Echocardiographic Detection of Cardiac Dysfunction in Childhood Cancer Survivors: How Long Is Screening Required?

Building a machine learning-assisted echocardiography prediction tool for children at risk for cancer therapy-related cardiomyopathy

BACKGROUND

Despite routine echocardiographic surveillance for childhood cancer survivors, the ability to predict cardiomyopathy risk in individual patients is limited. We explored the feasibility and optimal processes for machine learning-enhanced cardiomyopathy prediction in survivors using serial echocardiograms from five centers.

METHODS

We designed a series of deep convolutional neural networks (DCNNs) for prediction of cardiomyopathy (shortening fraction ≤ 28% or ejection fraction ≤ 50% on two occasions) for at-risk survivors ≥ 1-year post initial cancer therapy. We built DCNNs with four subsets of echocardiographic data differing in timing relative to case (survivor who developed cardiomyopathy) index diagnosis and two input formats (montages) with differing image selections. We used holdout subsets in a 10-fold cross-validation framework and standard metrics to assess model performance (e.g., F1-score, area under the precision-recall curve [AUPRC]). Performance of the input formats was compared using a combined 5 × 2 cross-validation F-test.

RESULTS

The dataset included 542 pairs of montages: 171 montage pairs from 45 cases at time of cardiomyopathy diagnosis or pre-diagnosis and 371 pairs from 70 at-risk survivors who didn't develop cardiomyopathy during follow-up (non-case). The DCNN trained to distinguish between non-case and time of cardiomyopathy diagnosis or pre-diagnosis case montages achieved an AUROC of 0.89 ± 0.02, AUPRC 0.83 ± 0.03, and F1-score: 0.76 ± 0.04. When limited to smaller subsets of case data (e.g., ≥ 1 or 2 years pre-diagnosis), performance worsened. Model input format did not impact performance accuracy across models.

CONCLUSIONS

This methodology is a promising first step toward development of a DCNN capable of accurately differentiating pre-diagnosis versus non-case echocardiograms to predict survivors more likely to develop cardiomyopathy.

Evaluation of cardiotoxicity of anthracycline-containing chemotherapy regimens in patients with bone and soft tissue sarcomas: A study of the FDA adverse event reporting system joint single-center real-world experience

OBJECTIVES

To assess the occurrence of cardiotoxicity in patients with tumors receiving anthracycline-based chemotherapy, especially for sarcomas.

METHODS

This study summarized the types and frequency of adverse events (AEs) for three anthracyclines from the FDA adverse event reporting system (FAERS) database. FAERS data from January 2004 to June 2022 were collected and analyzed. Disproportionality analyses, logistic regression, and descriptive analysis were used to compare the differences in cardiac disorders. A retrospective cohort study was conducted in a single center between December 2008 and May 2022. Our hospital-treated patients with bone and soft tissue sarcomas (BSTSs) with anthracycline-containing chemotherapy were analyzed. Serum markers, echocardiography, and electrocardiography have been used to evaluate cardiotoxic events.

RESULTS

One hundred thousand and seventy-five AE reports were obtained for doxorubicin (ADM), epirubicin (EPI), and liposome doxorubicin (L-ADM) from the FAERS database. ADM (OR = 3.1, p < 0.001), EPI (OR = 1.5, p < 0.001), and sarcomas (OR = 1.8, p < 0.001) may increase the probability of cardiac disorders. Cardiac failure, cardiotoxicity, and cardiomyopathy were anthracyclines' top 3 frequent AEs. Among patients receiving ADM-containing therapy, those with ADM applied at doses ≥75 mg/m2 /cycle were more likely to develop cardiac disorders than the other subgroups (OR = 3.5, p < 0.001). Patients younger than 18 are more likely to benefit from dexrazoxane prevention of cardiac failure. Six hundred and eighty-three patients with BSTSs receiving anthracycline-based chemotherapy were analyzed in our center. Patients receiving ADM-containing chemotherapy were likelier to experience abnormalities in serum troponin-T and left ventricular ejection fraction (p < 0.05). 2.0% (6/300) of patients receiving ADM-containing chemotherapy required adjustment of the chemotherapy regimen because of cardiotoxicity, whereas none were in the EPI or L-ADM groups.

