Itziar Irurzun-Arana 1, Chronis Kemos 2, Marc Goldfinger 3, MeeJ Kim 4, Tom Webb 2, Brandon Swift 5
1 GSK (Madrid, Spain), 2 GSK (Stevenage, UK), 3 GSK (London, UK), 4 GSK ( Rockville, , USA), 5 GSK (Durham, USA)
Introduction:
Loss-of-function (LOF) gene mutations in hydroxysteroid 17-beta-dehydrogenase-13 (HSD17B13) are associated with reduced risk of steatotic liver disease including metabolic dysfunction-associated steatohepatitis [MASH] [1]. GSK4532990 is an N-acetylgalactosamine (GalNAc) conjugated small interfering RNA (siRNA) which confers specificity to hepatocytes via the asialoglycoprotein (ASGPR) receptor and selectively targets HSD17B13 in hepatocytes to mimic the LOF mutation.
SKYLINE Phase 2a (NCT06104319) study investigated hepatic target engagement of single subcutaneous doses of GSK4532990 in patients with confirmed or suspected MASH. To this end, paired liver biopsies were collected pre- and post-dose (8, 12 or 28 weeks) after a single dose of 25, 200, 400, or 800 mg.
Objectives:
This analysis aimed to characterize the relationship between GSK4532990 dose, plasma exposure and reductions in HSD17B13 mRNA and protein over time as the primary endpoint from SKYLINE study, which consisted of a PK-PD model-based prediction of the percent change from baseline in HSD17B13 mRNA and protein levels.
Methods:
Due to the known importance of the ASGPR-mediated liver uptake for GalNAc-siRNAs, a target mediated drug disposition PK-PD model was developed to characterize GSK4532990 plasma concentrations over time and the plasma-to-liver siRNA transport [2]. Indirect response models were used to describe mRNA and protein data over time, where drug effects decreased the mRNA synthesis rate, which in turn decreased protein translation.
PK-PD model development was a sequential two-step process where the individual empirical Bayes estimates of PK parameters were fixed before estimating the PD parameters from the biopsy data. Both mRNA and protein data were modelled simultaneously, and mRNA measurements were transformed to relative change from baseline values. Model development was conducted via nonlinear mixed-effects modeling. Covariate analysis was performed by conventional stepwise forward and backward selection criteria. Models were evaluated with standard goodness of fit criteria.
Results:
The primary objective of the study was met with a successfully developed population PK-PD model. Of the 59 participants included in the population PK model, 48 were used for PK-PD model development, from which 45 and 40 had evaluable post-dose mRNA and protein observations, respectively.
Plasma PK was described by a one-compartment disposition model with two parallel first-order absorption mechanisms to the central compartment. Elimination from the central compartment occurred through a linear plasma clearance or nonlinear binding to ASGPR. The formed GSK4532990-ASGPR complex was internalized into a virtual liver (hepatocyte) compartment, where the drug cleaved from ASGPR and slowly degraded within the endolysosomal space. Liver siRNA inhibited the HSD17B13 mRNA synthesis rate though an Imax model, which consequently reduced HSD17B13 protein expression.
Model predictions demonstrated consistency with the observed data across the evaluated dose range, and it properly characterized the more than proportional increase in systemic exposure observed as GSK4532990 dose increased, which is indicative of transient ASGPR saturation.
The population PK-PD model predicted dose-dependent reductions in mRNA and protein levels in the liver after a single dose administration of GSK4532990, with only small differences for 200 mg, 400 mg and 800 mg doses at Weeks 8 and 12 due to near maximum mRNA and protein knockdown (mean reduction >80%) achieved by the 200 mg dose. At Week 28, the model predicted a reduced mRNA and protein knockdown compared to Weeks 8 and 12, reflecting protein recovery. Nevertheless, model predictions indicated that higher doses maintained a more prolonged target knockdown with a predicted mean change from baseline of -26.82%, -61.82%, -70.02%, and -75.27% for the 25, 200, 400, and 800 mg arms, respectively, by Week 28. None of the evaluated participant characteristics significantly influenced drug effects on hepatic mRNA and protein levels.
Conclusions:
The primary endpoint of the SKYLINE study was successfully met with the development of a population PK-PD model that described GSK4532990 mediated reductions in hepatic HSD17B13 mRNA and protein. The model incorporated an ASGPR-mediated liver uptake of GalNAc-siRNAs and a sustained liver PK compartment that captured the disconnect between the short plasma half-life of GSK4532990 and the prolonged PD effect observed on both HSD17B13 mRNA and protein expression 8, 12 and 28 weeks after dosing. The results from this analysis informed direct target engagement of GSK4532990, highlighted the model’s capability to elucidate key clinical pharmacology properties of GalNAc-siRNAs and strengthened the rationale for GSK4532990 dosing in MASH or other chronic liver diseases.
References:
1- Abul-Husn NS, Cheng X, Li AH, et al. A Protein-Truncating HSD17B13 Variant and Protection from Chronic Liver Disease. N Engl J Med. 2018; 378(12):1096-106.
2- Ayyar VS, Song D. Mechanistic Pharmacokinetics and Pharmacodynamics of GalNAc-siRNA: Translational Model Involving Competitive Receptor-Mediated Disposition and RISC-Dependent Gene Silencing Applied to Givosiran. J Pharm Sci. 2024;113(1):176-190.
Reference: PAGE 34 (2026) Abstr 11890 [www.page-meeting.org/?abstract=11890]
Poster: Drug/Disease Modelling - Other Topics