Paired PDX-PDXO models serve an integrated preclinical platform for high-throughput payload screening and ADC drug development

Introduction Antibody-drug conjugates (ADCs) offer an effective and targeted strategy to kill tumor cells by delivering potent cytotoxic payloads directly to tumor cells. However, payload activity can lead to the emergence of resistance through ADC mechanisms such as altered metabolism, efflux pumps, or impaired intracellular activation. Here, we describe an integrated strategy using patient-derived xenograft (PDX) models and paired PDX-derived organoids (PDXOs) for high-throughput payload screening to guide the selection of optimal ADCs and the preclinical validation. Methods PDX models were successfully established from a metastatic triple-negative breast cancer patient who showed resistance to Sacituzumab Govitecan (Trodelvy®), an ADC incorporating a topoisomerase I (TOP1) inhibitor payload (SN-38). The phenotypic and biomolecular characteristics of these models were thoroughly validated via RNA sequencing and whole-exome sequencing (WES). PDX-derived organoid (PDXO) model was developed from fresh PDX tumor tissue. In vivo efficacy was evaluated by tumor growth inhibition (TGI) ratio of each treatment group was calculated by formula: 1-Delta (Treatment group)/ Delta (Vehicle group). In vitro response to different payload of organoids were tested using CellTiter-Glo (CTG) assay and assessed by IC50 values. Results Biomarker analysis showed the PDXO models faithfully recapitulated key features of the corresponding PDX tumors, including high TROP2 expression and reduced sensitivity to Trodelvy. High-throughput in vitro cytotoxicity screening of various ADC payloads revealed a distinct sensitivity profile in these PDXOs: they were insensitive to TOP1 inhibitor payloads (SN-38: IC50>0.008 and DXD: IC50>0.007) but exhibited sensitivity to the tubulin-targeting agent monomethyl auristatin E (MMAE: IC50≤0.0004). Subsequent in vivo efficacy studies of the PDX models with ADCs with different payloads shown consistent sensitivity with the PDXO screening results. The PDX models showed less sensitivity to Trodelv Dato-Dxd and SKB264 which carry TOP1 inhibitor payloads (TGIs rang 24%-57%) but were more sensitive to Dato-MMAE, which showed obvious tumor regression with a TGI of more than 90%, thereby showing correlation with the in vitro predictions. Conclusion This study validates an efficient and integrated preclinical platform for early-stage ADC development. The combined PDX/PDXO approach significantly streamlines the ADC discovery process by enabling high-throughput payload screening, helping prioritize the most promising ADC candidates before advancing to more resource-intensive in vivo studies. This streamlined strategy accelerates data-driven decision-making for developing more effective, patient-tailored ADC therapies and for overcoming clinical resistance mechanisms.