CONCLUSIONS AND RELEVANCE

Among patients receiving anthracycline-containing therapy, patients with BSTSs were more likely to develop cardiac disorders than other tumors. In addition, patients with BSTSs receiving ADM chemotherapy had a higher likelihood of cardiotoxic events than those receiving EPI or L-ADM.

Myocardial Strain during Surveillance Screening Is Associated with Future Cardiac Dysfunction among Survivors of Childhood, Adolescent and Young Adult-Onset Cancer

Cardiovascular disease is a leading contributor to mortality among childhood, adolescent and young adult (C-AYA) cancer survivors. While serial cardiovascular screening is recommended in this population, optimal screening strategies, including the use of echocardiography-based myocardial strain, are not fully defined. Our objective was to determine the relationship between longitudinal and circumferential strain (LS, CS) and fractional shortening (FS) among survivors. This single-center cohort study retrospectively measured LS and CS among C-AYAs treated with anthracycline/anthracenedione chemotherapy. The trajectory of LS and CS values over time were examined among two groups of survivors: those who experienced a reduction of >5 fractional shortening (FS) units from pre-treatment to the most recent echocardiogram, and those who did not. Using mixed modeling, LS and CS were used to estimate FS longitudinally. A receiver operator characteristic curve was generated to determine the ability of our model to correctly predict an FS ≤ 27%. A total of 189 survivors with a median age of 14 years at diagnosis were included. Among the two survivor groups, the trajectory of LS and CS differed approximately five years from cancer diagnosis. A statistically significant inverse relationship was demonstrated between FS and LS -0.129, p = 0.039, as well as FS and CS -0.413, p < 0.001. The area under the curve for an FS ≤ 27% was 91%. Among C-AYAs, myocardial strain measurements may improve the identification of individuals with cardiotoxicity, thereby allowing earlier intervention.

Systematic review and updated recommendations for cardiomyopathy surveillance for survivors of childhood, adolescent, and young adult cancer from the International Late Effects of Childhood Cancer Guideline Harmonization Group

Survivors of childhood, adolescent, and young adult cancer, previously treated with anthracycline chemotherapy (including mitoxantrone) or radiotherapy in which the heart was exposed, are at increased risk of cardiomyopathy. Symptomatic cardiomyopathy is typically preceded by a series of gradually progressive, asymptomatic changes in structure and function of the heart that can be ameliorated with treatment, prompting specialist organisations to endorse guidelines on cardiac surveillance in at-risk survivors of cancer. In 2015, the International Late Effects of Childhood Cancer Guideline Harmonization Group compiled these guidelines into a uniform set of recommendations applicable to a broad spectrum of clinical environments with varying resource availabilities. Since then, additional studies have provided insight into dose thresholds associated with a risk of asymptomatic and symptomatic cardiomyopathy, have characterised risk over time, and have established the cost-effectiveness of different surveillance strategies. This systematic Review and guideline provides updated recommendations based on the evidence published up to September, 2020.

A review of the risks of long-term consequences associated with components of the CHOP chemotherapy regimen

