Robust gene expression and mutation analyses of RNA-sequencing of formalin-fixed diagnostic tumor samples.
Author(s): Graw S, Meier R, Minn K, Bloomer C, Godwin AK, Fridley B, Vlad A, Beyerlein P, Chien J
Publication: Sci Rep, 2015, Vol. 5, Page 12335
PubMed ID: 26202458 PubMed Review Paper? No
Purpose of Paper
This paper compared next-generation RNA sequencing (RNAseq) results obtained from case-matched formalin-fixed paraffin-embedded (FFPE) and frozen ovarian tumor specimens and normal fallopian tube specimens. Mutations identified by RNAseq were confirmed with whole exome sequencing in the same specimen cohort.
Conclusion of Paper
Although RNAseq quality control results were generally comparable between FFPE and frozen specimens, FFPE specimens displayed a peak GC content of 53%. FFPE specimens also displayed fewer exonic reads than frozen specimens, but the 3' bias observed in both FFPE and frozen specimens was independent of transcript length in FFPE specimens. The authors attribute the differences above to the mRNA enrichment performed in frozen but not FFPE specimens.
Case-matched FFPE and frozen specimens clustered by patient, and gene expression was very strongly correlated regardless of analysis by RNAseq or NanoString. Fold-change differences in expression between case-matched FFPE and frozen specimens were strongly correlated between ovarian tumor and normal fallopian tube specimens (R=0.744), with 63.3% of upregulated and 55.9% of downregulated genes in common between preservation methods.
Only 64% of mutations identified by RNAseq in FFPE specimens were confirmed in DNA from case-matched frozen specimens by whole exome sequencing, but more than 99% of mutations identified by RNAseq in frozen specimens were concordant with those identified by whole exome sequencing. However, after C>T or G>A substitutions with a variant allele frequency of less than 50% were filtered out, 91.14% of discordant variants were removed while 99.67% of concordant variants were retained.
Studies
-
Study Purpose
In this study, gene expression data generated with RNAseq was compared with NanoString in case-matched FFPE and frozen specimens. The specimen cohort consisted of six ovarian tumors and seven normal fallopian tube specimens. RNA was extracted from FFPE sections (5 µm) after deparaffinization with the FFPE All-Prep kit, and from frozen specimens with Trizol. Prior to library construction with the Illumina TruSeq Stranded Total RNA LT kit, ribosomal RNA was depleted in FFPE specimens with the RiboZero kit while mRNA was enriched in frozen specimens using Oligo-dT bound to magnetic beads. The authors state that mRNA enrichment was not possible with FFPE specimens.
Summary of Findings:
Although RNAseq quality control results were generally comparable between FFPE and frozen specimens, a peak in GC content of 53% was present in FFPE specimens but not frozen specimens; this peak in GC content was localized to reads within mRNA regions. FFPE specimens also had a greater number of overall reads but fewer reads mapped to exons than frozen specimens. The authors hypothesize that the increased number of intronic reads in FFPE specimens likely reflects a bias introduced by mRNA enrichment of frozen, but not FFPE, specimens. Further, there was a small bias in 3’ transcript coverage in both FFPE and frozen specimens. While that 3’ bias was independent of transcript length and expression level in FFPE specimens, it was dependent on transcript length (increased bias in transcripts greater than 4 kb) and RNA quality (no effect of transcript length was observed in the most intact specimen) in frozen specimens. The authors attribute the 3' bias observed in frozen specimens to the Oligo-dT priming method used for mRNA enrichment.
When data obtained with RNAseq and Nanostring were compared, case-matched FFPE and frozen specimens clustered within a technology by patient for all 7 patients, but data from 6 of 7 patients clustered by analysis method (RNAseq or Nanostring). Further, expression in case-matched FFPE and frozen specimens was very strongly correlated with both RNAseq and NanoString. Fold-change differences in expression between case-matched FFPE and frozen specimens were strongly correlated with gene expression in normal fallopian tube specimens (R=0.744).
Of the 742 genes that were upregulated in frozen tumor compared to normal specimens, 63.3% (470 genes) were also upregulated in case-matched FFPE specimens; similarly, of the 270 genes that were downregulated in frozen tumor compared to normal specimens, 55.9% (151 genes) were also downregulated in case-matched FFPE specimens.. Further, approximately 50% of the pathways that differed between tumor and normal specimens using frozen specimens also differed between FFPE specimens. Importantly, the magnitude of difference between tumor and normal specimens was greater for low expression transcripts among specimens that were frozen compared to those that were formalin-fixed and paraffin-embedded, indicating that short, low abundant transcripts are more likely to be lost during FFPE processing. Interestingly, the statistical strength of the differences in expression between tumor and normal specimens were only marginally different between frozen and FFPE specimens, and was not affected by the mean expression of the transcript.
Biospecimens
Preservative Types
- Frozen
- Formalin
Diagnoses:
- Normal
- Neoplastic - Carcinoma
Platform:
Analyte Technology Platform RNA DNA microarray RNA Next generation sequencing Pre-analytical Factors:
Classification Pre-analytical Factor Value(s) Biospecimen Preservation Type of fixation/preservation Formalin (buffered)
Frozen
Next generation sequencing Specific Length of gene fragment <4 kbp
>4 kbp
Next generation sequencing Specific Template modification mRNA enriched with OligodT
rRNA depleted
Next generation sequencing Specific Technology platform NanoString
Biospecimen Acquisition Biospecimen location Normal fallopian tube
Ovarian tumor
-
Study Purpose
This study investigated potential effects of formalin-fixation on the concordance of genotyping by RNAseq and whole exome sequencing. Specimens included case-matched FFPE and frozen specimens from six ovarian tumors. RNA was extracted from FFPE sections (5 µm) after deparaffinization using the FFPE All-Prep kit and from frozen specimens using Trizol. Before library construction using Illumina TruSeq Stranded Total RNA LT kit, frozen specimens underwent mRNA enrichment with oligo-dT bound to magnetic beads, while ribosomal RNA was depleted in FFPE specimens using the RiboZero kit. Libraries were prepared using the Illumina TruSeq Stranded Total RNA LT kit and the Nextera DNA Sample Prep Kit. DNA was extracted from frozen specimens using DNeasy.
Summary of Findings:
Only 64% of mutations identified by RNAseq in FFPE specimens were confirmed by whole exome sequencing of DNA from case-matched frozen specimens, however more than 99% of mutations identified by RNAseq in frozen specimens were concordant with whole exome sequencing. The majority of discordant calls in FFPE specimens were due to C>T or G>A substitutions; when these substitutions were filtered out based on ≥50% variant allele frequencies 91.14% of discordant variants were removed and 99.67% of concordant variants were retained.
Biospecimens
Preservative Types
- Frozen
- Formalin
Diagnoses:
- Neoplastic - Carcinoma
Platform:
Analyte Technology Platform RNA Next generation sequencing DNA Next generation sequencing Pre-analytical Factors:
Classification Pre-analytical Factor Value(s) Biospecimen Preservation Type of fixation/preservation Formalin (buffered)
Frozen
Next generation sequencing Specific Technology platform RNAseq
Whole exome sequencing