Clinical validation of KRAS, BRAF, and EGFR mutation detection using next-generation sequencing

Abstract

Objectives: To validate next-generation sequencing (NGS) technology for clinical diagnosis and to determine appropriate read depth.

Methods: We validated the KRAS, BRAF, and EGFR genes within the Ion AmpliSeq Cancer Hotspot Panel using the Ion Torrent Personal Genome Machine (Life Technologies, Carlsbad, CA).

Results: We developed a statistical model to determine the read depth needed for a given percent tumor cellularity and number of functional genomes. Bottlenecking can result from too few input genomes. By using 16 formalin-fixed, paraffin-embedded (FFPE) cancer-free specimens and 118 cancer specimens with known mutation status, we validated the six traditional analytic performance characteristics recommended by the Next-Generation Sequencing: Standardization of Clinical Testing Working Group. Baseline noise is consistent with spontaneous and FFPE-induced C:G→T:A deamination mutations.

Conclusions: Redundant bioinformatic pipelines are essential, since a single analysis pipeline gave false-negative and false-positive results. NGS is sufficiently robust for the clinical detection of gene mutations, with attention to potential artifacts.

Keywords: BRAF; Deamination; EGFR; KRAS; Next-generation sequencing; Read depth; Validation.

Figure 1

Analysis pipeline. Bioinformatics for point mutation detection using three parallel analysis tools: Torrent Variant Caller, Ion Reporter (Life Technologies, Carlsbad, CA), and Integrative Genomics Viewer (IGV; Broad Institute, Boston, MA). BAM, binary sequence alignment/map; COSMIC, Catalog of Somatic Mutations in Cancer.

Figure 2

Background noise. For each of the common mutations within KRAS (codons 12 and 13), BRAF (V600E), and EGFR (T790M and L858R) genes among the 16 formalin-fixed, paraffin-embedded (FFPE) lymph node control first library (A), a second FFPE library (B), and peripheral blood (C) specimens, the mean plus 3 SD is plotted. Note that the patterns of quiet bases and noisy bases in A and B are qualitatively similar and that background signal in FFPE is substantially higher than in the peripheral blood. Down arrows designate consistently quiet mutations. Asterisks indicate particularly noisy mutations that have arisen due to a C:G→T:A mutation (resulting in either a C→T or a G→A on the sense strand).

Figure 3

Required depth of coverage. Depth of coverage required as a function of the number of genomes analyzed and the percentage of cancer cells in the sample, at a diagnostic sensitivity of 99% (A) and 99.9% (B). The effect of formalin-fixed, paraffin-embedded fixation is modeled as a decrease in the number of functional genomes analyzed. Lines that terminate (eg, arrow for 30 genomes in A) do so because the required sensitivity cannot be achieved, as the percentage of cancer cells within the tumor decreases.

Figure 4

Accuracy of next-generation sequencing (NGS). Correlation of percent mutant allele of the KRAS (A) and BRAF (B) genes determined by the NGS platform (Personal Genome Machine [PGM] %) compared with the “gold-standard” pyrosequencing (Pyro %). A, y = 0.9786x – 0.0309; R² = 0.93421. B, y = 0.9819x – 2.2742; R² = 0.97221.

Figure 5

Precision between next-generation sequencing runs. Assessed by comparing the percent mutant allele (A) and allele frequencies of single-nucleotide polymorphism rs1050171 (B) from replicate runs. A, y = 1.0778x – 1.363; R² = 0.98916. B, y = 0.9946x – 0.603; R² = 0.99424.