Abstract
Background: Fludarabine (Flu) is a key conditioning agent to induce host lymphopenia and prevent rejection prior to allogeneic hematopoietic cell transplantation (alloHCT). Immunosuppression from Flu may increase non-relapse mortality (NRM) after alloHCT, particularly in patients receiving post-transplant cyclophosphamide (PTCy)-based graft-versus-host disease (GVHD) prophylaxis, where infections are more frequent. Previous studies have shown pharmacokinetic (PK)-guided dosing of agents like busulfan reduces toxicity and associated NRM. We hypothesized that over-exposure to Flu would increase NRM and lower survival following alloHCT. To evaluate this hypothesis, we used a population-level PK model to estimate Flu exposure following alloHCT in recipients of PTCy. The aims of the study were to evaluate the impact of Flu exposure on NRM and overall survival (OS), and to determine an optimal exposure window for prospective evaluation.
Methods: We used a validated Flu population PK model (Langenhorst et al., 2019) to estimate cumulative Flu exposure: the AUCt0−∞ and the AUC from time of transplantation until infinity (AUCtx−∞). Optimal cutoffs for Flu PK exposure were determined using maximally selected log-rank statistics for all survival outcomes. The p-values for comparisons between Flu exposure groups were approximated using maximally selected rank statistics for the respective outcome of comparison. Baseline characteristics between the two Flu exposure groups were compared using the Wilcoxon rank sum test for continuous covariates and Pearson's Chi-squared test along with Fisher's exact test for categorical covariates.
Results: A total of 399 alloHCT patients who received Flu-based conditioning and PTCy-based GVHD prophylaxis were included for analysis. Median age was 64 years (interquartile range (IQR), 56-69), and 60% were male. The median HCT-CI score was 3 (IQR, 1-4). Acute myelogenous leukemia and myelodysplastic syndrome accounted for 30% and 32% of patients, respectively. Most patients (71%) received RIC, while 29% received MAC. Haploidentical donor was the most common donor type at 40%, followed by matched unrelated donor at 31%. Among all patients, the estimated cumulative Flu AUC was 25.6 mg*h /L (mg*h/L) (IQR, 22.4-29.1). We identified 2 groups using maximally selected log-rank statistics: optimal Flu exposure (<28 mgh/L), and high Flu exposure (>28 mgh/L). Overall, 71% of patients were in the optimal exposure group, while 29% were in the high exposure group. The median follow-up was 35 months (IQR, 28-40). For the main outcomes of interest, the optimal exposure group had 1-year (OS) of 82% (95% confidence interval (CI), 77-86%) compared to 63% (95%CI, 54-73%) in the high exposure group (p=0.03). One-year NRM in the optimal exposure group was 7.9% (95%CI, 5.1-11%) vs. 23% (95%CI, 16-32%) in the high exposure group (p=0.006). There was no difference in relapse rates between optimal vs. high exposure (p=0.9), grade II-IV acute GVHD (p=0.7), grade III-IV acute GVHD (p=0.5), or chronic GVHD (p=0.085).
In a multivariable Cox regression analysis adjusted for age, gender, HCT-CI risk group, disease type, conditioning intensity, and donor type, high Flu exposure associated with lower OS (HR, 1.65; 95%CI,1.11 to 2.45; p=0.014) and higher NRM (HR, 2.26; 95%CI, 1.29 to 3.95; p=0.004) without difference in relapse.
In the high exposure group, infections and septic shock were the second leading cause of death (25% vs. 16% in optimal exposure) after relapse or progression (44% vs. 60%), followed by GVHD (15% vs. 9%).
Conclusion: Over-exposure to Flu, as determined using a readily available population PK model, defines a group of patients at higher risk for increased infections and inferior survival. Our data suggest that model-based Flu dosing to achieve an optimal exposure may be an easily modifiable strategy to improve survival in recipients of alloHCT receiving PTCy. External validation of these findings is planned.
