(W-056) Population Pharmacokinetics of Quemliclustat, a Potent, Selective, and Reversible CD73 Inhibitor, in Patients with Cancer

Ian Linsmeier, PhD – Clinical Pharmacologist, Arcus Biosciences Inc.; Lixia Jin, PhD – VP, DMPK & Bioanalytics, Arcus Biosciences Inc.; Elaine Ginn, PhD – Sr. Scientist, DMPK, Arcus Biosciences Inc.; Lisa Seitz, PhD – Sr. Director, Translational, Arcus Biosciences Inc.; Ayyappa Chaturvedula, PhD – Clinical Pharmacologist, Arcus Biosciences Inc.

Objectives: Quemliclustat (Q) is a potent, selective, small-molecule inhibitor of soluble and membrane-bound CD73 developed with the aim of blocking adenosine-mediated immunosuppression in the tumor microenvironment. The analysis objectives were to construct a population pharmacokinetic (Pop PK) model for Q and conduct simulations to recommend an optimal dose for a Phase 3 (Ph3) study.

Methods: The PopPK analysis included data from two phase 1 studies in cancer patients (ARC-6 and ARC-9), one in healthy volunteers (HVs) (AB680CSP0001) and one phase 1b/2 study in cancer patients (AB680CSP0002 or ARC-8). Peripheral CD73 activity was determined using data from HVs and patients. An Emax model, based on the PK-pharmacodynamic (PD) relationship between Q and CD73 activity, determined the C_trough required to achieve ≥ 90% inhibition (IC90). Model and graphical evaluations were conducted using NONMEM (v7.5) and R (v4.2.3). Selected covariates included weight, race, creatinine clearance (CrCL), AST, total bilirubin (TBIL), ECOG-PS, tumor size (TS), and dose. A full model was constructed using univariate selection to determine if a covariate improved the model fit (α=0.05). Backward elimination was performed until no further covariates could be removed without worsening model fit (α = 0.001). The stability and validity of the final model was evaluated using nonparametric bootstrapping and prediction-corrected visual predictive checks (pc-VPC). Covariate effects in the final model were evaluated using forest plots. The optimal dose, based on achieving a C_trough above the CD73 IC90, was identified using simulated PK profiles for 50, 100 and 200 mg Q2W dosing.

Results: The final dataset included 3189 PK observations from 271 patients with doses ranging from 0.1 mg-200 mg, as a single dose or every 1 or 2 weeks (QW or Q2W). The PK profile of Q was best described by a two-compartment model with both linear and non-linear clearance. The estimated linear clearance (CL) and central volume of distribution (V_d) were 0.065 L/hr and 5.18 L, respectively. Weight and CrCL were significant covariates on V_d and CL, respectively. For 100 mg Q2W, compared to the median, the C_max,ss at the 95^th percentile of the weight range was predicted to be 23% lower and 40% higher at the 5^th percentile. Moderate renal impairment (RI) (CrCL of 30 mL/min), relative to median CrCL, was predicted to increase the C_max,ss by 15% and AUC_ss by 77%. None of the other covariates (race, AST, TBIL, ECOG-PS, or TS) were significant. pc-VPC plots indicated no bias. PK simulations indicated 100 mg Q2W is the lowest dose yielding an average C_trough above the CD73 IC90.

Conclusions: A two-compartment model with parallel linear and nonlinear clearance adequately described the PK properties of Q in HVs and patients. No dose adjustment for weight or moderate RI is necessary based on available safety data. The overall PKPD profile of Q indicates the dose of 100 mg Q2W is an appropriate dose for Ph3 trials.

Citations: [1] Bergstrand M, Hooker AC, Wallin JE, Karlsson MO. Prediction-corrected visual predictive checks for diagnosing nonlinear mixed-effects models. AAPS J. 2011 Jun;13(2):143-51. doi: 10.1208/s12248-011-9255-z. Epub 2011 Feb 8. PMID: 21302010; PMCID: PMC3085712.