Long-Term Cognitive Impairment After CAR-T Therapy Versus Autologous Stem Cell Transplantation: A Propensity Score-Matched Cohort Study

Background/Objectives: Chimeric antigen receptor T-cell (CAR-T) therapy has transformed outcomes in relapsed or refractory hematologic malignancies, but long-term cognitive outcomes remain poorly understood. We compared the incidence and time course of cognitive impairment and associated neurological complications after CAR-T therapy compared with autologous stem cell transplantation (ASCT). Methods: This retrospective, propensity-matched cohort study utilized the TriNetX US Collaborative Network (January 2014–April 2025). To ensure concurrent comparisons, ASCT recipients were restricted to an index date beginning in August 2017 or later. CAR-T recipients were matched 1:1 to ASCT recipients for demographics, disease, comorbidities, prior and concomitant treatments, and laboratory parameters. The primary endpoint was time to cognitive impairment, as defined by ICD-10 codes. Results: After comparing 3067 CAR-T patients (median follow-up 634 days) with 3067 ASCT patients (median follow-up 713 days), CAR-T recipients had a higher risk of cognitive impairment (HR 1.58; 95% CI 1.39–1.80; p < 0.001). Because the risks were not proportional (Schaenfeld p < 0.001), the difference was also expressed as restricted median survival time (RMST): CAR-T recipients spent approximately 25 and 53 days fewer days without cognitive impairment at 1 and 2 years, respectively (both p < 0.001). The risk was greatest at 30 days (HR 4.22; 95% CI 3.23–5.53), but remained elevated in control analyses at 30 and 90 days that excluded the acute ICANS window (HR 1.30 and 1.25, respectively; both p < 0.05). Neurological dysfunction, particularly encephalopathy (HR 2.04; 95% CI 1.73–2.40), was more common after CAR-T. Conversely, CAR-T was associated with a reduced risk of secondary acute myeloid leukemia (HR 0.46; 95% CI 0.38–0.55; p < 0.001). Conclusions: CAR-T therapy is associated with a higher risk of cognitive impairment that persists beyond the acute phase. As these are observational, code-based data, they should be interpreted as associations rather than evidence of a specific mechanism, and they highlight the need for informed consent discussions, long-term neurocognitive monitoring, and the development of neuroprotective strategies.