Inpatient CAR-T safety in the United States: Severe ICANS and CRS by donor source and antigen target (NIS 2022–2023).

e19006 Background: The field of CAR-T is rapidly expanding, with emerging allogeneic (donor-derived) platforms and multiple commercial products being used across hematologic malignancies. Comparative real-world national inpatient estimates of severe cytokine release syndrome (CRS) and immune effector cell–associated neurotoxicity syndrome (ICANS) across donor source and antigen target class remain limited, with existing comparisons often indirect or confined to specific products, trials, or registries. We evaluated severe inpatient toxicities and mortality after CAR-T by donor source and antigen target class in a nationally representative U.S. inpatient cohort. Methods: We performed a retrospective cross-sectional analysis of the HCUP National Inpatient Sample (NIS) data from 2022 to 2023. CAR-T hospitalizations were identified using ICD-10-PCS codes and stratified by donor type (autologous vs. allogeneic) and commercial product, further stratified by target antigen (CD19 vs. BCMA). Severe toxicities were captured using ICD-10-CM diagnosis coding for CRS and ICANS, with high-grade defined as grade 3–5. In-hospital mortality was obtained from the NIS discharge disposition mortality indicator. The primary outcome was high-grade ICANS (3–5); the secondary outcomes were high-grade CRS (3–5) and in-hospital mortality. Group differences used χ² tests (two-sided). Results: We identified an estimated 10,105 CAR-T hospitalizations: 9,485 autologous and 620 allogeneic. High-grade CRS was uncommon and similar between cohorts (autologous 3.8% [365/9,485] vs allogeneic 3.2% [20/620]; p=0.43). High-grade ICANS was markedly higher with autologous CAR-T (9.7% [920/9,485]) than allogeneic CAR-T (2.4% [15/620]; p<0.001). In-hospital mortality was low and not statistically different (2.2% [205/9,485] autologous vs. 3.2% [20/620] allogeneic; p=0.082). Among the most common commercial products, utilization was highest for axicabtagene (Yescarta, 31.1%), idecabtagene (Abecma, 12.8%), ciltacabtagene (Carvykti, 11.3%), brexucabtagene (Tecartus, 9.8%), and tisagenlecleucel (Kymriah, 5.4%). High-grade ICANS clustered in CD19 therapies [Tecartus 18.2% (180/990), Yescarta 15.4% (485/3,140), Kymriah 10.9% (60/550) and were lower with BCMA therapies (Abecma 4.7% (60/1,290), Carvykti 1.3% (15/1,140)]. Pooled by target, high-grade ICANS was 15.5% for CD19 vs 3.1% for BCMA (p<0.001). In-hospital mortality remained low across products (0.8%–3.0%) and was similar by target class (CD19 2.0% vs. BCMA 1.4%; p=0.078). Conclusions: In national inpatient data, ICANS differed significantly by donor source and antigen target, with lower high-grade ICANS in allogeneic and BCMA-targeted CAR-T. The inpatient mortality remained low across therapies. These findings support ongoing post-marketing surveillance and risk-adjusted analyses as CAR-T adoption evolves.