{
  "abstract": "Background Bispecific antibodies (bsAbs) exhibit unique pharmacological and pharmacokinetic (PK) properties by simultaneously targeting two molecular epitopes. Predicting human PK and selecting the optimal clinical dose remains a significant challenge in early-stage development. Semi-mechanistic physiologically based pharmacokinetic (PBPK) models, which integrate system-level physiology with drug- and target-specific characteristics, offer a promising non-animal strategy aligned with model-informed drug development (MIDD) approaches. This study presents a modeling framework to predict human PK and optimal dose of a PD-L1/4-1BB-targeting bsAb using in vitro binding and pharmacodynamic (PD) data, with clinically translatable results that can be confirmed during clinical studies.Methods A semi-mechanistic physiologically based pharmacokinetic (PBPK) model was developed by integrating human physiological parameters such as organ volumes, blood and lymphatic flow rates, and FcRn-mediated recycling with drug-specific properties including molecular weight, binding affinities, and subcutaneous bioavailability. In addition, the model incorporated target biology characteristics such as PD-L1 and 4-1BB expression levels, turnover rates, and internalization kinetics. In vitro binding kinetics and T-cell expansion assay data were utilized to characterize trimer formation, which exhibited a bell-shaped concentration-response relationship. Using this information, the model was applied to simulate human PK profiles and to identify optimal dose levels that maximize trimer formation, prior to the availability of in vivo or clinical data.Results The model successfully reproduced dose-dependent human PK profiles consistent with known monoclonal antibody kinetics. Simulations accurately reflected the bell-shaped pharmacodynamic response of trimer formation observed in vitro. An optimal dosing range was identified that maximized trimer formation, providing a rational basis for dose selection prior to clinical studies and enabling confirmatory comparison with clinical data.Conclusions This study highlights the clinical utility of a semi-mechanistic PBPK modeling framework that incorporates in vitro kinetic and pharmacodynamic data to predict human PK and inform dose selection for bispecific antibodies. The approach serves as a viable alternative to animal testing, aligning with New Approach Methodologies (NAMs) and current regulatory expectation for biologics. With future validation using clinical data, the framework holds potential for further refinement and expansion into AI-driven predictive modeling for bsAb development.",
  "authors": [
    {
      "affiliations": [
        "AIMS BioScience, Co., Ltd., Seoul, Republic of Korea",
        "Sungkyunkwan University, Seoul, Republic of Korea"
      ],
      "name": "So Jin Lee"
    },
    {
      "affiliations": [
        "AIMS BioScience, Co., Ltd., Seoul, Republic of Korea"
      ],
      "name": "Yunjung Hong"
    },
    {
      "affiliations": [
        "AIMS BioScience, Co., Ltd., Seoul, Republic of Korea"
      ],
      "name": "Jinha Park"
    },
    {
      "affiliations": [
        "AIMS BioScience, Co., Ltd., Seoul, Republic of Korea"
      ],
      "name": "Soo Hyeon Bae"
    }
  ],
  "title": "1103 Semi-mechanistic PK modeling of bispecific antibodies for optimal dose prediction: a clinically confirmable and AI-expandable framework using bell-shaped PD responses",
  "uid": "e0b3101c-6ecf-5599-ba2d-81b3ac3d59fb"
}