Background
The aim of this case-control study was to assess the usefulness of three-dimensional (3D) speckle-tracking echocardiography in the evaluation of global left ventricular (LV) myocardial performance in adolescent and adult survivors of childhood cancers.
Methods
Fifty-three anthracycline-treated survivors of childhood cancers (mean age, 18.6 ± 5.1 years) and 38 controls were studied. Three-dimensional speckle-tracking echocardiography was performed to assess LV 3D global and segmental strain, time to peak segmental 3D strain, LV torsion, and ejection fraction. LV systolic dyssynchrony index (SDI) was calculated as the percentage of the standard deviation of times to peak strain of the 16 segments divided by the RR interval. A global performance index (GPI) was calculated as (global 3D strain × torsion)/SDI. The area under the receiver operating characteristic curve was calculated to determine the capability of various echocardiographic indices to discriminate between patients and controls.
Results
Compared with controls, patients had significantly reduced LV global 3D strain ( P < .001), torsion ( P < .001), and GPI ( P < .001) and greater SDI ( P < .001). All except the basal anteroseptal segment in patients had reduced regional 3D strain compared with controls ( P < .05 for all). Global 3D strain ( P = .018), SDI ( P = .003), and GPI ( P = .02) were correlated with cumulative anthracycline dose. The areas under the curves for GPI, global 3D strain, 1/SDI, torsion, and ejection fraction were 0.92, 0.79, 0.79, 0.79, and 0.78, respectively. A GPI cutoff of 10.6°/cm had sensitivity of 84.9% and specificity of 81.6% of differentiating patients from controls.
Conclusions
Three-dimensional speckle-tracking echocardiography enables the derivation of an index of LV global performance that incorporates LV 3D strain, dyssynchrony, and torsion for the sensitive detection of altered LV mechanics in childhood cancer survivors.
Although the incorporation of anthracyclines in the treatment regimen of childhood cancers has significantly improved patient survival, their long-term cardiotoxic side effects remain an issue of concern. The ability to detect myocardial dysfunction in the early subclinical phase may allow risk stratification and facilitate timely intervention.
Echocardiography remains the main imaging modality for the assessment of left ventricular (LV) function in children after anthracycline therapy. The past decade saw an evolution of echocardiographic imaging techniques in the evaluation of anthracycline cardiotoxicity. Although the conventional M-mode parameter of LV shortening fraction is commonly included in long-term follow-up protocols, its limitations, including the need for geometric assumption, an inability to take into account all myocardial segments, and variable reproducibility, are well recognized. Doppler tissue imaging, which permits direct interrogation of myocardial deformation, has demonstrated subclinical myocardial dysfunction in long-term survivors. Nonetheless, angle dependency, noise artifacts, confounding of interpretation by tethering of adjacent myocardium and cardiac translational movements, and assessment being confined to myocardial segments that move along the direction of the ultrasound beam have limited its application. We and others have recently shown potential advantages of the relatively angle independent two-dimensional (2D) speckle-tracking echocardiography (STE) in unveiling impairment of linear and twisting deformation of the LV myocardium in childhood cancer survivors. All of the previous cardiac functional studies have nonetheless primarily been focused on changes in LV cavity dimension, myocardial velocities, or tissue deformation in one or two dimensions.
Cardiac motion is three-dimensional (3D), and damage due to anthracyclines is not known to be limited to particular segments or territories of coronary perfusion. The recently developed 3D STE is a novel tool to noninvasively quantify LV function and dyssynchrony. Compared with the 2D modality, 3D STE can overcome the limitation of the failure of 2D STE to track out-of-plane speckle motion and allow efficient single acquisition with simultaneous assessment of all LV segments in the same 3D data set. The aim of this study was to investigate the usefulness of 3D STE and a LV global performance index (GPI) in the evaluation of global LV myocardial performance in adolescent and adult survivors of childhood cancers.
