectively; P ⬍ .001). This gradient persisted even for the same site of origin. In multivariate analysis, the only independent predictor of irreversible LV function was VPD QRS duration (odds ratio 5.07 [95% confidence interval 1.22–21.01] per 10-ms increase). CONCLUSION In patients with LV dysfunction and frequent VPDs, we identified VPD QRS duration as the only independent predictor for the recovery of LV function after ablation. This suggests that VPD QRS duration may be a marker for the severity of underlying substrate abnormality. KEYWORDS Cardiomyopathy; Catheter ablation; Electrocardiogram; Ventricular arrhythmia ABBREVIATIONS CC ⫽ coronary cusp; CHF ⫽ congestive heart failure; IQR ⫽ interquartile range; LV ⫽ left ventricle/ventricular; LVEF ⫽ LV ejection fraction; RV ⫽ right ventricle/ventricular; VPD ⫽ ventricular premature depolarization (Heart Rhythm 2012;9:1465–1472) © 2012 Heart Rhythm Society. All rights reserved.

found subclinical deterioration in LV function over 5 years in those with a high burden of VPDs (ⱖ10%–20%).9 Yet VPD burden alone may not explain the presence of LV dysfunction. Many patients with longstanding, frequent VPDs have no evidence of LV dysfunction. Furthermore, a significant proportion of patients with frequent VPDs and LV dysfunction show no or partial improvement in ventricular function despite the elimination of their ectopy. In clinical practice, it is the rare patient with frequent VPDs who has documented development of LV dysfunction over time. Rather, a more common scenario is the simultaneous discovery of frequent VPDs and LV dysfunction. Here, it is difficult to delineate the relative time course of the ectopy in relationship to the LV dysfunction, especially given the often nonspecific and insidious nature of symptoms caused by frequent VPDs. Consequently, patients with frequent VPDs frequently present a management challenge. Among patients with preserved LV function, it is unclear which patients are at risk for the development of LV dysfunction. In contrast, when

1547-5271/$ -see front matter © 2012 Heart Rhythm Society. All rights reserved.

http://dx.doi.org/10.1016/j.hrthm.2012.05.019

1466 patients present with LV dysfunction and frequent VPDs, it is uncertain whether the ectopy may be contributing to ongoing LV dysfunction, whether the VPD-associated LV dysfunction is now irreversible, or whether the VPDs are an epi-phenomenon of an underlying myopathic process. The purpose of this study was therefore twofold. First, we sought to examine factors associated with LV dysfunction among patients with frequent ventricular ectopy referred for ablation. Second, we sought to determine predictors of reversibility of LV dysfunction after the successful elimination of VPDs by ablation.

Methods Patient population We identified all patients who underwent successful ablation of VPDs, with a documented LV ejection fraction (LVEF) of ⬍50% on pre- or immediately postprocedure echocardiography, at the Hospital of the University of Pennsylvania between January 1, 2007, and April 30, 2011. Patients with frequent VPDs but in whom ablation was not performed were not systematically tracked. As existing data suggest that very frequent VPDs are required to contribute to LV dysfunction,9 we restricted our cohort to only patients with ⱖ10% VPDs on preprocedure 24-hour Holter monitoring. Successful ablation was defined as at least an 80% reduction in the 24-hour burden of VPDs, based on our previously published experience.10 Patients were further excluded if they (1) had a known secondary cause for LV dysfunction or (2) a history of sustained ventricular tachycardia, appropriate implantable defibrillator discharges, or sudden cardiac death. Only subjects with complete Holter and echocardiographic follow-up were eligible. Subjects with unsuccessful ablation were excluded from the analysis but tracked to evaluate procedural success. All antiarrhythmic agents were discontinued after ablation, except in the presence of ongoing supraventricular arrhythmias. All patients with ongoing LV dysfunction after ablation received therapy with beta blockade and renin-angiotensin inhibition. In order to evaluate characteristics associated with LV dysfunction in the presence of frequent VPDs, we also identified a reference group of patients with ⱖ10% VPDs on preprocedure Holter monitoring but with preserved LV function (echocardiographic LVEF of ⱖ50%) who underwent ablation during the study period.

