Background
Indications for prophylactic implantable cardioverter-defibrillator implantation in patients with nonischemic dilated cardiomyopathy (DCM) are based on left ventricular (LV) ejection fraction (LVEF), although LVEF has limited ability to predict arrhythmias. It has recently been shown that strain echocardiography can predict ventricular arrhythmias in patients after myocardial infarction. The aim of this study was to evaluate whether strain echocardiography may help in the risk stratification of ventricular arrhythmias in patients with DCM.
Methods
Ninety-four patients with nonischemic DCM were prospectively included. By speckle-tracking strain echocardiography, global longitudinal strain was calculated as the average of peak longitudinal strain from a 16-segment LV model. The time interval from electrocardiographic peak R to peak negative strain was assessed in each LV segment. Mechanical dispersion was defined as the standard deviation of time to peak negative strain from 16 LV segments.
Results
After a median of 22 months of follow-up (range, 1–46 months), 12 patients (13%) had experienced arrhythmic events, defined as sustained ventricular tachycardia or cardiac arrest. LVEF and global longitudinal strain were reduced in patients with DCM with arrhythmic events compared with those without (28 ± 10% vs 38 ± 13%, P = .01, and −6.4 ± 3.3% vs −12.3 ± 5.2%, P < .001, respectively). Global longitudinal strain showed greater area under the curve than LVEF to identify arrhythmic events in receiver operating characteristic curve analyses ( P = .05). Patients with arrhythmic events had increased mechanical dispersion (98 ± 43 vs 56 ± 18 ms, P < .001). Mechanical dispersion predicted arrhythmias independently of LVEF (hazard ratio, 1.28; 95% confidence interval, 1.11–1.49; P = .001).
Conclusions
Global longitudinal strain is a promising marker of arrhythmias. Mechanical dispersion predicted arrhythmic events in patients with DCM independently of LVEF. Strain echocardiography may help in the risk stratification of patients with DCM not fulfilling current implantable cardioverter-defibrillator indications.
The prediction of ventricular arrhythmias in patients with nonischemic dilated cardiomyopathy (DCM) is challenging, and questions about indications for prophylactic implantable cardioverter-defibrillator (ICD) treatment remain. Prophylactic ICD implantation in patients with nonischemic DCM has been shown to be efficient. Current indications for prophylactic ICD treatment in these patients are based on left ventricular (LV) ejection fraction (LVEF), with a threshold value < 35%. A large proportion of patients who experience ventricular arrhythmias, however, have LVEFs > 35%. A number of other diagnostic tests have been proposed to improve the accuracy of selection of patients with DCM for ICD therapy. However, currently available data are not sufficient to provide additional risk stratification tools beyond LVEF for the selection of ICD candidates in patients with DCM.
Myocardial strain derived by echocardiography can accurately quantify regional myocardial timing and function. Global longitudinal strain has been shown to be more accurate than LVEF in quantifying LV function and to possess prognostic impact. Myocardial mechanical dispersion by strain echocardiography is a measure of inhomogeneous ventricular contraction. We have recently shown that mechanical dispersion predicted ventricular arrhythmias in patients after myocardial infarction. In the present study, we hypothesized that global longitudinal strain may be a more accurate marker of ventricular arrhythmias in patients with DCM compared with LVEF. Furthermore, we hypothesized that mechanical dispersion may be a predictor of ventricular arrhythmias in these patients independently of LVEF and therefore may contribute to the risk stratification of these patients.
Methods
Study Population
We prospectively included 94 patients with nonischemic DCM at two different centers: the University Hospital of Jena (Jena, Germany) and Oslo University Hospital, Rikshospitalet (Oslo, Norway).
Inclusion criteria were LVEF < 50% and a dilated left ventricle, with LV end-diastolic diameter > 30 mm/m 2 (indexed to body surface area). The date of the echocardiographic examination was defined as the study start. Patients were followed for ≥6 months or to the date of a defined end point. End points were defined as all-cause mortality in addition to ventricular arrhythmic events, including sudden cardiac arrest, documented sustained ventricular tachycardia, appropriate therapy (antitachycardia pacing or shock) from ICDs implanted for primary prophylaxis, and syncope with probable cardiac cause. Patients with ventricular arrhythmias before inclusion and those with ICDs implanted for secondary prevention were not included in the study. Exclusion criteria were coronary artery disease with >50% stenosis on elective angiography, previous myocardial infarction, valvular disease more than moderate, mechanical valves, and previous treatment of malignancy. All patients were given optimal pharmacologic therapy, including angiotensin-converting enzyme inhibitors or angiotensin receptor blockers, β-blockers, and spironolactone, if tolerated.
Echocardiography
The echocardiographic studies were performed using Vivid 7 systems (GE Vingmed Ultrasound AS, Horten, Norway) and analyzed using commercially available software (EchoPAC; GE Healthcare, Milwaukee, WI). LVEF was assessed using Simpson’s biplane method. LV mass (LVM) was calculated from M-mode measurements and corrected for height 1.7 (LVM/height 1.7 ).
