Mapping Cancer Complexity with High-Resolution Proteomics.
Jong Bae Park
Kyung Hee University School of Medicine, Seoul 02447, Republic of Korea
Although genomic profiling has transformed precision oncology, it offers actionable therapeutic insights for only 30–40% of patients, underscoring the need for proteomics to bridge the gap between genotype and phenotype by capturing functional protein networks, signaling dynamics, and spatial interactions within tumors. This seminar introduces a next-generation, clinically scalable proteogenomic framework built on standardized, ultra-high-throughput mass spectrometry, enabled by automation, AI, and machine-learning–driven workflow optimization, allowing the processing of 80–300 samples per day. Central to this platform is ProteoSCOPE, which integrates H&E whole-slide images with AI-guided spatial proteomics to pinpoint tumor- and stroma-specific microregions within FFPE tissues. This approach yields an average of ~8,500 quantified proteins per region, enabling tissue-scale mapping of oncogenic pathways, microenvironmental states, and recurrence-associated features with unprecedented resolution. In parallel, an LC/MS plasma proteomics pipeline using engineered nanoparticles supports unbiased proteome capture from only 120 µL of blood, identifying >7,500 proteins and reflecting 68.7% of primary tumor protein profiles—more than twice the coverage of conventional antibody-based assays. Beyond profiling, functional mapping of cell-surface protein interactions using proximity labeling technologies revealed critical spatial proximities such as EGFR–HER2 complexes, offering mechanistic insight for optimizing bispecific ADC design through enhanced receptor internalization. Complementing this, ProteoMATCH, an AI-based drug response engine trained on 948 cancer cell lines, enables in silico prediction of personalized IC50 landscapes.
Together, these advances converge into a “Virtual Patient” ecosystem that unifies spatial proteomics, plasma proteomics, interactome mapping, and AI-driven therapeutic modeling—establishing a path toward precision oncology where mechanistic insight and clinical decision support are delivered at the speed of thought.
