The current work was conducted within the framework of the CTOXALL study. Briefly, CTOXALL is a cross-sectional study of a single-center cohort of childhood ALL survivors diagnosed between 1985 and 2015 and a control group composed of healthy siblings. The aim of the CTOXALL study is to evaluate the prevalence of long-term cardiotoxicity in CLSs with novel echocardiographic parameters and biomarkers. We have previously reported a high prevalence of subclinical LVSD in these patients.4 The present study aimed to assess the added value of LAS to conventional echocardiographic parameters (figure 1). The CTOXALL study protocol was approved by the local clinical research ethics committee according to institutional and good clinical practice guidelines. Written informed consent was obtained from all participants, parents, or legal guardians.

Figure 1.

Central illustration. Assessment of left atrial strain in a cohort of childhood ALL. A cohort of long-term childhood ALL showed impairment in diastolic function detected by left atrial strain but not by conventional parameters. This impairment was more marked in those at higher risk according to exposure to cardiotoxic treatment. ATC, anthracycline; IPW, inverse probability weighting; LA, left atrial; LACS, LA strain during the conduit phase; PACS, peak atrial contraction strain; PALS, peak atrial longitudinal strain; RT, radiotherapy.

(0.31MB).

Participants were evaluated from May 2019 to January 2022 at Hospital Universitario Reina Sofía (Cordoba, Spain). Survivors were eligible if they were diagnosed with ALL before the age of 18 years and had received the last anthracycline dose at least 3 years prior to their inclusion in the study. An individual with congenital heart disease (ventricular septal defect) was excluded. A sample of healthy siblings of survivors willing to participate was recruited as a comparison group.

All patients underwent clinical evaluation. Exposure to cardiotoxic treatments and doses were collected from medical records. Cumulative anthracycline doses were converted to doxorubicin equivalents using previously described conversion factors: 0.6 for daunorubicin, 0.8 for epirubicin, and 10.5 for mitoxantrone.9 Radiotherapy exposure was considered when the heart region was involved, including total body irradiation. According to the European guidelines on cardio-oncology, patients exposed to a cumulative doxorubicin dose<5Gy radiotherapy dose and <100mg/m2 were classified as low risk. The remaining patients were classified as ≥ moderate risk.10

All studies were performed by qualified echocardiographers using the same equipment (EPIQ CVx and iE33, Philips Medical Systems, United States). Standard echocardiographic parameters were obtained in accordance with the latest guidelines.11,12 Manual tracing of the left atrium (LA) borders in the apical 4-chamber and 2-chamber views was used to measure LA volumes with the modified biplane Simpson method. LV diastolic function was evaluated following the American Society of Echocardiography/European Association of Cardiovascular Imaging (ASE/EACVI) algorithm, which is based on the following variables: mitral flow velocities, mitral annular e’ velocity, E/e’ ratio, peak velocity of tricuspid regurgitation jet, and LA maximum volume index (LAVI).13

For the present study, we performed a retrospective post hoc analysis of LAS using a semiautomated assessment with AutoStrain (TomTec-Arena, TomTec Imaging Systems, Germany). LAS was measured from a nonforeshortened apical 4-chamber view, as recommended in the ASE/EACVI consensus report on the standardization of deformation imaging.14

Ventricular end-diastole and R-wave in the electrocardiogram were used as the time reference to define the zero-baseline for LAS curves. LA function can be defined by 3 components: reservoir (from end-diastole to mitral valve opening), conduit from mitral valve opening, throughout diastasis to onset of atrial contraction) and contraction phase (from onset of atrial contraction to end-diastole). LAS was assessed during these 3 phases and was represented by peak atrial longitudinal strain (PALS), peak atrial contraction strain (PACS), and LA strain during the conduit phase (LACS), representing the reservoir, contractile, and conduit function of the LA, respectively (figure 2). After automatic calculation using the software, the operator manually adjusted the contours if needed. For simplicity, LAS measurements are reported as their absolute values.

Figure 2.

Automated echocardiographic assessment of left atrial strain. Automatic tracing of the left atrium was performed from the apical 4-chamber view, as shown on the left of the image. The time reference to define the zero-baseline for left atrial strain (LAS) was set at ventricular end-diastole and R-wave in the electrocardiogram. The values of peak atrial longitudinal strain (PALS), left atrial strain during the conduit phase (LACS), and peak atrial contraction strain (PACS) are represented on the graphic on the right.

(0.08MB).

