Central nervous system disease status significantly impacts survival in pediatric acute myeloid leukemia: A report from Children's Oncology Group study AAML1831

Abstract Introduction: Standard treatment of pediatric AML (in contrast to adult AML) includes diagnostic lumbar puncture (LP) and intrathecal chemotherapy in multiple cycles for all patients. Results from Children's Oncology Group (COG) trials have demonstrated inferior event-free survival (EFS) in patients with CNS disease at diagnosis. Here we evaluate outcomes based on CNS status from recent COG trial AAML1831 compared to historical COG trials AAML0531/1031, to assess the impact of CNS diagnosis and management changes in AAML1831 including delayed evaluation of CNS status at diagnosis, incorporation of intrathecal triple (ITT; methotrexate/hydrocortisone/cytarabine) chemotherapy, and increased utilization of hematopoietic stem cell transplant (HSCT) in high-risk patients. Methods: Patients were characterized as CNS1 (no blasts), CNS2 (blasts with 5 CSF WBC), or CNS3 (blasts with 5 CSF WBC or signs/symptoms) based on diagnostic LP. On prior COG trials AAML0531/1031, diagnostic LP and intrathecal cytarabine was performed at initiation of systemic induction chemotherapy as well as with subsequent cycles, with exception of Capizzi high dose cytarabine. However, on AAML1831, patients had LP delayed to day 8 of induction 1 therapy to decrease the incidence of peripheral blood blast contamination of the CSF. In rare cases of patients with symptoms of CNS disease at diagnosis, patients were recommended to have an LP urgently. On AAML1831, patients received ITT at the time of diagnostic LP. Diagnostic CSF with myeloid blasts present (CNS2/3 status) resulted in treatment with weekly ITT during induction until 2 negative CSF samples obtained (minimum 4 and maximum 6). ITT was also administered with the start of each cycle, with exception of Capizzi high dose cytarabine. Patients excluded from the analysis included those with FLT3 mutations (TKD/ITD) due to ongoing trial accrual and those with no evaluation of CNS status (n=35) or evaluation of CNS status beyond day 10 of induction 1 (n=57). Results: Patients with evaluable CNS status on AAML1831 (n=628) and AAML0531/1031 (n=1810) were stratified into low, intermediate, and high-risk groups based on genetics and marrow disease induction response. The median (range) follow-up time for patients alive at last contact was 2.3 (0.04-4.73) years. At diagnosis, the rates of CNS1 (CNS negative) and CNS2/CNS3 (CNS positive) on AAML1831 were 82% and 18%. The new recommendation to delay LP to day 8 resulted in diagnosis of significantly less CNS disease on AAML1831 than on prior COG studies AAML0531/1031 (18% vs 31%; p0.001). CNS status on AAML1831 remained a significant predictor of survival with CNS1, CNS2 and CNS3 having 2-yr EFS rates of 60.4%, 69.5% and 45.6% (p=0.014), and 2-yr cumulative incidence (CI) of any CNS event of 2.2%, 3.5% and 20.8% (p0.001). Similar outcomes for EFS and CI of CNS relapse for CNS1 and CNS2 suggest that additional ITT therapy in induction was protective for CNS2, but patients with more severe CNS disease (CNS3) still had worse outcomes. Overall, patients on AAML1831 had a lower 2-year CI to any CNS event compared to patients on AAML0531/1031 (4.0% vs 9.4%, p0.001). Restricting to patients with CNS2/3 disease, this cohort on AAML1831 also had a lower 2-year CI to any CNS event compared to AAML0531/1031 (11.9% vs 19.1%, p=0.049). To further assess the impact of HSCT on CNS relapse risk we compared patients with high-risk features on AAML1831 who received HSCT consolidation to patients with similar risk features but not treated with HSCT on AAML0531/1031 (due to historical differences in HSCT recommendations on COG trials). This data suggests a strong protective effect of HSCT against CNS relapse with 2-year CI to any CNS event for HSCT vs Chemo alone of 1% vs 11.8% (p=0.002). Conclusions: This analysis from COG study AAML1831 indicates that delayed LP better identifies patients with CNS disease compared to prior studies AAML0531/1031. Despite fewer patients receiving additional IT therapy intensification due to CNS disease, the rates of CNS relapse were lower than prior COG studies AAML0531/1031, indicating a likely protective effect of using IT triple therapy and increased utilization of HSCT for high-risk patients. Unfortunately, patients with CNS3 disease on AAML1831 (similar to AAML0531/1031) had inferior EFS and high rates of CNS relapse compared to patients with CNS1/2 status, indicating a need for further optimization of therapy to prevent recurrence.