AUTHORS
Piyanuch Kongtim, MD, PhD, Piyatida Chumnumsiriwath, MD2, Pongthep Vittayawacharin, MD3, Dusit Jit-ueakul4, Nikkole Gaile Valdez5, Shawn Griffin, PharmD6, Jorge Ramos Perez7, Polina Bellman8, Erin A. Dean, MD9, Deepa Jeyakumar, MD10, Kiran Naqvi11, Alex C Minella, MD12, Stefan Ciurea, MD13
AFFILIATIONS
1 University of California Irvine, ORANGE, California, United States
2 University of California Irvine, ORANGE, California, United States
3 University of California Irvine, ORANGE, California, United States
4 University of California Irvine, ORANGE, California, United States
5 University of California Irvine, ORANGE, California, United States
6 University of California Irvine, ORANGE, California, United States
7 University of California Irvine, ORANGE, California, United States
8 University of California Irvine, ORANGE, California, United States
9 University of California Irvine, ORANGE, California, United States
10 University of California Irvine, ORANGE, California, United States
11 University of California Irvine, ORANGE, California, United States
12 University of California Irvine, ORANGE, California, United States
13 University of California Irvine, ORANGE, California, United States
ABSTRACT
Background: Fludarabine is a critical component of conditioning regimens for allogeneic hematopoietic stem cell transplantation (AHSCT). However, current standard dosing based on body surface area (BSA) results in profound pharmacokinetic (PK) variability because it fails to account for the drug’s primary reliance on renal clearance. While therapeutic drug monitoring (TDM) can optimize exposure, its clinical implementation is often hindered by high costs, logistical complexity, and slow turnaround times. In contrast, Population Pharmacokinetics offers a scalable, bedside alternative via model-informed precision dosing (MIPD), utilizing routine clinical parameters to predict individual exposure. Applying a data-driven optimization framework, we sought to define and validate the optimal fludarabine exposure window for adults with hematologic malignancies (HM) receiving AHSCT.
Methods: We evaluated 125 adult patients undergoing ASHCT for HM receiving fludarabine (40 mg/m²/day × 4 days), melphalan (100–140 mg/m²) and 2-4 Gy TBI. Cumulative 4-day fludarabine area under the curve (AUC) was calculated using the validated Langenhorst population PK model, which integrates estimated glomerular filtration rate (eGFR) and actual body weight (Langenhorst. Clin Pharmacokinet.2019). A two-cutpoint grid search optimization was used to identify the AUC pair that maximized the log-rank chi-square statistic for progression-free survival (PFS). This defined a three-tier therapeutic window, which was then internally validated via bootstrap resampling and permutation-based p-value correction. To assess the systemic failure of current BSA-based protocols, we performed a population simulation (400 scenarios) comparing BSA-based dosing against MIPD. The impact of exposure on longitudinal immune recovery was assessed through mixed-effects modeling of lymphocyte subsets through one-year post-AHSCT.
Results: Median patient age was 55 years (range 19–75); 65.6% were male. Donors included haploidentical (62.4%), matched unrelated (23.2%), and matched related (14.4%) sources. GVHD prophylaxis was predominantly post-transplant cyclophosphamide-based (89.6%).
Grid search optimization identified 19 mg·h/L and 24 mg·h/L as the optimal lower and upper AUC thresholds for PFS (log-rank chi2=8.33; p=0.016). PK analysis revealed that physiology was the primary determinant of exposure variability. The Under-exposed group (UEG, <19 mg·h/L; n=46) was significantly younger (median 37 years), heavier (mean 90 kg), and had superior renal function (mean eGFR 144.7 mL/min). Conversely, the Over-exposed group (OEG, >24 mg·h/L; n=43) was significantly older (median 65 years) with impaired renal function (mean eGFR 86.12 mL/min). Simulations confirmed that standard BSA-based dosing produces a 3-fold range in exposure (AUC 19.8–59.6 mg·h/L), with only 18.8% of scenarios achieving the target window.
The Optimal AUC group (n=36) achieved a 3-year PFS of 67.2%, significantly superior to the UEG (44.9%) and OEG (57.3%) (p=0.032). In multivariable analysis, deviating from this window was an independent predictor of inferior PFS (UEG: HR 2.95, p=0.024; OEG: HR 2.17, p=0.041) and overall survival (UEG: HR 4.92, p=0.008; OEG: HR 3.63; ;p=0.048). Median neutrophil engraftment was significantly faster in the UEG compared to the OEG (14 vs. 18 days; p=0.0007).
The failure mechanisms were exposure-dependent: Under-exposure was the primary driver of disease relapse (subdistribution hazard ratio; sHR 3.25, p=0.031). Conversely, Over-exposure was the primary determinant of transplant-related mortality (TRM) (sHR 3.27, p=0.039). Longitudinal analysis revealed that Over-exposure fundamentally altered the trajectory of immune reconstitution, resulting in significantly stunted absolute lymphocyte (p=0.045) and CD4+ T-cell recovery through one-year post-HCT (p=0.032). Decision Curve Analysis confirmed that a model-informed dosing strategy provided a superior clinical net benefit over standard BSA-based dosing across all clinical risk thresholds.
Conclusion: A cumulative fludarabine AUC of 19–24 mg·h/L optimizes AHSCT success in patients with hematologic malignancies by balancing the competing hazards of recurrence and lethal toxicity. Precision dosing using model-informed algorithms is required to correct the systematic exposure variability inherent in standard BSA-based dosing and maximize the curative potential of allogeneic transplantation.
DISCLOSURES
Piyanuch Kongtim, MD, PhD: CareDx - Consultancy (Includes expert testimony)