Quantitative detection of mutation alleles derived from lung cancer in plasma cell-free DNA by use of anomaly detection with deep sequencing data

Rare mutation detection with next generation sequencing has a considerable potential for diagnostics. Detection of circulating tumor DNA is the forefront application of this approach. The major obstacle is the high read error of next-generation sequencers. Instead of increasing accuracy of final sequences, we detected rare mutations with Ion Torrent PGM, using anomaly detection criteria set based on a statistical model of read error rate at each error position. Statistical models were deduced from sequence data of normal samples. We detected epidermal growth factor receptor (EGFR) mutation in plasma DNA of lung cancer patients. Single-pass deep sequencing (>100,000 reads) could detect one activating mutation allele in 10,000 normal alleles. We confirmed the method with 22 prospective and 155 retrospective samples, mostly DNA purified from plasma. The temporal analysis suggested potential application for disease management in addition to therapeutic decision for epidermal growth factor receptor tyrosine kinase inhibitors (EGFR-TKI).