We prospectively enrolled consecutive AL patients (≥18 years of age) who were referred to Wuhan Union Hospital before any chemotherapy between March 2021 and April 2022. Patients with reduced LVEF (<52% for male and <53% for female) (22), more than mild valvular heart disease, atrial fibrillation, congenital heart disease, previous chemotherapy history and inadequate echocardiographic images were excluded from this study. The AL cohort consisted of a total of 72 patients, of whom 44 were diagnosed with AML and 28 with ALL. All of them had an echocardiogram performed before initiation of chemotherapy. 72 AL patients and 58 cases of control group were matched for age and gender.

This study was approved by the Ethics Committee of Tongji Medical College, Huazhong University of Science and Technology. Written informed consent was obtained from all patients.

Transthoracic echocardiographic examinations were performed by an experienced echocardiographic doctor using Philip EPIQ 7C (Philips Medical Systems, Andover, USA) equipped with S5-1 and X5-1. All measurements were performed according to the current recommendations of the American Society of Echocardiography (23). 2D echocardiographic acquisitions were recorded from the basal, middle and apical levels of parasternal short axis views and the apical 4-chamber (Ch), 2-Ch and 3-Ch views with 4 consecutive cardiac cycles. The frame rate was set between 55 and 90 frames per second. LVEF was calculated by biplane Simpson's method from apical 4-Ch and 2-Ch views (24). Tricuspid annular plane systolic excursion (TAPSE) was measured as the systolic displacement of the tricuspid lateral annulus, recorded on M-mode imaging. Right ventricular fractional area change (RVFAC) was calculated as: (RV end-diastolic area−RV end-systolic area)/end-diastolic area × 100%. Tricuspid lateral annular systolic velocity (S') was assessed using tissue Doppler imaging from the apical 4-ch view.

3D echocardiographic acquisitions were recorded from the four-chamber apical view in heart model mode with volume rate of 19–25 volumes/s, and were gathered over four cardiac cycles, during a breath-hold lasting for a few seconds. Left ventricular end-systolic volume (LVESV), end-diastolic volume (LVEDV), right ventricular end-systolic volume (RVESV), end-diastolic volume (RVEDV) and RV ejection fraction (RVEF) were measured by 3D echocardiography. The 2D and 3D echocardiographic datasets were stored digitally for offline analysis.

The off-line 3D-STE analyses were performed using the vendor-independent software TomTec (4D LV-Analysis 3.1 for 3D-STE Tom Tec Imaging Systems, Unterschleissheim, Germany) by an experienced researcher who was blinded to the groups. The software semi-automatically aligned and displayed the LV in the short axis view at level of the aortic valve and 3 standard apical views (25). To avoid foreshortening, manual adjustment was necessary to obtain the maximal LV long axis view. The endocardial surface of the ventricle in end-diastolic and end-systolic frames was identified automatically for initial contour detection. The operator could manually adjust the ventricular endocardial surface if necessary. Subsequently, the software automatically tracked the endocardium throughout the entire cardiac cycle, then the software provides the time-volume and time-strain curves (Figure 1, Supplementary Figure S1).

The 3D LV global longitudinal strain (LVGLS) and global circumferential strain (LVGCS) were quantified as the mean of the strain values from all 16 segments of LV. The 3D RV free wall longitudinal strain (RVFWLS), RV septal wall longitudinal strain (RVSWLS) and RV global longitudinal strain (RVGLS) were acquired simultaneously. The patients with persistent inadequate tracking were excluded from the study.

Intraobserver and interobserver variability of the 3D-STE parameters were analyzed in 20 randomly selected patients from AL group. Intraobserver reproducibility was assessed by the same observer performed 4 weeks later again. Interobserver reproducibility was assessed by the second observer who was blinded to the measurements of the first observer in the same 20 patients.

