Comparison of Accuracy of Whole-Exome Sequencing with Formalin-Fixed Paraffin-Embedded and Fresh Frozen Tissue Samples.
Author(s): Oh E, Choi YL, Kwon MJ, Kim RN, Kim YJ, Song JY, Jung KS, Shin YK
Publication: PLoS One, 2015, Vol. 10, Page e0144162
PubMed ID: 26641479 PubMed Review Paper? No
Purpose of Paper
This paper compared DNA integrity, next generation sequence quality, and somatic mutation identification in matched formalin-fixed and paraffin-embedded (FFPE) and snap-frozen tumor specimens of an unspecified type.
Conclusion of Paper
Compared to case-matched snap-frozen specimens and blood, DNA from FFPE tumor specimens was clearly degraded as determined by agarose gel electrophoresis and resulted in lower numbers of unique reads, percentages of properly mapped reads, covered target regions, on-target average depths, and 100 bp flanking region depths as well as higher percentages of unmapped reads and discordantly mapped reads. However, frozen specimens had a longer average insert length resulting in fewer double sequenced bases and a lower percentage of soft clipped bases, but a higher percentage of off-target bases than FFPE specimens. There was no difference in nucleotide quality between mapped and unmapped reads among FFPE specimens, but the number of low quality reads (≤20 on phred scale) was 2-fold higher in unmapped than mapped reads among frozen specimens, indicating an alteration due to fixation may have occurred in FFPE specimens. Discrepancies in sequencing results, specifically changes from C to T or G to A, occurred more frequently among FFPE than frozen specimens. The frequency of detection of somatic mutations was up to 10-fold higher in FFPE than in matched frozen specimens, but as the log odds in tumor score (LODT) increased to 50 concordance of somatic mutations in FFPE and frozen specimens reached 80% . Thus the authors recommend an LODT of greater than 50 for FFPE specimens aligned using MuTect.
Studies
-
Study Purpose
This paper compared DNA integrity, next generation sequence quality, and somatic mutation identification in matched formalin-fixed and paraffin-embedded (FFPE) and snap-frozen tumor specimens of an unspecified type. Paired specimens of 3 tumors were formalin-fixed for 24 h and paraffin-embedded or snap-frozen in liquid-nitrogen and stored at -80˚C. A matched blood specimen was available for one of these one tumors. Specimens from an additional patient with a reoccurring tumor were formalin-fixed in 2007 and 2009 and snap-frozen in liquid nitrogen in 2010. DNA was extracted using non-specified methods, fragmented using Covaris, and exome capture was performed using NimbleGen exome 2.1M array and SureSelect All Human exon V5. DNA was then sequenced on Illumina HiSeq 2000 and aligned to the UCSChg19 assembly.
Summary of Findings:
Compared to case-matched snap-frozen specimens and blood, DNA from FFPE tumor specimen was clearly degraded as determined by agarose gel electrophoresis. Compared to DNA from case-matched frozen specimens, DNA from FFPE specimens had lower numbers of unique reads (106-127 million versus 154-181 million), percentages of properly mapped reads (87.4-93.8% versus 98.4-98.7%), covered target regions (96-98% versus 97-99%), on-target average depth (62-110x versus 105-152x), and 100 bp flanking region depth (18-59x versus 55-68x) as well as higher percentages of unmapped reads (0.3-0.9% versus 2.6-7%) and discordantly mapped reads (0.4-1.2% versus 3.3-5.6%). However, snap-frozen specimens had a higher percentage of off-target bases than FFPE specimens (48-63% versus 32-42%), which was attributed to the longer insert length for frozen (150-200 bp versus 100 bp). The longer inserts observed in blood and frozen specimens, in comparison to FFPE specimens, resulted in lower percentages of double sequenced bases (12.22%, 5.28% and 29.28%, respectively) and soft clipped bases (11.56%, 4.65% and 28.66%). The nucleotide quality of more than 80% of mapped reads from FFPE and frozen specimens was ≥30 on the phred scale. Interestingly, there was no difference in nucleotide quality between mapped and unmapped reads for FFPE specimens (4.8% versus 4.3%), but the number of low quality reads (≤20 on phred scale) was 2-fold higher in unmapped than mapped reads for frozen specimens (9.5% versus 3.3%). Thus the authors state that while for frozen specimens low quality NGS data results from mismatches, for FFPE specimens mapping failure is not just from poor quality reads but is also due to base alterations caused by fixation. Discrepancy rates at homozygous sites were low for control blood or frozen specimens (100-300 alterations per MB) indicating a low level of DNA damage or sequencing errors. Discrepancies in sequencing results, specifically changes from C to T or G to A, occurred more frequently in FFPE (0.04-0.15% and 0.04-0.13%, respectively) than frozen specimens (0.01-0.02% both), but the frequency of most other discrepancies was 0.00-0.02% in both frozen and FFPE specimens. The frequency of somatic mutation detection was up to 10-fold higher in FFPE than in matched frozen specimens, but as the LODT increased to 50, concordance of somatic mutations reached 80%. Thus the authors recommend an LODT of greater than 50 for FFPE specimens aligned using MuTect.
Biospecimens
Preservative Types
- Frozen
- Formalin
Diagnoses:
- Neoplastic - Not specified
Platform:
Analyte Technology Platform DNA Next generation sequencing DNA Electrophoresis Pre-analytical Factors:
Classification Pre-analytical Factor Value(s) Biospecimen Preservation Type of fixation/preservation Formalin (buffered)
Snap frozen