By speckle-tracking echocardiography, longitudinal strain was obtained from apical four-chamber, two-chamber, and long-axis views. Three cardiac cycles from each view were recorded for offline analyses with a frame rate > 70 frames/sec. Peak negative longitudinal strain was assessed in 16 LV segments as the most negative value during the cardiac cycle. The time interval from electrocardiographic peak R to peak negative strain was assessed in each LV segment ( Figure 1 ). Mechanical dispersion was defined as the standard deviation of time to peak negative strain from the 16 LV segments. The maximum longitudinal strain amplitude, either negative or positive, was assessed as peak strain and averaged to global longitudinal strain from a 16-segment LV model.
Mitral inflow E velocity was recorded using pulsed Doppler. The e′ velocity by tissue Doppler, averaged from the septal and lateral mitral annuli, was recorded and the E/e′ ratio calculated.
Electrocardiography
QRS durations and QT intervals were measured on 12-lead electrocardiograms recorded at 50 mm/sec. The electrocardiograms were enlarged two times, and QRS durations were measured manually in all 12 leads. The longest QRS duration was selected as the QRS duration. QRS dispersion was calculated as the difference between the longest and shortest QRS durations and as the standard deviation from the 12 QRS durations. QT intervals were corrected for heart rate using Bazett’s formula. No patients underwent ventricular pacing during electrocardiographic or echocardiographic recordings.
Echocardiographic data (K.H.H., B.G., and T.D.) and electrocardiographic data (K.H.H. and B.G.) were analyzed by independent observers blinded to clinical data.
Written informed consent was given by all participants. The study was approved by the regional committee for medical research ethics.
Statistical Analyses
Continuous data are presented as mean ± SD or as median (range). Comparisons of means were analyzed using unpaired t tests (SPSS version 16; SPSS, Inc., Chicago, IL). Proportions were compared using χ 2 or Fisher’s exact tests. Correlations between continuous variables were assessed using linear regression. Cox regression analysis was performed to identify predictors of arrhythmic events. Multivariate Cox regression analyses were performed by including significant variables from the univariate model. A strong correlation was observed between LVEF and global longitudinal strain. Global longitudinal strain and LVEF were therefore not included in multivariate analyses together. Receiver operating characteristics (ROC) curves were created and compared with each other by using nonparametric U tests (Analyse-it; Analyse-it Software Ltd., Leeds, United Kingdom). The value closest to the upper left corner of the ROC curve was defined as the cutoff value for optimal sensitivity and specificity for the ability of the chosen parameters to identify arrhythmic events. Reproducibility was expressed as intraclass correlation coefficients. P values were two sided, and results < .05 were considered significant.
Results
Study Population
Clinical characteristics are presented in Table 1 . During a median of 22 months of follow-up (range, 1–46 months), 12 patients (14%) experienced severe arrhythmic events. Of these, one patient died of electrical storm, two were resuscitated from ventricular fibrillation, six received appropriate therapies for ventricular tachycardia from a cardiac resynchronization therapy (CRT) with a defibrillator (CRT-D) device/primary-prophylaxis ICD, and two had repetitive sustained ventricular tachycardia before heart transplantation. One patient had syncope with probable cardiac cause. All-cause mortality, including cardiac transplantation, was 6% ( n = 6). In addition to the three patients described above (one with electrical storm and two with heart transplantation), two patients died and one underwent heart transplantation because of progressive heart failure.
| Variable | DCM without arrhythmias ( n = 82) | DCM with arrhythmias ( n = 12) | P ∗ |
|---|---|---|---|
| Age (y) | 50 ± 11 | 44 ± 18 | .20 |
| Heart rate (beats/min) | 74 ± 20 | 74 ± 18 | .93 |
| Men/women | 66/16 | 10/2 | .99 |
| NYHA class | 2.6 ± 0.8 | 2.8 ± 0.7 | .27 |
| QRS duration (msec) | 115 ± 35 | 141 ± 31 | .02 |
| QRS dispersion (msec) | 36 ± 12 | 38 ± 21 | .22 |
| SD QRS dispersion (msec) | 11 ± 3 | 9 ± 4 | .33 |
| Corrected QT interval (msec) | 428 ± 83 | 493 ± 51 | .01 |
| BSA (m 2 ) | 2.03 ± 0.22 | 2.09 ± 0.20 | .38 |
| BMI (kg/m 2 ) | 27 ± 4 | 28 ± 5 | .36 |
| Echocardiographic findings | |||
| LVEDD (mm) | 63 ± 9 | 77 ± 10 | <.001 |
| LVESD (mm) | 51 ± 12 | 66 ± 14 | <.001 |
| LVEDV (mL) | 165 ± 68 | 251 ± 113 | <.001 |
| LVESV (mL) | 105 ± 58 | 186 ± 102 | <.001 |
| LVEF (%) | 38 ± 13 | 28 ± 10 | .01 |
| Global longitudinal strain | −12.3 ± 5.2 | −6.4 ± 3.3 | <.001 |
| Mechanical dispersion (ms) | 56 ± 18 | 98 ± 43 | <.001 |
| LVM/height 1.7 (g/m 1.7 ) | 119 ± 38 | 163 ± 42 | <.001 |
| E/e′ ratio | 12.5 ± 6.6 | 22.5 ± 12.2 | <.001 |
At inclusion, five patients had previously implanted CRT-D devices ( n = 2) or ICDs for primary prophylaxis ( n = 3). Sixteen patients received CRT-D devices or ICDs at the time of inclusion, and 13 patients received CRT-D devices or ICDs during follow-up.
