Applying a quantitative system pharmacology (QSP) model to inform KRASG12C covalent inhibitors’ PK driver level needed for efficacy in patients with non-small cell lung cancer (NSCLC)
Siak-Leng, Choi, Delphine Valente, Valerie Czepczor, Florence Fassy, Loreley Calvet, Isabelle Meaux, Christine Mauriac
Sanofi DMPK France, Chilly-Mazarin
Introduction: KRAS mutation occur in 30% of all human cancers [1], however, this target is known to be undruggable until recently with approvals of KRASG12C covalent inhibitors sotorasib [2] and adagrasib [3] for NSCLC indication. In patients, occurrence of KRAS nucleotide cycling equilibrium between unbound KRASG12C, KRASG12C bound with GDP and KRASG12C bound with GTP forms was found [1]. Among them, KRASG12C with GTP form, elicits oncogenic signals [4]. As a treatment, the KRASG12C inhibitors bind covalently with mutated GDP-bound form of KRASG12C at switch II pocket (SIIP) for inhibition to disrupt the KRAS oncogene equilibrium state and thereby its KRASG12C signaling pathway [1]. We developed a QSP model describing the covalent binding kinetics of KRASG12C inhibitors in tumor, tumor KRASG12C target occupancy rate (TOR), phosphorylated extracellular-signal-regulated kinase (pERK) downstream effect, and tumor growth inhibition (TGI) in H358 tumor-bearing mice. The model informed the physchem properties of a KRAS covalent inhibitor to be potent, the PK driver for efficacy, and the level of PK driver needed for efficacy of preclinical candidates in patients with NSCLC.
Objectives:
- Develop a QSP model of KRASG12C using in vitro data and in vivo PK/PD/efficacy data in H358 tumor-bearing mice
- Validate the model prediction with the reported minimum effective plasma exposure from adagrasib
- Estimate the level of PK driver of KRASG12C inhibitors needed for efficacy of the preclinical candidates in patients with NSCLC
Methods: Serial sampling of plasma and tumor PK, tumor TOR, pERK/ERK, and TGI (Tumor growth Inhibition) in H358 tumor-bearing mice model following daily dosing (30, 100, 150 mg) of KRASG12C inhibitors (adagrasib, and two preclinical candidates [Compounds A and B]) for either 3 or 10 days. The model development was performed using Stochastic Approximation Expectation Maximization (SAEM) in Monolix version 2020 (Lixoft®). The simulations were performed using R-based Monolix Suite Simulx version 2020 (Lixoft®).
Results: The model constituted of a one-compartmental model for KRASG12C inhibitors plasma PK with a tumor penetration model [5] adapted from Wittrup et al. [6, 7, 8]. In the tumor, the nucleotide cycling model [1] was adopted and free KRASG12C inhibitors assumed to engage covalently with KRASG12C GDP form, depicted by a covalent binding model which consists of initial reversible binding followed by formation of covalent bond [9]. The key binding parameters, reversible binding equilibrium (Ki) and first-order constant of the chemical step (Kinact) were fixed with estimations from in vitro assays. In the model, different forms of KRASG12C were used to calculate TOR that links with downstream signal of pERK/ERK using a Emax inhibition model and followed by the link with TGI model [10] via a power function. The model successfully fitted PK/PD/efficacy data of adagrasib, compounds A, and B for all the tested doses in mice. The model predicted average total plasma concentration (Cave) as the PK driver. Moreover, we predicted a consistent adagrasib total Cave needed for efficacy as compared to reported value (predicted 1.4 µM vs. 1.6 µM [11]). Our precandidates, compounds A and B were predicted requiring 3x and 6x higher total plasma Cave for efficacy than adagrasib, despite similar average free tumor concentrations. This suggested that a candidate with a higher fraction unbound in tumor would be favorable for its potency, which is warranted to be confirmed.
Conclusion: The model development in mice for KRASG12C covalent inhibitors was well validated to predict exposure needed in patients using clinically validated molecule adagrasib. It informed the level of PK driver needed for efficacy and thus for active dose projection of preclinical candidates in patients with NSCLC.
References:
[1] Patricelli MP et al. Cancer Discov. 2016 Mar;6(3):316-29.
[2] https://www.fda.gov/drugs/resources-information-approved-drugs/fda-grants-accelerated-approval-sotorasib-kras-g12c-mutated-nsclc
[3] https://www.fda.gov/drugs/resources-information-approved-drugs/fda-grants-accelerated-approval-adagrasib-kras-g12c-mutated-nsclc
[4] Ostrem JM et al. Nature 2013 ; 503 : 548 – 51 .
[5] Aman PS and Dhaval KS. AAPS J. 2017 Jul;19(4):1054-1070.
[6] Thurber GM, Schmidt MM, Wittrup KD. Adv Drug Deliv Rev. 2008; 60(12):1421–34.
[7] Thurber GM, Schmidt MM, Wittrup KD. Trends Pharmacol Sci. 2008; 29(2):57–61.
[8] Schmidt MM, Wittrup KD. Mol Cancer Ther. 2009; 8(10):2861–71.
[9] Hansen R et al. Nat Struct Mol Biol. 2018 Jun;25(6):454-462.
[10] Simeoni M et al. Cancer Res. 2004 Feb 1;64(3):1094-101.
[11] Hallin J et al AACR-NCI-EORTC International Conference on Molecular Targets (2019) Poster CO69
