Abstract 3282: Standardization and coordination of variant interpretation knowledgebases improves clinical genome actionability

Abstract

Abstract The promise of precision medicine in which a cancer patient's treatment is tailored to their specific cancer variants requires concise, standardized, and searchable clinical interpretations–biomarker-disease associations that provide therapeutic or prognostic value. While many institutions curate interpretations for clinically relevant cancer variants, these efforts are siloed and non-interoperable. Through the Global Alliance for Genomics and Health (GA4GH), we formed a Variant Interpretation for Cancer Consortium (VICC, cancervariants.org). We have evaluated six established sources of cancer variant interpretations hosted by members of the consortium: the Clinical Interpretations of Variants in Cancer (CIViC) knowledgebase, OncoKB, the Jackson Laboratories Clinical Knowledgebase (JAX-CKB), Precision Medicine Knowledgebase (PMKB), MolecularMatch, and the Cancer Genome Interpreter (CGI). We compared the coverage of the biomedical literature across them, and found that from 6,034 publications across the 6 resources, 86% of them were unique (by PubMed ID) across the collective. This illustrates both the value to be gained from aggregating the knowledge of these resources and the enormity of the biomedical literature describing cancer variants that is not yet queryable. To create a framework and resource for sharing these interpretations, we have integrated guidelines and ontologies describing genes, variants, diseases, drugs, and evidence to harmonize and describe clinical interpretations of cancer variants. From this, the VICC has collectively established a set of minimum elements required for describing a cancer variant interpretation, and used this to guide harvesting and harmonizing knowledge from these founding resources. Our results demonstrate that the knowledge from the constituent sources of the VICC are highly heterogeneous across every element defining these interpretations (genes/variants, drugs, diseases, evidence). We have developed a harmonized database and an alpha-stage web interface for exploring and querying this breadth of knowledge (http://g2p-ohsu.ddns.net). We next evaluated the 110,320 somatic mutations of 17,005 patient donors in the AACR Project GENIE cohort against our VICC aggregate interpretation database. When separated out by evidence level (as described by the joint consensus recommendation of the Association for Molecular Pathology, American Society of Clinical Oncology, and College of American Pathologists), the interpretations within the VICC aggregate knowledgebase provided 2-4x breadth of coverage across each level of evidence when compared to the median coverage from the constituent resources. Finally, we demonstrate how the clinically actionable interpretation of a patient mutation may change dramatically when considering the knowledge from the VICC resources in aggregate instead of individually. Citation Format: Alex H. Wagner, Brian Walsh, Dmitriy Sonkin, Rodrigo Dienstmann, Xuan S. Li, Jacques Beckmann, Georgia Mayfield, David Tamborero, Núria López-Bigas, Jeremy Goecks, Adam Margolin, Malachi Griffith, Obi Griffith. Standardization and coordination of variant interpretation knowledgebases improves clinical genome actionability [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 3282.