At inclusion, four patients were in New York Heart Association class I, 37 were in class II, 44 were in class III, and nine were in class IV.
LV Function, LV Volume, and LVM
Patients with arrhythmic events had lower LVEF compared with those without ( P = .01; Table 1 ). In all, 45 patients had LVEFs < 35%, and nine of these (20%) had arrhythmic events during follow-up. In patients with LVEFs > 35% ( n = 49), three patients (6%) had arrhythmic events ( P = .06). By ROC analyses, the optimal cutoff value for LVEF in this study was 40%, which detected patients with arrhythmias with sensitivity of 92% (95% confidence interval [CI], 61%–100%) and specificity of 45% (95% CI, 34%–56%) ( Figure 2 ). For the clinically used criterion of LVEF < 35%, sensitivity was 75% (95% CI, 43%–95%) and specificity 54% (95% CI, 42%–65%).
Global longitudinal strain was reduced in DCM patients with arrhythmic events ( P < .001; Table 1 ). By ROC analyses, global longitudinal strain had greater area under the curve compared with LVEF ( P = .05; Figure 2 ). The optimal cutoff value for global longitudinal strain was −7.1% and detected arrhythmic events with sensitivity of 67% (95% CI, 35%–90%) and specificity of 85% (95% CI, 75%–92%). Nineteen patients had global longitudinal strain worse than −7.1%. Of these, eight (42%) had arrhythmic events, while only four of 75 patients (5%) with global longitudinal strain better than −7.1% had arrhythmias ( P < .001). Global longitudinal strain and LVEF were significantly correlated ( r = 0.84, P < .001). Global longitudinal strain and mechanical dispersion were significant predictors of arrhythmic events in the multivariate Cox regression model ( Table 2 ). By including LVEF instead of global longitudinal strain in the model, only mechanical dispersion remained a significant predictor (hazard ratio, 1.3; 95% CI, 1.1–1.5; P = .001), while LVEF ( P = .18) and QRS duration ( P = .40) did not reach significance.
| Parameters | Univariate | P | Multivariate | P |
|---|---|---|---|---|
| Age (per year increase) | 0.97 (0.92–1.01) | .12 | ||
| QRS duration (per 10-msec increase) | 1.22 (1.04–1.44) | .02 | 1.10 (0.91–1.33) | .40 |
| LVEF (per 5% decrease) | 1.52 (1.11–2.08) | .01 | ||
| Global longitudinal strain (per 1% increase) | 1.37 (1.15–1.62) | <.001 | 1.26 (1.03–1.54) | .02 |
| Mechanical dispersion (per 10-msec increase) | 1.39 (1.21–1.58) | <.001 | 1.20 (1.03–1.40) | .02 |
LVM/height 1.7 was higher in patients with DCM and arrhythmias compared with those without ( P < .001). LVM/height 1.7 was increased in those with QRS > 120 msec compared with those with normal QRS durations (138 ± 50 vs 118 ± 30 g/m 1.7 , P < .01).
The E/e′ ratio as a parameter of increased filling pressures was increased in the total study population (13.6 ± 7.9) and markedly increased in patients with arrhythmic events ( P < .001; Table 1 ). The area under the curve for E/e′ ratio by ROC analyses was 0.72 (95% CI, 0.49–0.96). E/e′ ratio was significantly correlated with global longitudinal strain ( r = 0.52, P < .001) and LVEF ( r = 0.41, P < .001) and was weakly correlated with mechanical dispersion ( r = 0.35, P = .003).
Mechanical and Electrical Parameters of Dispersion
Mechanical dispersion was increased in patients with DCM with arrhythmias compared with those without ( P < .001; Table 1 ). In multivariate analysis, mechanical dispersion was an independent predictor of arrhythmias ( P = .001; Table 2 ). Mechanical dispersion > 72 msec detected arrhythmic events with sensitivity of 67% (95% CI, 35%–90%) and specificity of 89% (95% CI, 80%–95%). Kaplan-Meier analyses showed that patients with mechanical dispersion > 72 msec had a significantly higher event rate compared with those with mechanical dispersion < 72 msec ( P < .001; Figure 3 ). Mechanical dispersion was weakly correlated with QRS duration ( r = 0.23, P = .03).