A common chemotherapy regimen in post-transplant lymphoproliferative disease (PTLD) following solid organ transplants (SOT) is cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP). This study reviews the quantitative evidence for long-term consequences associated with components of CHOP identified from the Children's Oncology Group Long-Term Follow-Up Guidelines. Cited references were screened using prespecified criteria (English, systematic review, randomized controlled trial n > 100, observation study n > 100, case series n > 20). Relevant data were extracted and synthesized. Of 61 studies, 66% were retrospective cohort studies, 28% were in the US, and 95% enrolled pediatric patients. No study focused specifically on the CHOP regimen. Long-term consequences for CHOP components observed in >3 studies included cardiac toxicity (n = 14), hormone deficiencies/infertility (n = 14), secondary leukemia (n = 7), osteonecrosis (n = 6), and bladder cancer (n = 4). These effects are significant, impact a high percentage of patients, and occur as early as one year after treatment. Although none of the studies focused specifically on the CHOP regimen, 30%, 23%, and 15% evaluated alkylating agents (e.g. cyclophosphamide), anthracyclines (e.g. doxorubicin), and corticosteroids (e.g. prednisone), respectively. All three product classes had a dose-dependent risk of long-term consequences with up to 13.2-fold, 27-fold, 16-fold, 14.5-fold, and 6.2-fold increase in risk of heart failure, early menopause, secondary leukemia, bladder cancer, and osteonecrosis, respectively. Lymphoma patients had significantly elevated risks of cardiac toxicity (up to 12.2-fold), ovarian failure (up to 3.8-fold), and osteonecrosis (up to 6.7-fold). No studies were found in PTLD or SOT. Safe and effective PTLD treatments that potentially avoid these long-term consequences are urgently needed.

Advanced Imaging and New Cardiac Biomarkers in Long-term Follow-up After Childhood Cancer

OBJECTIVES

Pathologic ejection fraction (EF), shortening fraction (FS), and standard heart failure biomarkers (high sensitive troponin T and N-terminal brain natriuretic peptide) during follow-up after childhood cancer have been associated with irreversible cardiac damage. We aimed to evaluate strain imaging values by echocardiography and new biomarkers for heart failure with preserved ejection fraction (HFpEF) as potential more sensitive parameters for cardiac deterioration in childhood cancer survivors (CCS).

MATERIALS AND METHODS

Prospective study with 50 CCS (median 16.2 y) at a median follow-up of 13 years. In addition to standard echo and laboratory parameters for heart failure, strain measurements and new biomarkers, including myocardial inflammation (interleukin 6), extracellular matrix (ECM) remodeling (C-telopeptide for type I collagen, intact N-terminal propeptide of type III procollagen), and other heart failure biomarkers (galectin 3, solutable ST2, growth differentiation factor 15), were obtained and compared with 50 healthy controls.

RESULTS

No significant differences in EF, FS, high sensitive troponin T, N-terminal brain natriuretic peptide, interleukin 6, solutable ST2, and galectin 3 were found between study and control groups. In contrast, strain imaging showed significant differences between both groups (global longitudinal strainGLS -16.1% vs. -20.4%, P<0.0001; global circumferential strain -14.3 vs. -21.4%, P<0.0001), detecting 66% (global longitudinal strain) and 76% (global circumferential strain) of patients with pathologic values in contrast to 6% (EF) and 16% (FS) for standard parameters. Markers for disturbances of ECM remodeling (C-telopeptide for type I collagen, intact N-terminal propeptide of type III procollagen, each P<0.0001) and growth differentiation factor 15 (P<0.0001) were significantly different between the groups.

CONCLUSION

Strain imaging and new cardiac biomarkers used in HFpEF focusing on ECM remodeling appear to be more sensitive in detecting early remodeling processes in CCS than standard echo and laboratory parameters.

