Abstract 1295: A translational approach to ADC evaluation: preclinical single-mouse basket trial with Bayesian modelling analysis

Abstract Introduction: Antibody-drug conjugates (ADCs) have revolutionized cancer therapy, exemplified by the clinical success of HER2-targeting ADCs like Trastuzumab-Deruxtecan (T-Dxd), which demonstrates efficacy across a spectrum of HER2 expression levels. However, ADC development remains challenging due to translational gaps between preclinical and clinical findings. This study introduces a robust translational approach for ADC evaluation, combining a single-mouse preclinical model using MiniPDX® technology with Bayesian statistical analysis—a widely recognized and commonly applied statistical approach in clinical research. Bayesian analysis offers the advantage of integrating prior knowledge with experimental data to generate probabilistic predictions, providing a dynamic and nuanced understanding of treatment efficacy. Our study mimics clinical trials of T-Dxd and Trastuzumab-Emtansine (T-DM1) using this robust, translational framework to bridge the gap between preclinical and clinical evaluation. Methods: We selected 20 PDX models over six cancer types, representing HER2 expression levels (high, medium, low, and negative control). Tumor tissues were dissociated into single-cell suspensions and implanted subcutaneously in mice via MiniPDX capsules. Each mouse received a single 10 mg/kg dose of T-Dxd or T-DM1 over 10 days. Tumor cell viability was assessed using CellTiter-Glo, and tumor relative proliferation rate was calculated as Treatment/Control ratios (T/C%). A T/C ratio <50% indicated significant response. Bayesian modelling was applied to estimate the probability of tumor inhibition across HER2 cohorts. Results: T-Dxd demonstrated a 100% overall response rate (ORR) in HER2-high models, with significant tumor inhibition. Additionally, T-Dxd exhibited substantial efficacy in HER2-medium (ORR 75%) and HER2-low (ORR 60%) groups. In contrast, T-DM1 displayed strong tumor inhibition in HER2-high models but limited efficacy in HER2-medium and HER2-low groups. In addition, Bayesian analysis yielded posterior probabilities of 83%, 66%, and 54% that T-Dxd slows down tumor growth by 50% when compared to control for the HER2-high, HER2-medium, and HER2-low groups, respectively. Conclusion: Our study demonstrates the feasibility and promising results of preclinical single-mouse trials using MiniPDX®. The Bayesian statistical analysis provides a rich summary of the effects of T-Dxd and T-DM1. This approach bridges the translational gap, offering a platform to predict ADC efficacy and guide candidate selection in the preclinical setting. Citation Format: Diandong Jiang, Yuan Ji, Jiaxin Liu, Bin Xie, Hongkui Chen, Shizhu Zhao, Le Li, Yu Di, Xinlei Chen, Xiaorong Gu, Huanhuan Zhang, Zhenle Bi, Kaimeng Hu, Loc Van, Josh Caggiula, Yulan Tang, Danyi Wen. A translational approach to ADC evaluation: preclinical single-mouse basket trial with Bayesian modelling analysis [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 1295.