Michael Williams 1, Sara Asimus 1, Kristin M D'Silva 2, Eduard Molins 3, Andre Santa Maria 4, Joachim Almquist 1
1 Clinical Pharmacology & Quantitative Pharmacology, BioPharmaceuticals R&D, AstraZeneca (Gothenburg, Sweden), 2 BioPharmaceuticals R&D, Respiratory and Immunology, AstraZeneca (Boston, United States of America), 3 Late Respiratory & Immunology, BioPharmaceuticals R&D (Barcelona, Spain), 4 Clinical Operations, Respiratory & Immunology, BioPharmaceuticals R&D, AstraZeneca (Barcelona, Spain)
Objectives
AZD6912 is a GalNAc-conjugated double-stranded small interfering RNA (siRNA) [1] which has been administered in a first-in-human dose escalation study, while being considered for the treatment of rheumatoid arthritis (NCT06115967). It targets expression of Complement Factor B (CFB) in the liver to effectively suppress the downstream complement alternative pathway (CAP) with a long duration of action, enabling infrequent dosing.
We describe a semi-mechanistic K-PD model [2] which links the unobserved AZD6912 liver amount to CAP activity, mediated through CFB. The model first links unobserved AZD6912 liver amounts to CFB suppression with a sigmoidal Emax expression, then links CFB levels to CAP activity suppression through another sigmoidal Emax expression.
The objective of this analysis is to inform clinical development by describing the dose-response of CAP activity inhibition to AZD6912. There was a particular emphasis on the level and duration of CAP activity inhibition, which are important signals of both target engagement and safety. Specifically, 95% CAP activity inhibition was the prespecified efficacy target and inhibition below 50% was considered outside of healthy range, necessitating prophylactic antibiotic use in healthy volunteers.
Methods
The model was based on clinical data from a Phase 1 randomized, placebo controlled, double-blind multicenter single ascending dose study conducted in healthy volunteers. The trial included a single subcutaneous-administered dose and variable follow-up time based on target engagement recovery. Sampling includes assays for CFB and CAP activity expression in plasma. PK data was disregarded in favor of a K-PD model due to the uncertain relationship between AZD6912 plasma concentrations and activated siRNA in the liver. This methodology allows for efficient use of relevant biomarker data to predict efficacy time-course of possible additional cohorts in the phase I program.
A K-PD model was estimated using NONMEM version 7.4.4. Predictions were derived for higher dose levels during the phase I program to assess potential target engagement benefits as well as operational and safety risks.
Results
A population K-PD model was used to describe the time-dependent relationship between AZD6912 subcutaneous dosing, CFB concentrations, and CAP activity. The half-life of AZD6912 in the liver was estimated to be 153 days [95% CI: 114 – 206].
A sigmoidal Emax model described the inhibition of CFB based on AZD6912 liver amount; the amount of AZD6912 in the liver associated with 50% of maximal inhibition of CFB synthesis was 27.6 mg [95% CI: 18.5 – 41.1, IIV CV%: 103] with low sigmoidicity (0.987) [95% CI: 0.762 – 1.26]. The half-life of CFB was estimated to be 11.1 days [95% CI: 9.34 – 13.2, IIV CV%: 31]. A further Emax model described the direct relationship between CFB expression and CAP activity; the CFB concentration producing 50% of maximal inhibition of CAP activity was 43.9 ug/mL [95% CI: 38.6 – 49.7, IIV CV%: 21] with higher sigmoidicity (2.42) [95% CI: 2.04 – 2.98].
Conclusions
The implementation of the K-PD model with CFB serving as an intermediary between liver exposure and CAP Activity adequately described the CAP activity in the phase 1 cohorts. Predictions are being used to support key decisions during progress of the Phase I program based on the predicted extent and duration of CAP Activity inhibition in healthy volunteers.
The analysis provides insight into the relationship between CFB inhibition and downstream inhibition of CAP activity, which can inform the development of other compounds targeting the same pathway. This modeling approach can be considered in clinical studies where both the mechanistic target and the downstream effective target are assayed in the study.
References:
[1] Springer AD, Dowdy SF. GalNAc-siRNA Conjugates: Leading the Way for Delivery of RNAi Therapeutics. Nucleic Acid Ther. 2018;28(3):109-118. doi:10.1089/nat.2018.0736
[2] Jacqmin P, Snoeck E, van Schaick EA, et al. Modelling Response Time Profiles in the Absence of Drug Concentrations: Definition and Performance Evaluation of the K–PD Model. J Pharmacokinet Pharmacodyn. 2007;34(1):57-85. doi:10.1007/s10928-006-9035-z
Reference: PAGE 34 (2026) Abstr 11926 [www.page-meeting.org/?abstract=11926]
Poster: Drug/Disease Modelling - Other Topics