PII-076 - MODEL-ESTIMATED TISSUE PHARMACOLOGY AS A SURROGATE OF MONOCLONAL ANTIBODY CLINICAL EFFICACY DOSE RESPONSE: CASE STUDY USING DUPILUMAB IN PATIENTS WITH ATOPIC DERMATITIS.

K. Bartlette¹, A. MacDonald¹, J. Clary¹, J. Roberts¹; ¹Allucent, Cary, NC, USA.

Background: The development of monoclonal antibodies (mAbs) for treatment of various diseases has rapidly increased over the past several decades. High potency and target specificity, long half-lives, and convenient dosing regimens have made mAbs a very effective treatment option [1]. Pharmacokinetic/pharmacodynamic (PK/PD) models are often used to support clinical dose selection [2]. However, published examples demonstrating the predictive value of PK/PD models to predict clinical efficacy dose response are lacking. As dupilumab demonstrated a clinical dose response in an atopic dermatitis (AD) phase IIb study [3], it was selected as a case study. Dupilumab is approved for treatment of AD, asthma, and several other type 2 inflammatory diseases. By binding to the IL-4Rα receptor, dupilumab inhibits the downstream signaling of inflammatory cytokines IL-4 and IL-3. Using in vitro measures of dupilumab binding affinity, estimates of IL-4Rα receptor density within the skin, and phase II clinical data, a minimal physiologically-based pharmacokinetic/pharmacodynamic (mPBPK/PD) model was developed to predict receptor occupancy (RO), to correlate this with clinical efficacy data.

Methods: A mPBPK/PD model was constructed, using R v4.1.3 and package mrgsolve, by combining population pharmacokinetic (PopPK) models [4,5], with physiological skin and lymph compartments. Free and drug-bound IL4Rα within skin were predicted based on skin dupilumab concentrations and binding affinity parameters of the drug to receptor. The final model predicted skin RO time-course and exposure dose response following the 16-week multi-dose regimens investigated in the phase IIb study [3]. Emax models were fit to both RO dose response and observed eczema area and severity index (EASI) disease activity scores at end of treatment and compared.

Results: Emax dose response models for RO and EASI scores adequately described the data. Predicted RO Emax curve was shifted to the right when compared to the observed efficacy dose response and the ED50s were approximately two-fold apart.

Conclusion: This demonstrates the simulated dose response model would have been informative to guide dose selection in this phase II study for dupilumab. Further investigation with similar mAbs is warranted to see if this work can be replicated across compounds, targets and diseases.