Principal Scientist - I Vantage Research LLC Lewes, Delaware, United States
Disclosure(s):
Kannan Thiagarajan: No financial relationships to disclose
Objectives: To identify key translational principles for first-in-human (FIH) dose prediction of antibody-drug conjugates (ADCs) by analyzing publicly available preclinical and clinical data from FDA-approved ADCs.
Methods: We curated preclinical and clinical datasets for 12 FDA-approved ADCs (up to 2022) [1], covering in vitro properties (e.g., binding affinity, internalization, antigen expression), ADC/payload physicochemical characteristics (e.g., linker stability, diffusion), and in vivo PK/PD data across species. Clinical data included human PK and tumor response metrics across dose levels. An existing mechanistic PK/PD model [2] was extended to incorporate key ADC-specific features: target-mediated drug disposition (TMDD), tumor-specific growth kinetics, and payload-mediated cytotoxic effects. Dose translation was approached via three main steps: 1. Calibration of ADC PK using animal data (mouse, rat, or monkey); 2. Estimation of tumor growth and therapy effect parameters using CDX/PDX tumor growth inhibition data; 3. Projection of ADC and payload PK/PD parameters to human. When detailed preclinical time-course data were unavailable, reverse translation (starting from human data) was used for model validation. Model performance was evaluated through comparison of predicted vs observed human PK profiles and visual predictive checks (VPCs) for tumor response.
Results: Most ADCs had robust clinical PK/PD data and in vitro characterization, though gaps remain in linker stability and tissue distribution data. ADC PK was generally predictable using standard allometric scaling (exponent ~0.8–1.0), with < 20% deviation from observed human data. Surprisingly, both monkey and rodent species provided comparable PK prediction accuracy. In contrast, payload PK translation was less reliable; standard scaling methods often failed, except in cases like DXd and DM1. For PD translation, CDX and PDX models both yielded clinically relevant dose–response projections. Notably, tumor doubling time was a key parameter requiring species-specific scaling to match observed human responses. Parameter correlation analysis also revealed potential synergistic patterns between antibody and payload properties that correlated with clinical success.
Conclusions: This study provides the first comprehensive, model-informed evaluation of translational strategies across all FDA-approved ADCs. While ADC PK is reasonably predictable using traditional scaling, payload PK requires clinical anchoring for reliable dose projection. PD translation is sensitive to tumor growth kinetics and preclinical model choice. These insights can guide more informed, efficient dose selection strategies in future ADC programs.
Citations: [1] López de Sá A, Díaz-Tejeiro C, Poyatos-Racionero E, Nieto-Jiménez C, Paniagua-Herranz L, Sanvicente A, Calvo E, Pérez-Segura P, Moreno V, Moris F, Ocana A. Considerations for the design of antibody drug conjugates (ADCs) for clinical development: lessons learned. J Hematol Oncol. 2023 Dec 12;16(1):118. doi: 10.1186/s13045-023-01519-0. [2] Singh AP, Seigel GM, Guo L, Verma A, Wong GG, Cheng HP, Shah DK. Evolution of the Systems Pharmacokinetics-Pharmacodynamics Model for Antibody-Drug Conjugates to Characterize Tumor Heterogeneity and In Vivo Bystander Effect. J Pharmacol Exp Ther. 2020 Jul;374(1):184-199. doi: 10.1124/jpet.119.262287.