Data collection Baseline demographic, historical, and clinical characteristics were collected prospectively at ablation. Additional clinical and electrophysiologic parameters were assessed through detailed review. All electrocardiographic measurements were performed, blinded to outcomes, by 1 of 2 authors (MWD or KMP) by using digital calipers at 100 mm/s on CardioLab (version 6.5.4.1858, GE Medical Systems, Waukesha, WI, USA). All electrocardiographic measurements were repeated on 5 separate VPDs, occurring prior to the introduction of intracardiac catheters. The mean of the 5 measurements was used for analysis. In the case of multiple VPD morphologies,

Heart Rhythm, Vol 9, No 9, September 2012 the dominant/targeted VPD was measured or, if more than 1 dominant/targeted VPD was present, the measurements were averaged between morphologies. As quantitative echocardiographic methods of assessment of LV function may be inaccurate in the setting of frequent VPDs, visual estimation was used to determine LVEF. All echocardiographic assessments were blinded to electrophysiologic and procedural characteristics. Where a range was reported by the interpreter, the lower value was used for all analyses. Blinded assessment of intra- and interobserver agreement for electrocardiogram and echocardiographic measurements was performed on subsets of 20 patients. Preablation echocardiographic measurements were made by using stored loops of 3 cardiac cycles without ectopy, where possible. The site of origin of the VPDs was classified by the site of successful ablation. In the absence of a standard classification scheme, we initially classified VPDs into the following groups on the basis of the anatomic structure giving rise to the ectopy: (1) right ventricular outflow tract, right coronary cusp (CC), or pulmonary artery; (2) other right ventricular (RV) site; (3) left CC, left-right CC commissure, aorto-mitral continuity, or great cardiac vein/anterior interventricular vein; (4) other LV site; and (5) multiple dominant/targeted morphologies. Given the lack of consensus regarding the classification of the site of origin, in sensitivity analyses we also alternatively dichotomized as (1) septal vs nonseptal, (2) outflow tract vs nonoutflow tract, and (3) RV vs LV.

Patient classification There is also no accepted disease definition for VPD-induced cardiomyopathy. We therefore a priori characterized patients with impaired LV function into 3 groups on the basis of their response to therapy with ablation: (1) reversible (an improvement in LVEF of ⱖ10% to a final LVEF of ⱖ50%), (2) partially reversible (an improvement in LVEF of ⱖ10% but a final LVEF of ⬍50%), and (3) irreversible (an improvement in LVEF of ⬍10%). The minimum change in LVEF of 10% to be considered reversible was chosen as the interobserver variability in determining LVEF by echocardiography ranges from 9% to 14%.11 Given the small number in the partially reversible group, these were included with the irreversible group for the primary analysis. In a sensitivity analysis, they were instead included with the reversible group to assess the robustness of the findings.

Statistical methods Pearson’s product-moment correlation coefficient was calculated to quantify intra- and interrater variability for the electrocardiogram and echocardiographic measurements. Differences in baseline characteristics across the groups of interest were carried out first in a univariate fashion by using the Fisher exact test for categorical and the KruskalWallis test for continuous variables. To determine independent predictors of the presence of both LV dysfunction and reversibility of LV dysfunction, mul-

Deyell et al

Frequent VPDs and LV Dysfunction

1467

tivariate logistic regression was used. All covariates were initially assessed in a univariate fashion, and variables were introduced into the multivariate model if the likelihood ratio test P value was ⱖ.20. The percent VPDs/24 hours was a priori included in the preliminary main effects models, given its clinically plausible influence on outcomes. For the model evaluating reversibility of LV dysfunction, baseline LV ejection fraction was also a priori included in the preliminary main effects model. All variables from the univariate analysis were added in a stepwise fashion, and interaction variables were introduced to identify effect modification. The influence of each variable on the model was assessed by using a likelihoodratio test. For all continuous variables, the linear assumption was tested by performing a lowess smoothed plot of the continuous variable against the logit of the dependent variable. Goodness of fit was assessed by using the Hosmer-Lemeshow test, and model accuracy was assessed by calculating the area under the receiver-operating characteristic curve. All analyses were performed by using STATA version 9.2 (STATACorp, College Station, TX).

