Predictors of severe hematotoxicity after CAR T-cell therapy.

7024 Background: The EHA/EBMT immune effector cell-associated hematotoxicity (ICAHT) grading system characterizes hematotoxicity after CAR T-cell therapy based on depth and duration of neutropenia (Rejeski et al, Blood, 2023). We evaluated pre- and post-infusion factors associated with grade 3-4 ICAHT and developed a predictive model in 602 patients with hematologic malignancies undergoing CAR T-cell therapy at Fred Hutch Cancer Center. Methods: Grading of early hematotoxicity (day-0-30 after CAR T-cell cell infusion) was automated using the heatwaveR package. Associations with 50 patient, disease-related, and laboratory factors, and grade 3-4 ICAHT were modeled using univariate and multivariable logistic regression. Results: The most common disease types were aggressive non-Hodgkin lymphoma (NHL; n = 293; 49%), indolent NHL (n = 150; 25%), and acute lymphoblastic leukemia (ALL; n = 94; 16%). The most common CAR T-cell products were the investigational CD19 CAR T-cell product JCAR014 (n = 197; 33%), axicabtagene ciloleucel (n = 129; 21%), and lisocabtagene maraleucel (n = 73; 12%). Incidences of early ICAHT grades 1, 2, 3, and 4 were 319 (53%), 96 (16%), 60 (10%), and 35 (6%) patients, respectively. Baseline patient factors associated with grade 3-4 ICAHT included ALL (reference: aggressive NHL, OR = 4.18, 95% CI, 2.41-7.26, p < 0.001) , Hispanic or Latino ethnicity (OR = 2.17, 95% CI, 1.07-4.20, p = 0.025), and lower age (OR = 1.03, 95% CI, 1.02-1.05, p < 0.001). Pre-lymphodepletion (LD) laboratory factors associated with grade 3-4 ICAHT in univariate analyses included lower ANC (OR = 6.67 per log10 cells/μL, 95% CI, 4.0-11.1, p < 0.001), LDH (OR = 8.70per log10U/L, 95% CI, 4.03-19.4, p < 0.001), CRP (OR = 3.18 per log10mg/L, 95% CI, 2.06-5.07, p < 0.001), and ferritin (OR = 6.07 per log10mg/L, 95% CI, 3.65-10.6, p < 0.001). Peak CRP (OR = 15.6, 95% CI, 7.27-36.4, p < 0.001), peak ferritin (OR = 7.41, 95% CI, 5.02-11.3, p < 0.001), peak CRS grade (OR = 2.01, 95% CI, 1.59-2.56, p < 0.001), and peak ICANS grade (OR = 1.51, 95% CI, 1.29-1.77, p < 0.001) after CAR T-cell infusion were also strongly associated with grade 3-4 ICAHT. A multivariable model using restricted cubic splines and including disease type, pre-LD ANC, pre-LD LDH, peak CRP, peak ferritin, and CRS grade demonstrated high discrimination to predict grade 3-4 ICAHT (C-index = 0.89). Internal validation using bootstrapping showed near-perfect calibration without overfitting. Sensitivity and specificity based on the Youden criteria was 74% and 90%, respectively. The sensitivity and specificity of the CAR-HEMATOTOX score in predicting grade 3-4 ICAHT was 93% and 30%, respectively. Conclusions: We identified pre- and post-infusion predictors of grade 3-4 ICAHT and internally validated a multivariable logistic regression model including disease-type, pre-LD ANC, pre-LD LDH, peak CRP, peak ferritin, and CRS grade. We plan to further evaluate our model in an external cohort.