The present study consisted of a cross-sectional analysis of clinical and echocardiographic data from ambulatory CHF patients, followed in an outpatient “Heart Failure, Cardiomyopathy and Pre-Transplant” Clinic, included consecutively, between September and November 2012. Eligible patients were CHF ambulatory patients, with EF <45% under optimal medical therapy for at least three months. Exclusion criteria included recent acute coronary syndrome, myocardial revascularization in the last three months, recent or planned cardiac surgery within six months, significant primary valvular disease, recent onset cardiomyopathy (<6 months), early stage of potentially reversible LV dysfunction (<6 months) and the presence of concomitant serious illness.

Patients in atrial fibrillation (AF) were not excluded from the study.

Patient demographic, clinical information and laboratory data were collected during the clinical visit at the time of the echocardiographic study.

Application of the SHFM was performed in all patients, using the online module.12 There were no missing covariates for score calculation. Estimated SHFM-predicted life expectancy was used as the prognostic status indicator.

A standardized complete echocardiographic examination was performed using a commercially available Vivid 7 system (GE Vingmed, Horton, Norway), with a M4S (2.5-MHz) probe. All data were stored digitally for off-line analysis on Echo-Pac PC software (V8.1 GE, Horton Norway).

Standard echocardiographic techniques were used to obtain two-dimensional (2D) and Doppler measurements in accordance with European Association of Echocardiography guidelines.13,14

LV volumes and LVEF were calculated by the biplane Simpson's method from apical 4- and 2-chamber views.

Peak tricuspid annular systolic velocity (S′), by tissue Doppler imaging (TDI) was used to assess right ventricle systolic function (RVSF).

The ratio of early transmitral velocity (E wave by conventional pulsed Doppler) to tissue Doppler mitral annular early diastolic velocity (e′ by TDI) was used to assess LV end-diastolic pressure (LVEDP).

GLS was assessed in the 18 LV segments from the three standard apical views (4-chamber, 2-chamber, 3-chamber), using a speckle tracking software. For strain processing, the peak of the R wave on the electrocardiogram was used as the reference time point for end-diastole. The endocardial border was traced manually in the end-diastolic frame. The software subsequently and automatically traced the borders in the other frames. Segments which failed to track were manually adjusted by the operator. Any segments which subsequently failed to track were excluded from the analysis.

In AF patients, cine loops were analyzed to determine which index beat should be used to GLS calculation. The speckle tracking software was able to successfully and appropriately track the frame-to-frame movement of LV segments if cardiac cycle of the index beat was >500 ms and if the preceding and pre preceding intervals of the index beat were of nearly equal duration (<60 ms).15

GLS was obtained only in the case of adequate tracking quality ≥5 of the six segments per view.

Graphical displays of deformation parameters for each segment were then generated automatically. GLS was obtained by averaging the peak strain values from the 18 regional longitudinal strain curves.

Fifteen patients were randomly selected for evaluation of the interobserver variability of GLS measurement by two independent observers. To get the intraobsever variability, the same measurement was repeated three to six months apart. Reproducibility was expressed as the mean absolute difference.

Continuous variables with normal distribution were expressed as mean±SD and skewed variables as median and quartiles. Categorical variables were expressed as percentages. The comparison of continuous variables was performed by an independent sample t test. For categorical variables, the chi-square test was used.

A Pearson correlation analysis was used to evaluate the strength of the association between continuous variables.

A logistic regression model was used for multivariate analysis to investigate which prognostic factors identified using univariate analysis were independently associated with a lower survival (<10 years of life expectancy derived from SHFM).

Receiver-operating characteristic (ROC) curve analysis was used to determine optimal cutoff value of continuous variable. The best cutoff value was defined as the point with the highest sum of sensitivity and specificity.

Two-tailed p-values <0.05 were considered statistically significant. Statistical analysis was performed using SPSS Statistics version 19.0 (SPSS Inc., Chicago, Illinois).

