Clinical Factors Associated with Prolonged Cytopenias after CD19 Chimeric Antigen Receptor T-Cell Therapy for Pediatric B-Cell Acute Lymphoblastic Leukemia

Introduction: Cytopenias are among the most common CD19 CAR T cell therapy (CAR19)-related adverse events. Although the frequency, severity, risk factors and outcomes of CAR19 related cytopenias are well described in adult lymphoma and multiple myeloma, data for pediatric B-lymphoblastic leukemia (B-ALL) is limited. As a bone marrow infiltrative disease, cytopenia characteristics in B-ALL differ from lymphoma, making this an especially relevant data gap to address. Methods: We conducted a multicenter, retrospective review of children and young adults treated with 4-1BB-based CAR19 constructs for B-ALL between 2012-2022 at 6 US centers in the CAR-Multicenter Analysis cohort. The cytopenia analysis was limited to patients in a complete remission (CR) at Day 28 (D28) after infusion. Later timepoints were further limited to those who remained in remission without hematopoietic cell transplant (HCT) or alternative therapy. The primary outcome was a composite endpoint of at least 1 severe cytopenia (grade [gr] 3-4 neutropenia, anemia or thrombocytopenia) present at D28 after infusion per CTCAE v5.0. Secondary outcomes included individual and composite cytopenias at months 3 and 6 after infusion, as well as relapse-free survival (RFS) and overall survival (OS), stratified by pre-infusion disease burden (DB; high, ≥25% marrow blasts; low, <25% blasts) and by cytopenia severity at D28. Results: Of 379 patients treated with CAR19, 350 (92%) achieved a CR at D28 and were included in the analysis. Median age at infusion was 13 years (range, 1-29); 39% received a prior HCT and 29% had high DB pre-infusion. At baseline, prior to lymphodepletion, 16% and 24% had gr 3 and 4 composite cytopenias, respectively. At D28, 155/313 (50%) patients with complete laboratory data had severe composite cytopenias (gr 3, 22%; gr 4, 27%). This proportion decreased to 60/224 (26%) at month 3 (gr 3, 18%; gr 4, 19%) and 25/153 (17%) at month 6 (gr 3, 14%; gr 4, 3%). The respective frequencies of severe neutropenia, anemia and thrombocytopenia were 43%, 15%, and 27% at D28; 25%, 1%, and 10% at month 3; and 14%, 0%, and 6% at month 6. GCSF was used in 5% of patients at D28, 1% at month 3, and 2% at month 6. One patient received a stem cell boost at month 4 due to persistent cytopenias. To evaluate predictors of severe composite cytopenias at D28, we developed a multivariable logistic regression model incorporating baseline factors associated with cytopenias by univariate analysis at p<0.1. In the multivariate analysis, baseline cytopenias (gr 2: odds ratio [OR] 3.6, p<0.001; gr 3: OR 3.4, p=0.002; gr 4: OR 7.9, p<0.001), gr 3-4 cytokine release syndrome (CRS) (OR 9.8, p<0.001), prior HCT (>1 year pre-CAR19: OR 1.9, p=0.04; within 1 year: OR 3.7, p=0.003), and high DB (OR 2.5, p=0.009) were all significant risk factors. We constructed similar multivariable models at months 3 and 6. At month 3, baseline gr 4 cytopenias (OR 3.9, p=0.003) and prior HCT (>1 year: OR 1.9, p=0.07; within 1 year: OR 4.2, p=0.002) remained associated with composite cytopenias, but baseline gr 2-3 cytopenias, CRS, and DB did not. The addition of D28 cytopenias to the model demonstrated the strongest association with month 3 cytopenias (gr 3: OR 4.3, p=0.002; gr 4: OR 7.2, p<0.001). At month 6, only prior HCT (>1 year: OR 3.0, p=0.03; within 1 year: OR 2.7, p=0.14) was associated with composite cytopenias, along with D28 gr 4 cytopenias (OR 11.3, p=0.003). To determine if D28 severe composite cytopenias are associated with worse survival, we evaluated RFS and OS by cytopenias using Kaplan-Meier methods. Since pre-infusion DB is the strongest predictor of CAR19 outcomes, we further stratified by low versus high DB. Median follow-up was 26 months. For low DB, RFS (p=0.22) and OS (p=0.96) were similar by cytopenia severity. For high DB, RFS was similar by cytopenia severity (p=0.33), but there was a trend towards worse OS in patients with severe cytopenias (p=0.09). Conclusion: In the largest analysis of cytopenias in children after CAR19 for B-ALL to date, we demonstrate that 50% have persistent, severe cytopenias at D28, with 17% maintaining gr 3-4 cytopenias at month 6. Baseline cytopenias, HCT within 1 year pre-CAR19, high DB, and severe CRS are all highly associated with cytopenias at D28. However, severe persistent cytopenias at months 3 and 6 are best predicted by D28 cytopenias. These results provide novel predictors of CAR19 hematotoxicity in B-ALL and insights into its underlying pathophysiology.