A comparative study of curated contents by knowledge-based curation system in cancer clinical sequencing.

Medical oncologists are challenged to personalize medicine with scientific evidence, drug approvals, and treatment guidelines based on sequencing of clinical samples using next generation sequencer (NGS). Knowledge-based curation systems have the potential to help address this challenge. We report here the results of examining the level of evidence regarding treatment approval and clinical trials between recommendations made by Watson for Genomics (WfG), QIAGEN Clinical Insight Interpret (QCII), and Oncomine knowledge-based reporter (OKR). The tumor samples obtained from the solid cancer patients between May to June 2018 at Kindai University Hospital. The formalin-fixed paraffin-embedded tumor samples (n = 31) were sequenced using Oncomine Comprehensive Assay v3. Variants including copy number alteration and gene fusions identified by the Ion reporter software were used commonly on three curation systems. Curation process of data were provided for 25 solid cancers using three curation systems independently. Concordance and distribution of curated evidence levels of variants were analyzed. As a result of sequencing analysis, nonsynonymous mutation (n = 58), gene fusion (n = 2) or copy number variants (n = 12) were detected in 25 cases, and subsequently subjected to knowledge-based curation systems (WfG, OKR, and QCII). The number of curated information in any systems was 51/72 variants. Concordance of evidence levels was 65.3% between WfG and OKR, 56.9% between WfG and QCII, and 66.7% between OKR and QCII. WfG provided great number of clinical trials for the variants. The annotation of resistance information was also observed. Larger differences were observed in clinical trial matching which could be due to differences in the filtering process among three curation systems. This study demonstrates knowledge-based curation systems (WfG, OKR, and QCII) could be helpful tool for solid cancer treatment decision making. Difference in non-concordant evidence levels was observed between three curation systems, especially in the information of clinical trials. This point will be improved by standardized filtering procedure and enriched database of clinical trials in Japan.