Optimizing fludarabine exposure in pediatric CAR T-cell patients through an adaptive dosing approach

Background: B-cell acute lymphoblastic leukemia (B-ALL) is the most common cancer in children and young adults. While initial response rates are high (~90%), 15-20% of patients relapse within 2 years, and outcomes for relapsed or refractory (r/r) disease are poor. CD19-directed CAR T-cell therapy offers a potential cure for r/r B-ALL, but relapse currently occurs in approximately 50% of patients. Lymphodepletion with fludarabine (Flu) and cyclophosphamide is essential for CAR T-cell expansion and efficacy, with recent data showing that Flu exposure strongly correlates with outcome. Cumulative area under the curve (cAUC) values <14 mg.h/L were associated with shorter leukemia-free survival (1.8 vs 12.9 months) and higher CD19⁺ relapse (100% vs 27.4%). Notably, 40% of patients achieved suboptimal exposures (Dekker et al 2022), suggesting that therapeutic drug monitoring (TDM) to individualize Flu exposure may be beneficial. Methods: r/r B-ALL patients receiving CAR T-cell therapy were recruited to the NCCPG TDM 2018 clinical trial (ISRCTN10139334). Patients received 30 mg/m 2 /day Flu for 4 days and ≥4 plasma samples were taken per patient over the first 24 hours post Flu infusion for pharmacokinetic (PK) analysis. Plasma concentrations were analyzed by liquid chromatography-tandem mass spectrometry. PK analysis was conducted using a 3-compartment population PK model in InsightRX. Analysis was conducted in real time with dose adjustments implemented to target a cAUC within the therapeutic window of 16-20 mg.h/L. For a proportion of patients, further PK analysis was conducted on day 4 post dose adjustment. PK analysis was conducted with and without the inclusion of patient specific concentration-time data to assess the appropriateness of model-based dosing (MBD). Results: To date, analysis has been conducted for 21 patients. Estimated cAUCs were variable (10.2-26 mg.h/L) with no trends between exposure and age or weight. 81% of patients (17/21) required dose adjustments to achieve exposures within the therapeutic window. Without dose adjustments, nine patients would have achieved exposures <16mg.h/L, putting them at risk of CAR-T failure, with eight patients achieving exposures >20mg.h/L, increasing risk of drug toxicity. Three patients underwent additional TDM to validate dose adjustments, with final cAUCs within 6% of target exposures. Model-predicted exposures were significantly different from exposure predictions incorporating concentration-time data, with exposures outside the therapeutic window predicted in 47% of patients using MBD alone. Conclusions: Preliminary data indicate considerable interpatient variability in Flu exposure in r/r B-ALL patients receiving CAR T-cell therapy. BSA-based dosing and MBD approaches may be inadequate, demonstrating the need for adaptive dosing. TDM is feasible with Flu dosed across 4 days and leads to accurate target exposure attainment.