AbstractBACKGROUNDDetecting single-nucleotide variations and insertions/deletions in circulating tumor DNA is challenging because of their low allele frequency. The clinical use of circulating tumor DNA to characterize tumor genetic alterations requires new methods based on next-generation sequencing.METHODSWe developed a method based on quantification of error rate of each base position [position error rate (PER)]. To identify mutations, a binomial test was used to compare the minor-allele frequency to the measured PER at each base position. This process was validated in control samples and in 373 plasma samples from patients with lung or pancreatic cancer.RESULTSMinimal mutated allele frequencies were 0.003 for single-nucleotide variations and 0.001 for insertions/deletions. Independent testing performed by droplet digital PCR (n = 231 plasma samples) showed strong agreement with the base-PER method (κ = 0.90).CONCLUSIONSTargeted next-generation sequencing analyzed with the base-PER method represents a robust and low cost method to detect circulating tumor DNA in patients with cancer.