Doxorubicin-induced delayed-onset subclinical cardiotoxicity in mice

Subclinical cardiotoxicity at low total cumulative doxorubicin (DOX) doses can manifest into cardiomyopathy in long-term cancer survivors. However, the underlying mechanisms are poorly understood. In male B6C3F₁ mice, assessment of cardiac function by echocardiography was performed at 1, 4, 10, 17, and 24 weeks after exposure to 6, 9, 12, and 24 mg/kg total cumulative DOX doses or saline (SAL) to monitor development of delayed-onset cardiotoxicity. The 6- or 9-mg/kg total cumulative doses resulted in a significant time-dependent decline in systolic function (left ventricular ejection fraction (LVEF) and fractional shortening (FS)) during the 24-week recovery although there was not a significant alteration in % LVEF or % FS at any specific time point during the recovery. A significant decline in systolic function was elicited by the cardiotoxic cumulative DOX dose (24 mg/kg) during the 4- to 24-week period after treatment compared to SAL-treated counterparts. At 24 weeks after DOX treatment, a significant dose-related decrease in the expression of genes and proteins involved in sarcoplasmic reticulum (SR) calcium homeostasis (Ryr2 and Serca2) was associated with a dose-related increase in the transcript level of Casp12 (SR-specific apoptosis) in hearts. These mice also showed enhanced apoptotic activity in hearts indicated by a significant dose-related elevation in the number of apoptotic cardiomyocytes compared to SAL-treated counterparts. These findings collectively suggest that a steady decline in SR calcium handling and apoptosis might be involved in the development of subclinical cardiotoxicity that can evolve into irreversible cardiomyopathy later in life.

Late Morbidity among Survivors of Childhood Cancers; Experience at Tertiary Care Cancer Hospital

Background

Long-term survivors of childhood malignancies are at increased risk of experiencing treatment-related morbidities. Survival into late adulthood in these children provides ample time for the acquisition of long-term sequelae. This study aimed to determine the late adverse effects among long-term survivors of childhood cancer from a low-income country perspective.

Materials and Methods

Data were retrospectively collected from review of charts of patients aged under 18 years at the time of their primary diagnosis between 1 January, 1995, and 31 December, 2008, and who survived for at least 5 years after completion of their treatment. Analysed data included demographics, cancer type, treatment modality, types of chemotherapy agents administered and specific late morbidities including frequency of azoospermia, oligospermia, endocrine abnormalities, hearing and pulmonary function impairment and cardiac dysfunction among the long-term survivors of cancer.

Results

The total number of patients was 300 with a mean age of 18 ± 2 years. The male to female ratio was 2.7:1. Median follow-up duration was 18 years (range: 5-25 years). Seventy-seven percent of patients were from Punjab, 20% were from Khyber Pakhtunkhwa and 3% were from other provinces. Fifty percent had a diagnosis of Hodgkin lymphoma, 17% had acute lymphoblastic leukaemia, 13% had non-Hodgkin lymphoma, 10% had germ cell tumours and 10% had other tumours. Fifty-seven percent received chemotherapy, 23% had chemotherapy and radiotherapy, 15% had chemotherapy and surgery, 3% had chemotherapy, surgery and radiotherapy and 2% had only surgery. Notable long-term documented sequelae were; azoospermia/oligospermia in 64%, endocrine abnormalities in 25% with hypothyroidism in 13.5% and follicle-stimulating hormone and luteinizing hormone abnormalities in 11.5%, ototoxicity in 6.5%, impaired pulmonary function tests in 4.6%, cardiotoxicity in 2.4% and second malignancies (acute myeloid leukaemia and myelodysplastic syndrome) in 1%.

Conclusion

Childhood cancer survivors are at increased risk of adverse treatment-related sequelae and a long-term follow-up plan should be in place in centres where they receive treatment for their primary disease.

Treatment exposures stratify need for echocardiographic screening in asymptomatic long-term survivors of hematopoietic stem cell transplantation

