Orwa Albitar, Sylvie Retout, Bastian Zinnhardt, Nima Shariati, Krisztina Kajzinger, Kezia Kobana, Gabriel Schnetzler, Gennaro Pagano, Silke Simon, Benedicte Ricci
1 Roche Pharma Research and Early Development, Roche Innovation Center (Basel, Switzerland)
Objectives:
The release of pro-inflammatory cytokines, including interleukin (IL)-18 and IL-1β, is potentially driven by the Nucleotide-binding oligomerization domain, leucine-rich repeat and pyrin domain containing 3 (NLRP3) inflammasome [1]. Therefore, NLRP3 inhibition could be highly relevant for both peripheral inflammatory [2,3] and central neurodegenerative diseases, where microglial activation is a key component of the pathophysiology [4]. Selnoflast (RG6418) is an orally active, selective and reversible small molecule NLRP3 inhibitor. The current study aimed to characterize the PKPD profile of selnoflast in the Parkinson’s disease (PD) population and assess the potential dose selection targeting NLRP3 inhibition.
Methods:
Data: Rich PK data obtained from the entry-into-human healthy volunteer (HV) study, including single ascending dose, food effect, multiple ascending dose and relative bioavailability parts, as well as Ph1b studies, including ulcerative colitis (UC) and PD participants, were used to establish the PopPK model. Peripheral target engagement data, ex-vivo stimulated IL-1β release inhibition in whole blood and downstream plasma biomarker effects, high-sensitivity C-reactive protein (hsCRP), in PD participants were used to establish the PKPD models.
Base models: PopPK models, including one or two-compartments, with first-order absorption, an absorption lag time or a transit compartment model with and without a depot, were tested. hsCRP data were log transformed. Direct, effect compartment and indirect PKPD models with linear, loglinear, and Emax were tested. Nonlinear mixed-effect modeling NONMEM 7.4.3 software was used for all models [5].
Covariate models and evaluations: Influences of scientifically plausible covariates were investigated, e.g., food, age, and creatinine clearance. Prediction corrected visual predictive checks (pVPCs) were produced to check the predictive validity and simulation properties of the models. Sampling-importance resamplings (SIRs) were performed to assess the uncertainty of models’ parameters [6,7].
Dose simulation: a shiny app with mrgsolve simulation based on the final models was developed and used to simulate doses with clinically relevant PK exposure, and related peripheral target engagement and effect on downstream biomarker.
Results:
Data: 1952 PK observations were pooled from 52 HV, receiving Selnoflast as a single dose (20-600 mg) and multiple doses (180-450 mg once daily (OD)) for 7 days, as well as 13 UC, and 39 PD participants receiving Selnoflast 200 twice daily (BID) for four weeks. Ex-vivo stimulated IL-1β release inhibition (188 observations), and hsCRP data (294 observations) were obtained in PD participants pre-dose and 2 h post dose on days 1, 15 and 28.
Base models: A one-compartment model with 1st order absorption and elimination and a reduced transit absorption model with six compartments was sufficient to describe the PK with interindividual variability (IIV) on clearance, volume of distribution and rate of transit absorption. Interoccasion variability was also introduced on the rate of transit absorption. A combined residual error model was used. Bodyweight effect was prespecified using allometric scaling [8]. A direct PKPD emax model best described ex-vivo stimulated IL-1β inhibition with IIVs introduced on the concentration achieving 50% of the maximal effect (EC50) and baseline; an indirect effect emax model best described the hsCRP reduction with IIV full block introduced on EC50 and baseline, and a half life of the effect estimated to 30 h.
Covariate models and evaluations: The rate of transit absorption was 66% lower in the fed state. Clearance decreased with age. For example, clearance was 13% lower in 65 vs 40 years old participants. All models showed good predictive performance and parameters with relative standard errors of less than 35%.
Dose simulation: a range of dosing regimens was simulated, accounting for plasma PK exposure, peripheral target inhibition (based on ex vivo IL1b) and downstream biomarker, e.g., 200 mg BID was associated with a median of 90% target inhibition maintained at C-trough and an unadjusted 45% decrease in hsCRP after four weeks of treatment.
Conclusions:
The developed PopPK and PKPD modeling framework for the NLRP3 inhibitor selnoflast successfully integrated PK and PD biomarker data across different clinical studies and populations, with the aim of predicting clinically relevant levels of target engagement and downstream biomarker reduction for future clinical studies with NLRP3 inhibitors.
References:
References:
[1] Coll et al. Nature Reviews Immunology. 2025
[2] Klughammer et al. Clinical and Translational Medicine. 2023
[3] Klughammer et al. Journal of Crohn’s and Colitis. 2023
[4] D’Urso et al. Neurology. 2023
[5] Beal et al. NONMEM Users Guides. 2008
[6] Bergstrand et al. AAPS Journal. 2011
[7] Dosne et al. Journal of Pharmacokinetics and Pharmacodynamics. 2016
[8] González-Sales et al. Journal of Pharmacokinetics and Pharmacodynamics. 2022
Reference: PAGE 34 (2026) Abstr 12186 [www.page-meeting.org/?abstract=12186]
Poster: Drug/Disease Modelling - CNS