Pan-cancer proteogenomics expands the landscape of therapeutic targets
Introduction
Molecularly targeted therapies are critical for improving cancer treatment. Since proteins are the primary targets of these therapies and functional effectors of genetic aberrations, proteogenomics, the integration of unbiased mass-spectrometry (MS)-based proteomics with genomics, epigenomics, and transcriptomics, provides a powerful framework for exploring targets for therapeutic intervention in cancer research. The Clinical Proteomics Tumor Analysis Consortium (CPTAC) has performed comprehensive proteogenomic characterization for over 1,000 prospectively collected, treatment naïve primary tumors spanning 10 cancer types, many with matched normal adjacent tissues. Recently harmonized proteogenomics data of these samples provide an unprecedented opportunity to characterize existing and future therapeutic targets for cancer treatment.
Methods
CPTAC proteogenomics data consisting of mutation, copy number variation, methylation, transcript abundance, protein abundance, and phosphosite abundance from >1,000 cancer patients spanning 10 cancer types was used to evaluate current and potential therapeutic targets curated from DrugBank, Guide to Pharmacology, Drug Gene Interaction Database, and the in silico human surfaceome. Data harmonization across all cancer types and omics data types was achieved through the use of standardized computational pipelines and the same versions of genome assembly and gene annotation. Cell line data from DepMap was further integrated to distinguish causations from associations. Drug response data from the profiling relative inhibition simultaneously in mixtures (PRISM) primary screen in cancer cell lines was used to evaluate the quality of our target prioritization strategy. Computational pipelines were deployed to identify synthetic lethality for targeting tumor suppressor loss and to prioritize tumor associated antigens as immunotherapy targets.
Results
We identified >2,800 druggable proteins and classified them into five tiers to facilitate different applications such as companion diagnostics, drug repurposing, and new therapy development. Four hundred fifty-seven druggable proteins showed both overexpression in tumors compared to normal and significant dependency in cell line CRISPR-Cas9 screens of the same lineage. Notably, 51 proteins demonstrated both overexpression and dependency in five or more cancer types. A similar analysis of phosphoproteomics data focusing on known activating sites of druggable proteins further revealed targetable dependencies driven by protein hyperactivation. The highest proportion of druggable protein dependencies was found for Tier 1 targets compared to other drug target tiers. Tier 1 are targets of already approved oncology drugs and these results support the relevance and potential effectiveness of our identified targets. Our predicted drug target dependencies also helped to improve the identification of successful drug responses from the PRISM primary drug screen with a high specificity of 83% and accuracy of 76%, two attributes critical for nominating promising targets for further experimental and clinical validation. We also identified synthetic lethality for difficult-to-target tumor suppressor losses, revealing TP53 mutations as a candidate biomarker to select endometrial cancer patients for treatment with doxorubicin. Finally, our tumor associated antigens analysis followed by experimental confirmation nominated peptides as promising immunotherapy targets.
Conclusion
Pan-cancer analysis depicts a comprehensive landscape of protein and peptide targets for developing companion diagnostics, drug repurposing, and new therapies.