Background
The risk stratification of patients with left ventricular (LV) dysfunction can be performed using echocardiographic parameters such as the ejection fraction (EF). Recently, new technologies based on deformation measurements have been shown to identify early myocardial dysfunction before EF decrease. Consequently, tools such as two-dimensional strain have been incorporated into echocardiographic systems, allowing for fast, reliable, and reproducible calculation of longitudinal components of LV systolic deformation. The hypothesis in this study was that as a more sensitive marker of LV dysfunction, longitudinal strain would allow for the risk stratification of patients with heart failure.
Methods
This multicenter study included 147 patients with heart failure with LV EFs ≤ 45% (mean age, 64 ± 14 years; 74% men; mean LV EF, 29.9 ± 8.9%). Conventional echocardiographic parameters as well as global and segmental longitudinal strain were measured and compared with these values in a control population. Patients were monitored for cardiac events, defined as a composite criterion, over 12 months.
Results
Clinical events were observed in 20% of patients during the 12-month follow-up period. On receiver operating characteristic curve analysis, global longitudinal strain had the highest prognostic value (area under the curve, 0.83) and the highest combination of sensitivity (73%) and specificity (83%), using a cutoff value of −7%.
Conclusion
Strain assessment is highly feasible and reliable in patients with LV dysfunction and allows for cardiovascular risk stratification in patients with heart failure with greater accuracy than LV EF.
Two-dimensional (2D) echocardiography is widely used in clinical routine to assess left ventricular (LV) function, and its value is acknowledged by guidelines. Among the proposed validated parameters, the ejection fraction (EF) is recognized as an important criterion for heart failure (HF) prognosis and for pharmacologic, defibrillator, and resynchronization therapies. A strong relationship between the LV EF and mortality was established in patients with HF. However, EF quantification requires the manual endocardial tracking of end-diastolic and end-systolic frames from 2D imaging, which demands experience, is time consuming, and is restricted by a high level of measurement variability. The need for formal quantitative assessment of myocardial systolic function remains a significant challenge. Doppler tissue imaging (DTI) is emerging as a useful echocardiographic tool for the quantitative assessment of LV systolic and diastolic function. Recent studies have explored the prognostic role of DTI-derived parameters in major cardiac diseases such as HF, acute myocardial infarction, and hypertension. Myocardial mitral annular or basal segmental systolic (S′) and early diastolic (Ea or E′) velocities were shown to predict mortality or cardiovascular events in these cardiac diseases. Patients with reduced S′ or E′ values, in particular, have a very poor prognosis. However, for a predefined value of the EF in patients with intermediate-stage HF, a large range of DTI values may be observed, highlighting the discrepancy between the two parameters.
Recent advances in three-dimensional echocardiography are likely to improve the accuracy of LV volumes, but these techniques not yet feasible in clinical routine.
Other emerging techniques have been developed and provide new hope by directly targeting myocardial contraction or deformation rather than its consequences (i.e., DTI velocities). The first approaches are based on velocity imaging data sets, with encouraging experimental and clinical results. However, feasibility and reproducibility have proven to be problematic in practical routine. More recently, deformation analyses were switched from color-derived DTI to pure grayscale imaging, applying the speckle-tracking method. This method was validated in experimental studies and implemented in conventional echocardiographic systems for direct clinical applications. For example, longitudinal strain was proven to be effective in the early detection of contraction abnormalities in cardiac diseases such as hypertrophic cardiomyopathy with preserved EF.
On the basis of these reflections, our hypothesis was that the stratification of patients with abnormal LV function would be better predicted using global longitudinal strain (GLS) than conventional echocardiographic parameters. The aims of the study were (1) to assess GLS in a population of patients with HF, (2) to compare GLS and other parameters in subgroups of patients with or without cardiac events, and (3) to determine a GLS cutoff value associated with poor cardiac outcomes.
