Marie Lambert1, Cedric Vinson1, Taryn Sleger2, Sylvain Fouliard1
1Quantitative Pharmacology, Translational Medicine, Servier, 2Research, Servier
Objectives: The deletion of metabolic gene methylthioadenosine phosphorylase (MTAP) occurs, globally, in 15% of human cancers [1] with unmet needs. Deletion of MTAP gene with synthetic lethal inhibition of methionine adenosyl-transferase IIA (MAT2A) leads to the death of tumor cells. S095035 is a potent and selective of MTAP null-type cancer cells small molecule antagonist of the MAT2A. Ahead of S095035 entering clinical development, the objective of this study is to translate PK & PKPD and predict efficacious exposure and dose in patients. Methods: First, S095035 plasma PK data from single dose IV/PO studies across species (mouse, rat, dog, monkey) were modeled using a joint IV/PO compartmental model including absorption. Based on single-dose IV data, compartments models were tested to describe the distribution and elimination process. Single-dose PO data were added to evaluate several absorption processes (zero or first-order absorption constant with and without lag-time or transit compartments). The human distribution and elimination PK parameters were predicted using allometry principles, and several sets of assumptions were considered for absorption. With this translational PK model, simulations of human PK profiles were performed ranging from 10mg to 500mg. Then, PKPD of S095035 was based on 2 experiments with xenograft-tumor bearing mice models (pancreatic cancer KP4K and colon cancer HCT-116), treated with several S095035 dose levels ranging from 0.3 mg/kg to 20mg/kg. The tumor volumes were measured every 3-4 days on each animal (control group and treated group, n = 12 mice per groups) over the duration of treatment. On the last day of the study, PK samples were collected at 1h, 6h,12h,24h after administration. The modeling was performed using sequential approach (PK modelling, then tumor growth modelling in the absence of treatment, then PK/PD). Different tumor growth models were tested: simple, double exponential and exponential with linear growth model (e.g Simeoni 2004 [2]). A proportional and a saturable effect of S095035 plasma concentrations were tested on tumor growth. Based on final model, the efficacious steady-state exposure in mice, as measured by AUC0-24h at 90% of the maximal simulated TGI (EAUC90[0-24h]) was obtained in each study using simulations. Finally, the efficacious EAUC90[0-24h] was translated and combined with the predicted human PK profiles to determine an expected therapeutic dose. Nonlinear mixed effect modeling was performed using Monolix (version2019 R2). Allometric scaling were performed using MicrosoftExcel365 (2006 FDA guidance [3]) and Simcyp v2018R1 allometric tools. The distribution of the individual PK was assumed to be log-normal or logit normal. The evaluation of PK and PKPD models have been assess by precision of estimation (RSE), goodness-of-fit (GOF) plots, and advanced evaluation methods for final models (Normalized Prediction Distribution Error (NPDE) and Visual Predictive Check (VPC)). Results: For all species, the best joint IV/PO model was a linear 2-compartmental model with first order absorption and a proportional residual error. VPC showed that all PK models described correctly the observed data. A low clearance (0.0597 L/h/kg) and a moderate volume of distribution at steady-steate (2.39 L/kg) in humans were predicted, associated with a predicted effective half-life of 27 hours. For both xenograft-tumor bearing mice models, a linear 2-compartment PK with a saturable effect of S095035 (EC50 [RSE%] = 25.6 [7.26] and EC50 [RSE%] = 220 [43.2] ng/mL) better described the observed data. The PD model using a tumor double exponential growth rate constant was chosen based on RSE and validation plots to describe tumor growth. An EAUC90[0-24h] of 5 444 h.ng/mL and 30 888 h.ng/mL were found in the two studies. Heterogeneity in sensitivity to S095035 was accounted for when target exposure based on both xenograft models were included to predict therapeutic dose in human. After plasma protein binding correction, and across the different considered scenarios, the efficacious dose range predicted in human is approximately 20 mg to 200 mg once daily. Conclusions: It is important to provide an active dose range to support FIH study design and translational PK/PD modeling is the right tool for this assessment. In oncology, the pre-clinical PK and tumor volume data can inform on the efficacious exposure and dose in human.
[1] Kalev P, Hyer ML, Gross S, Konteatis Z, Chen CC, Fletcher M, Lein M, Aguado-Fraile E, Frank V, Barnett A, Mandley E, Goldford J, Chen Y, Sellers K, Hayes S, Lizotte K, Quang P, Tuncay Y, Clasquin M, Peters R, Weier J, Simone E, Murtie J, Liu W, Nagaraja R, Dang L, Sui Z, Biller SA, Travins J, Marks KM, Marjon K. MAT2A Inhibition Blocks the Growth of MTAP-Deleted Cancer Cells by Reducing PRMT5-Dependent mRNA Splicing and Inducing DNA Damage. Cancer Cell. 2021 Feb 8;39(2):209-224.e11. doi: 10.1016/j.ccell.2020.12.010. Epub 2021 Jan 14. PMID: 33450196. [2] Simeoni M, Magni P, Cammia C, De Nicolao G, Croci V, Pesenti E, Germani M, Poggesi I, Rocchetti M. Predictive pharmacokinetic-pharmacodynamic modeling of tumor growth kinetics in xenograft models after administration of anticancer agents. Cancer Res. 2004 Feb 1;64(3):1094-101. doi: 10.1158/0008-5472.can-03-2524. PMID: 14871843. [3] 2006 FDA guidance’s
Reference: PAGE 33 (2025) Abstr 11521 [www.page-meeting.org/?abstract=11521]
Poster: Drug/Disease Modelling - Oncology