therefore excluded. Of the remaining 37 patients with successful ablation, 24 patients had reversible LV dysfunction, 2 had partially reversible LV dysfunction, and 11 had irreversible LV dysfunction. Those with partially reversible LV dysfunction were combined with the irreversible LV dysfunction group for the primary analysis. The acute procedural success rate (no observed targeted VPDs during a 30-minute waiting period after ablation) was achieved in 99 of 110 (90.0%) in the overall study population and 41 of 44 (93.2%) in those with impaired LV function. Long-term success (at least an 80% reduction in VPD burden) for ablation among patients with LV dysfunction was 84.1% (37 of 44), with second procedures (within our institution) required for success in 8 of 44 (18.2%). Among those with impaired LV function, 16 of 44 (36.3%) had failed prior attempts at ablation at other institutions. The median echocardiographic follow-up among patients with impaired LV function was 10.6 months (interquartile range [IQR] 6.4 –19.3). Postprocedure Holter monitoring was performed at a median of 2.7 months (IQR 1.1–15.9).

Results

Baseline characteristics

We identified 114 patients with ⱖ10% VPDs who underwent ablation (Figure 1). Of these, 48 (42%) had impaired LV function while the remaining 66 (58%) had normal LV function and served as controls. Of those with decreased LV function, 7 patients had failed ablations and were excluded from the primary analysis but are described below. Four patients had frequent ectopy and LV dysfunction but did not have formal postprocedure Holter monitoring and were

The baseline characteristics across the 3 groups of (1) normal LV function, (2) reversible LV function, and (3) partially reversible/irreversible LV function are presented in Table 1. Among the historical variables, the presence of prior, symptomatic congestive heart failure (edema, orthopnea, or paroxysmal nocturnal dyspnea) was seen only in the impaired LV function groups, but the prevalence was low (⬍10%).

Total popula
on with ≥10% VPDs/24 hours undergoing abla
on: 114 pa
ents Normal LV func
on (reference group): 66 pa
ents (58%) Impaired LV func
on: 48 pa
ents (42%) Exclusions: Unsuccessful abla
on - 7 pa
ents Missing post-procedure Holter monitor - 4 pa
ents

Successful abla
on of VPDs: 37 pa
ents

Reversible LV dysfunc
on: 24 pa
ents

Par
ally reversible/irreversible LV dysfunc
on: 13 pa
ents

Figure 1 The patient flow for the study is depicted. Note that successful ablation was defined as at least an 80% reduction in the burden of ectopy on postablation 24-hour Holter monitoring. LV ⫽ left ventricular; VPD ⫽ ventricular premature depolarization.

1468 Table 1

Heart Rhythm, Vol 9, No 9, September 2012 Baseline characteristics prior to ablation

Characteristic Demographics Age (y), mean ⫾ SD Sex: Woman, n (%) Medical history History of hypertension, n (%) History of diabetes, n (%) History of atrial fibrillation, n (%) History of CHF, n (%) Medications Beta-blocker, n (%) Antiarrhythmic, n (%) ACE inhibitor/ARB, n (%) Loop diuretic, n (%) Symptoms, n (%) Asymptomatic Palpitations/skipped beats Other symptoms* Holter monitoring % VPDs, mean ⫾ SD VPDs/24 h, mean ⫾ SD NSVT ⱖ5 beats, n (%) Echocardiography, mean ⫾ SD LV ejection fraction LV diastolic dimension LV systolic dimension Cardiac MRI, n (%) Performed Abnormal†

Normal LV function (n ⫽ 66)

Reversible LV dysfunction (n ⫽ 24)

Irreversible/partially reversible LV dysfunction (n ⫽ 13)

48.3 ⫾ 15.4 34 (51.5)

56.5 ⫾ 15.0 8 (33.3)

53.2 ⫾ 15.8 5 (38.5)

P for comparison across all groups .044 .297

22 (33.3)

8 (33.3)

5 (38.5)

.953

7 (10.6) 4 (6.1)

2 (8.3) 4 (16.7)

1 (7.7) 3 (23.1)

1.000 .077

0 (0.0)

2 (8.3)

1 (7.7)

.044

36 7 14 2

(54.6) (10.6) (21.2) (3.0)

18 4 16 2

(75.0) (16.7) (66.7) (8.3)

8 4 9 1

(61.5) (30.8) (69.2) (7.7)

.217 .133 ⬍.001 .286

5 (38.5) 4 (30.8) 4 (30.8)

.004 .002 .390

8 (12.2) 45 (68.2) 13 (19.7)

