PK-PD Modeling Suggests Tumor Heterogeneity Limits Preclinical to Clinical Translation of Enzalutamide in Prostate Cancer

Purpose: To assess the translational performance of preclinical models to anticipate enzalutamide clinical efficacy in prostate cancer using a cross-species physiologically-based pharmacokinetic (PBPK)-pharmacodynamic (PD) model and published enzalutamide preclinical and clinical PK-PD data. Methods: A mouse PBPK model was developed and linked to a tumor growth inhibition (TGI) model to describe tumor volume profiles in 5 cell line-derived and 3 patient-derived xenografts. Estimated preclinical PD model parameters were fixed to simulate clinical progression free survival (PFS) using a scaled-up human PBPK model and a calculated prostate cancer tumor growth rate. Tumor static concentrations (TSCs) were calculated and assessed relative to predicted steady-state concentrations from a standard dosing regimen. Results: Estimated mouse model parameters were precise (CV% < 15% and 45% for most PK and PD parameters), and the estimated hepatic intrinsic clearance was comparable to conventional allometry (allometric exponent = 0.722). The 8 TSC values were near or below the simulated steady-state plasma drug concentration (18,000 ng/mL) achieved with a standard enzalutamide regimen (TSC range 203–20,740 ng/mL). The TGI model failed to predict clinical PFS unless a tumor heterogeneity model with sensitive and resistant tumor cell populations and individually calibrated growth rates were explicitly incorporated. Conclusions: The potential and the limitations of using preclinical PD parameters for clinical translation were assessed, and more biologically accurate TGI models are needed to enhance the use of translational modeling in oncology.