We sought to define the prevalence of echocardiographic abnormalities in long-term survivors of paediatric hematopoietic stem cell transplantation and determine the utility of screening in asymptomatic patients. We analysed echocardiograms performed on survivors who underwent hematopoietic stem cell transplantation from 1982 to 2006. A total of 389 patients were alive in 2017, with 114 having an echocardiogram obtained ⩾5 years post-infusion. A total of 95 patients had echocardiogram performed for routine surveillance. The mean time post-hematopoietic stem cell transplantation was 13 years. Of 95 patients, 77 (82.1%) had ejection fraction measured, and 10/77 (13.0%) had ejection fraction z-scores ⩽-2.0, which is abnormally low. Those patients with abnormal ejection fraction were significantly more likely to have been exposed to anthracyclines or total body irradiation. Among individuals who received neither anthracyclines nor total body irradiation, only 1/31 (3.2%) was found to have an abnormal ejection fraction of 51.4%, z-score -2.73. In the cohort of 77 patients, the negative predictive value of having a normal ejection fraction given no exposure to total body irradiation or anthracyclines was 96.7% at 95% confidence interval (83.3-99.8%). Systolic dysfunction is relatively common in long-term survivors of paediatric hematopoietic stem cell transplantation who have received anthracyclines or total body irradiation. Survivors who are asymptomatic and did not receive radiation or anthracyclines likely do not require surveillance echocardiograms, unless otherwise indicated.

Utility of Echocardiography as Screening for Late-onset Anthracycline-induced Cardiotoxicity in Pediatric Cancer Survivors: Observations from the First Decade After End of Therapy

Current screening guidelines are available for anthracycline-induced cardiotoxicity. However, the utility of echocardiogram screening for late-onset anthracycline cardiotoxicity especially in the decade immediately after end of therapy is debatable. A retrospective chart review of patients seen in the Thriving after Cancer Clinic at Rady Children's Hospital January 2006 to December 2013 was performed. Treatment data, echocardiogram results, cardiology referral notes and cardiac medication data were abstracted from anthracycline-exposed survivors. Descriptive and univariate comparative statistics were performed. Of 368 patients (45% female, median 5.3 y old at diagnosis [range 0 to 18.3], median 5.0 y from end of therapy [EOT] [range 0 to 18.2]), a total of 4 patients (10-year cumulative incidence after EOT 1.3%; 95% confidence interval, 0.1%-19.7%) required cardiac medication for late-onset cardiotoxicity (>1 y after EOT). Those requiring medication for late-onset cardiotoxicity were exposed to more anthracyclines than survivors without cardiotoxicity (median, 360 mg/m [range, 300 to 375 mg/m] vs. 182 mg/m [range, 26 to 515 mg/m], P=0.009). None had neck or chest radiation. In this population, medication initiation for late-onset anthracycline cardiotoxicity was limited predominantly to the first 3 years after EOT, with the next >13 years after EOT. These findings add to the growing body of literature assessing current guidelines to inform improvements in screening practices of survivorship providers.

Electrophysiological effects of anthracyclines in adult survivors of pediatric malignancy

BACKGROUND

Anthracycline use is limited by cardiotoxicity, including arrhythmias and left ventricular (LV) dysfunction. We aim to characterize the association between electrophysiological changes and LV dysfunction.

METHODS

A retrospective chart review was conducted, including all 147 pediatric cancer survivors at our institution over 18 years of age and treated with an anthracycline. One hundred thirty-four patients who had at least one electrocardiogram (ECG) and echocardiogram were analyzed. The association between dysfunction and baseline characteristics, treatment history, and electrocardigraphic parameters were analyzed using multivariable logistic regression. Additionally, a longitudinal generalized estimating equation (GEE) model was used to examine the temporal association between repeated measure corrected QT (QTc) intervals and subsequent LV function.

RESULTS

In our population, 24% of patients had LV dysfunction. The initial posttreatment QTc interval was longer in patients with LV dysfunction (438 ± 35 vs. 420 ± 20 msec, P = 0.002). In logistic regression analysis, QTc interval (P < 0.001) and cumulative radiation dose (P = 0.027) were associated with LV dysfunction. On ECGs performed prior to evidence of LV dysfunction, the QTc was longer than on ECGs preceding a normal echocardiogram (451 ± 32 msec vs. 423 ± 25 msec, P < 0.001). Mean time from QTc ≥ 450 msec to evidence of LV dysfunction was 1.8 ± 2.9 years. In the longitudinal GEE model, QTc prolongation was associated with subsequent decreased fractional shortening.

