Li, Yimei

Abstract INTRODUCTION Outcomes for 1st relapse of pediatric B-lymphoblastic leukemia (B-ALL), especially high/intermediate-risk (HR/IR) relapse, remain unacceptably poor. Limited enrollment of this population on cooperative group clinical trials has impeded our ability to study potentially effective regimens. In the current era, patients can access innovative immunotherapies commercially. However, there is no mechanism to capture outcomes for these off-study patients and no comparative effectiveness data to guide treatment selection. METHODS We conducted a multicenter, retrospective study of patients <30 years who initiated 1st relapse therapy at a participating center from 2018-2022 (ReCALL-1 Study). Objectives were to define real-world survival outcomes, compare response rates of reinduction strategies, describe the post-reinduction treatment landscape, and, for patients with HR/IR relapse, assess the comparative effectiveness (CE) of blinatumomab to hematopoietic cell transplant (blina/HCT) v CD19 CAR T cells (CART). For the CE analysis, we emulated an open-label, randomized study of blina/HCT v CART who achieved an M1/M2 marrow at the end of an intensive reinduction (EoRI). Patients were classified by their intended treatment at EoRI. Cox regression models were used to adjust for unbalanced patient and disease features by arms. The primary outcome was overall survival (OS) from EoRI. RESULTS The full cohort included 479 patients from 33 centers (31 US, 2 international). Median time to relapse was 31 months (m); 77% had bone marrow (BM) involvement and 31% central nervous sytem involvement. With a median follow-up of 49m from relapse, 4 year (y) OS, event-free survival (EFS), and non-relapse mortality (NRM) were 64%, 43%, and 13%, respectively. Among patients with very early BM (<18m, n=79), early BM (18-35m, n=76), late BM (≥36m, n=162), and isolated CNS (n=85) relapses, 4y OS/EFS were 29%/20%, 57%/35%, 78%/49%, and 84%/61%, respectively. Only 19% received reinduction therapy on a clinical trial. Among patients with BM disease, R3 reinduction (n=179) and VXLD (n=61) had identical MRD-negative (<0.01%) CR rates of 38% (p=1.00) and equivalent treatment failure (TF; M3 or extramedullary disease) rates: R3, 8% v VXLD 7%, p=0.79. Blina (n=33) and inotuzumab-based (n=21) reinductions achieved MRD-negative CR rates of 39% and 52%, respectively, but higher TF rates of 39% and 24%. Reinduction mortality was 2%: R3, n=5; VXLD, n=3; 3-drug reinduction, n=1. Overall, as part of 1st relapse therapy, 36% received CART (78% commercial product), 44% HCT, 40% blina, and 11% inotuzumab. The emulated trial included 159 patients, 95 on blina/HCT and 64 on CART. Study arms were well matched by demographics, cytogenetics, and relapse sites, but CART patients were more likely to have relapsed <18m from diagnosis (34% v 24%) and have EoRI M2 marrows (31% v 11%), and less likely to have private insurance (36% v 51%). Unadjusted OS was comparable between arms (log rank p=0.18): CART, 60% at 4y; blina/HCT, 71%, as was NRM (10% v 13% at 4y). EFS was lower for CART (p=0.049): 4y EFS 35% v 49% for blina/HCT. After adjustment for unbalanced covariates, the hazard of death was equivalent for CART as compared to blina/HCT (aHR 1.1, 95% CI 0.6-2.1, p=0.67) and EFS became more comparable (aHR 1.3, 95% CI 0.8-2.1, p=0.22). 91% of blina/HCT and 97% of CART received their intended cell therapy; time from EoRI to cell therapy was shorter for CART (67 v 86 days, p<.001). In the CART arm, 34% underwent consolidative HCT; 1/3 were preplanned and 2/3 due to loss of B cell aplasia or NGS-MRD positivity. Of those with a 2nd relapse, CART patients were more likely to be CD19 negative (47% v 10%, p=0.003). CONCLUSION In this multicenter, real-world cohort of children treated for 1st relapse in the immunotherapy era, suboptimal outcomes for early BM relapse persisted, with particularly dismal 4y OS of 29% for relapses <18m after diagnosis. In contrast, survival for late BM and isolated CNS relapses compared favorably to historic data. Only 1 in 5 patients initiated relapse therapy on a clinical trial and although 1st relapse is not an approved indication for commercial CART in pediatrics, 36% received CART. In an emulated trial of post-reinduction blina/HCT v CART, adjusted EFS and OS were comparable, suggesting that CART may be a viable alternative to HCT for select patients. Post hoc analyses will identify whether specific subgroups may benefit more from one strategy.