10 (41.7) 7 (29.2) 7 (29.2)

26.6 ⫾ 12.0 31,276.3 ⫾ 16,722.0 20 (30.3)

31.6 ⫾ 11.5 34,202.6 ⫾ 15,493.0 6 (25.0)

24.0 ⫾ 8.1 28,583.8 ⫾ 10,633.4 4 (30.8)

58.5 ⫾ 6.0 5.1 ⫾ 0.7 3.4 ⫾ 0.1

38.2 ⫾ 6.8 5.6 ⫾ 0.7 4.2 ⫾ 0.1

35.8 ⫾ 8.9 6.1 ⫾ 1.1 4.5 ⫾ 0.3

4 (16.7) 0 (0.0)

3 (23.1) 2 (66.7)

25 (37.9) 2 (8.0)

.077 .533 .921 .001 .001 ⬍.001 .140 .050

ACE ⫽ angiotensin-converting enzyme; ARB ⫽ angiotensin receptor blocker; CHF ⫽ congestive heart failure; LV ⫽ left ventricle; MRI ⫽ magnetic resonance imaging; NSVT ⫽ nonsustained VT; SD ⫽ standard deviation; VPD ⫽ ventricular premature depolarization; VT ⫽ ventricular tachycardia. *Includes fatigue, shortness of breath, chest pain, or dizziness. †Defined as any area of delayed enhancement or regional wall motion abnormality.

The presence and nature of symptoms differed markedly across the groups. Subjects with LV dysfunction were more likely to be asymptomatic (12.2% with normal LV function, 41.7% with reversible LV function, and 38.5% with partially reversible/irreversible LV function; P ⫽ .008). Only 32 of 103 (31.1%) underwent preprocedure cardiac magnetic resonance imaging. Notably, no patients in the reversible group, but 2 of 3 patients in the partially reversible group/irreversible group, had an abnormal magnetic resonance imaging (defined as a wall motion abnormality or delayed enhancement). Among electrophysiologic characteristics (Table 2), there was a significant difference in mean sinus QRS duration (P ⫽ .012) and VPD QRS duration (P ⬍ .001), with longer durations seen in the LV dysfunction groups. There was a gradient of VPD QRS duration across the 3 groups, even for PVCs with the same site of origin (Figures 2 and 3). The shortest median VPD QRS duration was observed for nonoutflow tract (“other”) RV sites at 126 ms, which included para-Hisian sites. The longest median VPD QRS duration was observed for VPDs arising from the LV outflow tract (excluding the right CC) at 150 ms. Septal sites had a shorter VPD QRS duration than nonseptal sites (135

vs 149 ms). The distribution of the site of origin of the VPDs was similar across the 3 groups with the exception of nonoutflow tract LV sites, which were significantly higher in the reversible and partially reversible/irreversible groups (33.3% and 23.8%, respectively) compared with controls (10.6%; P ⫽ .035). When the VPD site was dichotomized by septal/nonseptal, outflow tract/nonoutflow tract, or LV/RV sites, there was a trend toward more LV sites in those with impaired LV function (P ⫽ .076). A documented history of decline in LVEF in the presence of frequent ventricular ectopy was present for only 5 of 24 (20.8%) and 1 of 13 (8.7%) patients in the reversible and partially reversible/irreversible groups, respectively (P ⫽ .294). In the remainder, the relative time course of onset of LV dysfunction in relation to the VPDs could not be determined (typically because of concomitant presentation).

Characteristics associated with the presence of LV dysfunction In the multivariate analysis comparing normal with abnormal LV function, the only variable independently associated with the presence of LV dysfunction was VPD QRS dura-

Deyell et al Table 2

Frequent VPDs and LV Dysfunction

1469

Electrophysiologic characteristics

Characteristic Electrocardiographic parameters Sinus QRS width (ms), mean ⫾ SD VPD QRS width (ms), mean ⫾ SD Sinus cycle length, mean ⫾ SD VPD coupling interval (ms), mean ⫾ SD VPD site of origin, n (%) RVOT/right CC/PA Other RV Left CC/AMC/left-right CC commissure/AIV Other LV Multiple VPDs VPD classification,* n (%) Septal (vs nonseptal) Outflow tract (vs nonoutflow tract) LV site (vs RV)

Normal LV function (n ⫽ 66)

