Rapid peak CAR-T expansion is associated with delayed neurotoxicity following ciltacabtagene autoleucel in multiple myeloma

Abstract Introduction: Ciltacabtagene autoleucel (cilta-cel) has demonstrated deep and durable responses in relapsed/refractory multiple myeloma (RRMM). However, its broader adoption is limited by unique toxicities, particularly delayed neurotoxicity (DNT). The relationship between CAR T-cell expansion and DNT is not well understood. In this study, we evaluated whether peak CAR T-cell expansion, measured by flow cytometry, is associated with DNT risk. We also investigated absolute lymphocyte count (ALC) as a reliable surrogate for CAR T-cell expansion and identified clinically actionable ALC thresholds to stratify DNT risk. Methods: We included 256 patients receiving commercial cilta-cel from 2022 to 2024 for RRMM across 3 institutions (ALC cohort). We assessed associations between peak ALC following infusion (day 0) and DNT. In a subset of 54 patients (CAR expansion cohort), weekly flow cytometry quantified CAR T-cell expansion. Results: In the ALC cohort (n=256), the median age was 64 years (IQR: 57.8, 70), 54% were male and 38% were classified as penta-refractory. Median follow-up was 14.7 months, and median progression-free survival (PFS) was 28.7 months. DNT occurred in 11% (n=29) of patients, including 8% (n=20) with cranial nerve palsy and 3% (n=8) with Parkinsonism. In the CAR expansion cohort (n=54), baseline characteristics were similar. Peak CAR T-cell expansion did not significantly differ between patients achieving a response (=PR) vs not at 3 months (median 155 vs. 66 cells/uL; P=0.27) or at 6 months (median 128 vs 90 cells/uL; P=0.65). Additionally, peak CAR T-cell expansion was not significantly associated with time-to-progression (TTP; HR 1.0; P=0.82). Similarly, peak CAR T-cell expansion was not significantly different between patients who developed CRS vs not (P=0.13). However, it was significantly higher in patients who developed ICANS (median 961 vs 91; P=0.001). DNT was observed in 16% (n=8), including Parkinsonism in 7.5% (n=4). Patients who developed DNT had significantly higher peak CAR-T expansion compared to those who did not (p=0.04). In the ALC cohort, patients who developed DNT had a significantly higher median peak ALC compared to those who did not (5780/μL vs. 2200/μL; p0.001). Among patients with Parkinsonism, peak ALC was significantly higher compared to those without (13,335/μL vs. 2270/μL; p0.001). The early rise in ALC from day 7 to 12 was significantly greater in patients with DNT (5360/μL vs 1040/μL: p0.001). The following peak ALC thresholds as optimal and clinically implementable for DNT risk stratification: either (1) ≥3000/μL between days 7-21, and/or (2) ≥2500/μL between days 8-12 with a ≥2-fold increase from the prior value. These thresholds collectively captured 81% of DNT while excluding 59% of patients who did not develop DNT. Conclusion: Rapid and robust CAR T-cell expansion was associated with an increased risk of DNT. Additionally, ALC may serve as a practical surrogate for early identification of high-risk patients. ALC thresholds of (1) ≥3000/μL between days 7-21, and/or (2) ≥2500/μL between days 8-12 with a ≥2-fold increase from the prior value collectively identified the majority of DNT cases and may guide the implementation of preemptive interventions.