The efficacy of uracil DNA glycosylase pretreatment in amplicon-based massively parallel sequencing with DNA extracted from archived formalin-fixed paraffin-embedded esophageal cancer tissues.
Author(s): Serizawa M, Yokota T, Hosokawa A, Kusafuka K, Sugiyama T, Tsubosa Y, Yasui H, Nakajima T, Koh Y
Publication: Cancer Genet, 2015, Vol. 208, Page 415-27
PubMed ID: 26194062 PubMed Review Paper? No
Purpose of Paper
This paper evaluated the predictive power of a real-time PCR quality control (QC) assay on next generation sequencing performance for formalin-fixed paraffin-embedded (FFPE) specimens. Pretreating DNA isolated from FFPE specimens with uracil DNA glycosylase (UDG) was also examined to determine if it improved sequencing quality and reduced detection of false single nucleotide variants (SNVs). Effects of FFPE block storage on DNA integrity were also examined.
Conclusion of Paper
Of the 134 FFPE specimens with sufficient DNA, 101 had difference in cycle threshold values (ΔCT) from the control DNA in the real-time PCR based Illumina quality control assay of ΔCT≤2, indicating good quality. Specimens with a ΔCT≤2 had significantly higher quality control metrics (total reads, reads after quality-based trimming, mapped reads of targeted regions, proportion of mapped reads, and average coverage) than those with ΔCT>2, regardless of UDG pretreatment. Although UDG pretreatment did not significantly affect any of the quality control metrics, it did reduce variability noted for the quality metrics and strengthen the negative correlation between ΔCT and quality control metrics. Greater than 100X coverage was achieved for 126 of the 134 FFPE specimens, and 7 of the 8 specimens that did not yield 100X coverage had a ΔCT>4. The average storage duration of FFPE blocks was comparable among specimens classified by ΔCT (ΔCT ≤2 vs ΔCT>2).
The number of detectable SNVs was positively correlated with ΔCT, indicating that more sequencing artifacts are associated witha higher ΔCT. Importantly, C:G > T:A were the predominant mutations elevated in specimens with ΔCT≥1.55 compared to those with a ΔCT<1.55, but this increase was attenuated by UDG pretreatment. While 97% of SNVs detected in both untreated and UDG pretreated specimens were confirmed by sequencing in both directions, the majority of SNVs identified in only untreated or UDG pretreated specimens were false positive. The number of false positive SNVs was 63% lower in UDG pretreated than untreated specimens. UDG pretreatment reduced the number of C>T SNVs in CpG, CpA, CpC, and CpT dinucleotide sites, but a greater percentage of C>T SNVs remained in CpG dinucleotide sites than in CpA, CpC and CpT sites.
Studies
-
Study Purpose
This study evaluated the predictive power of a real-time PCR quality control assay on NGS performance using 135 archival FFPE esophageal cancer specimens. UDG pretreatment of isolated DNA was also investigated to determine if it improved sequencing quality and reduced detection of false single nucleotide variants (SNVs). The effect of FFPE block storage on DNA integrity was also examined. Specimens were fixed in 10% formalin and stored as FFPE blocks for 1.3-10.3 years before extraction, but specifics of fixation, embedding and storage were not provided. A punch (2 mm in diameter) that contained tumor cells was collected from each FFPE block and deparaffinized in room temperature xylene for 4 h. DNA was extracted using the QIAamp DNA FFPE Tissue Kit. DNA was quantified using Quant-iT PicoGreen dsDNA assay kit and integrity was assessed by real-time PCR using the Illumina FFPE QC Kit. DNA was treated with UDG or left untreated and then sequenced using the TruSeq Amplicon Cancer Panel.
