84. Benefits of integrating an open-source knowledgebase in a precision oncology workflow

Aly Abdelkareem

Cameron Grisdale

Dr. Cameron Grisdale PhD is a Research Associate in the Clinical Informatics group at Canada’s Michael Smith Genome Sciences Centre in Vancouver, Canada. He conducts genomic and bioinformatic analyses of high-throughput tumour sequencing data, in particular for BC Cancer’s Personalized OncoGenomics (POG) program, a translational research effort that uses in-depth genome and transcriptome sequencing to guide personalized treatment decision-making. Dr. Grisdale is focussed on the growth and development of the POG knowledgebase, which plays a critical role in identifying clinically relevant alterations in patient tumours that are ultimately used for tumour characterization and therapy selection. A key part of this effort includes collaboration with the open-source CIViC knowledgebase team, working towards increasing curation of cancer variant interpretations relevant to precision oncology and developing knowledge models and curation standards for emerging biomarkers.


Cameron Grisdalea, Erin Pleasancea, Caralyn Reislea, Laura Williamsona, Kilannin Krysiakb, Jason Salibab, Arpad Danosb, Adam Coffmanb, Susanna Kiwalab, Joshua McMichaelb, Malachi Griffithb, Obi L. Griffithb, Steven Jonesa

aCanada’s Michael Smith Genome Sciences Centre, Vancouver, BC, CA; bMcDonnell Genome Institute, St. Louis, MO, USA

Next-generation sequencing is revolutionizing precision oncology, but vast amounts of sequence data must be processed and mined to reveal clinically relevant alterations that can inform patient care. The Personalized OncoGenomics (POG) program at BC Cancer utilizes whole genome and transcriptome analysis, providing a comprehensive view of advanced cancer patient tumours. This analysis relies on curated clinical knowledgebases linking cancer variants and their clinical relevance, but the breadth and utility of these can be limited by proprietary access or limited coverage. CIViC (Clinical Interpretation of Variants in Cancer) is an open-access, crowd-sourced knowledgebase of clinically relevant cancer variants that aims to address these limitations, and is one of several sources used for variant interpretation in POG. Based on a retrospective cohort of POG cases, we evaluated the knowledgebase coverage of genes and variants involved in treatment recommendations from the molecular tumour board (MTB). In a preliminary analysis, we found more than 95% of patients had an alteration considered clinically actionable by the MTB, demonstrating the benefit of combined genome and transcriptome analysis. We found that knowledge from CIViC represented nearly 50% of therapeutic evidence reported at the MTB, emphasizing the role of open-access knowledge in precision oncology. Additionally, by prospectively examining the levels and sources of evidence in patient reports, we identified areas where additional knowledgebase curation would most impact alignment with clinical utility, highlighting genome signatures as a critical area for future curation efforts. We also considered the impact of quality of evidence on MTB recommendations and patient treatments.