Mechanistically Coupled Physiologically Based Pharmacokinetic-Pharmacodynamic Modeling for Analysis and Optimization of Lu-177-PSMA Therapies: Patient-Specific Model Fitting
Abstract
Purpose
Radiopharmaceutical therapies (RPTs) effectively treat metastatic castration-resistant prostate cancer, yet injected radioactivities remain empirically prescribed. Although physiologically-based pharmacokinetic (PBPK) and pharmacodynamic (PD) models can separately predict biodistribution and radiation response, they are rarely mechanistically coupled, limiting prediction of how tumor response influences uptakes and absorbed doses across treatment cycles. Here, we couple PBPK and PD models for 177Lu-PSMA therapies to quantify cycle-dependent changes in tumor uptake and response and to characterize inter-patient variability in lesion-level outcomes, providing a framework for personalized RPT.
Methods
We mechanistically coupled a 177Lu-PSMA-617 PBPK model with a radiobiological PD model representing viable and damaged tumor cells. The PBPK model included compartments for simulated lesions, remaining tumor burden, kidneys, and salivary glands. Each compartment tracked radiolabeled and unlabeled ligand, with PSMA-specific binding and internalization in lesion, kidney, and salivary gland compartments. PD-predicted cell death dynamically updated lesion volumes, enabling bidirectional PBPK-PD feedback. We fit PSMA receptor densities (Rden) to clinical data from Peters et al. (2022) for patients receiving 2-cycle 177Lu-PSMA-617 therapy (3, 6 GBq).
Results
Compared with uncoupled PBPK simulations, the coupled PBPK-PD model predicted progressive reductions in tumor uptake across cycles, yielding cycle-dependent tumor time-activity curves (TACs). In the coupled model, later cycles showed substantially reduced tumor uptake, with rapid post-injection peaks followed by washout as radioligands redistributed to blood. In contrast, uncoupled simulations produced nearly identical TACs across cycles because lesion volumes and volume-dependent model parameters remained fixed. Fitting Rden to clinical data reproduced measured lesion mass changes and activity uptake (mean square error<0.002). Estimated lesion Rden exhibited inter-patient and inter-lesion variability, reflecting tumor response variability.
Conclusion
This coupled PBPK-PD framework captures bidirectional feedback between tumor response and radiopharmaceutical distribution. By quantifying inter-patient variability in model parameters, this approach provides a quantitative foundation for evaluating multi-cycle regimens and optimizing patient-specific 177Lu-PSMA therapies.