Paper Details

ABSTRACT

Introduction: Cardiotoxicity, manifesting as left ventricular systolic dysfunction (LVSD), is a common complication of pediatric acute myeloid leukemia (AML) therapy and is associated with inferior event free and overall survival. Dexrazoxane (DRZ) can reduce but not eliminate the occurrence of early LVSD. Liposomal anthracycline preparations are associated with reduced cardiotoxicity in adults; however, pediatric data are lacking. The Children’s Oncology Group (COG) phase III trial, AAML1831, randomized children with newly diagnosed AML to standard induction daunorubicin and cytarabine (DA) with dexrazoxane or CPX-351, a liposomal form of DA. In this pre-specified substudy, we compared cardiotoxicity rates across these two arms using prospectively collected, centrally quantified echocardiograms.

Methods: AAML1831 enrolled 720 eligible patients < 22 years of age with de novo non-FLT3-mutant AML. This analysis is restricted to patients included in a pre-specified sub-cohort (n=510) of patients enrolled from sites with a cardiac sonographer certified to provide echocardiographic images of adequate quality. Patients were randomized to receive standard DA with DRZ (10+3 in Induction 1, 7+3 in Induction 2) vs. CPX-351 (60mg/m2 daunorubicin days 1,3,5 of induction 1, 50mg/m2 on days 1,3,5 of induction 2). All patients received gemtuzumab ozogamicin in induction 1. Risk adapted post-induction therapy was as previously described; patients stratified as low risk 2 received Mitoxantrone (48 mg/m2) with dexrazoxane during intensification on both arms. Echocardiograms were performed and centrally analyzed at baseline, prior to each chemotherapy cycle, at end of treatment, and 1-year post-therapy. LVSD was defined as a decline in left ventricular ejection fraction (LVEF) to < 50% (or if LVEF not available, shortening fraction < 24%). Global longitudinal strain (GLS), a sensitive measure of cardiac function, was considered significantly reduced if > 15% relative decline from baseline or abnormal if > -16%. Cumulative incidence (CI) of LVSD was compared across arms using a time to event model with induction failure, relapse, and non-cardiac death treated as competing events. Patients were censored at withdrawal if prior to completion of planned protocol therapy.

Results: 509 subjects (DA+DRZ Arm n=247, CPX-351 Arm n=262; median age at diagnosis 10.1 years) were included and 2,233 echocardiograms were centrally quantified by a sonographer blinded to patient characteristics. Baseline demographics and clinical factors were balanced between arms. 18-month LVSD incidence was similar across arms: DA+DRZ 21.0% (15.7-26.9) vs. CPX-351 18.1% (13.2-23.5), p=0.40. The majority of LVSD events occurred during induction 1 and 2: 28/46 (60.9%) on DA+DRZ Arm vs. 24/40 (60.0%) on CPX-351 Arm. Of 87 first LVSD events, 16 (18.6%) occurred in the setting of a bloodstream infection/sepsis. The 18-month incidence of infection-associated LVSD was similar across arms: DA+DRZ 4.6% (2.3-8.2) vs. CPX-351 6.3% (3.5-10.2), p=0.39. The occurrence of abnormal GLS > -16% was also similar across arms (18-month CI 28.3 (DA+DRZ) vs. 28.2% (CPX-351), p=0.93). GLS decline of > 15% from baseline occurred in more than half of patients (18-month CI 53.2% (DA+DRZ) vs. 51.2% (CPX-351 Arm), p=0.38).

Conclusion: In the first randomized comparison of cardioprotective strategies in pediatric AML, prospective core lab-quantified echocardiographic phenotyping identified substantial cardiotoxicity despite cardioprotective measures. Rates are higher than previously described based on site-reported assessment on the prior COG trial AAML1031, wherein 10.9% of dexrazoxane treated patients experienced LVSD. Cardiotoxicity was similar across induction cardioprotective strategies, with CPX-351 demonstrating comparable cardioprotection to standard anthracycline with dexrazoxane. Ongoing analyses aim to identify high-risk subgroups and the most effective cardioprotective interventions.