Abstract Acute lymphoblastic leukemia (ALL) and acute myelogenous leukemia (AML) are the two most common childhood leukemias. Novel drug therapies and allogeneic hematopoietic cell transplant (alloHCT) have improved ALL and AML treatment outcomes over the past two decades. However, alloHCT may be associated with significant morbidity and mortality that results in increased healthcare utilization. To date, no multi-center comparative effectiveness studies have been performed specifically evaluating alloHCT in children with ALL or AML. This study uses a novel methodology to investigate the hypothesis that matched sibling donor (MSD) alloHCT has significantly lower healthcare utilization compared to unrelated donor (URD) and that among URD, umbilical cord blood transplants (CBT) will have higher initial utilization but lower long-term utilization. Clinical and transplant outcomes data from the Center for International Blood and Marrow Transplant Research (CIBMTR) database were merged with inpatient resource utilization and standardized cost data from the Pediatric Health Information System (PHIS) database using a probabilistic merge methodology. This merged dataset contained U.S. patients age 1-21 years who received alloHCT for ALL or AML from 2004-2013 with comprehensive CIBMTR data at a PHIS reporting hospital. The primary outcomes of overall survival (OS), leukemia free survival (LFS), and inpatient resource utilization were evaluated using Kaplan-Meier analysis, Cox proportional hazards, and unadjusted Poisson regression analysis. Cost and resource comparisons for all donor types were made through day+100 and focused on URD only comparisons up to 2-year follow-up. Donor procurement costs were not directly included in this analysis. A total of 797 patients (54% of CIBMTR eligible patients and 89% of CIBMTR patients at PHIS centers) were successfully identified in both CIBMTR and PHIS; 433 had ALL and had 364 AML. LFS for ALL at 3 years was 68% for MSD, 53% for MUD, and 53% for CBT. The LFS in ALL was significantly impacted by donor type, age, disease status, and CMV (Table). The associated total costs were significantly higher for well-matched unrelated bone marrow (MUD) vs MSD at 100d ($246,149 vs $147,547, p<0.01), but CBT costs were not significantly higher compared to MUD at 100d ($275,673 vs $246,149, p=0.09). When median total costs per day were compared across donor source, findings were similar with well- MUD was significantly higher than MSD at 100d ($3,388/d vs $2,511/d, p=0.012), and no significant difference between CBT and MUD ($3,881/d vs $3,388/d, p=0.53). The corresponding average length of stay (ALOS) was 35 days for MUD, 34 days MSD, and 53 days CBT. Among URD alloHCT, median CBT costs per day were not significantly higher compared to MUD at 2yrs ($2,472/d vs $2,294/d, p=0.49). AlloHCT for AML had a 3-yr LFS of 56% for MSD, 46% for MUD, and 41% for CBT. LFS were significantly different by disease status and performance status (Table). The associated total costs were significantly higher for MUD at 100d than MSD ($259,293 vs $143,870, p<0.001), but CBT costs were not significantly higher compared to MUD at 100d ($301,160 vs $259,293, p=0.055). Comparisons of median total costs per day were similar, MUD vs MSD at 100d was significantly higher ($3,627/d vs $2,478/d, p<0.01), but CBT costs were not significantly higher compared to MUD at 100d ($3,891/d vs $3,627/d, p=0.95). The corresponding ALOS was 38 days for MUD, 35 days MSD, and 66 days CBT. Among URD alloHCT, median CBT costs per day were not significantly higher compared to MUD at 2yrs ($3,232/d vs $3,102/d, p=0.49). This study recapitulates previously published outcomes of alloHCT for acute leukemia AND adds key findings on the impact of alloHCT on inpatient resource utilization and costs. Specifically, MSD and MUD alloHCT provide similar survival outcomes; however, MSD alloHCT has a significant advantage in cost and resource utilization for both ALL and AML. Among URD transplants, CBT does not show any substantial survival or resource utilization advantage over MUD. Ongoing research will need to be performed to determine if the difference among URD alloHCT becomes significant with a larger sample size and/or beyond the 2 years following alloHCT. Table Table. Disclosures Pulsipher: Novartis: Consultancy, Other: Study Steering Committee; Jazz Pharmaceutical: Consultancy; Medac: Other: Housing support for conference; Chimerix: Consultancy. Hahn:Novartis: Equity Ownership; NIH: Research Funding.

