Population pharmacokinetic analysis of guanabenz in children with Vanishing White Matter: Evaluating scaling adult drug safety and pharmacokinetics data
Bart van Lieshout (1), Renate Verbeek (1), Marije Voermans (1), Pierre M. Bet (1), Ron A. A. Math˘t (1), Marjo S. van der Knaap (1)*, Imke Bartelink (1)* *equal contribution
(1) Amsterdam University Medical Centers, Amsterdam, the Netherlands
Introduction: Vanishing White Matter (VWM) is an inherited leukodystrophy caused by mutations that affect the eIF2B protein, which has a key role in the regulation of protein synthesis and the integrated stress response (ISR). In VWM patients the ISR is continuously activated and, with no current effective treatment, leads to the degradation of white matter and death. Guanabenz, an old α2-adrenergic receptor agonist, reduces ISR mRNA expression in VWM mouse models. Pharmacokinetic and pharmacodynamic (PK/PD) modelling was performed using rodent brain exposure, toxicity, and PK data and existing literature on adult PK and α2 effects to determine the safe dosages for an ongoing open-label phase 1/2 clinical trial for children with VWM.
Objectives: The aim of this study was to assess the adult extrapolated PK/PD model in VWM children receiving guanabenz orally. Subsequently, to develop a population PK-model using the prospective pediatric data. Additionally, to perform covariate analyses to explain the variability in the PK-profiles and the association between exposure and safety and tolerability.
Methods: The pediatric PK was predicted with a one-compartment model with first order absorption and linear clearance, using extrapolated adult PK parameters from literature through allometric scaling: clearance (CL) = 770.0 L/h * (WT/70)0.75; volume (V) = 4970.0 L * (WT/70)1; rate of absorption (Ka) = 1.30 h-1. This model determined the minimum, target, and maximum dose to be 1, 2, and 10 mg/kg/day respectively. Exceeding the maximum dose was expected to cause severe cardiovascular effects. Initial oral doses up to 0.5 mg/kg/day were given, followed by dose titration every three days until the maximum tolerated dose was achieved. Plasma PK-samples were taken during the first dose and annual visits. Blood pressure, heart rate, temperature, and adverse events were monitored during the trial. Population PK modelling was performed on the analyzed samples in NONMEM® after including intra-individual variability (IIV) on CL, V, and Ka. Stepwise covariate modelling (SCM) was conducted to find predictors of PK variability. The models were evaluated with bootstrap analysis (n=2000), Akaike information criterion (AIC), and visual predictive checks (VPC).
Results: In 2023, plasma PK samples were determined in 23 participants (aged 2.4-12.9 years old) at onset with doses of 0.15-0.49 mg/kg/day and 6 at year one. Current doses range from 1.02-1.70 mg/kg/day, not reaching the target dose due to non-cardiovascular side-effects. Observed plasma concentrations were best described by a one-compartment model with first order absorption and linear CL. The extrapolated PK-model underpredicted these concentrations with a median prediction error of -50.3%. Re-estimation decreased CL by 51.9% and V by 35.0%. Girls had a clinically relevant 38.8% (95%CI: 8.1-77.6%) higher CL than boys (ΔAIC: -3.4). SCM analysis confirmed the relationship between bodyweight and CL, and V with exponents of 1.00 (95%CI: 0.70-1.38) and 1.24 (CI95%: 0.91-1.62) respectively. The extrapolated versus final bootstrapped SCM model population parameters were: CL 770.0 L/h vs 448.7 L/h (RSE 26.9%, IIV 30.7%); V 4970.0 L vs 4343.0 L (RSE 26.2%, IIV 25.9%); Ka 1.30 h-1 vs 0.91 h-1 (RSE 16.1%, IIV 60.2%). VPCs of the SCM model corresponded better to observations than the re-estimated extrapolated PK-model.
Unexpectedly, only transient decreases in blood pressure, heart rate, and temperature were observed shortly after the predicted Tmax. These were described using a non-linear concentration response association with a time-dependent decrease in Emax. Common side-effects were nausea, dry mouth, tiredness, and disturbed sleep. Less frequent yet notable side-effects were hallucinations, which occurred predominantly during titration in younger patients and did not relate to any PK-parameters, dose, or exposure in this preliminary dataset.
Conclusions: Extrapolation of adult guanabenz PKPD was able to effectively predict safe starting dosages for a pediatric population, with an overprediction within 2-fold of the observed pediatric CL. Its allometric exponents were adequate, within the 95%CI of the final SCM model. The cardiovascular effects in children were overpredicted; however, other side-effects limited the dose below the target, yet above the minimum. Tolerance developed for the α2 effects at these doses; hallucinations justified a slower titration in younger patients.
 Holmes, B., Brogden, R. N., Heel, R. C., Speight, T. M., & Avery, G. S. (1983). Guanabenz a review of its pharmacodynamic properties and therapeutic efficacy in hypertension. Drugs, 26(3), 212–229. https://doi.org/10.2165/00003495-198326030-00003
 Witkamp, D., Oudejans, E., Hu‐A‐Ng, G. V., Hoogterp, L., Krzywańska, A. M., Änidaršič, M., Marinus, K., de Veij Mestdagh, C. F., Bartelink, I., Bugiani, M., van der Knaap, M. S., & Abbink, T. E. (2022). Guanabenz ameliorates disease in vanishing white matter mice in contrast to sephin1. Annals of Clinical and Translational Neurology, 9(8), 1147–1162. https://doi.org/10.1002/acn3.51611
 Lasseter, K. C., Shapse, D., Pascucci, V. L., & Chiang, S. T. (1984). Pharmacokinetics of guanabenz in patients with impaired liver function. Journal of Cardiovascular Pharmacology, 6. https://doi.org/10.1097/00005344-198400065-00008