Optimized methods for detecting somatic variants in plasma of advanced colorectal cancer patients with pooled normal dataset.

Genotyping plasma cell-free DNA(cfDNA) has been to be prevalence method in cancer care. However, the somatic mutant fragments from peripheral blood mononuclear cell(PBMC) due to clonal hematopoiesis in plasma are the one of the main reason for false positives in tumor monitoring. In order to identify and overcome these false positives, plasma cfDNA and DNA from PBMC and available tumor tissues of 51 advanced colorectal cancer patients were analyzed by 38kb size panel of ultra-deep sequencing. For somatic variant calling, we used the optimal sized pooled PBMC data set as normal control. The mutation of PBMC in plasma were counted as 2.0%(358/17,749), and ARID1A, TP53 and ATM were highly frequently detected. Using our somatic variant calling method after filtering raw data, the rate of these biological false positives was reduced from 1.2% to 0.7% compared with the conventional method. Meanwhile, the mutant fragments in cfDNA derived from tumor were detected identically. Across all genes, the concordance rate between plasma cfDNA and tumor tissue-based DNA was calculated as 90.4%. Sensitivity and specificity were 67.1% and 91.2% respectively. In conclusion, our study suggests the method for suppressing biological false positives due to clonal hematopoiesis by using our somatic variant calling strategy.