Mélanie Wilbaux (1), Heller Chen (2), Mirek Dostalek (2), Aiyang Tao (2), Fang Xiang (2), Nancy Lewis (2), Christophe Meille (1)
(1) Novartis, Switzerland; (2) Novartis, US
Introduction:
MIW815 (ADU-S100) is a novel immuno-oncology cyclic dinucleotide compound targeting STING (Stimulator of Interferon Genes). It is currently being evaluated in a Phase I first-in-human single-agent dose-escalation study and in combination with PDR001, a checkpoint inhibitor. MIW815 is administered via intra-tumoral (IT) injection and only plasma concentrations are determined.
Objective:
The objective is to combine pre-clinical tumor concentration data and clinical data through a similar structural model to simulate intra-tumoral PK in human and to enable future support of dose and schedule selection of MIW815.
Methods:
Plasma concentrations (n=740) were available from 47 patients with solid tumors and lymphomas enrolled in the single-agent study. They received IT injection of MIW815 with doses ranging from 50 µg to 6400 µg, weekly administered for 3 weeks followed by one week-off. Full PK profiles up to 4 hours after injected dose were collected at cycle 1 day 1, cycle 1 day 15 and cycle 3 day 1.
Pre-clinical intra-tumoral concentrations (n=32) and plasma concentrations (n=64) up to 5 hours were available in 37 syngeneic mice after a single administration of MIW815 via: (i) IT injection of 20 mg/kg, (ii) sub-cutaneous (SC) administration of 20 mg/kg, or (iii) intra-venous (IV) administration of 1 mg/kg. A multilevel pre-clinical semi-mechanistic PK model was established to describe all these data and to characterize drug distribution between tumor and plasma. Several model structures were probed to describe the intra-tumoral distribution and diffusion to plasma.
The established pre-clinical PK model structure was applied by estimating parameters based on the clinical plasma PK data and used to simulate the intra-tumoral time course of MIW815.
A minimal empirical PK model was also developed in human to describe plasma concentrations and to estimate key PK parameters.
Both pre-clinical and clinical data analyses were developed using non-linear mixed-effects modeling implemented in Monolix 2018R1. Model selection and evaluation were based on statistical criteria, goodness-of-fit plots and simulations-based diagnostics.
Results:
The selected pre-clinical model combined two main components: tumor and systemic circulation. IT injection of MIW815 was described by a transfer from a depot compartment to tumor tissue and a leakage to plasma with estimation of the fraction of drug amount distributed to tissue. Intra-tumoral concentration kinetics were best described by a two-compartment structure, while plasma PK were described by one-compartment model with linear elimination. An additional tumor-plasma distribution was estimated to describe observed intra-tumoral and plasma concentrations after SC administration. The model was able to describe both plasma and intra-tumoral PK profiles after IT, SC and IV injection.
The same structural model was applied to human plasma PK data. The following assumptions were made: (i) applicability of pre-clinical experimental models (mice) to humans, (ii) similar process describing PK and tumor distribution. Observed human plasma PK data were used to re-estimate all model parameters. This model showed a good prediction of clinical plasma PK. It was also used to perform simulations of human MIW815 PK profiles in the injected tumor for different doses.
Plasma PK in human were also described by an empirical one-compartment model with linear elimination. An alpha-order absorption process was necessary to describe the data, suggesting that a complicated distribution process might be involved between tumor and plasma. Plasma half-life of MIW815 was estimated to be very short.
Conclusions:
A translational modeling approach was applied to develop a semi-mechanistic model and allowed projection of PK profiles in the injected tumor in humans by combining pre-clinical and clinical data. The future plan is to use this model to support human dosing regimen selection by simulating human intra-tumoral PK that matches pre-clinical efficacious concentrations, or that lead to biomarker changes.
Reference: PAGE 28 (2019) Abstr 9136 [www.page-meeting.org/?abstract=9136]
Poster: Drug/Disease Modelling - Oncology