PLoS ONE (Jan 2015)

Amplicon sequencing of colorectal cancer: variant calling in frozen and formalin-fixed samples.

  • Johannes Betge,
  • Grainne Kerr,
  • Thilo Miersch,
  • Svenja Leible,
  • Gerrit Erdmann,
  • Christian L Galata,
  • Tianzuo Zhan,
  • Timo Gaiser,
  • Stefan Post,
  • Matthias P Ebert,
  • Karoline Horisberger,
  • Michael Boutros

DOI
https://doi.org/10.1371/journal.pone.0127146
Journal volume & issue
Vol. 10, no. 5
p. e0127146

Abstract

Read online

Next generation sequencing (NGS) is an emerging technology becoming relevant for genotyping of clinical samples. Here, we assessed the stability of amplicon sequencing from formalin-fixed paraffin-embedded (FFPE) and paired frozen samples from colorectal cancer metastases with different analysis pipelines. 212 amplicon regions in 48 cancer related genes were sequenced with Illumina MiSeq using DNA isolated from resection specimens from 17 patients with colorectal cancer liver metastases. From ten of these patients, paired fresh frozen and routinely processed FFPE tissue was available for comparative study. Sample quality of FFPE tissues was determined by the amount of amplifiable DNA using qPCR, sequencing libraries were evaluated using Bioanalyzer. Three bioinformatic pipelines were compared for analysis of amplicon sequencing data. Selected hot spot mutations were reviewed using Sanger sequencing. In the sequenced samples from 16 patients, 29 non-synonymous coding mutations were identified in eleven genes. Most frequent were mutations in TP53 (10), APC (7), PIK3CA (3) and KRAS (2). A high concordance of FFPE and paired frozen tissue samples was observed in ten matched samples, revealing 21 identical mutation calls and only two mutations differing. Comparison of these results with two other commonly used variant calling tools, however, showed high discrepancies. Hence, amplicon sequencing can potentially be used to identify hot spot mutations in colorectal cancer metastases in frozen and FFPE tissue. However, remarkable differences exist among results of different variant calling tools, which are not only related to DNA sample quality. Our study highlights the need for standardization and benchmarking of variant calling pipelines, which will be required for translational and clinical applications.