Continuous variables are presented as mean ± standard deviation or as median (interquartile range), and categorical variables as number (%). Comparisons of clinical and echocardiographic parameters between AL patients and control group were performed using the Student's t test or the Wilcoxon rank sum tests for continuous variables and the χ² or Fisher's exact test for categorical variables. Comparisons of echocardiographic parameters between AML and ALL patients were performed using one-way analysis of covariance. Correlations between LVGLS and clinical, laboratory, and echocardiographic variables were estimated using Pearson correlation coefficients or Spearman ρ test for nonparametric values. A multivariable linear regression was also performed to explore the independent clinical and echocardiographic parameters associated with GLS in patients with AL. All variables reaching P values of < 0.1 on univariate analysis were included in the multivariate analyses. The intraobserver and interobserver variability of 3D-STE parameters were assessed by the intraclass correlation coefficients (ICCs) and the Bland–Altman analyses in a random sample of 20 patients. All statistical analyses were performed on SPSS version 26.0 (Statistical Package for the Social Sciences, Chicago, Illinois) and GraphPad Prism version 8.0.1. A 2-sided P value of < 0.05 was considered as statistically significant.

There was no difference in age, sex distribution, BMI, diabetes, prevalence of hypertension, dyslipidemia, smoking, prior coronary artery disease (CAD) and chronic kidney disease (CKD) between AL patients and controls. The history of medications did not differ between the two groups (Table 1).

LV volume and LV volume indexed to body surface area were slightly higher in patients with AL (P < 0.05). Systolic dyssynchrony index (SDI) was similar between the two groups. Although LVEF was similar between AL patients and controls (61.1 ± 4.2% vs. 62.4 ± 4.9%, P > 0.05), a statistically significant difference was noted with regard to LVGLS (−23.0 ± 1.4% vs. −24.1 ± 1.3%, P < 0.001). There was no significant difference in LVGCS (Table 2).

Conventional parameters of RV systolic function (FAC, TAPSE and S’) were similar between two groups. Compared with controls, the RVGLS and RVFWLS were lower in AL patients (−25.9 ± 4.7% vs. −28.2 ± 4.3%, P < 0.05 and −27.9 ± 7.1% vs. −33.0 ± 4.6%, P < 0.001, respectively) (Table 2). There was no significant difference in RVSWLS.

After adjustment for hemoglobin counts, only LVGLS and RVFWLS were lower in patients with AML (−22.7 ± 1.3% vs. −23.5 ± 1.6% and −26.2 ± 7.6% vs. −30.4 ± 5.5%, respectively, P < 0.05 for all) (Table 3). LVGLS was lower in ALL patients compared with controls (−23.5 ± 1.6% vs. −24.7 ± 1.4%, P < 0.05). However, there was no difference in RV systolic function parameters between the two groups (Figure 2).

Among all AL patients, LVGLS was correlated with LVEF (r = −0.430, P < 0.001), BMI (r = 0.263, P = 0.026), age, (r = 0.272, P = 0.021), the absolute number of C reactive protein (CRP) (r = 0.298, P = 0.011), the absolute number of circulating lymphocytes (r = −0.524, P < 0.001) and hemoglobin counts (r = 0.304, P = 0.009; Figure 3). Therefore, a decrease in LVGLS was associated with a reduction in the absolute number of lymphocytes and LVEF and with an increase in age, BMI, the absolute number of CRP and hemoglobin counts. In a multivariable linear regression analysis, the factors independently associated with LVGLS were LVEF (β = −0.085, P = 0.019) and the absolute number of circulating lymphocytes (β = −0.814, P = 0.004). However, RVFWLS was only correlated with BMI (r = 0.323, P = 0.005) and hemoglobin counts (r = 0.267, P = 0.021) (Supplementary Table S1).

Intraobserver and interobserver variability of the 3D-STE parameters are presented in Table 4. Strain parameters measured by 3D-STE showed excellent reproducibility, as evidenced by the high ICC, small bias, and narrow limits of agreement.

This study has several limitations. First, this was a single-center study and the sample size was relatively small. Second, we did not acquire all the laboratory biomarker tests of our patients, such as high-sensitivity troponin I (hs-TNI) and IL-6, which reflect the level of inflammation in the body. Third, our study lacks histological confirmation of interstitial inflammatory infiltrates of myocardial tissue. Fourth, 3D-STE technique was dependent on image quality, and the value of strain in our study may not apply to other software algorithms due to inter-vendor variability. Therefore, future studies with multicenter involvement may strengthen the study power.