Methods
Patient Population
One hundred eighty-one patients with chronic cardiomyopathy and LV EFs ≤ 45% were enrolled in a prospective multicenter study between January and November 2007 (Cardiologic Hospital and INSERM 828, Bordeaux University Hospital Center, Pessac, France; Timone Hospital, Cardiology Department, Marseille, France; and Pontchaillou University Hospital Center, Rennes, France). The exclusion criteria were recent acute HF (<3 months with clinical congestive HF), unstable patients (cardiac shock), severe valvular diseases (mitral and aortic regurgitation grades 3 and 4 and severe aortic and mitral stenosis ), atrial fibrillation, cardiac resynchronization therapy, valvular prosthesis, and poor window quality, defined by at least three nonvisualized segments. Overall, 8% of patients were excluded because of incomplete image data sets or insufficient image quality. In total, 147 patients for whom clinical examinations, biologic characteristics (low-density lipoprotein and serum creatinine), electrocardiography, and transthoracic echocardiographic studies were obtained were included in our study.
Analyses of echographic parameters were performed at an independent core laboratory center. Conventional echocardiographic parameters as well as GLS and segmental longitudinal strain were measured.
Follow-Up and End Points
A follow-up screening for cardiac events and death was carried out after 12 months. Cardiac events and death were defined as a composite criterion including rehospitalization for acute HF (including the worsening of functional status) and death from cardiac events (ventricular tachycardia, sudden death, acute HF, or myocardial infarction). The time to the first event was taken into account for the survival analysis. In case of death, the reasons were checked on hospital records or death certificates. The causes of death were identified in all patients.
Reproducibility
The studies were analyzed offline at a specialized core laboratory center and, for 20 patients, by a second blinded observer. In total, 20 randomly selected patients were evaluated twice by the same observer (for intraobserver variability), and 20 patients were evaluated by two different observers (for interobserver variability). Interobserver variability was calculated as the absolute difference divided by the average of the two observations times 100 for GLS. Intraobserver variability was calculated using the average differences among the 20 measurements and was computed as the absolute difference divided by the average of the two measurements times 100.
Echocardiographic Measurements
Conventional echocardiographic analysis was performed offline using an EchoPAC workstation (GE Healthcare, Milwaukee, WI).
Standard transthoracic echocardiographic examinations were recorded in all patients using a Vivid 7 digital ultrasound system (GE Vingmed Ultrasound AS, Horten, Norway). Three cardiac cycles were stored in cine loop format for offline analysis. LV and left atrial dimensions were measured according to the American Society of Echocardiography’s recommendations. The LV EF was measured using the biplane Simpson’s method. S′ systolic velocity at the lateral mitral annulus was assessed using pulsed-wave DTI. Diastolic function was evaluated by the analysis of the mitral Doppler inflow, pulsed-wave DTI at the lateral mitral annulus, and mitral propagation velocity using color M-mode echocardiography. Mitral regurgitation (MR) was obtained using the proximal isovelocity surface area method. Right ventricular function was assessed with the S′ systolic velocity in the lateral wall. Systolic pulmonary arterial pressure was obtained from the tricuspid regurgitation flow. In patients with adequate Doppler MR signals, dP/dt was determined noninvasively from the rate of change of MR velocity, according to American Society of Echocardiography guidelines.
Strain Measurements
Two-dimensional grayscale images were acquired in the standard apical four-chamber, three-chamber, and two-chamber views at a frame rate of ≥80 frames/sec (mean, 89 ± 4 frames/sec). The dimensions of the computation area were 80° for the angle and 13 cm for the depth. The left ventricle was divided into 17 segments, and each segment was analyzed individually. Using a dedicated software package (EchoPAC PC), GLS and segmental longitudinal strain were obtained as previously described by Leitman et al. Average longitudinal strains were calculated automatically by the software. Two-dimensional strain is a non-Doppler-based method for the evaluation of systolic strain using standard 2D acquisitions. After placing three endocardial markers in an end-diastolic frame, the software automatically tracks the contour on subsequent frames. Adequate tracking can be verified in real time and corrected by adjusting the region of interest or by manually correcting the contour to ensure optimal tracking. Two-dimensional longitudinal strain was assessed in apical views. Average longitudinal strains were calculated for the 17 segments.
