Implementation of next generation sequencing technology for somatic mutation detection in routine laboratory practice.

The introduction of next generation sequencing (NGS) in the routine diagnostic setting is still in the development phase and has been limited by its complexity. Targeted NGS offers an attractive alternative to performing multiple single target assays and is very useful in meeting the increasing clinical demand for testing of multiple genetic aberrations in cancer specimens. To this end, we carried out a blinded validation study on 113 tumours in a diagnostic laboratory and compared mutation results from targeted NGS with those from Sanger sequencing, pyrosequencing, competitive allele specific TaqMan polymerase chain reaction (CAST PCR) and Cobas assays. DNA was extracted from formalin fixed, paraffin embedded (FFPE) tissue samples that included core biopsies, resections and cytology samples from three common and one rare cancer types [non-small cell lung cancer (NSCLC), colorectal cancer (CRC), malignant melanoma (MM) and gastrointestinal stromal tumour (GIST)]. Libraries were prepared using the TruSight Tumour 26 gene panel and NGS was carried out on the MiSeq instrument. Results from NGS were concordant with the mutational status determined by other platforms in 107 of the 113 cases tested (94.7%). The sequencing quality for NGS failed in four of the six false negative cases, while a further two samples gave false negative results because the c-KIT mutations were located outside the range of the NGS panel. One NSCLC sample contained an EGFR mutation previously detected by the Cobas assay. Reanalysis of the NGS data for this sample using a cut-off allele frequency of 1% revealed the mutation had an allele frequency of 2%, which was below the recommended software-determined threshold of 3%. NGS detected 113 additional mutations that were not previously known from analysis by the conventional methods. Twenty-six of these have known clinical importance, 37 have potential clinical significance, while 50 were novel mutations with unknown clinical significance. NGS detected variants using inputs of 10-20 ng of FFPE extracted DNA and from specimens with a tumour cell content less than 50%, for which when possible we recommend microdissection. We conclude that results from targeted NGS are highly concordant with those from other mutation testing platforms. Targeted NGS is suitable for a range of sample types received in the diagnostic pathology laboratory, including those with limited material or with low tumour cell content (TCC). This work has allowed us to determine the quality parameter settings required in order to obtain robust mutation data by NGS.