Skip to main content

Targeted Next Generation Sequencing as a Reliable Diagnostic Assay for the Detection of Somatic Mutations in Tumours Using Minimal DNA Amounts from Formalin Fixed Paraffin Embedded Material.

de Leng WW, Gadellaa-van Hooijdonk CG, Barendregt-Smouter FA, Koudijs MJ, Nijman I, Hinrichs JW, Cuppen E, van Lieshout S, Loberg RD, de Jonge M, Voest EE, de Weger RA, Steeghs N, Langenberg MH, Sleijfer S, Willems SM, Lolkema MP

Publication: PLoS One (2016) • Vol 11 • Page e0149405PMID: 26919633View on PubMed

Purpose

This paper compared next generation sequencing (NGS) results in specimens that were formalin-fixed paraffin-embedded (FFPE) and those that were frozen. The detection of sequence variants using NGS, conventional sequencing or high resolution melting was also compared among and between FFPE and frozen specimens.

Conclusion

Compared to FFPE specimens, DNA from frozen specimens had a higher percentage of on-target reads, fewer specimens with less than 800x coverage, lower coverage of amplicons greater than 100 bp, and less variability in sequencing depth and percentage of mapped reads.  However, there were no significant differences in the percentage of mapped reads, the depth of coverage, or C to T or G to A ratio between FFPE and frozen specimens. The authors defined good quality specimens as those that had more than 800x coverage and more than 80% on-target reads. NGS was able to reliably detect variants that were present in a minimum of 20 reads in specimens that contained 10% tumor content with a coverage of 800x. Importantly, concordance for variant detection by NGS and conventional sequencing or high resolution melting analysis was 98.5%, but specimen type was not specified. Discordance between methods of analysis included  22 variants identified by Sanger sequencing or high resolution melting that were unable to be confirmed by NGS (19 were not tested and 3 were not found), another whose identification depended on the method analysis , and a new variant with a frequency of 7.5% identified by NGS.

Studies (1)

Purpose

This study compared NGS results among tumor and normal tissue specimens that were FFPE or frozen. The detection of sequence variants using NGS, conventional sequencing or high resolution melting was also compared among and between FFPE frozen specimens. NGS was performed on 96 FFPE tumor and 290 unmatched snap-frozen specimens using the Ion AmpliSeq Cancer Panel. NGS  was also conducted using the Ion AmpliSeq Cancer Panel on 227 FFPE tumor specimens and 23 normal or normal adjacent FFPE specimens with available mutation data. Among the 227 tumor specimens were 15 specimens taken from additional tumors within the same patient, of which 8 were clonal. DNA was extracted from 10 µm FFPE sections that contained at least 1 cm2 of tissue with a tumor content of 10% or greater using the Cobas method. DNA was extracted from frozen specimens that contained tumor content greater than 30% using NorDiag Arrow. DNA concentration was determined using Qubit and 10 ng of frozen or 20 ng of FFPE DNA was used for sequencing. Although colorectal cancer, non-small cell lung cancer and melanoma are mentioned as tissue sources, tissue-specific investigations were not performed.

Summary of Findings

DNA from frozen specimens produced 7.1% more reads that passed quality filters than DNA from FFPE specimens (P=0.0009). This difference was likely attributable to the higher percentage of on-target reads in sequenced frozen specimens compared to FFPE specimens (P=0.002). Overall, 7.7% of specimens were excluded from analysis due to insufficient coverage (< 800x), the majority of which (79%) also had an insifficient percentage of on-target reads (< 80%). The percentage of specimens for which coverage was insufficient (<800x) was greater for FFPE specimens than frozen (13% vs  6%).  There were no significant differences in the percentage of mapped reads, the depth of coverage, or C to T or G to A ratio between FFPE and frozen specimens; however, FFPE specimens had a lower coverage of amplicons that were longer than100 bp, more variability in sequencing depth and the percentage of mapped reads than frozen specimens. The authors defined good quality specimens as those that had more than 800x coverage and greater than 80% on-target reads.

Dilution of DNA from four frozen specimens displayed linearity for the frequency of the TP53 allele down to a dilution of 1% (R2=94.53%). Thus, NGS could reliably predicted variants when they occurred in at least 20 reads in specimens containing 10% tumor content with a coverage of 800x. Interestingly, the percentage of tumor content was not correlated with the frequency of the variant allele. Of 328 variants identified, 323 (98.5%) were concordant between NGS and conventional sequencing or high resolution melting analysis. For 2 specimens, variants in exon 8 of TP53 that were originally identified by Sanger sequencing were not observed by NGS. For  1 specimen a variant in TP53 exon 7 identified by Sanger sequencing was only confirmed by NGS with the use of Integrative Genomics Viewer (IGV) but not Torrent variant caller (TVC). A different variant that was believed by the authors to be false was identifid using TVC but was not observed using IGV. A final variant was identified with NGS and determined to have a frequency of 7.5%, but was not identifiable by HRM possibly due to low frequency. Additionally, 19 mutations identified by Sanger sequencing were not included on the AmpliSeq panel and thus could not be identified by NGS.

  • Biospecimens

    TissueColorectal
    TissueLung
    TissueSkin
  • Preservation Types

    FormalinFrozen
  • Diagnoses

    NeoplasticCarcinoma
    NeoplasticNot specified
    NeoplasticNormal Adjacent
    Normal
    Not specified
  • Platforms

    DNANext generation sequencing
    DNASNP assay
    DNADNA sequencing
  • Pre-analytical Factors

    ClassificationPre-Analytical FactorValue(s)
    Biospecimen PreservationType of fixation/preservation
    Formalin (buffered)
    Frozen
    Next generation sequencing SpecificTechnology platform
    High resolution melting
    Sanger sequencing