Adaptation, refinement and clinical correlation of a pharmacokinetics model for patients receiving anti-T-lymphocyte globulin as part of their conditioning regimen: Advances towards individualized dosing.

Abstract Introduction Despite the advantages of including anti-thymocyte globulin (ATG) or anti-T-lymphocyte globulin (ATLG) in the conditioning regimen for graft-versus-host disease prophylaxis (GVHD, Admiraal et al. Blood 2016, Turki et al. Am. J. Transplant. 2020), concerns remain relating to higher relapse rates (Soiffer et al. JCO 2017) or incidences of viremia after allogeneic stem cell transplantation (HCT). Individualized dosing of ATG and ATLG may offer the best solutions to this dilemma and have been successfully implicated in the pediatric setting using ATG (Admiraal et al. Lancet Hematol. 2022) and beginning using ATLG (Oostenbrink et al. Haematologica 2024 and Maier et al. Blood Advances 2024). However, for adult patients with ATLG such data has not yet been extensively explored. Aim: Here, we assembled both adult (n=298) and pediatric (n=44) cohorts undergoing HCT with ATLG as GVHD prophylaxis to model ATLG pharmacokinetics and explore individualized ATLG dosing. Methods: We first performed multiple simulation studies using several pretrained models for T-cell depleting agents' pharmakokinetics (Admiraal et al. 2014 and 2018) on real-word patient data exposed to ATLG to identify the most appropriate differential equation model for ATLG pharmacokinetics. These models were adapted to the ATLG use case and further refined e.g. by including random effects modeling. Next, we simulated these models on new adult and pediatric ATLG data and explored potential clinically-relevant cutoff points in the variance of ATLG pharmacokinetics peak dosing titers. Mean values and 95% confidence interval spreads were used as primary threshold. Peak ATLG titers after three days of ATLG application and baseline dosing (10mg/kg or 20mg/kg bodyweight) were associated with clinical outcome statistics of overall survival, nonrelapse mortality (NRM) and GVHD after HCT. Results: Acute myeloid leukemia was the most frequent disease (34.7%) in the adult cohort (n=298, median age 52.6, range: 18 – 76; 43.2% females). Patients received HCT primarily for hematological malignancies and ATLG as part of the conditioning (59.5% reduced intensity conditioning protocols, e.g. busulfan/fludarabine) for additional GVHD prophylaxis together with cyclosporine and methotrexate. ATLG was applied at a median dosage of 10mg/kg. Peripheral blood stem cells were infused in 283 patients (96.3%). The pediatric cohort included 44 patients with HCT. Median weight was 32kg and the median ATLG dosing was 20mg/kg. Starting the adaptation with initial simulations, the ATG model matched better to the requirements of our setting than the alemtuzumab pharmacokinetics model, as the latter resulted in very high peak levels in repeated simulations. Following further model refinements including random effects, we simulated ATLG pharmacokinetics with mean dosing peaks at about 45µg/ml following the third dosage of ATLG and rapid clearance to about 20µg at day +1 and about 10µg at day+8 after HCT. However, the spread of individual peak ATLG levels after three infusion days in this model was substantial ranging from -251µg/ml to +921 µg/ml. The negative values as lower extremes are not real, but result of random effect simulations. Recent pharmacokinetics measurement data of ATLG exposed pediatric patients (Oostenbrink et al. Haematologica 2024; Maier et al. Blood Advances 2024) described ATLG mean values and clearance times similar to our final model. For the analysed paediatric cases (n=44), the model fitted well and produced similar results with mean dosing peaks at about 49µg/ml. Finally, we explored the associations between the ATLG peak levels and clinical outcome. For ATLG peak values above or below the median, general significant differences in NRM, relapse or survival were not observed. Still, adults with relapse in the highest or lowest ATLG pharmacokinetics quartiles associated with different times to relapse (10 versus 60 months, p=0.0055). However, this association of ATLG levels was non-linear and not observed across all levels, reason why individualized outcomes analyses and subgroups are explored. For acute GVHD rates, adult patients with acute lymphoblastic leukemia had linearly decreasing incidence (p=0.0029) with increasing ATLG levels. Conclusions: Several pharmacokinetics models for ATLG dosing in adult and pediatric patients with HCT revealed substantial variance in individual ATLG levels that support individualized dosing strategies.