BACKGROUNDAnti-CD19 CAR T-cells have revolutionized outcomes in relapsed/refractory large B-cell lymphomas. Long-term follow-up underscored the role of hematological toxicity in non-relapse mortality, largely driven by infections, leading to the development of the CAR-HEMATOTOX (HT) score for predicting neutropenia. The European scientific community (EHA/EBMT) later reached a consensus, defining a new entity: immune effector cell-associated hematotoxicity (ICAHT).AIMSTo validate the ability of the HT score to predict ICAHT and survival.METHODSThe CART-SIE is an ongoing multicenter prospective observational study collecting data on patients affected by B-cell lymphoma treated with commercial anti-CD19 CAR T-cells (ClinicalTrials.gov ID: NCT06339255).RESULTSSince 2019 to 2024, 1002 consecutive patients were enrolled. Out of 746 patients infused, the HT score at infusion was evaluable in 389. Median age was 59 years (48-66). Patients with high HT score had greater disease burden and a greater need for bridge therapy. Patients with a HTHIGH score had a 4-fold higher risk of experiencing late ICAHT of grade≥3 (OR=3.99, 95%CI=1.16-13.77, p=0.03). Patients with a HTHIGH score also showed lower overall response rates (ORR) and complete response rates (CRR) at 90 days (CRR at 90 days: 59% HTLOW vs 38% HTHIGH, OR=0.42, 95%CI=0.27-0.66, p=0.0002; ORR at 90 days: 67% HTLOW vs 49% HTHIGH, OR=0.47, 95%CI=0.29-0.74, p=0.001). Adjusted logistic models confirmed that the effect of HT score was independent from baseline characteristics. With a median follow-up of 18 months, patients with a HTHIGH score have lower OS and PFS (1-year OS: 78% HTLOW versus 62% HTHIGH, p=0.0002; 1-year PFS: 49% versus 39%, p=0.003). Adjusted Cox models confirmed that HT was an independent prognostic factor for OS. A high HT-score was found to be associated with higher risk of secondary primary malignancy (HR=2.8, 95%CI=1.03-7.8, p=0.04). A simplified version of HT (simpleHT), based solely on the platelet count and C-reactive protein at infusion, was calculated for 560 patients and proved significant in predicting both OS and PFS (1-year was 72% simpleHTLOW vs 37% simpleHTHIGH, p<0.0001, 1-year PFS was 48% simpleHTLOW vs 22% simpleHTHIGH, p<0.0001).CONCLUSIONIn our prospective real-world study, we validated the ability of the HT score to predict ICAHT and survival. SimpleHT identified a population at very high risk with an impaired progression free and overall survival.