To test intraobserver and interobserver variability, 20 echocardiograms were randomly selected, and LAS was measured by the same investigator who performed the analysis and a second investigator, respectively.

Categorical variables are presented as count (percentage) and continuous variables as mean±standard deviation or median [interquartile range] according to their distribution, which was assessed using the Shapiro-Wilk test and QQ plots. Survivors and controls were compared using the chi-square test or Fisher exact test for categorical data and the Student t-test or Mann-Whitney U test for continuous data, as appropriate.

General linear models were used to compare echocardiographic measurements between groups. Inverse probability of treatment weighting (IPW) was used to balance patient characteristics in the 2 groups.15 Propensity scores were computed using logistic regression with age, sex, body mass index, heart rate, and diastolic blood pressure as covariates. Standardized mean differences before and after the weighting were used to evaluate the balance. A difference of <10% was considered to indicate good balance. The distribution of the propensity score before and after the weighting was plotted to assess the degree of overlap between the 2 groups. Standard errors of the IPW linear regression coefficients were obtained using robust sandwich-type variance estimators.16

Univariable and multivariable linear regression models were used to compare echocardiographic parameters between ≥ moderate risk CLSs, low-risk CLSs and controls, with age and sex as covariables in the multivariable models.

Intraobserver and interobserver agreement were assessed using intraclass correlation coefficients and the Bland-Altman method, plotting the difference of 2 measurements (y-axis) against their mean (y-axis) for each subject. The limit of agreement was computed as the mean difference±1.96 standard deviation.

Statistical analyses were performed using the SPSS software (version 24; IBM Corp., United States) and R software (version 4.0.3; R Foundation for Statistical Computing, Austria).

During the above-mentioned period, 170 patients with age <18 years were diagnosed with ALL at our center. A total of 52 patients had died before the start of the recruitment period. Among the 118 long-term survivors, 28 were not included (17 could not be reached, 10 refused to participate, and 1 had been diagnosed with congenital heart disease). Thus, the CLSs group was integrated by 90 patients.

The CLSs group had a median age at diagnosis of 4 [3-8] years and 34 (37.8%) were women. The mean age at recruitment was 24.6±9.7 years. The median time from diagnosis was 18 [11-26] years. The control group was composed of 58 healthy siblings. The characteristics of the unweighted groups are shown in table 1. Both groups were comparable in age, body measurements, and prevalence of risk factors, but there was a higher proportion of women in the control group (37.8% vs 56.6%; P=.018). The prevalence of sedentarism was high in both groups (41.1% vs 34.5%; P=.525). Smoking was more frequent in the CLSs group (15.6% vs 1.7%; P=.005). The variables used in the IPW models for the echocardiographic comparisons between the groups, including sex, were well balanced after the weighting, with standardized mean differences <10% for all the covariates (figure 1 of the supplementary data).

The details of the cardiotoxic treatments received by the CLSs are shown in table 2. All of them were exposed to anthracyclines. The mean isotoxic cumulative anthracycline dose was 138 [72-192mg/m2]. Three (3.3%) patients received thoracic radiation and 17 (18.9%) underwent hematopoietic stem cell transplantation. In all, 23 (25.6%) survivors were classified as low risk according to current recommendations. The prevalence of asymptomatic LVSD (left ventricle ejection fraction [LVEF] <50%) was 12.2% and of subclinical LVSD (global longitudinal strain [GLS] <18.5%) was 26.6%.

Conventional echocardiographic diastolic measurements were available and feasible in all patients. There were no cases of DD or undetermined diastolic function according to the ASE/EACVI algorithm.

