KRAS mutation: comparison of testing methods and tissue sampling techniques in colon cancer

Abstract

Treatment of colon carcinoma with the anti-epidermal growth factor receptor antibody Cetuximab is reported to be ineffective in KRAS-mutant tumors. Mutation testing techniques have therefore become an urgent concern. We have compared three methods for detecting KRAS mutations in 59 cases of colon carcinoma: 1) high resolution melting, 2) the amplification refractory mutation system using a bifunctional self-probing primer (ARMS/Scorpion, ARMS/S), and 3) direct sequencing. We also evaluated the effects of the methods of sectioning and coring of paraffin blocks to obtain tumor DNA on assay sensitivity and specificity. The most sensitive and specific combination of block sampling and mutational analysis was ARMS/S performed on DNA derived from 1-mm paraffin cores. This combination of tissue sampling and testing method detected KRAS mutations in 46% of colon tumors. Four samples were positive by ARMS/S, but initially negative by direct sequencing. Cloned DNA samples were retested by direct sequencing, and in all four cases KRAS mutations were identified in the DNA. In six cases, high resolution melting abnormalities could not be confirmed as specific mutations either by ARMS/S or direct sequencing. We conclude that coring of the paraffin blocks and testing by ARMS/S is a sensitive, specific, and efficient method for KRAS testing.

Figure 1

Schematic diagram for testing protocol. Tissue blocks were tested using two tissue block sampling methods and three mutation assays including direct sequencing, ARMS/scorpions, and high resolution melting.

Figure 2

Raw amplification curves for KRAS mutant tumor sample tested by the ARMS/S method. Each curve represents an amplification product of the same DNA template for one of eight specific primer sets. The curve on the farthest left is for wild-type control primers. The next curve to right represents amplification products for mutant template (GGT→GAT, G12D). The remaining curve is for the wild-type template. A positive result is determined by the cross point, the point on each curve where the slope of the curve becomes linear. There is a clear separation between mutant product (CP 29.6) and non-mutant product (CPs >38.5). The difference between CP for the wild-type control DNA and the mutant DNA (ΔCP) is small (2.4 cycles), indicating a mutation.

Figure 3

High resolution melting (HRM) difference point plot for KRAS. A: Normalized high resolution melt curves. PCR products are labeled with a fluorescent dye and the fluorescent signal is plotted as the temperature increases. Strand melting results in a decrease in fluorescent signal. B: The difference plot displays the melting curve of each sample subtracted from a reference curve to visually accentuate the melt curve differences and aid in genotype grouping. Products of three mutant templates are shown in red and pink. Products of wild-type templates are shown in blue. The upper (pink curve) represents the codon 12 mutation G12S, whereas the red curves represent amplification products for a single mutation at codon 12 (G12V) at two different concentrations. The latter two curves are similar shapes but have different heights due to variation in the quantity of PCR amplicon.

Figure 4

Mutation Surveyor readout for KRAS codon 12 mutation (GGT→GAT, G12D), forward primer set. Top tracing is for wild-type control. Mutant tracing is shown in the center. The chomatogram peak differences are calculated by the Mutation Surveyor program and plotted in the bottom tracing. Numbers refer to areas under the curve. The horizontal green line indicates the calculated threshold level for the mutation call. Visual inspection of chromatograms are required to indentify mutation peaks that fall below the threshold of detection of the software, as well as for false positives due to band compression or other sequencing artifacts.

Figure 5

Example of discrepancy between ARMS/S and direct sequencing. On the left (A), ARMS/S amplification curve for mutant tumor DNA (G12D) is clearly separated from that of non-mutant control DNA with ΔCP of 7.3, a value that is within the range indicating mutation. On the right (B) is a chromogram for direct sequencing results for KRAS codons 12 and 13. In frames 1 and 6, tracings of forward and reverse control DNA strands are shown. Tracings for forward and reverse test DNA strands are shown in frames 2 and 5. Frames 3 and 4 show difference plots for forward and reverse strands calculated by Mutation Surveyor software. There is a low peak (A) corresponding to the expected mutation, but the height of the peak is below the level background and cannot be confidently interpreted as a mutation.