Summary of Findings:
Quantifiable DNA was extracted from 134 of the 135 FFPE specimens evaluated. The average difference in cycle threshold values (ΔCT) from the control DNA in the Illumina QC assay was 1.0, with 101 of the 134 specimens considered to be of sufficient integrity (ΔCT≤2) for sequencing. The average storage duration was comparable among specimens Δ≤2 and those with ΔCT>2 (5.7 versus 6.4 years P=0.1813). When FFPE specimens were sequenced without UDG pretreatment greater than 100 X average coverage was obtained from 126 of the 134 specimens. Importantly, 7 of the 8 specimens that failed to generate 100X average coverage had ΔCT>4 per the QC assay. In comparison to specimens with ΔCT≤2, FFPE specimens with ΔCT>2 had significantly lower numbers of total reads (0.523 x 106 versus 1.37 x 106, P=0.004), reads after quality-based trimming (0.522 x 106 versus 1.396 x 106, P=0.0004), mapped reads of targeted regions (0.451 x 106 versus 1.332 x 106, P=0.002), and a lower proportion of mapped reads (81.74% versus 94.29%, P<0.0001), and lower average coverage (1.758 versus 5.434 reads/base P=0.002). Similarly when FFPE specimens underwent UDG pretreatment, all quality control metrics were significantly lower in specimens with ΔCT≤2 than those with ΔCT>2 (P<0.001, all). Both with and without UDG pretreatment, ΔCT was significantly and negatively correlated with the number of total reads (R=-0.4396, P<0.0001 and R=-0.7517, P<0.0001, respectively) , the number of reads after quality-based trimming (R=-0.4397, P<0.0001 and R=-0.7514, P<0.0001, respectively), the number of mapped reads of targeted regions (R=-0.4601, P<0.0001 and R=-0.7553, P<0.0001, respectively), the proportion of mapped reads (R=-0.6237, P<0.0001 and R=-0.5520, P<0.0001, respectively), and average coverage (R=-0.4673, P<0.0001 and R=-0.7566, P<0.0001, respectively). UDG pretreatment did not significantly affect any of the quality control metrics, but it did reduced their variability and strengthen negative correlations with ΔCT.
The number of SNVs detected was positively correlated with ΔCT in untreated (R=0.6939, P<0.0001) and UDG pretreated (R=0.5687, P<0.0001) FFPE specimens, indicating ΔCT is proportional to the number of sequencing artifacts detected. Importantly, C:G > T:A were the predominant mutations elevated in specimens with ΔCT≥1.55 compared to those with a ΔCT<1.55 in untreated (218 versus 37) and UDG pretreated (47 versus 37) specimens. The fact that the difference in the number of detected mutations between specimens classified by ΔCT value was much smaller when FFPE specimens underwent UDG pretreatment indicates that these artifacts can be attenuated by UDG treatment, but the lack of effect following UDG pretreatment among specimens with a ΔCT<1.55 indicates it is not necessarily in more intact specimens. Ninety-seven percent (97%) of SNVs detected in both untreated and UDG pretreated specimens were confirmed by sequencing both strands, but only 4% (7 of 158) and 7% (4 of 60)of SNVs detected only in untreated or UDG pretreated specimens, respectively, were confirmed. Importantly, the number of false positive SNVs was 63% lower among UDG pretreated specimens than untreated specimens (151 versus 56), with only 7 false negatives detected in specimens subjected to UDG pretreatment. Reductions in C>T SNVs when FFPE specimens with a ΔCT≥1.55 underwent UDG pretreatment occurred in CpG, CpA, CpC, and CpT dinucleotide sites, but a greater percentage of C>T SNVs remained in CpG dinucleotide sites in comparison to CpA, CpC and CpT dinucleotide sites(18.8% versus 3.6%, 5.9%, and 0.0%, respectively).
Biospecimens
Preservative Types
- Formalin
Diagnoses:
- Neoplastic - Carcinoma
Platform:
Analyte Technology Platform DNA Fluorometry DNA Real-time qPCR DNA Next generation sequencing Pre-analytical Factors:
Classification Pre-analytical Factor Value(s) Storage Storage duration 1.3-10.3 years
Next generation sequencing Specific Technology platform Real-time PCR
Next generation sequencing Specific Template modification UDG treated
untreated