Methods
Subjects
This was a case-control study. Fifty-three anthracycline-treated survivors of childhood cancers who had been off treatment for ≥1 year were consecutively recruited from the oncology clinic. The following data were collected from the case notes: diagnosis, duration of follow-up, cumulative dose of anthracyclines, clinical evidence of heart failure, and the need for cardiac medications. Thirty-eight control subjects were studied. The healthy volunteers were invited to participate in the study by the research staff. They were screened by history and physical examination as appropriate to be free of cardiac disease and medical illnesses affecting cardiac function. These included healthy siblings of patients, subjects with functional heart murmurs, and those with nonspecific chest pain or palpitations for which organic causes were excluded after formal cardiac assessment. The body weights and heights of all subjects were measured, and the body surface areas and body mass indices were calculated accordingly. The institutional review board approved the study, and all patients and parents of minors gave informed written consent.
Three-Dimensional Speckle-Tracking Echocardiographic Assessment
All echocardiographic studies were performed using a commercially available system (Artida; Toshiba Medical Systems, Tokyo, Japan). Real-time 3D echocardiographic imaging was performed at the cardiac apex using a matrix-array transducer, with patients lying in the left lateral decubitus position. Full-volume acquisition, in which four adjacent subvolumes were captured over four consecutive cardiac cycles, was performed during a breath-hold to minimize artifacts between subvolumes. A wide sector was used to ensure that the entire LV cavity was included within the full-volume data set. Offline line analysis of the data sets was performed using 3D tracking software (Advanced Cardiology Package; Toshiba Medical Systems).
On the basis of the 3D volume data set, five planes were displayed for tracing of the endocardium and epicardium: the apical four-chamber view, the orthogonal two-chamber view, and three short-axis planes at near the apex, the midlevel, and the base of the left ventricle. The LV epicardium and endocardium were traced semiautomatically with refinement by further manual adjustments. After generation of the left ventricle 3D cast, the LV end-systolic and end-diastolic volumes and ejection fraction were derived. The software further automatically divided the left ventricle into 16 segments as defined by the American Society of Echocardiography. On the basis of the time-strain curves for each of the LV segments, the following parameters were determined: global and segmental 3D strain, LV twist and torsion (twist/LV length), and a 3D strain–derived systolic dyssynchrony index (SDI). The SDI was calculated as the standard deviation of the 16 times to the highest peak segmental systolic 3D strain, as a percentage of the RR interval.
The LV GPI, which takes into account the different aspects of LV mechanics, was calculated as (global 3D strain × torsion)/SDI.
Statistical Analysis
All data are expressed as mean ± SD. LV volumes were normalized to body surface area. Intraobserver and interobserver variability is reported as the coefficient of variation, calculated by dividing the standard deviation of the differences between measurements by the mean and expressed as a percentage. Ten subjects, five patients and five control subjects, were analyzed offline at different times for the determination of intraobserver and interobserver variability. Blinding was ensured by measuring the global 3D strain, twist, torsion, and SDI parameters using selected data sets on different days by the same observer (H.Y.) for intraobserver variability and by independent assessments by two observers (H.Y. and S.J.W.) for interobserver variability. Differences in demographic and echocardiographic parameters between patients and controls were compared using unpaired Student’s t tests or Fisher’s exact tests as appropriate. Pearson’s correlation analysis was used to assess for relationships between cumulative anthracycline dose and 3D strain, torsion, SDI, GPI, and LV ejection fraction. The area under the receiver operating characteristic (ROC) curve was calculated to determine the capability of various echocardiographic indices to discriminate between patients and controls. Two-tailed P values < .05 were considered statistically significant. All statistical analyses were performed using SPSS version 16.0 (SPSS, Inc., Chicago, IL).
Results
Subjects
The 53 patients were aged 18.6 ± 5.1 years at the time of study. The diagnoses included acute lymphoblastic leukemia ( n = 26), acute myeloid leukemia ( n = 10), osteosarcoma ( n = 8), Burkitt lymphoma ( n = 3), Hodgkin lymphoma ( n = 2), non-Hodgkin lymphoma ( n = 1), synovial sarcoma ( n = 1), neuroblastoma ( n = 1), and hepatoblastoma ( n = 1). All patients had received anthracycline-based regimens, and the median cumulative anthracycline dose was 229 mg/m 2 (range, 40–644 mg/m 2 ). Patients were studied a median of 7.2 years (range, 2.4–16.4 years) after the completion of chemotherapy. None had overt clinical heart failure at the time of study, and none required cardiac medications. The 38 control subjects were aged 19.9 ± 6.4 years ( P = .28). Table 1 summarizes the demographic data of patients and controls. Their sex distribution, weights, heights, and body mass indices were similar ( P > .05 for all).
