Abstract 205: Astraea: A first-in-class biomarker database integrating genomic, transcriptomic, and tumor microenvironment properties for precision oncology

• Published in: Cancer research (Chicago, Ill.) Ill.), 2021-07, Vol.81 (13_Supplement), p.205-205
• Main Authors: Gafurov, Azamat, Mamichev, Ivan, Vasileva, Elena V., Sagaradze, Georgy D., Shitova, Maria S., Nos, Grigorii, Kotlov, Nikita, Brown, Jessica H., Bagaev, Alexander, Fowler, Nathan
• Format: Article
• Language: English
• ISSN: 0008-5472; 1538-7445
• DOI: 10.1158/1538-7445.AM2021-205

Along with advances in precision oncology, checkpoint inhibitors and targeted therapies have substantially improved outcomes for cancer patients. However, many patients still demonstrate a limited response to these therapies due to many biological factors, including genetic heterogeneity, unique molecular profiles, and the complex features of the tumor microenvironment (TME). Therefore, the selection of personalized effective treatment requires a comprehensive source of therapy response biomarkers, enabling precision medicine strategies for therapy selection. Here, we present a first-in-class automated biomarker analysis database, Astraea, that comprehensively describes genomic, transcriptomic, and TME biomarkers across a wide array of cancers. Automated daily literature reviews of the therapeutic efficacy of biomarkers provided the foundation of Astraea. To date, the database contains a total of 4,116 published biomarkers associated with genomic events, the TME, and targeted proteomic, transcriptomic, and gene signatures. To ensure accuracy of the final inclusion of biomarkers in the database, a multi-step quality control process was implemented that includes an automatic validation step and manual review. After selection, each biomarker is organized into a unique profile in the database which includes assay specifics, the biomarker-associated cancer type, therapy, primary study design, and statistical analysis. Data available from The Cancer Genome Atlas (TCGA) was then used to aggregate interrelated biomarkers into 25 biologically meaningful clusters, with the most prominent clusters identified as components of the TME (i.e., cytotoxic T cells, B cells, fibroblasts) and proliferation rate signatures. The aggregation enabled an easier interpretation and understanding of potentially actionable molecular findings as well as insight into unique neoplastic drivers. To apply Astraea in a clinical setting, we then developed a platform to match therapies to patients based on 1) identified biomarkers prioritized according to level of evidence, including both number of associated publications, statistical strength of individual studies, and cohort size and 2) therapies scored according to supporting biomarkers and associated relevance (resistance/response). By providing comprehensive, up-to-date biomarker identification and matching through utilization of a large automated multi-platform database, this technique aids in the identification and application of bioma