Physiologically-based pharmacokinetic/pharmacodynamic modeling to predict tumor growth inhibition and the efficacious dose of selective estrogen receptor degraders in humans.

GDC-9545 (giredestrant) is a highly potent, nonsteroidal, oral selective estrogen receptor antagonist and degrader that is being developed as a best-in-class drug candidate for early-stage and advanced drug-resistant breast cancer. GDC-9545 was designed to improve the poor absorption and metabolism of its predecessor GDC-0927, for which development was halted due to a high pill burden. This study aimed to develop physiologically-based pharmacokinetic/pharmacodynamic (PBPK-PD) models to characterize the relationships between oral exposure of GDC-9545 and GDC-0927 and tumor regression in HCI-013 tumor-bearing mice, and to translate these PK-PD relationships to a projected human efficacious dose by integrating clinical PK data. PBPK and Simeoni tumor growth inhibition (TGI) models were developed using the animal and human Simcyp V20 Simulator (Certara) and adequately described each compound's systemic drug concentrations and antitumor activity in the dose-ranging xenograft experiments in mice. The established PK-PD relationship was translated to a human efficacious dose by substituting mouse PK for human PK. PBPK input values for human clearance were predicted using allometry and in vitro in vivo extrapolation approaches and human volume of distribution was predicted from simple allometry or tissue composition equations. The integrated human PBPK-PD model was used to simulate TGI at clinically relevant doses. Translating the murine PBPK-PD relationship to a human efficacious dose projected a much lower efficacious dose for GDC-9545 than GDC-0927. Additional sensitivity analysis of key parameters in the PK-PD model demonstrated that the lower efficacious dose of GDC-9545 is a result of improvements in clearance and absorption. The presented PBPK-PD methodology can be applied to support lead optimization and clinical development of many drug candidates in discovery or early development programs.

Preclinical optimization of Ly6E-targeted ADCs for increased durability and efficacy of anti-tumor response.

Early success with brentuximab vedotin in treating classical Hodgkin lymphoma spurred an influx of at least 20 monomethyl auristatin E (MMAE) antibody-drug conjugates (ADCs) into clinical trials. While three MMAE-ADCs have been approved, most of these conjugates are no longer being investigated in clinical trials. Some auristatin conjugates show limited or no efficacy at tolerated doses, but even for drugs driving initial remissions, tumor regrowth and metastasis often rapidly occur. Here we describe the development of second-generation therapeutic ADCs targeting Lymphocyte antigen 6E (Ly6E) where the tubulin polymerization inhibitor MMAE (Compound 1) is replaced with DNA-damaging agents intended to drive increased durability of response. Comparison of a seco-cyclopropyl benzoindol-4-one (CBI)-dimer (compound 2) to MMAE showed increased potency, activity across more cell lines, and resistance to efflux by P-glycoprotein, a drug transporter commonly upregulated in tumors. Both anti-Ly6E-CBI and -MMAE conjugates drove single-dose efficacy in xenograft and patient-derived xenograft models, but seco-CBI-dimer conjugates showed reduced tumor outgrowth following multiple weeks of treatment, suggesting that they are less susceptible to developing resistance. In parallel, we explored approaches to optimize the targeting antibody. In contrast to immunization with recombinant Ly6E or Ly6E DNA, immunization with virus-like particles generated a high-affinity anti-Ly6E antibody. Conjugates to this antibody improve efficacy versus a previous clinical candidate both in vitro and in vivo with multiple cytotoxics. Conjugation of compound 2 to the second-generation antibody results in a substantially improved ADC with promising preclinical efficacy.