| Variable | Patient ( n = 53) | Controls ( n = 38) | P |
|---|---|---|---|
| Age (year) | 18.6 ± 5.1 | 19.9 ± 6.4 | .28 |
| Male/female | 37/16 | 23/15 | .39 |
| Weight (kg) | 53.7 ± 15.4 | 52.0 ± 13.9 | .60 |
| Height (cm) | 162 ± 13 | 161 ± 14 | .77 |
| Body surface area (m 2 ) | 1.6 ± 0.3 | 1.5 ± 0.3 | .68 |
| Body mass index (kg/m 2 ) | 20.2 ± 4.0 | 19.6 ± 3.0 | .48 |
Three-Dimensional Speckle-Tracking Echocardiographic Parameters
For global 3D strain, SDI, twist, and torsion, intraobserver variability was 7.3%, 6.9%, 7.5% and 7.7%, respectively, and interobserver variability was 8.2%, 10.1%, 10.3%, and 10.7%, respectively.
Table 2 summarizes the 3D speckle-tracking echocardiographic parameters in patients and controls. Compared with controls, patients had significantly lower LV global 3D strain ( P < .001), twist ( P < .001), and torsion ( P < .001) and greater SDI ( P < .001). Representative illustrations of reduced LV 3D strain, twist, and torsion in a patient compared with a control subject are shown in Figure 1 . On the other hand, LV end-diastolic and end-systolic volumes were similar between the two groups ( P > .05 for both). Although LV ejection fraction was significantly lower in patients than in controls ( P < .001), 47 of the 53 patients (89%) had ejection fractions > 50%.
| Variable | Patient ( n = 53) | Controls ( n = 38) | P |
|---|---|---|---|
| LV end-diastolic volume (mL/m 2 ) | 51.1 ± 8.0 | 54.4 ± 8.9 | .06 |
| LV end-systolic volume (mL/m 2 ) | 22.8 ± 4.8 | 21.8 ± 4.8 | .30 |
| Ejection fraction (%) | 55.6 ± 4.2 | 60.1 ± 4.2 | <.001 ∗ |
| Global 3D strain (%) | 35.4 ± 7.5 | 44.6 ± 7.8 | <.001 ∗ |
| Twist (°) | 6.6 ± 2.5 | 9.9 ± 3.2 | <.001 ∗ |
| Torsion (°/cm) | 1.3 ± 0.5 | 1.9 ± 0.7 | <.001 ∗ |
| SDI (%) | 7.8 ± 3.1 | 4.9 ± 1.9 | <.001 ∗ |
| GPI (°/cm) | 6.7 ± 3.9 | 20.2 ± 10.6 | <.001 ∗ |
Cumulative anthracycline dose was correlated with LV global 3D strain ( r = −0.32, P = .018) and SDI ( r = 0.41, P = .003) but not with LV torsion ( P = .57). It also tended to correlate with LV ejection fraction ( r = −0.27, P = .051).
With regard to LV regional deformation, all except the basal anteroseptal segments in patients had significantly reduced segmental 3D strain compared with controls ( P < .05 for all; Figure 2 ).
Global LV Performance
Global LV performance was quantified by the GPI, a composite index of LV mechanics that takes into account of global 3D deformation, torsion, and dyssynchrony. The GPI was significantly lower in patients than controls ( P < .001; Figure 3 ). This was contributed to in patients by the reduced global 3D strain and torsion and greater LV dyssynchrony ( Table 2 ). The GPI was correlated with cumulative anthracycline dose ( r = −0.32, P = .02).
ROC Curve Analysis
Figure 4 shows the ROC curves generated using five 3D speckle-tracking echocardiographic parameters to determine the capability of the various echocardiographic indices to discriminate between patients and controls. Of the five parameters, the GPI had the greatest area under the ROC curve, while LV ejection fraction had the smallest ( Table 3 ). A GPI cutoff of 10.6°/cm had sensitivity of 84.9% and specificity of 81.6% in differentiating between patients and controls.