Statistical Analyses
All values are expressed as mean ± SD or as percentages, as appropriate. Comparisons of continuous variables were performed using unpaired t tests. For categorical variables, the χ 2 test was used. A p value < .05 was considered to be significant. Receiver operating characteristic curve analysis was used to determine optimal cutoff values for continuous variables. The best cutoff value was defined as the point with the highest aggregate of sensitivity and specificity.
A backward stepwise multivariate logistic regression analysis was performed, with cardiac events occurring during the follow-up as dependent variables. The selection cutoff was set at 0.20. For model 1, the independent variables were age, gender, a pathologic LV EF, a pathologic average GLS, and a pathologic E/E′ ratio. Model 2 combined model 1 with New York Heart Association (NYHA) functional classes III and IV. The cutoff for continuous variables was set on the basis of the receiver operating characteristic analyses.
Survival analysis was performed using the Kaplan-Meier method, and p values were calculated using the log-rank test. All statistical analyses were performed using Statel software (AdScience, Paris, France).
Results
Patient Characteristics at Baseline
The study population’s baseline characteristics are shown in Table 1 . The mean age was 64.5 ± 14 years, 74% were men, and 42.8% presented with ischemic cardiomyopathy. Most patients were in NYHA class II at the time of inclusion. The mean QRS interval was 135 ± 35 msec, while the mean LV EF was 29.9 ± 8.9%.
| Variable | Value |
|---|---|
| Demographics | |
| Age (years) | 64.5 ± 14 |
| Men | 109 (74%) |
| BMI (kg/m 2 ) | 25 ± 5 |
| NYHA classes III and IV symptoms | 47 (32%) |
| Cardiovascular risk factors | |
| Hypertension | 70 (48%) |
| Dyslipidemia | 60 (41%) |
| Diabetes | 25 (17%) |
| Smoking (current) | 55 (37%) |
| CrCl (mL/min) | 65.63 ± 37 |
| LDL (g/L) | 1.0 ± 1.0 |
| LV EF | 29.9 ± 8.9 |
| QRS duration (ms) | 135 ± 35 |
| HR (beats/min) | 70 ± 15 |
| Chronic cardiomyopathy | |
| ICM | 63 (42.8%) |
| NICM | 84 (57.1%) |
Patient Characteristics and Echocardiographic Parameters According to Cardiac Events
Characteristics of patients with HF and conventional echocardiographic parameters according to cardiac events are shown in Table 2 . Overall, 20% of patients had cardiac events at the end of the follow-up period ( n = 30). Among them, 15 patients presented with worsening functional status, two were rehospitalized for acute HF, and one was rehospitalized for acute myocardial infarction. Twelve patients died: eight from acute HF, one from acute myocardial infarction, and three from ventricular tachycardia.
| Variable | Events ( n = 30) | No events ( n = 117) | P |
|---|---|---|---|
| Age (years) | 67 ± 14 | 63 ± 17 | .27 |
| Men | 24 (80%) | 85 (72%) | .41 |
| BMI (kg/m 2 ) | 25 ± 5.4 | 26 ± 5 | .32 |
| Hypertension | 16 (53%) | 54 (46%) | .48 |
| Dyslipidemia | 12 (40%) | 48 (41%) | .92 |
| Diabetes | 7 (23.3%) | 18 (15%) | .30 |
| Smoking (current) | 15 (50%) | 40 (34%) | .11 |
| NYHA classes I and II | 12 (40%) | 88 (75%) | <.005 |
| NYHA classes III and IV | 18 (60%) | 29 (25%) | <.005 |
| NICM | 14 (47%) | 47 (41%) | .52 |
| LV EDD (mm) | 69.2 ± 8.3 | 67 ± 10 | .40 |
