Optimal design of a reduced sampling schedule for the characterization of prolactin dynamics in healthy male volunteers and its evaluation to the describe the pharmacodynamic effect of a prolactin antagonist in a single rising dose study
Anders Kristoffersson(1) Karla Allebrandt(2)
(1) Pharmetheus AB, Uppsala, Sweden (2) Boehringer Ingelheim Pharma GmbH & Co. KG, Biberach, Germany
Objectives: Due to its involvement in dopaminergic pathways prolactin can be used as an endogenous biomarker for brain target engagement of certain antipsychotics, allowing proof of pharmacological principle in early clinical studies. The high inter-individual variability (IIV) and diurnal cycle of prolactin must be considered in the study design to estimate a drug effect. In this work we optimize a constrained sampling design to estimate the diurnal cycle of prolactin in the absence of drug in healthy male subjects and evaluate the design performance to describe the pharmacodynamic (PD) effect of a prolactin antagonist in a single rising dose (SRD) study.
Methods: A prolactin antagonist model of a somatostatin-dopamine chimera in healthy male volunteers after single or multiple dose administration, and with extensive sampling, (Esdonk model) [1] was implemented using mrgsolve [2] and the PopED [3,4] optimal design package. The Esdonk model features a pool model with two cosine functions of 12h and 24h periods respectively to describe the diurnal pattern of prolactin. In the absence of drug, the model has been verified to result in similar description of prolactin levels over a day as circadian agonist-antagonist interaction models for remoxipride, and risperidone and paliperidone [5]. A sampling schedule optimization of four samples between 08:00 and 20:00 (ie no night sampling) was performed with the aim of characterizing the prolactin diurnal pattern in the absence of drug (representing a screening visit, day -1) with only the prolactin formation rate, the phase shifts and amplitudes of the 12 and 24h cosine functions, the residual variability and the associated IIV variances as unfixed parameters. The optimized sampling times were then evaluated on a hypothetical SRD study where the same sampling pattern was repeated at day 1 (drug administration assumed at 08:00) in addition to a morning sample at screening, day 1, 2 and 6. In the evaluation all parameters in the Esdonk prolactin model were unfixed, 10 active dose groups were assumed in addition to placebo with 6 subjects per dose group, and maximum suppression of prolactin release was observed within the dose range.
Results: With the addition of two fixed morning samples at 08:00 and the same time the following day, a schedule with four samples at 10:30, 14:00, 18:00 and 20:00 was found to yield acceptable precision in the prolactin diurnal model (highest RSE predicted for the amplitude of the 24h cosine at less than 42% for typical and 45% for IIV variance. When the sampling design was repeated on the hypothetical SRD study, all parameters in the Esdonk prolactin model were predicted to be estimable (highest RSE again predicted for the amplitude of the 24h cosine at 36% for both typical and IIV variance). However, the actual performance of the design is highly dependent on the PK/PD properties of the study drug. The relatively poor precision in the amplitude of the 24h cosine function is not surprising given the limitation of no sampling during half of the time interval but could be improved by the addition of a single night sample at the screening day at the cost of convenience.
Conclusions: We suggest a practical sampling design for capturing prolactin diurnal dynamics without nighttime sampling. The same sampling design when applied in an SRD study allows estimation of all system and PD parameters of a prolactin antagonist, demonstrating the practicability of the design for prolactin biomarker analysis in early clinical studies.
References:
[1] van Esdonk, M. J., Burggraaf, J., Dehez, M., van der Graaf, P. H., & Stevens, J. (2020). Quantification of the endogenous growth hormone and prolactin lowering effects of a somatostatin-dopamine chimera using population PK/PD modeling. Journal of pharmacokinetics and pharmacodynamics, 47, 229-239
[2] Baron K (2023). mrgsolve: Simulate from ODE-Based Models. R package version 1.3.0, https://github.com/metrumresearchgroup/mrgsolve.
[3] Nyberg J, Ueckert S, Stroemberg EA, Hennig S, Karlsson MO, Hooker AC (2012). “PopED: An extended, parallelized, nonlinear mixed effects models optimal design tool.” Computer Methods and Programs in Biomedicine, 108.
[4] Foracchia M, Hooker AC, Vicini P, Ruggeri A (2004). “POPED, a software for optimal experiment design in population kinetics.” Computer Methods and Programs in Biomedicine, 74.
[5] Ma, G., Friberg, L. E., Movin‐Osswald, G., & Karlsson, M. O. (2010). Comparison of the agonist‐antagonist interaction model and the pool model for the effect of remoxipride on prolactin. British journal of clinical pharmacology, 70(6), 815-824.
