Paper Details

ABSTRACT

Background: Fludarabine (Flu) is a cornerstone of allogeneic hematopoietic cell transplantation (alloHCT) conditioning, but substantial interpatient pharmacokinetic (PK) variability produces heterogeneous drug exposure with standard dosing. Excessive Flu exposure may prolong immune recovery and increase the risk of non-relapse mortality (NRM), particularly among recipients of post-transplant cyclophosphamide (PTCy)-based graft-versus-host disease (GVHD) prophylaxis. We previously showed that high Flu exposure was associated with increased infection-related mortality, NRM, and inferior overall survival (OS) in a large cohort of PTCy-treated alloHCT recipients at Memorial Sloan Kettering (MSK) (Blood (2025) 146 (Supplement 1): 113). Here, we sought to externally validate these findings in participants of the ACCESS trial, a phase II, multicenter evaluation of PTCy-based GVHD prophylaxis after HLA-mismatched (MMUD) alloHCT with peripheral blood stem cell allografts (JCO, 2025; NCT04904588). The primary aim was to determine whether model-estimated Flu exposure reproducibly identifies patients at increased risk for adverse alloHCT outcomes.

Methods: We used a validated Flu population PK model to estimate cumulative Flu exposure (in mg*h/L) among all patients(Langenhorst et al., 2019). In the MSK training cohort, optimal Flu exposure thresholds associated with survival outcomes were identified using maximally selected log-rank statistics. This statistical approach was also independently applied to the multicenter ACCESS cohort to derive cohort-specific optimal Flu exposure thresholds and evaluate their association with clinical outcomes. Baseline characteristics between Flu exposure groups were compared using the Wilcoxon rank-sum test for continuous variables and Pearson's chi-square or Fisher's exact test for categorical variables.

Results: A total of 399 alloHCT recipients from the MSK training cohort and 268 recipients from the ACCESS validation cohort were included. The two cohorts were comparable with respect to age (median 64 vs. 62 years), disease (AML/MDS: 62% vs. 67%), conditioning intensity (RIC/NMA: 71% vs. 60%), and cumulative Flu exposure (median AUC 25.6 vs. 26.2 mg*h/L). All ACCESS patients received MMUD allografts, whereas 62% of MSK patients received haploidentical or MMUD donor allografts. All patients received conventional-dose PTCy (50 mg/kg/day on days +3 and +4 after alloHCT).

An optimal Flu exposure threshold of 25 mg*h/L was independently identified in the ACCESS cohort, closely approximating the 28 mg*h/L threshold previously derived at MSK. Across both cohorts, higher Flu exposure (above the observed thresholds) was consistently associated with worse outcomes. One-year NRM was 7.9% versus 23% in the optimal- and high-exposure groups at MSK (P=0.006) and 5.9% versus 17% in ACCESS (P=0.05). One-year OS was 82% versus 63% at MSK (P=0.03) and 88% versus 75% in ACCESS (P=0.07). Flu exposure was not associated with relapse, grades II–IV or III–IV acute GVHD, or chronic GVHD in either cohort.

In a multivariable Cox regression analysis adjusted for age, sex, HCT-comorbidity index, disease, conditioning intensity, and donor type, high fludarabine exposure was independently associated with increased NRM (HR, 2.26; 95% CI, 1.29–3.95; P=0.004) and inferior OS (HR, 1.65; 95% CI, 1.11–2.45; P=0.014), without difference in relapse in the MSK cohort. Consistent with these findings, multivariable analyses in the ACCESS cohort demonstrated that high fludarabine exposure remained associated with increased NRM (HR, 4.60; 95% CI, 1.85–11.5; P=0.001) and inferior OS (HR, 2.25; 95% CI, 1.10–4.62; P=0.022), with no difference in relapse.

The higher infection-related mortality observed in the MSK cohort was consistent in the ACCESS cohort. One-year infection-related mortality was 15% vs. 4.3% in the high- and optimal-exposure groups at MSK (P=0.05) and 8.9% vs. 1.7% in ACCESS (P=0.08).

Conclusion: This validation cohort from the ACCESS trial corroborates findings from our MSK training cohort, demonstrating that model-informed Flu exposure reproducibly identifies alloHCT recipients at increased risk of NRM and inferior survival largely driven by infection-related mortality after alloHCT in patients receiving conventional dose PTCy. These findings across two independent cohorts provide a strong rationale for prospective validation of model-informed Flu dosing to optimize alloHCT outcomes in patients receiving PTCy.