Abstract 7679: Advanced proteomic profiling of patient-derived tumor models reveals high cross-model concordance and improved translational fidelity

Abstract Introduction: High-throughput proteomic profiling of preclinical tumor models is increasingly critical in oncology studies and translational research for anti-cancer drug development. However, the murine stromal compartment in patient-derived xenografts (PDXs) confounds accurate proteomic quantification, as previously demonstrated. For instance, 30% of PDX tumor samples contain over 25% mouse stromal cells, which causes false results in downstream proteomic analysis. Methods: To overcome this issue, we developed an optimized "separate-then-run" methodology that effectively separates human and mouse cells before proteomics profiling by mass spectrometry, thus enabling the acquisition of precise human tumor proteomic signatures from PDX samples. For organoid models, no mouse cell removal was performed before the mass spectrometry experiment. Results: Using an optimized data-analysis pipeline, we profiled 418 PDX models and 89 organoids, including 59 patient-derived organoid (PDO) and 30 PDX-derived organoid (PDXO) models, quantifying a median of 9,916 human proteins per sample (range: 8,514-11,206) with high reproducibility after batch correction and normalization (Pearson correlation coefficient range: 0.93-0.98). To evaluate cross-platform fidelity, we also start to perform proteomic profiling of paired in vivo and in vitro tumor models; preliminary results show high consistency of proteome expression between paired models. For example, Pearson correlation coefficients are 0.93-0.96 for 3 PDO-PDOX pairs and 0.91-0.96 for 7 PDX-PDXO pairs; both are much higher than inter-model correlation within PDO, PDOX, PDX, and PDXO (Wilcox test p-value 0.01). Further ongoing studies will provide more evidence on within/between tumor model proteome similarity. Conclusions: In summary, we have established an advanced proteomic profiling platform that significantly enhances protein detection coverage in preclinical tumor models over conventional methods and validates the high concordance of protein expression between paired in vitro and in vivo tumor models. Citation Format: Jia Xue, Hengyuan Liu, Xiaobo Chen, Sheng Guo. Advanced proteomic profiling of patient-derived tumor models reveals high cross-model concordance and improved translational fidelity abstract. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 7679.