Clinical interpretation of genomic variants in childhood cancers is heavily reliant upon our ability to precisely locate and apply structured biomedical knowledge. This exercise is uniquely challenging for pediatric cancers. Even compared to other cancers, childhood cancers are highly heterogeneous, often involving genes not typically attributed to adult cancers and often involving complex, large-scale variations. Childhood cancers also tend to arise in less differentiated cell lineages, progress rapidly, and have much lower incidence and mutation burden. These characteristics are compounded by unique clinical challenges related to developmental factors and treatment-related secondary cancers, resulting in significantly fewer targeted therapy options available to pediatric patients. These challenges have led to a significant under-representation of genes and variants relevant to childhood cancers in public cancer databases and knowledgebases. Thus, there is a critical unmet need for structured genetic variant level data in publicly accessible databases which document diagnostic, prognostic and therapeutic biomarkers for childhood cancers. The Childhood Cancer Data Initiative (CCDI) focuses on accelerating research on childhood cancers by developing and enhancing tools and methods to extract knowledge and enable sharing of childhood cancer data. The CIViC resource is a knowledge extraction and data sharing platform for clinical interpretation of cancer variants. Expert curators extract knowledge and evidence from the literature to produce rigorous and high-quality assertions following widely-recognized clinical variant classification guidelines. The variant classifications are made freely available in the public domain and are widely utilized in clinical workflows. However, to date, CIViC knowledge curation has focused almost entirely on adult malignancies. Other key variant knowledgebases either do not provide clinical relevance, lack childhood-specific interpretations, and/or have restrictive licenses. We propose to close this knowledge gap in CIViC by forging new and expanded collaboration with pediatric cancer experts, and building domain-specific tools for curation and dissemination of high-quality variant interpretations for childhood cancers. The CIViC visual interface will be adapted to better support curation and dissemination of childhood cancer variants interpretations. A modification of our existing natural language processing approach will be used to prioritize literature relevant to childhood cancers and an expert panel will prioritize childhood variants and subsequently perform and also evaluate curation on taskforce calls. Based on the experience and products of this work, we will develop childhood-cancer-specific curation guidelines and conduct pilot development of a childhood cancer variant panel that is capable of evolving rapidly as childhood cancer variant knowledge accumulates in CIViC.
Childhood cancers present many unique challenges for interpreting the clinical relevance of genetic variations found in each patient. For this reason, genes and variants associated with childhood tumors are significantly under-represented in public cancer databases and knowledgebases. Building on the success of CIViC (www.civicdb.org), a widely adopted resource for interpretation of adult cancer variants, we will create a free, publicly accessible companion resource of expert curated diagnostic, prognostic and therapeutic biomarkers for childhood cancers.