Dyslipidemia, hypertension, and diabetes were present in 43 (80%), 23 (43%) and 16 (30%) patients, respectively. Ischemic etiology was present in 30% and 59% of the patients were in NYHA functional class II. Median BNP level was relatively low: 186.3 [P25-75 55.9–560.1] pg/ml.

Most patients were treated with beta-blockers (94%) and angiotensin converting enzyme inhibitors or angiotensin-receptor blockers (94%). Diuretics, spironolactone and statins were also frequently used in this patient population (Table 1).

In addition, 11 (20%) patients had an implantable cardioverter defibrillator (ICD) device and 6 (11%) had a cardiac resynchronization therapy defibrillator (CRT-D) device.

End diastolic and systolic volumes averaged 116.0±40.25 ml/m2 and 86.9±38.56 ml/m2, respectively. Mean LVEF was 27.0±8.8% and E/e′ ratio averaged 12.7±5.4. RVSD was observed in 11% of the patients and mean peak systolic velocity of the tricuspid annulus by tissue Doppler imaging (RV S′ TDI) averaged 11.8±2.3 cm/s.

The mean GLS was −10.4±3.1%.

GLS was found to be highly associated with various clinical and echocardiographic variables; relevant correlations are summarized in Table 2.

GLS was significantly related to NYHA functional class (r=0.41, p=0.002) and BNP levels (r=0.47, p=0.001).

GLS presented a good correlation with LV end-diastolic volume index (r=0.60, p<0.001), LV end-systolic volume index (r=0.68, p<0.001) and with LVEF (r=-0.69, p<0.001).

The magnitude of GLS showed a moderate correlation with RVSF assessed by TDI-derived peak systolic tricuspid velocity (r=−0.32, p=0.031) and LV filling pressure, assessed by E/e′ ratio (r=0.35, p=0.014) and LA maximal volume index (r=0.57, p<0.001).

Finally, a significant correlation was found between GLS and SHFM prognostic estimates for life expectancy (r=−0.41, p=0.002).

Univariate analysis using a logistic regression model (Table 3) showed that GLS predicted a life expectancy <10 years, derived from SHFM (OR 1.544, p=0.008). Other relevant variables found to be significantly associated to a worse survival included BNP levels (OR 1.001, p=0.013), E/e′ ratio (OR 1.241, p=0.012) and LA maximal volume (OR 1.071, p=0.021).

In a multivariate analysis adjusted to E/e′ ratio and BNP (significant variables not contributing to the SHFM), GLS remained a significant independent predictor of survival (OR 2.614, 95% CI: 1.010–6.763, p=0.048).

The area under the receiver operating characteristic curve (AUC) for GLS to predict a life expectancy <10 years was highly relevant (AUC: 0.802 [0.653–0.951]) and a GLS cut-off >−9.5% identified patients with poor prognosis (80% sensitivity, 65% specificity, p=0.003) – Figure 1.

Figure 1.

The area under the receiver operating characteristic curve for GLS to predict a life expectancy <10 years. AUC: area under curve; ROC: receiver operating characteristic.

(0,09MB).

The intraobserver and interobserver mean difference for GLS measurement were 0.47±0.40 and 0.86±0.56, respectively.

All conclusions are supported by a developed prognostic model and not on the real evolution and follow-up of a CHF population. Nevertheless, SHFM is a widely validated model for predicting survival and assessing disease severity in outpatients with CHF, based on multiple clinical variables known to have prognostic impact in CHF patients.

Speckle tracking 2D strain is highly dependent on image quality and the inherent limitation therein. In addition, radial and circumferential strains were not explored. However, longitudinal deformation is a particular marker of subendocardial function, which is frequently the first layer to be affected when myocardial function deteriorates. Thus, longitudinal deformation appears to be a more sensitive marker of cardiac function compared to radial or circumferential strain.22

Finally, our conclusions apply to CHF outpatients and should not be extrapolated for hospitalized patients or acute heart failure patients.

The authors declare that no experiments were performed on humans or animals for this study.

The authors declare that they have followed the protocols of their work center on the publication of patient data

The authors declare that no patient data appear in this article