| LV ESD (mm) | 60.1 ± 8.9 | 57 ± 11 | .09 |
| LV EDV (mL 3 ) | 184 ± 75.5 | 176 ± 76 | .62 |
| LV ESV (mL 3 ) | 141.3 ± 60.8 | 123 ± 59 | .12 |
| LV EF (%) | 24.3 ± 8.8 | 31 ± 8 | <.001 |
| LV S′ velocity (cm/sec) | 4.7 ± 1.3 | 5.3 ± 1.4 | .04 |
| dP/dt (mm Hg/sec) | 579 ± 174 | 695 ± 240 | .07 |
| E (cm/sec) | 81.5 ± 37.1 | 77 ± 26 | .45 |
| A (cm/sec) | 40.3 ± 31.9 | 57 ± 32 | .01 |
| E/E′ | 12.8 ± 7.4 | 10.9 ± 5.5 | .13 |
| Vp (cm/sec) | 37 ± 11 | 36 ± 11 | .62 |
| EORA of MR (mm 2 ) | 13 ± 24 | 6 ± 10 | .009 |
| LA area (cm 2 ) | 26 ± 8 | 23 ± 7 | .01 |
| RA area (cm 2 ) | 24 ± 9 | 18 ± 6 | <.0002 |
| RV S′ cm/sec | 8.5 ± 2 | 11 ± 3 | .0036 |
| SPAP (mm Hg) | 37 ± 5 | 33 ± 13 | .10 |
| Average GLS (%) | −5.72 ± 3.25 ∗ | −9.90 ± 3.07 †‡ | <.001 |
There was no statistical difference in terms of ischemic or nonischemic chronic cardiomyopathy. Among the patients with cardiac events, NYHA status and renal function were more impaired than in those without cardiac events, with a mean creatinine clearance of 53.6 ± 29.5 versus 75.1 ± 3 mL/min ( p < .0007). Patients with cardiac events at 12 months had statistically greater LV dysfunction than those without cardiac events: the mean LV EF was 24 ± 8% versus 31 ± 8% ( p < .0001), and the mean S′ systolic velocity at the mitral lateral annulus was 4.7 ± 1.3 versus 5.3 ± 1.4 cm/sec ( p = .04). The mean dP/dt was 667 ± 230 mm Hg/s, and the dP/dt was <600 mm Hg/s in 18% of patients with HF (51% of patients had no significant MR for dP/dt analysis). MR was statistically more severe in patients with cardiac events than in those without, with a mean effective regurgitant orifice area of 13 ± 24 versus 6 ± 10 mm 2 ( p = .0094). Additionally, right ventricular function was statistically more depressed in patients with cardiac events, as demonstrated by the mean S′ value of 8.5 ± 2.4 cm/sec ( p = .0035).
In patients with HF, GLS in the apical two-chamber, four-chamber, and three-chamber views was −9.8 ± 4.1%, −8.8 ± 3.7%, and −9.0 ± 4.3%, respectively. No significant difference was observed in patients with HF between strain averages in the apical, mid, and basal levels (−9 ± 6%, −8 ± 5%, and −8 ± 4%, respectively).
Last, GLS was significantly more depressed in patients with cardiac events than in those without, with a value of −5.7% versus −9.9% at 12 months.
Predictors of Cardiac Events on the Basis of Echocardiography
On the basis of the receiver operating characteristic curve analysis, each parameter was compared using the area under the curve to identify an LV dysfunction predictor of cardiac events. GLS showed the highest area under the curve (0.83), even compared with the LV EF (area under the curve, 0.72). The best combination of sensitivity and specificity was obtained for the mean GLS, with a cutoff of −7% (sensitivity, 73%; specificity, 83%; Figure 1 , Table 3 ).
| Parameter | AUC | Cutoff | Sensitivity (%) | Specificity (%) |
|---|---|---|---|---|
| LV EF (%) | 0.72 | 29.5 | 73 | 58 |
| EROA of MR (mm 2 ) | 0.60 | 15 | 27 | 88 |
| LV S′ (cm/sec) | 0.70 | 6 | 92 | 21 |
| RV S′ (cm/sec) | 0.61 | 8 | 50 | 80 |
| E/E′ | 0.58 | 11.3 | 57 | 66 |
| LA area (cm 2 ) | 0.65 | 19 | 96 | 30 |
| SPAP (mm Hg) | 0.62 | 31 | 76 | 47 |
| Average GLS (%) | 0.83 | −7 | 73 | 83 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
