A comparison of four methods for detecting KRAS mutations in formalin-fixed specimens from metastatic colorectal cancer patients.

There is currently no standard method for the detection of Kirsten rat sarcoma viral oncogene homolog (KRAS) mutation status in colorectal tumors. In the present study, we compared the KRAS mutation detection ability of four methods: direct sequencing, Scorpion-ARMS assaying, pyrosequencing and multi-analyte profiling (Luminex xMAP). We evaluated 73 cases of metastatic colorectal cancer (mCRC) resistant to irinotecan, oxaliplatin and fluoropyrimidine that were enrolled in an all-case study of cetuximab. The KRAS mutation detection capacity of the four analytical methods was compared using DNA samples extracted from tumor tissue, and the detection success rate and concordance of the detection results were evaluated. KRAS mutations were detected by direct sequencing, Scorpion-ARMS assays, pyrosequencing and Luminex xMAP at success rates of 93.2%, 97.3%, 95.9% and 94.5%, respectively. The concordance rates of the detection results by Scorpion-ARMS, pyrosequencing and Luminex xMAP with those of direct sequencing were 0.897, 0.923 and 0.900 (κ statistics), respectively. The direct sequencing method could not determine KRAS mutation status in five DNA samples. Of these, Scorpion-ARMS, pyrosequencing and Luminex xMAP successfully detected three, two and one KRAS mutation statuses, respectively. Three cases demonstrated inconsistent results, whereby Luminex xMAP detected mutated KRAS in two samples while wild-type KRAS was detected by the other methods. In the remaining case, direct sequencing detected wild-type KRAS, which was identified as mutated KRAS by the other methods. In conclusion, we confirmed that Scorpion-ARMS, pyrosequencing and Luminex xMAP were equally reliable in detecting KRAS mutation status in mCRC. However, in rare cases, the KRAS status was differentially diagnosed using these methods.