Abstract 3113: Pan-cancer stemness defined by CPTAC proteogenomics guides personalized therapies

Cancer progression involves the gradual loss of a differentiated phenotype and acquisition of progenitor and stem cell-like features. Here, we provide new stemness indices for assessing the degree of oncogenic dedifferentiation. We used machine learning approaches to extract epigenetic, transcriptomic, and proteomic features from human stem cells and applied the computed result to index the Clinical Proteomic Tumor Analysis Consortium (CPTAC) tumor samples by their proteogenomic hallmarks of stemness. Leveraging the resource generated by Human Induced Pluripotent Stem Cells Consortium (HipSci), we extracted stemness signatures from the coherent epigenetic, transcriptomic, and proteomic datasets, using one-class logistic regression machine learning algorithm. The newly obtained stemness scores based on the DNA methylation and gene expression are significantly more robust compared to our previous published work based on the dataset obtained from a smaller number of samples with only genomics data. The stemness score computed by machine learning algorithms using proteomic datasets is novel and original. The obtained proteomic score is able to classify stem cells and non-stem cell classes. Indexing of CPTAC tumors with proteomic stemness score brought us with previously unappreciated findings. We have used the stemness scores computed using gene expression, DNA methylation, and proteins and their modifications to interrogate the coherent proteogenomic CPTAC datasets. The initial analysis of over 2000 tumor samples obtained from twelve types of primary carcinomas of breast, ovary, lung, kidney, uterus, brain (pediatric and adult), head and neck, liver, stomach, colon, and pancreas has confirmed our previously published results. Importantly, our original proteomic-based score brought the analyses to a novel dimension, far beyond previously described results. The initial findings of our work identified proteins and phospho-proteins as active nodes of signaling pathways and transcriptional networks that drive aggressiveness of the primary tumors that cause resistance to existing therapies. Our results indicate that cancer stemness is associated with the engagement of developmental pathways shaping tumor plasticity. Progressive oncogenic de-differentiation of cancer cells impacts tumor microenvironment and is associated with “cold” tumors thus impairing anti-tumor immune response and limiting the efficacy of immunotherapies. Hallmarks of cancer stemness correlate with increased tumor pathology grade and clinical stage and results in worse survival of the cancer patients across analyzed tumor types. Targeting identified proteins and cellular mechanisms that drive lethal phenotype of de-differentiated tumors with existing or novel drugs may pave the way for clinical development of effective cures for cancer patients. Citation Format: Maciej Wiznerowicz, Antono Colaprico, Erik Storrs, Francesca Petralia, Iga Kolodziejczak, Felipe da Veiga Leprevost, Weiping Ma, Daniel Cui Zhou, Bo Wen, Alexander Lazar, Pietro Pugliese, Michele Ceccarelli, Bozena Kaminska, Jan Lubinski, Alexey Nesvizhskii, Bing Zhang, Henry Rodriguez, Ana L. Robles, Mehdi Mesri. Pan-cancer stemness defined by CPTAC proteogenomics guides personalized therapies [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 3113.