CONCLUSIONS

Among adult survivors of pediatric cancer treated with anthracyclines, prolongation of the QTc interval was associated with subsequent LV dysfunction.

Yield of screening echocardiograms during pediatric follow-up in survivors treated with anthracyclines and cardiotoxic radiation

BACKGROUND

Guidelines published by the Children's Oncology Group recommend screening echocardiograms for childhood cancer survivors exposed to anthracyclines and/or cardiotoxic radiation. This study aims to assess risk factors for cardiac late effects while evaluating the overall yield of screening echocardiograms.

PROCEDURE

Demographics, exposures, and echocardiogram results were abstracted from the medical records of survivors diagnosed at ≤ 21 years old and ≥ 2 years off therapy who were exposed to anthracyclines and/or potentially cardiotoxic radiotherapy. Descriptive statistics and logistic regressions were performed and the yield of screening echocardiograms was calculated.

RESULTS

Of 853 patients, 1,728 screening echocardiograms were performed, and 37 patients had an abnormal echocardiogram (overall yield 2.1%). Yields were only somewhat higher in more frequently screened patients. Risk factors for an abnormal result included anthracycline dose of ≥300 mg/m² (adjusted odds ratio [aOR] 3.1; 95% confidence interval [CI]: 1.3-7.2; P < 0.01) with a synergist relationship in patients who also received radiation doses ≥30 Gy (aOR 7.0; 95% CI: 1.6-31.9; P = 0.01), as well as autologous bone marrow transplant (OR 3.3; 95% CI: 1.3-8.5; P = 0.01). Sex, race, age at diagnosis, and cyclophosphamide exposure were not statistically significant risk factors, and no patient receiving <100 mg/m² anthracycline dose without concomitant radiation had an abnormal echocardiogram.

CONCLUSIONS

Dose-dependent and synergist anthracycline and cardiotoxic radiotherapy risks for developing cardiomyopathy were confirmed. However, previously identified risk factors including female sex, black race, and early age at diagnosis were not replicated in this cohort. The yields showed weak correlation across frequency categories. Echocardiographic screening recommendations for low-risk pediatric patients may warrant re-evaluation.

Cardiac Mortality in Children and Adolescents with Hodgkin's Lymphoma: A Surveillance, Epidemiology and End Results Analysis

PURPOSE

The purpose of this study was to evaluate the risk of cardiac death in pediatric Hodgkin's lymphoma (HL) survivors and identify high-risk groups that may need additional surveillance.

METHODS

The Surveillance, Epidemiology and End Results program database was queried to analyze the rates of radiation therapy (RT) use and cardiac-specific mortality (CSM) in HL patients, aged 0-21 years, treated from 1973 to 2007. Primary endpoint was cardiac mortality.

RESULTS

A total of 6552 patients were included. Median follow-up was 12 years (range, 0-40). Median age at diagnosis was 17 years (range, 0-21). The majority were white (85.5%), from western states (41.2%), had nodular sclerosis HL (73.2%), presented with stage I or II disease (51.5%), and received RT (56.1%). Death from cardiac disease occurred in 114 patients (9.2% of all deaths). CSM for the entire cohort at 10-, 20-, and 30-year time points was 0.3%, 1.6%, and 5.0%, respectively. Median age at the time of cardiac death was 39 years (range, 18-58 years). Under multivariate analysis (MVA), adolescent patients (ages 13-21) had higher rates of CSM (hazard ratio [HR], 3.05; p = 0.005). Female gender (HR, 0.43; p < 0.001), patients treated from 1998 to 2007 (HR, 0.19; p = 0.018), and those with lymphocyte-rich histology (HR, 0.14; p = 0.047) had significantly lower rates of CSM. Use of RT was not associated with CSM under MVA (HR, 1.18, p = 0.452).

CONCLUSION

The cumulative incidence of CSM in this population analysis of pediatric HL was 9.2%, with a steady decline over the past several decades. Adolescent patients at diagnosis and males were more likely to die of cardiac-related causes.