A germline and somatic mutation sorting (GeMSort) algorithm for extracting presumed germline pathogenic variants in liquid genomic profiling: Insights from Database of Center for Cancer Genomics and Advanced Therapeutics (C-CAT)

Liquid biopsy comprehensive genomic profiling (LB-CGP) testing is performed on circulating tumor DNA to detect tumor recurrence, predict prognosis, and select therapeutic agents. Pathogenic variants of germline origin in genes associated with hereditary tumor syndrome (HTS) can be simultaneously detected by CGP testing. Nonetheless, it is often challenging to differentiate whether the variants are of somatic or germline origin. The differentiation criteria were primarily based on the variant allele frequencies (VAFs). However, more evidence is needed to establish clear criteria, and it is often difficult to differentiate between variants based on VAF alone. In this study, using the national database of the Center for Cancer Genomics and Advanced Therapeutics, which accumulates real-world data on CGP testing in Japan, we analyzed 169,370 variants detected in 11,399 patients registered with FoundationOne Liquid CDx testing. By extracting the predominantly presumed somatic and germline variants, we established a criterion for VAF that could achieve high specificity and sensitivity. Further investigation into the detection status of other variants led to the development of an algorithm for differentiating somatic/germline variants of genes associated with HTSs. Based on this algorithm, 726 variants were extracted as presumed germline pathogenic variants among the 26 genes with high germline conversion rates in 710 patients in the study. This algorithm should help to discriminate with high accuracy whether the variants detected in LB-CGP tests are of somatic or germline origin, although further analyses are required to confirm the validity of this algorithm.