84.7 ⫾ 10.6

Reversible LV dysfunction (n ⫽ 24)

90.7 ⫾ 16.0

Partially reversible/irreversible LV dysfunction (n ⫽ 13)

P for comparison across all groups

102.8 ⫾ 25.6

.018

134.7 ⫾ 12.3

158.2 ⫾ 8.6

173.2 ⫾ 12.9

.001

838.5 ⫾ 163.3

905.7 ⫾ 183.8

838.2 ⫾ 197.2

.322

468.7 ⫾ 65.4

514.3 ⫾ 76.4

527.1 ⫾ 100.7

.006

24 (36.4) 5 (7.6) 25 (37.9)

6 (25.0) 0 (0.0) 7 (29.2)

3 (23.1) 0 (0.0) 7 (53.9)

.520 .388 .340

7 (10.6) 5 (7.6)

8 (33.3) 3 (12.5)

3 (23.1) 0 (0.0)

.035 .482

24 (36.4) 49 (74.2)

6 (25.0) 13 (54.2)

3 (23.1) 10 (76.9)

.486 .211

41 (62.1)

18 (75.0)

10 (76.9)

.214

AIV ⫽ anterior interventricular vein; AMC ⫽ aorto-mitral continuity; CC ⫽ coronary cusp; LV ⫽ left ventricle; PA ⫽ pulmonary artery; RV ⫽ right ventricle; RVOT ⫽ right ventricular outflow tract; SD ⫽ standard deviation; VPD ⫽ ventricular premature depolarization. *Subjects with multiple VPD morphologies were excluded.

tion (Table 3). The univariate analysis is provided in Appendix 1. The presence of congestive heart failure symptoms perfectly predicted LV dysfunction and therefore could not be included in the multivariate model. The odds ratio for the presence of LV dysfunction for every 10-ms increase in VPD QRS duration was 12.08 (95% confidence interval 4.12–35.46; P ⬍ .001). The goodness-of-fit test was not significant (P ⫽ .966). The area under the receiveroperating characteristic curve for the final model (shown in Table 3) was 0.964. The sensitivity and specificity of various cutoffs of VPD QRS duration to distinguish normal LV function from LV dysfunction were assessed. The best diagnostic accuracy was obtained by using a cutoff of 151 ms, yielding a sensitivity of 79.2% and a specificity of 90.9%. No patient with LV dysfunction had a VPD QRS duration of ⬍140 ms while no patient with normal LV function had a VPD QRS of ⱖ164 ms.

Characteristics associated with irreversibility of LV dysfunction The median improvement in LV ejection fraction was 17.5% (IQR 10 – 40) for patients with reversible LV dysfunction. Between the 2 patients with partially reversible LV dysfunction, the improvements in ejection fraction were 10% and 25%. The median change in LV ejection fraction in the irreversible group was 0% (range 11 to 9). The median change in LV diastolic dimension was 0.6 cm (IQR 0.3 to 0.9) and 0.3 cm (IQR 0.0 to 0.7) in the reversible and partially reversible/irreversible groups, respectively (P ⫽

.291). The median change in LV systolic dimension was 0.5 cm (IQR 0.1 to 0.7) and 0.2 cm (IQR 0.1 to 0.4), respectively (P ⫽ .235). While echocardiography was not performed at prespecified time points after ablation, the majority of patients underwent serial studies. Among those with reversible LV dysfunction, 17 of 24 (70.8%) and 19 of 24 (79.2%) underwent echocardiography by 3 and 6 months after ablation, respectively. By 3 months, 13 of 17 (76.5%) patients showed complete reversibility and 3 of 17 (11.8%) had a ⱖ10% improvement in LVEF but still had an LVEF of ⬍50%. The remaining patient showed slow, but gradual normalization of LV function over 3 years of follow-up. By 6 months, 15 of 19 (78.9%) patients showed complete reversibility and the same 3 of 19 (15.8%) patients showed partial reversibility (ⱖ10% improvement in LVEF). Two patients in the irreversible group had ⬍6 months of echocardiographic follow-up (each had over 4 months of follow-up). In the univariate analysis (Appendix 2), none of the demographic, historical, or medication usage variables were associated with irreversible LV dysfunction. In the final multivariate model (Table 4), adjusted for baseline LVEF, only VPD QRS duration was significantly associated with irreversible LV dysfunction. For every 10-ms increase in QRS duration, the adjusted odds ratio for partially reversible or irreversible LV dysfunction was 5.07 (95% confidence interval 1.22–21.01). The area under the receiver-operating

