Fludarabine exposure predicts outcome after CD19 CAR T-cell therapy in children and young adults with acute leukemia

Abstract

The addition of fludarabine to cyclophosphamide as a lymphodepleting regimen prior to CD19 chimeric antigen receptor (CAR) T-cell therapy significantly improved outcomes in patients with relapsed/refractory (r/r) B-cell acute lymphoblastic leukemia (B-ALL). Fludarabine exposure, previously shown to be highly variable when dosing is based on body surface area (BSA), is a predictor for survival in allogeneic hematopoietic cell transplantation (allo-HCT). Hence, we hypothesized that an optimal exposure of fludarabine might be of clinical importance in CD19 CAR T-cell treatment. We examined the effect of cumulative fludarabine exposure during lymphodepletion, defined as concentration-time curve (AUC), on clinical outcome and lymphocyte kinetics. A retrospective analysis was conducted with data from 26 patients receiving tisagenlecleucel for r/r B-ALL. Exposure of fludarabine was shown to be a predictor for leukemia-free survival (LFS), B-cell aplasia, and CD19-positive relapse following CAR T-cell infusion. Minimal event probability was observed at a cumulative fludarabine AUCT0-∞ ≥14 mg*h/L, and underexposure was defined as an AUCT0-∞ <14 mg*h/L. In the underexposed group, the median LFS was 1.8 months, and the occurrence of CD19-positive relapse within 1 year was 100%, which was higher compared with the group with an AUCT0-∞ ≥14 mg*h/L (12.9 months; P < .001; and 27.4%; P = .0001, respectively). Furthermore, the duration of B-cell aplasia within 6 months was shorter in the underexposed group (77.3% vs 37.3%; P = .009). These results suggest that optimizing fludarabine exposure may have a relevant impact on LFS following CAR T-cell therapy, which needs to be validated in a prospective clinical trial.

Graphical abstract

Figure 1.

Fludarabine exposure and functional form. (A) Histogram (gray area) and density plot (black solid line) of the determined cumulative fludarabine AUC_T0−∞. Density is the proportion of patients with fludarabine exposure within the specified limits. (B) Scatterplot of cumulative fludarabine AUC_T0−∞ vs martingale residuals of null Cox proportional hazard model for LFS.

Figure 2.

Cumulative fludarabine AUC_T0−∞ <14 mg*h/L associates with worse outcome. (A-C) Kaplan Meier plots of LFS and the occurrence of CD19-positive relapse and B-cell recovery from the day of CAR T-cell infusion stratified by cumulative fludarabine AUC_T0−∞ of 14 mg*h/L. Groups were compared using the log-rank test. P values < .05 were considered statistically significant.

Figure 3.

Cellular kinetics after CAR T-cell infusion in groups stratified by cumulative fludarabine AUC_T0−∞ of 14 mg*h/L. (A) The AUCs of CAR T-cell numbers from the first 28 days after CAR T-cell infusion. The high and low fludarabine exposure groups consisted of 9 and 7 patients, respectively. Groups were compared using the Mann-Whitney U test. (B) LOESS regression curves of CD4+ T-cell recovery within the first month after CAR T-cell infusion in the whole cohort. (C) LOESS regression curves of CD8+ T-cell recovery within the first month after CAR T-cell infusion in the whole cohort. Groups were compared with linear mixed effect models. P < .05 considered statistically significant.

Figure 4.

Distribution of determined and predicted fludarabine AUC_T0−∞. Boxplots of the cumulative fludarabine AUC_T0−∞ determined in our cohort (determined) and predicted using a previously published population pharmacokinetic model taking renal function and body weight as covariates (predicted). Determined and predicted AUC_T0−∞ of every patient (n = 26) is linked with a black solid line. Groups were compared with the Mann-Whitney U test. P < .05 considered statistically significant.