The risk of LVSD and HF in long-term survivors of childhood cancer is clearly established, but there is a lack of agreement regarding LV diastolic function impairment in these patients.26–28 Indeed, there is marked variation among the studies assessing the prevalence of DD in childhood cancer survivors, ranging from 0% to 29%.27,29,30 This variation seems to be fundamentally determined by 3 factors: the age of the survivors at the time of the echocardiographic evaluation, the percentage of patients exposed to mediastinal radiotherapy, and the parameters used to define DD. Christiansen et al.27 found a prevalence of DD of 29% in a cohort of 125 survivors of childhood lymphoma (median age at exam 33 years). Diastolic dysfunction definition was based on a single parameter (reduction in septal and/or lateral e’) and was particularly frequent in patients exposed to mediastinal radiotherapy, which were as high as 50%. In contrast, Slieker et al.31 failed to find diastolic function impairment in 546 pediatric survivors of childhood cancer. Survivors were younger (median age at assessment 14 years) and only 12% of them received mediastinal radiotherapy. In line with the latter study, we found no cases of DD based on conventional echocardiographic parameters, which is consistent with the relatively young age of the participants (mean age 25 years) with exceptional exposure to mediastinal radiotherapy (only 3 patients). Through a mechanism of microvascular damage and generation of reactive oxygen species, radiation might lead to interstitial fibrosis causing a compliance reduction in the LV walls. This leads to a 7-fold higher risk of developing of DD in patients treated with radiotherapy compared with the general population.32 Through the production of reactive species and free radicals and the inhibition of topoisomerase IIβ, anthracyclines lead to myocyte cell death and interstitial fibrosis. These disruptions generate ventricular wall stiffness, impaired relaxation and, eventually, elevated filling pressure and DD.33,34 Indeed, according to the theory of senescent cell accumulation, childhood cancer survivors are believed to have an accelerated aging-like phenotype.35 The varying prevalence of DD reported in the literature could therefore be interpreted as pictures of different stages of a dynamic phenomenon that is mainly driven by cellular senescence and that seems to be enhanced by different exposures, such as mediastinal radiotherapy and cumulative anthracycline dose. Due to the limitations of the multiparametric assessment of diastolic function36 and the overlap between normal and pathologic values, conventional echocardiographic parameters might be insufficient to detect the earliest stages of DD in childhood cancer survivors.

LAS has been proposed as a single, sensitive and reproducible measure of DD.37,38 Although this measurement is influenced by LV GLS, it has been proven to be independently associated with LV filling pressures.39,40 Accordingly, we found that there was a mild to moderate correlation between PALS/LACS and GLS, and that these measurements remained significantly reduced in the survivors, regardless of the presence of abnormal GLS. Additionally, LA wall fibrosis has been associated with reduced PALS41 and is a predictor of mortality and hospitalizations in patients with HF.42 PACS corresponds to active atrial contraction and seems to be less associated with filling pressures and diastolic function.43 We speculate that the thinner left atrial wall thickness and the differential atrial cardiomyocyte phenotype might be associated with a lower susceptibility of atrial myocardium to anthracycline-induced injury.44

There are limited data on LAS in long-term survivors of childhood cancer. Loar et al.45 evaluated LAS in a cohort of 45 pediatric survivors of childhood cancer (median age 11.8 years) and found a reduction in PALS and LACS compared with a group of healthy controls. No differences in conventional diastolic measurements or PACS were found between the groups. Patients exposed to ≥ 250mg/m2 anthracycline doses had the lowest values of PALS. Despite the younger age of the participants in the study, these findings are in line with those observed in our study and altogether support the hypothesis that there are subtle alterations in the diastolic function of childhood cancer survivors that can be detected by LAS in adolescence and early adulthood. We speculate that this subtle diastolic impairment, which seems to be more pronounced in patients exposed to higher doses of anthracyclines, might constitute an early, subclinical stage of DD that might worsen with aging-related changes in LV filling, ultimately leading to an increased risk of DD and overt HF over time.46,47

To our knowledge, this is the largest study to evaluate DD in long-term survivors of childhood cancer with a comprehensive approach applying both the ASE/EACVI multiparametric algorithm and LAS measurements. Although the cross-sectional design of the study prevented us from evaluating the prognostic value of LAS impairment in these patients, it allowed us to depict novel findings in the context of continuous advancement in cardiac imaging. Of note, that despite a mean follow-up of almost 20 years since diagnosis, most participants were young adults at the time of the evaluation, which may limit the accuracy of conventional diastolic parameters in our sample. Conversely, a strength of the study is the use of more robust strain measurements. Furthermore, although echocardiograms were not analyzed by a core lab, automated software was used for quantification. Another strength of the study is the inclusion of a homogeneous population of ALL survivors with similar protocol-based treatments, reducing confusion bias secondary to heterogeneity in age of presentation, treatment regimens, and other unmeasurable disease-related factors. However, this came at the cost of a low percentage of patients exposed to radiotherapy, a factor that is strongly associated with DD. In this respect, the composition of the control group of healthy siblings reduced unmeasurable genetic and environmental factors, but resulted in sex differences between the 2 groups, which could have led to bias. Irrespective of this consideration, inverse probability weighting resulted in a good balance of sex and the other relevant covariates.