1470

Heart Rhythm, Vol 9, No 9, September 2012 odds ratio of irreversible LV dysfunction, per 10-ms increase in VPD QRS duration, was 7.21 (95% confidence interval 1.34 –38.78). The sensitivity and specificity of various cutoffs of VPD QRS duration for the presence of partially reversible/irreversible LV dysfunction at various cutoffs were assessed (Table 5). The best diagnostic accuracy was obtained with a cutoff of 165 ms, yielding a sensitivity of 76.92% and a specificity of 70.83% for the presence of partially reversible or irreversible LV dysfunction. Nonetheless, 92.31% of the patients with reversible LV dysfunction had a VDP QRS duration of ⱕ160 ms. Conversely, a VPD QRS duration of ⱖ170ms was observed in 91.67% of the patients with partially reversible/irreversible LV dysfunction (91.67% specific).

Intra- and interrater reliability of measurements The intra- and interrater correlation coefficients for VPD QRS duration were 0.87 and 0.92, respectively. The intraand interrater correlation coefficients for VPD coupling interval were 0.94 and 0.96, respectively. The intra- and interrater correlation coefficients for LV ejection fraction were 0.94 and 0.89, respectively. There was no disagreement regarding the presence of LV dysfunction between raters. The intra- and interrater correlation coefficients for LV diastolic dimension were 0.97 and 0.96. The intra- and interrater correlation coefficients for LV systolic dimension were 0.98 and 0.95.

Discussion

Figure 2 VPD QRS duration, by the presence and reversibility of LV dysfunction. A: The box plots of mean VPD QRS duration, by category, for the overall study population. B: The mean VPD QRS duration, by category, for all VPDs originating from the RVOT, right CC, or PA. C: The mean VPD QRS duration, by category, for all VPDs originating from the left CC, AMC, or AIV. Mann-Whitney U test P values presented. AIV ⫽ anterior interventricular vein; AMC ⫽ aorto-mitral continuity; CC ⫽ coronary cusp; LV ⫽ left ventricular; PA ⫽ pulmonary artery; RVOT ⫽ right ventricular outflow tract; VPD ⫽ ventricular premature depolarization.

characteristic curve was 0.82 for the final model. The goodness-of-fit test was not significant (P ⫽ .514). In a sensitivity analysis, the partially reversible patients were instead grouped with those with reversible LV dysfunction. The VPD QRS duration remained a significant, independent predictor of irreversible LV dysfunction. The

Among patients with impaired LV function and frequent ventricular ectopy, we identified that the VPD QRS duration and the presence of overt congestive heart failure were the only independent predictors of lack of recovery of LV function after successful ablation. Furthermore, there was a gradient in VPD QRS duration from those with normal LV function to reversible and partially reversible/irreversible LV dysfunction. Notably, other variables, such as VPD burden, sinus QRS duration, multifocal VPDs, nonsustained ventricular tachycardia, VPD coupling interval, and baseline LV ejection fraction, were not predictive. Many of these variables have previously been associated with the presence of baseline LV dysfunction at ablation in cross-sectional studies,4,7,10 but their relationship with outcomes has not previously been assessed. Prior studies have demonstrated an improvement in the majority of patients with LV dysfunction and frequent VPDs undergoing ablation.4 – 8 However, the lack of improvement in a substantial proportion of patients (21% of the patients in this cohort) remained unexplained. While no single cutoff value completely discriminated reversible from irreversible LV dysfunction, this study suggests that those with a VPD QRS duration of ⱖ170ms are unlikely to normalize their LV function, even with elimination of their ectopy. Importantly, the site of origin of the VPDs did not appear to influence this relationship.

Deyell et al

Frequent VPDs and LV Dysfunction

1471

Figure 3 Sample VPDs originating from the same site of origin (left-right coronary cusp commissure) in patients with normal LV function and reversible or irreversible cardiomyopathy. Two VPD examples for each condition are shown. LV ⫽ left ventricular; VPD ⫽ ventricular premature depolarization.

