PII-104 - UTILIZING A MINIMAL PBPK (MPBPK) MODEL TO PREDICT INTRA-TUMORAL CONCENTRATIONS OF MONOCLONAL ANTIBODIES.

l. wang¹, Y. Chang², L. Gaohua³, E. Gibson⁴, P. Statkevich³; ¹Bristol Myers Squibb, Bristol Myers Squibb, Princeton, NJ, USA, ²University at Buffalo, Buffalo, NY, USA, ³Bristol Myers Squibb, Princeton, NJ, USA, ⁴Bristol Myers Squibb, Princeton NJ, USA.

Background: Monoclonal antibodies (mAbs) have advanced cancer treatment; however, their therapeutic efficacy is reliant on their ability to penetrate tumor tissues to reach their target. Therefore, accurately estimating mAb tumor exposure is crucial for optimizing the dosing strategy.

Methods: A minimal Physiologically-Based Pharmacokinetic (mPBPK) model was developed using Monolix (2024), incorporating three lumped well-stirred compartments to represent tumor tissue, and normal tight and leaky tissues (Figure 1). System-specific parameters were sourced from the literature, while drug-specific parameters such as reflection coefficient (σ), percentage of the drug that does not penetrate tissue, for two BMS mAbs (anti-IL8, anti-NKG2A) were determined by fitting clinically observed plasma and tumor PK data. The model performance was validated using PK data from anti-TIGIT mAb, and then the model was expanded to estimate the anti-CTLA4 probody tumor PK by integrating the additional unmasking mechanism of action (MoA) process of the probody.

Results: Diagnostic plots from the anti-IL8 and anti-NKG2A demonstrated that the mPBPK model well captured both plasma and tumor PK. The simulated anti-TIGIT plasma and tumor PKs were well aligned with observed data which further validated the model. In addition, the expanded model with the additional MoA incorporated also well characterized the observed anti-CTLA4 probody plasma and tumor PKs. The predicted σ showed that the penetration ratio of mAb from plasma to tumor ranged from 8 % to 24 %, which falls between that in normal tight and leaky tissues across all the investigated drugs.

Conclusion: Although application of the normal tissue penetration ratio to predict tumor PK has been widely used, it may result in an over- or under-estimation. Incorporating the penetration differences between normal and tumor tissues improves the prediction accuracy of tumor mAb concentrations, which are then used to further support dose selection. This modeling approach may serve as a conceptual framework and workflow process to evaluate the tumor exposures of other biologics including bispecific antibodies and antibody-drug conjugates (ADC) considering its flexibility to adapt to the unique MoA of various mAbs as demonstrated from the CTLA4 probody with unmasking process.