借助高通量双偶联平台和预测性耐药疾病模型进行多样性导向的dpADC发现

Acquired resistance to antibody-drug conjugate (ADC) therapy remains a major clinical challenge, often leading to diminished efficacy of subsequent ADCs that share the same payload class even based on different targets. For instance, reduced response has been observed in patients receiving sequential treatment with the Top1 inhibitor-based ADCs sacituzumab govitecan and trastuzumab deruxtecan, irrespective of treatment sequence. This underscores payload cross-resistance as an emerging limitation in ADC-based regimens. Dual-payload ADCs (dpADCs) represent a novel therapeutic modality with the potential to overcome such resistance. However, conventional dpADC discovery is often constrained by limited molecular designs, as the vast structural complexity-arising from variations in payload pairing, stoichiometric ratios, linker release mechanisms and kinetics, and antibody properties-poses significant challenges for systematic synthesis and evaluation. To address this challenge, we developed a comprehensive linker-payload (LP) library featuring diverse linker designs and multiple payload classes-including LPs based on Topoisomerase I inhibitors, Topoisomerase II inhibitors, PARP1 inhibitors, ATR inhibitors, and CHK1/2 inhibitors. Using our automated, high-throughput dual-conjugation platform (iScreener), we efficiently constructed a diversity-oriented dpADC library targeting HER2 and TROP2 respectively. These dpADCs were systematically evaluated in resistant in vitro and in vivo models, including patient-derived organoid/xenograft (PDXO/PDX) systems. Notably, several candidates with novel designs demonstrated significantly enhanced therapeutic efficacy while maintaining favorable safety profiles compared to benchmark and conventional mono-payload ADCs, revealing clear enhanced effects between payload classes. In summary, our shift from a purely rational design paradigm to a high-throughput screening approach-enabled by efficient dpADC library construction and predictive resistant disease models-offers a robust discovery framework. This strategy identifies potent dpADC candidates through empirical screening rather than traditional design perception, and we are currently expanding our evaluation of additional payload combinations in dpADC format using resistant preclinical models that recapitulate unmet clinical needs.