D93-09 Temperature Trajectory Subphenotypes After Chimeric Antigen Receptor T-Cell Treatment Identify Patients With Distinct Critical Illness Outcomes Independent of Toxicity Grade

Abstract Rationale Chimeric Antigen Receptor T-cell (CAR-T) therapy has transformed B-cell malignancy treatment but carries significant risk for toxicities, including cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS). Identifying clinically relevant toxicity patterns could improve prognostication and lead to more personalized management. We hypothesized that a previously validated group-based temperature trajectory subphenotyping model could categorize CAR-T recipients into subgroups with distinct clinical characteristics and risk for critical illness due to severe CRS or ICANS. Methods We performed a retrospective cohort study of adult patients with B-cell malignancies hospitalized between 2018 - 2024 for CAR-T therapy. We excluded patients with fevers before infusion and those without 24 hours of post-infusion data. Using group-based temperature trajectory models through 72 hours, we assigned patients into one of four groups: normothermic (NT), hypothermic (HT), hyperthermic fast-resolver (HFR), and hyperthermic slow-resolver (HSR). Groups were based on trajectories, not specific temperature thresholds. Co-primary outcomes were incidence and maximum grade of CRS or ICANS. Secondary outcomes were hospital length of stay, intensive care admission, vasoactive or respiratory support (invasive and noninvasive ventilation, high flow nasal cannula), receipt of tocilizumab, anakinra, steroids, or broad-spectrum antibiotics, bloodstream infection, and 90-day mortality. Results Among 249 patients receiving inpatient CAR-T therapy, 240 met inclusion criteria. Most patients received CD19 targeted CAR-T (n = 194, 81%, for B-cell lymphoma/leukemia) vs. BCMA targeted CAR-T (n = 46, 19%, for multiple myeloma). Fewer than five patients fit the HFR trajectory, thus we recategorized them as NT. Most patients fell into the NT (n = 93) or HT (n = 97) groups, followed by HSR (n = 50). There was no difference in malignancy type nor CAR-T product between subgroups. Neither the incidence nor the maximum grade of CRS or ICANS differed between groups (Table). However, HSR patients had higher rates of ICU admission (n = 14, 28%; n = 7, 7.2% in HT and n = 5, 5.4% in NT, p 0.001), vasoactive support (n = 7, 14%; n 5 in HT or NT, p = 0.011), and bloodstream infection (n = 22, 44%; n = 22, 23% in HT and n = 27, 29% in NT, p = 0.029), and received more antibiotics, corticosteroids, and anakinra. HSR patients had higher 90-day mortality (n = 12, 24%) than NT (n = 11, 11%) or HT patients (n = 8, 8.6%, p = 0.03). Conclusions Temperature trajectory subphenotypes were not clearly associated with CRS or ICANS but the HSR subphenotype differentiated patients who experienced more frequent ICU admission, organ support and higher mortality. Temperature-based trajectory subphenotyping may identify patients with worse clinical outcomes independent of toxicity grade. This abstract is funded by: NIH/NCI K08CA270383, NIH/NIGMS K23GM144867