(483) Predictive Performance of Early Warning Scoring Systems for ICU Transfer Following CAR-T Therapy

Introduction: Chimeric antigen receptor T-cell therapy (CAR-T) has shown durable remissions in patients with advanced onco-hematological malignancies. However, CAR-T treatment patients are at risk of severe complications, with cytokine release syndrome (CRS) being the most frequent. Early detection of these complications is essential for rapid transfer to the ICU and the implementation of appropriate therapies. Therefore, in this study, we evaluated the predictive performance of Early Warning Systems (EWS), including NEWS 2, MEWS, PARS, and qSOFA scores, in predicting Intensive Care Unit (ICU) admission among patients undergoing CAR-T cell therapy.

Methods: A retrospective cohort study was conducted across Mayo Clinic sites from 2019 to 2021, analyzing data from adult patients receiving CAR-T therapy. EWS scores were computed using vital signs recorded from electronic health records, comparing scores 24 hours before ICU admission to those in non-ICU patients within 24 hours post-therapy. Discrimination (AUROC) and calibration (Brier Score) were assessed for each EWS.

Results: During the study period, 214 adult patients underwent CAR-T therapy. All treated patients were included, with 42% (n = 91) having Diffuse Large B-cell Lymphoma (DLBCL) and 28% (n = 61) diagnosed with refractory Multiple Myeloma (MM). Twenty-three patients were transferred to the ICU within 21 days after CAR-T treatment, with a median time of 8 days (IQR 5-13) post-treatment. All EWSs showed an AUROC value above 0.6, but NEWS 2 and PARS demonstrated the highest AUROC values of 0.72 (95% CI 0.61, 0.83) and 0.74 (95% CI 0.63, 0.84), respectively. MEWS and qSOFA showed moderate discrimination. NEWS 2 and MEWS, showed the highest calibration with Brier scores of 0.087 and 0.084, respectively.

Conclusions: NEWS 2 was the most effective EWS in predicting ICU admission post-CAR-T therapy, demonstrating both good discrimination and calibration, with potential usefulness for monitoring and early warning for complicated patients, indicating an early transfer to the ICU. These findings underscore its utility in the early identification of deteriorating patients and timely escalation of care.