Roberto Visentin 1, Chiara Roversi 1, Simone Zannoni 1, Davide Ronchi 1, Johannes Fröhlich 2, Jonas Steinhauser 2, Alessia Tagliavini 1
1 Evotec SE (Verona, Italy), 2 Akroswiss AG (Zurich, Switzerland)
Objectives
Midazolam is a short-acting benzodiazepine widely used for indications such as conscious sedation and treatment of epileptic seizures [1]. Several routes of administration are available, including intravenous (IV) and intranasal (IN) ones. IN midazolam is particularly attractive in paediatric settings, as it enables rapid drug absorption, avoids invasive administration, and can benefit from user-friendly nasal spray devices. However, dose selection for IN midazolam in children remains challenging due to substantial pharmacokinetic (PK) variability associated with age-related physiological changes and differences in body size [2].
Population pharmacokinetic (popPK) modelling provides a quantitative framework to characterize drug PK variability and support dose optimization.
The aim of this work was to develop a popPK model of two nasal-spray midazolam formulations (MDZ-1 and MDZ-2.5) using nonlinear mixed-effects (NLME) modelling based on adult PK data, to translate it to paediatric subjects using allometric scaling, and subsequently evaluate its predictive performance against independent paediatric PK data reported in the literature.
Methods
The dataset consisted of plasma concentration-time profiles for midazolam and its primary 1-OH metabolite from 28 healthy volunteers (16 males; age 32±10 years; body weight (BW) 68±12 kg) undergoing five PK sessions in a Latin square crossover design (EudraCT-Number 2022-002154-16). In each session, subjects received a single dose of one of five treatments – IV midazolam (2.5 mg), IN MDZ-1 (1 mg, 2 mg, or 4 mg), or IN MDZ-2.5 (2.5 mg) – and blood samples were collected for PK assessment up to 12 hours post-dose. Each PK session was separated by a minimum washout period of 3 days.
PopPK modelling was performed using NLME implemented in MonolixSuite™. Structural models evaluated parent-metabolite transformation, number of compartments, type of absorption for IN formulation, and elimination processes. Allometric scaling was applied to clearance (CL) and volume (V) parameters using BW as a covariate with fixed exponents (CL ~ BW⁰’⁷⁵; V ~ BW¹) [3]. Age and sex were evaluated as possible additional covariates. Inter-individual and inter-occasion variability was assumed to follow log-normal distributions. Residual variability was described using appropriate error models. Model selection was guided by parameter relative standard error (RSE), goodness of fit (GoF) and corrected Bayesian Information Criterion (BICc).
The final popPK model was then scaled to paediatric population using a refined allometric scaling of CL that accounts for organ size and maturation [4].
PopPK model ability to well describe midazolam PK in the paediatric population was evaluated by simulating clinical experiments with both IV and IN midazolam in the target population (available from literature) and comparing the resulting PK profiles with observed data.
Results
The final popPK model included first-order absorption for both MDZ-1 and MDZ-2.5 formulations, linear elimination, a two-compartment structure for parent compound and one compartment for 1-OH metabolite, unidirectional parent-to-metabolite transformation, and a proportional residual error model. BW was the only covariate retained across all clearance and volume parameters. The model adequately captured observed adult PK data, with satisfactory GoF and accurate estimated parameters (RSE < 42%).
The refined allometric scaling on CL assumed a BW-dependent exponent model identified on both children and adult literature data, and improved characterization of the BW-CL relationship in the paediatric population.
Simulated paediatric PK profiles following both IV and IN midazolam administration were consistent with literature data, as the simulated 90% prediction interval successfully captured most of the observed PK concentrations across 15 independent studies.
Conclusions
A parent-metabolite popPK model describing two intranasal midazolam formulations was successfully developed using NLME and translated to the paediatric population using maturation-informed allometric scaling.
This modelling approach demonstrated robust predictive performance for midazolam PK following both IV and IN administration and represents a quantitative framework to explore alternative dosing regimens.
Its application may support model-informed dose optimization of IN midazolam for children in clinical practice.
References:
[1] Nordt SP, Clark RF. Midazolam: a review of therapeutic uses and toxicity. J Emerg Med. 1997 May-Jun;15(3):357-65.
[2] Pacifici GM. Clinical pharmacology of midazolam in neonates and children: effect of disease-a review. Int J Pediatr. 2014;2014:309342.
[3] Kleiber M, Body size and metabolism. Hilgardia. 1932;6(11):315–53.
[4] Germovsek E, Barker CI, Sharland M, Standing JF. Scaling clearance in paediatric pharmacokinetics: All models are wrong, which are useful? Br J Clin Pharmacol. 2017 Apr;83(4):777-790.
Reference: PAGE 34 (2026) Abstr 12070 [www.page-meeting.org/?abstract=12070]
Poster: Drug/Disease Modelling - Other Topics