We also identified several important differences in baseline characteristics between patients with impaired and preserved LV function. Although the rate of symptoms among patients with preserved LV function may have been overestimated (symptoms being the main indication for ablation in this group), only a minority of patients with impaired LV function presented with palpitations. This finding is consistent with prior studies.5,7 The frequent lack of symptoms makes longitudinal studies among patients with frequent VPDs challenging, and it is difficult to establish the relative onset of the VPDs and LV dysfunction.

Table 3 Multivariate analysis of characteristics associated with the presence of LV dysfunction

Covariate VPD QRS duration, per 10-ms increase %VPDs on Holter monitoring, per 1% increase

Adjusted OR for LV dysfunction (95% CI)

Likelihood ratio test P

12.09 (4.12–35.46)

⬍.001

1.05 (0.99–1,12)

.094

Note that the presence of prior symptoms of congestive heart failure perfectly predicted the presence of LV dysfunction and therefore could not be included in the final model. CI ⫽ confidence interval; LV ⫽ left ventricle; OR ⫽ odds ratio; VPD ⫽ ventricular premature depolarization.

An association between VPD QRS duration and the presence of baseline LV dysfunction among patients with frequent VPDs has been previously reported,7 though recovery of LV function after ablation was not specifically evaluated. A similar association between VPD QRS duration and LV function has also been shown among patients being evaluated with coronary artery disease.12 The underlying nature of the association between VPD QRS duration and reversibility of LV dysfunction is not known. Although dyssynchrony may play a role, this study did not demonstrate that septal sites (with narrower QRS durations) had

Table 4 Multivariate analysis for predictors of partially reversible or irreversible LV dysfunction

Covariate

Adjusted OR for partially reversible/irreversible LV Likelihood ratio dysfunction (95% CI) test P

VPD QRS duration, 5.07 (1.22–21.01) per 10-ms increase Sinus QRS duration, 1.01 (0.96–1.06) per 10-ms increase LV ejection fraction 1.00 (0.89–1.03) prior to ablation, per 1% increase

.003

.783

.962

CI ⫽ confidence interval; LV ⫽ left ventricle; OR ⫽ odds ratio; VPD ⫽ ventricular premature depolarization.

1472

Heart Rhythm, Vol 9, No 9, September 2012

Table 5 Sensitivity and specificity of VPD QRS cutoff values for the presence of partially reversible/irreversible LV dysfunction VPD QRS duration (ms)

Sensitivity (%)

Specificity (%)

155 160 165 170 175

100 92 77 54 31

33 42 71 92 100

LV ⫽ left ventricle; VPD ⫽ ventricular premature depolarization.

less propensity toward irreversible LV dysfunction than did other sites. We favor the hypothesis that VPD QRS duration is a manifestation of the degree of myofibril disarray and fibrosis of the underlying myocardium, the dominant pathologic findings in idiopathic LV cardiomyopathy.13,14 Patients with increased VPD QRS durations may have a greater substrate abnormality and consequently are less likely to recover after VPD elimination. Sinus QRS duration would also be affected, but rapid conduction along the His-Purkinje system may mask subtle differences. VPD QRS duration is likely a better barometer of underlying substrate abnormality as conduction does not (typically) utilize the His-Purkinje system. Identification of microscopic substrate abnormality is not readily possible by other techniques, including imaging modalities. Whether such fibrosis is the direct consequence of VPDs or rather the result of an independent myopathic process with concurrent VPDs cannot be answered by this study alone. A major limitation of this, and many prior studies, is that subjects were enrolled at the time of ablation. As longitudinal follow-up prior to ablation was not systematic, we were unable to examine the relative time course and risk factors for the development of LV dysfunction in the presence of frequent VPDs. Another limitation was that only subjects undergoing ablation were included and therefore the findings may be influenced by referral bias. We were unable to track patients not referred for or who refused ablation. We attempted to minimize the influence of referral bias by restriction of our cohort to only those with ⱖ10% VPDs as there is likely a bias against ablation among patients with impaired LV function who have infrequent (⬍10%) VPDs. The proportion of patients with VPDs of LV origin in this study was higher (67%) than would be expected in the general VPD population. This likely reflects a referral bias toward more complex patients seen in our practice and may limit the generalizability of our findings. Finally, we were unable to evaluate clinical outcomes such as heart failure or arrhythmic death. Unfortunately, the rarity of frequent VPDs in the general population combined with a low clinical event rate means that the sample size required to evaluate such outcomes would be impossible to achieve from a single center alone.

