Δ-β2M as a dynamic biomarker for predicting ICU admission after CAR T-cell therapy

CAR T-cell therapy is associated with life-threatening toxicities requiring intensive care unit (ICU) admission. Reliable biomarkers for early risk stratification are lacking. We evaluated Δ-β2-microglobulin (Δ-β2M) as a dynamic biomarker for predicting ICU admission after CAR T-cell therapy. Single-center retrospective study with a derivation cohort (2020-2023, n = 97) and two temporally independent validation cohorts (2024, n = 38; 2025, n = 23). Δ-β2M was calculated as the change between infusion (T2) and day +3 (T3). Primary endpoint was ICU admission. Predictive performance was assessed using receiver operating characteristic (ROC) analysis, calibration (Hosmer-Lemeshow test, Brier score), and decision curve analysis. Complete Δ-β2M data were available in 54 (derivation), 17 (2024), and 17 (2025) patients. Δ-β2M demonstrated consistent discrimination for ICU admission across all cohorts (derivation AUC 0.67; 2024 AUC 0.70; 2025 AUC 0.82), but not for immune effector cell associated neurotoxicity syndrome (ICANS) (AUC 0.53). The model showed good calibration (Hosmer-Lemeshow p = 0.073; Brier score 0.232), confirmed by cross-validation (Brier score 0.236). Decision curve analysis revealed positive net benefit across clinically relevant thresholds (0.20-0.70). Baseline parameters (albumin, CRP, NT-proBNP) lacked consistent validation. Δ-β2M is an easily measurable dynamic biomarker associated with ICU admission after CAR T-cell therapy, with good calibration and clinical utility. These findings support further evaluation of early biomarker kinetics in toxicity risk prediction models.