KRAS genotyping of paraffin-embedded colorectal cancer tissue in routine diagnostics: comparison of methods and impact of histology

Abstract

KRAS mutation testing before anti-epidermal growth factor receptor therapy of metastatic colorectal cancer has become mandatory in Europe. However, considerable uncertainty exists as to which methods for detection can be applied in a reproducible and economically sound manner in the routine diagnostic setting. To answer this question, we examined 263 consecutive routine paraffin slide specimens. Genomic DNA was extracted from microdissected tumor tissue. The DNA was analyzed prospectively by Sanger sequencing and array analysis as well as retrospectively by melting curve analysis and pyrosequencing; the results were correlated to tissue characteristics. The methods were then compared regarding the reported results, costs, and working times. Approximately 40% of specimens contained KRAS mutations, and the different methods reported concordant results (kappa values >0.9). Specimens harboring fewer than 10% tumor cells showed lower mutation rates regardless of the method used, and histoanatomical variables had no influence on the frequency of the mutations. Costs per assay were higher for array analysis and melting curve analysis when compared with the direct sequencing methods. However, for sequencing methods equipment costs were much higher. In conclusion, Sanger sequencing, array analysis, melting curve analysis, and pyrosequencing were equally effective for routine diagnostic KRAS mutation analysis; however, interpretation of mutation results in conjunction with histomorphologic tissue review and on slide tumor tissue dissection is required for accurate diagnosis.

Figure 1

Different detection methods for KRAS mutations in codon 12/13. A and B: Sanger sequencing. Example of a wild-type (A) and a mutated (B) case (p.G12D). C and D: Pyrosequencing. Example of a wild-type (C) and a mutated (D) case (p.G12D). E–H: Array analysis. Examples of a wild-type case (E) and cases harboring a p.G13D (F), a p.G12D (G), and a p.G12C (H) mutation are presented. I and J: Melting curve analysis. Examples of a wild-type (I) and a mutated (J) case (p.G13D). Arrows indicate location of each mutation.

Figure 2

Percentage of mismatches for all detection methods. The percentage was calculated as the number of cases with discordant results (when compared with the other three assays) divided by the total number of tests performed per method.

Figure 3

Tissue characteristics and frequency of mutation detection as determined by Sanger sequencing (n = 260).