Characterization and prediction of hematotoxicity in pediatric patients receiving tisagenlecuecel Open Access

• PSN occurred in 15% and 21% of CAR T recipients in training and validation cohorts respectively was associated with inferior survival

• The CytoRisk score predicted patients that would develop PSN with high specificity (93%), sensitivity (74%), PPV (74%), and NPV (92%)

Hematotoxicity is the most frequent severe toxicity following chimeric antigen receptor (CAR) T therapy. However, limited data exist on risk factors and outcomes for hematotoxicity for children and young adults (CAYA) with B-acute lymphoblastic leukemia (B-ALL) treated with tisagenlecleucel. We conducted a multi-institutional study involving 326 CAYA, with 144 evaluable in an initial training cohort and 141 evaluable in a validation cohort, through the Pediatric Real World CAR Consortium to characterize the incidence and outcomes of prolonged severe neutropenia (PSN) and to develop a predictive risk score for PSN, tailored for use in this population. The incidence of PSN, defined as an absolute neutrophil count (ANC) of <500 cells/µL for ≥ 30 days, was 15.3% in the initial training cohort and 21% in the validation cohort. Development of PSN was associated with inferior overall survival (p<0.001), relapse-free survival (p=0.01), higher non-relapse mortality (p=0.003), and a greater risk of infections within 30 days (p=0.03). Multivariable penalized regression analysis identified key risk factors for PSN which included pre-infusion C-reactive protein and bone marrow disease burden, and post-infusion peak ferritin and occurrence of severe CRS, which were used to create the CytoRisk score. In the validation cohort, the CytoRisk score discriminated between patients with and without PSN (area under the curve: 0.90, specificity: 93%, sensitivity: 71%, positive predictive value: 74%, negative predictive value: 92%). The CytoRisk score may be used to a priori identify patients at highest risk of PSN and overall worse outcomes.