Conclusion This study identified the VPD QRS duration as an independent predictor of lack of recovery of LV function after successful ablation. Furthermore, a gradient of VPD QRS duration was observed between those with normal LV function and those with reversible and irreversible LV dysfunction. This preliminary data suggest that VPD QRS duration may have a potential role in risk stratifying and guiding therapy in patients presenting with frequent VPDs. We hypothesize that the VPD QRS duration is a marker of the presence and severity of underlying substrate abnormality and may provide physicians with a useful and simple metric to help guide clinical decision making in this disorder.

Appendix Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.hrthm.2012.05.019.

References 1. Duffee DF, Shen WK, Smith HC. Suppression of frequent premature ventricular contractions and improvement of left ventricular function in patients with presumed idiopathic dilated cardiomyopathy. Mayo Clin Proc 1998;73:430 – 433. 2. Betensky BP, Park RE, Marchlinski FE, et al. The V(2) transition ratio: a new electrocardiographic criterion for distinguishing left from right ventricular outflow tract tachycardia origin. J Am Coll Cardiol 2011;57:2255–2262. 3. Mountantonakis SE, Frankel DS, Gerstenfeld EP, et al. Reversal of outflow tract ventricular premature depolarization-induced cardiomyopathy with ablation: effect of residual arrhythmia burden and preexisting cardiomyopathy on outcome. Heart Rhythm 2011;8:1608 –1614. 4. Baman TS, Lange DC, Ilg KJ, et al. Relationship between burden of premature ventricular complexes and left ventricular function. Heart Rhythm 2010;7:865– 869. 5. Hasdemir C, Ulucan C, Yavuzgil O, et al. Tachycardia-induced cardiomyopathy in patients with idiopathic ventricular arrhythmias: the incidence, clinical and electrophysiologic characteristics, and the predictors. J Cardiovasc Electrophysiol 2011;22:663– 668. 6. Kanei Y, Friedman M, Ogawa N, Hanon S, Lam P, Schweitzer P. Frequent premature ventricular complexes originating from the right ventricular outflow tract are associated with left ventricular dysfunction. Ann Noninvasive Electrocardiol 2008;13:81– 85. 7. Del Carpio Munoz FD, Syed FF, Noheria A, et al. Characteristics of premature ventricular complexes as correlates of reduced left ventricular systolic function: study of the burden, duration, coupling interval, morphology and site of origin of PVCs. J Cardiovasc Electrophysiol 2011;22:791–798. 8. Takemoto M, Yoshimura H, Ohba Y, et al. Radiofrequency catheter ablation of premature ventricular complexes from right ventricular outflow tract improves left ventricular dilation and clinical status in patients without structural heart disease. J Am Coll Cardiol 2005;45:1259 –1265. 9. Niwano S, Wakisaka Y, Niwano H, et al. Prognostic significance of frequent premature ventricular contractions originating from the ventricular outflow tract in patients with normal left ventricular function. Heart 2009;95:1230 –1237. 10. Mountantonakis SE, Frankel DS, Gerstenfeld EP, et al. Reversal of outflow tract ventricular premature depolarization-induced cardiomyopathy with ablation: effect of residual arrhythmia burden and preexisting cardiomyopathy on outcome. Heart Rhythm 2011;8:1608 –1614. 11. Malm S, Frigstad S, Sagberg E, Larsson H, Skjaerpe T. Accurate and reproducible measurement of left ventricular volume and ejection fraction by contrast echocardiography: a comparison with magnetic resonance imaging. J Am Coll Cardiol 2004;44:1030 –1035. 12. Moulton KP, Medcalf T, Lazzara R. Premature ventricular complex morphology: a marker for left ventricular structure and function. Circulation 1990;81:1245–1251. 13. Roberts WC, Siegel RJ, McManus BM. Idiopathic dilated cardiomyopathy: analysis of 152 necropsy patients. Am J Cardiol 1987;60:1340 –1355. 14. Unverferth DV, Baker PB, Swift SE, et al. Extent of myocardial fibrosis and cellular hypertrophy in dilated cardiomyopathy. Am J Cardiol 1986;57:816 – 820.