Cancer Diagnosis Program | Biorepositories and Biospecimen Research Branch

Search BRD Papers

A Comparison of RNA-Seq Results from Paired Formalin-Fixed Paraffin-Embedded and Fresh-Frozen Glioblastoma Tissue Samples.

Author(s): Esteve-Codina A, Arpi O, Martinez-García M, Pineda E, Mallo M, Gut M, Carrato C, Rovira A, Lopez R, Tortosa A, Dabad M, Del Barco S, Heath S, Bagué S, Ribalta T, Alameda F, de la Iglesia N, Balaña C

Publication: PLoS One, 2017, Vol. 12, Page e0170632

PubMed ID: 28122052 PubMed Review Paper? No

Purpose of Paper

Conclusion of Paper

Studies

1. Study Purpose

   This study compared the RNA yield, RNA integrity, library metrics, and NGS results in matched FFPE and frozen glioblastoma specimens. Specimens from eleven primary glioblastomas were obtained, but no details of specimen handling were included. RNA was extracted using the RNeasy mini kit and quantified by spectrophotometry. RNA integrity was determined with a bioanalyzer. Libraries were constructed using the TruSeq Stranded Total RNA kit with a shortened fragmentation time. Libraries were evaluated on a bioanalyzer and sequenced using the TruSeq SBS kit v3-HS. 

   Summary of Findings:

   Of the eleven frozen specimens, nine had RNA of sufficient yield (>50 ng/µL or >1 µg total RNA) but only four of those had RNA of satisfactory integrity (RIN >6.0) for RNA-seq. The RNA from the four matched FFPE specimens had RINs ≤2.6. RNA.  The storage duration had no apparent effect on the RIN or RNA yield from either frozen or FFPE specimens. Compared to RNA from frozen specimens, RNA from FFPE specimens had shorter paired-end inner distances between read pairs (P<0.001) and lower and more variable transcript integrity numbers (median of 72-73 versus median of 1-60, P<0.000); indicating high and variable degrees of RNA fragmentation. The degradation was most abundant when GC content was <35% or >55% and at the 5’ end of the transcript. As expected, RNA from the FFPE specimens compared to RNA from frozen specimens yielded a significantly higher percentage of duplicate reads (P=2.2 x 10^^-16), a higher percentage of uniquely mapped reads (P=2.2 x 10^^-16, all), a lower percentage of unmapped reads (P=2.2 x 10^^-16, all), a lower percentage of split mapped reads (P=2.2 x 10^^-16, all), a higher percentage of intron mapped reads (P=2.2 x 10^^-16, all), and more G>A and C>T transitions. For the three highly degraded FFPE RNA specimens, fewer genes used 25% of the sequencing effort than for their matched frozen specimen. For the most degraded of these, one gene accounted for 25% of the sequencing effort and only had ~8000 genes total (compared to 25000-30000 in matched frozen). The increased number of reads mapping to introns reflects a relative increase in nuclear RNA (pre-mRNA or lincRNA) relative to spliced RNA in the cytosol, the percent spliced index, and percentage reads mapping to lincRNA and snRNA was higher in FFPE specimens than frozen specimens indicate a protective effect of nuclear localization. Approximately 90% of annotated pair-end reads mapped to protein-coding genes, but the percentage was slightly higher for RNA from frozen specimens than FFPE specimens (P=2.2 x 10^^-16, all). In the most degraded FFPE specimen, 90% of the reads were mapped to the mitochondria rRNA and only 5% to protein-coding RNA, indicating that the mitochondria may be protective. A total of 2,133 genes were found to have differential expression at a false discovery rate of less than 0.05, of which 908 protein-coding and 26 non-coding were under-represented in the FFPE specimen compared to the matched frozen specimen and 169 protein-coding and 1030 non-coding were overrepresented in the FFPE specimen. The RNAs that were over-represented in FFPE specimens were either non-coding genes found in the nucleus or genes transcribed in the mitochondria. Despite these differences, the gene expression was strong between three matched FFPE and frozen specimens (R2=0.8679, 0.8987, and 0.907), but the correlation in gene expression was weak for the highly degraded FFPE specimen (R2=0.3579).  Three of the four FFPE specimens were assigned a glioblastoma molecular subtype, of which, two matched the assigned subtype based on the frozen specimen. However, the predictive ability was low with a mean cross-validated error of 33.7%, which the authors state may be due to the heterogeneity of glioblastoma and may account for discrepancy of one FFPE/frozen pair.

   Biospecimens

   • Tissue - Brain

   Preservative Types

   • Frozen
   • Formalin

   Diagnoses:

   • Neoplastic - Other

   Platform:

   Analyte	Technology Platform
   RNA	Automated electrophoresis/Bioanalyzer
   RNA	Spectrophotometry
   RNA	Next generation sequencing

   Pre-analytical Factors:

   Classification	Pre-analytical Factor	Value(s)
   Storage	Storage duration	Collected in 2006 Collected in 2007 Collected in 2008 Collected in 2009 Collected in 2010 Collected in 2011
   Biospecimen Preservation	Type of fixation/preservation	Frozen Formalin (buffered)

   View more