Integrating a Tumor Size-Overall Survival Model Into a Spatial Quantitative Systems Pharmacology Framework for Antibody-Drug Conjugates in Non-Small Cell Lung Cancer

Abstract

Antibody-drug conjugate (ADC) development in solid tumors faces efficacy and safety challenges from heterogeneous target expression, limited tumor penetration, and off-target payload exposure. We developed a spatial quantitative systems pharmacology (QSP) framework integrating ADC-analyte physiologically based pharmacokinetics (PBPK), a Krogh-cylinder tumor model, and a non-small cell lung cancer (NSCLC) tumor size-overall survival (TS-OS) relationship developed without ADC-treated populations. Applied to trastuzumab deruxtecan (T-DXd) in HER2-mutant NSCLC and datopotamab deruxtecan (Dato-DXd), a TROP2-directed ADC, in EGFR-mutated NSCLC, all parameters came from literature or independent assays; only nonspecific antibody uptake and unconjugated-payload hepatic clearances were estimated from population-level T-DXd serum PK. Using a common structure and virtual-patient framework with independently informed ADC-specific inputs, simulations generated PK, objective response rate (ORR), disease control rate (DCR), and exploratory OS trends consistent with clinical data. Heterogeneous tumor trajectories emerged without fitting individual trajectories, adding ad hoc response compartments, or retuning tumor growth solely to reproduce ORR/DCR. Tumor growth kinetics and target-expression characteristics, represented by histological score and antigen-positive cell fraction, were associated with response, but responder/non-responder distributions overlapped, indicating limited discrimination and need for predictive biomarkers. The TS-OS sub-model reproduced median OS and survival trends, supporting transportability despite absent individual-level tumor and survival data. Tumor growth rate was the strongest survival-associated input, consistent with TS-OS covariate structure. Across doses, 5.4 mg/kg Q3W T-DXd and 6 mg/kg Q3W Dato-DXd were preferred based on predicted efficacy and off-tumor tissue payload exposure, consistent with approved doses. The framework supports ADC design and dose evaluation despite limited individual-level data.