Abstract Introduction: Current advances in allogeneic hematopoietic cell transplant (alloHCT) may warrant a paradigm shift in managing children with sickle cell disease (SCD). This study characterizes the clinical outcomes and health care utilization (HCU) of alloHCT for pediatric SCD. We hypothesize that early alloHCT will have improved clinical outcomes, and decreased HCU. Methods: The Center for International Blood and Marrow Transplant Research database was used to identify children 21 years or less with alloHCT for SCD in the United States. Patient data included comprehensive research forms (CRF) from 2000-13 and transplant essential data (TED) forms from 2000-11. CRFs provided clinical risk factors associated with overall survival, graft failure, grade III-IV acute GVHD, and GVHD related event free survival (GREFS) - the survival free of graft failure, chronic GVHD, or death. Risk factors included age, gender, performance status, year of alloHCT, prior SCD complications and therapy, CMV status, donor type, conditioning regimen, and GVHD prophylaxis. Due to low event rates and sample size, only univariate analysis of risk factors was performed. TED data was merged with Pediatric Health Information System (PHIS) inpatient data using a probabilistic merging algorithm to determine risk factors and clinical outcomes associated with HCU. Available PHIS adjusted cost data was used to determine the total adjusted cost for all inpatient admissions per patient per hospital day. To standardize these costs, the total adjusted cost per 30 hospital days was calculated for each patient and used as the primary HCU outcome. HCU outcomes were analyzed for the alloHCT year, day 0 to day +365. Results: CRF data for 161 patients showed an overall survival at 2 years of 90% (95% confidence intervals [CI] 85-95%): 96% (95% CI 89-100%) for related and unrelated cord blood transplant (CBT), 94% (95% CI 86-98%) matched siblings (MSD), and 74% (95% CI 54-90%) matched unrelated donors (MUD, p=0.002). All deaths occurred among children with pre-alloHCT complications of SCD, and deaths were due to organ failure (37.5%), infections (25%), GVHD (6.25%). Risk of death was significantly higher for children ≥10 years old (HR 21, p=0.003) and MUD compared to MSD (HR 5.88, p=0.005) but lower with cyclosporine A (CSA) GVHD prophylaxis versus FK506 (HR 0.33, p=0.031). 75% of deaths occurred before day +42. Cumulative acute GVHD incidence at day 100 was 14% (95% CI 9-20%)and was associated with age ≥10yrs (HR 2.63, p=0.035). Chronic GVHD incidence was 31% (95% CI 23-38%) at 2yrs, and factors associated were age ≥10yrs (HR 1.92, p=0.034), MUD vs MSD (HR 2.53, p=0.017), and CSA vs FK506 prophylaxis (HR 0.48, p=0.018). Chronic GVHD risk increased significantly after 2006 (HR 2.81, p=0.018). The 2yr GREFS was 64% (95% CI 56-71%). Age ≥10yrs (HR 2.2, p=0.005), MUD (vs MSD, HR 3.00, p=0.002) and CSA prophylaxis (vs FK506, HR 0.49, p=0.011) were significantly associated with this outcome. Among the 175 patients with combined TED and PHIS data, the median total adjusted cost was $117,393 per 30 hospital days per patient (range: $36,244-$515,640) during the alloHCT year with a median of 53 hospital days per patient (range: 16-304). Age ≥10yrs and HCU were not significantly associated (p=0.775). MSD had the lowest HCU compared to CBT and unrelated transplants (p<0.001). CBT and peripheral blood stem cells were associated with higher HCU compared to bone marrow (p=0.004). Increased HCU was associated with prior stroke (p=0.0004) and pain crises (p=0.0094) but not acute chest syndrome (p=0.2913). Overall SCD complication and severity indices correlated with increased HCU (p=0.052, p=0.0219, respectively). Conclusions: AlloHCT outcomes in children with SCD were linked to age and donor type suggesting that early alloHCT before age 10 years is preferred. Specifically, SCD severity and MUD alloHCT are associated with poorer outcomes and increased HCU. This supports the recommendation of early alloHCT, prior to onset of SCD complications, for children with SCD and an available MSD. Donor source and type had a significant impact on both outcomes and HCU. CBT outcomes were similar to MSD bone marrow; yet CBT had higher HCU suggesting additional analysis is needed to determine if the clinical benefit outweighs the cost. Further analysis is also needed to better understand and mitigate risk factors associated with poor outcomes and increased HCU following MUD alloHCT. Disclosures Arnold: Robert Wood Johnson Foundation Harold Amos Medical Faculty Development Program: Other: award. Hahn:NIH/NHLBI: Research Funding; Novartis: Equity Ownership.