PII-211 - A QUANTITATIVE SYSTEMS PHARMACOLOGY (QSP) MODEL FOR PRECLINICAL TO CLINICAL TRANSLATION OF ANTIBODY-DRUG CONJUGATE (ADC) EFFICACY AND THROMBOCYTOPENIA.

K. McGirr¹, K. Ghusinga², S. Gruver², A. Betts³, F. Hua², J. Apgar²; ¹Applied BioMath, Applied BioMath - Concord, MA, USA, ²Applied BioMath, ³Takeda.

Background: ADCs are a complex drug modality that employ a highly specific monoclonal antibody to deliver a potent cytotoxic payload. Development of ADCs remains challenging because clinical results are influenced by numerous drug-specific properties, such as affinity and potency, and system-specific information, such as target expression and cell growth rate. We present a multiscale platform model for ado-trastuzumab emtansine (Kadcyla), an ADC targeting HER2 positive cancers, to facilitate preclinical to clinical translation of drug disposition, efficacy, and thrombocytopenia (TCP), a common dose-limiting toxicity for ADCs.

Methods: In vitro, mouse, and human models were consecutively built to inform one another. For efficacy translation, an in vitro model replicated experimental data including intracellular drug disposition, mechanistic antibody binding to HER2, and receptor-mediated internalization. The mouse model was trained on pharmacokinetic and tumor growth inhibition (TGI) data. For TCP translation, an in vitro model of thrombopoiesis simulated experiments measuring non-specific ADC uptake into megakaryocytes, which induces platelet count drops characteristic of TCP in vivo. Parameters calibrated from these models were translated to the human model, which was used to project clinical outcomes.

Results: In vitro and mouse models captured published data well. Without further calibration, human TGI predictions aligned with the approved dose and virtual clinical trial simulations using the human model agreed with progression-free survival. Emtansine concentration-driven death of megakaryocytes replicated published platelet dynamics.

Conclusion: Integration of preclinical data and QSP modeling successfully projected ado-trastuzumab emtansine efficacy and TCP in humans. With the proper data, these models could inform first-in-human and efficacious dose predictions for other developing ADCs. The model can be used to predict therapeutic window and guide optimal dose selection.