Cancer Diagnosis Program | Biorepositories and Biospecimen Research Branch

Search BRD Papers

Comparison of whole transcriptome sequencing of fresh, frozen, and formalin-fixed, paraffin-embedded cardiac tissue.

Author(s): Jacobsen SB, Tfelt-Hansen J, Smerup MH, Andersen JD, Morling N

Publication: PLoS One, 2023, Vol. 18, Page e0283159

PubMed ID: 36989279 PubMed Review Paper? No

Purpose of Paper

Conclusion of Paper

Studies

1. Study Purpose

   This paper compared RNA quality and gene expression profiles of protein-coding mRNAs, mitochondrial RNA (mtRNA), and long non-coding RNA (lncRNA) among case-matched fresh, frozen (-80°C), and FFPE cardiac tissue by whole transcriptome sequencing (WTS); concordance between RNA and DNA variants detected in fresh, frozen, and FFPE tissue was also determined. Biopsies of cardiac muscle (from the right atrial appendage) were collected from 10 patients undergoing cardiac surgery and divided into three pieces; one piece immediately underwent RNA and DNA extraction (fresh), another was frozen at -80°C and stored for 20-40 d, and the third was fixed in 4% buffered formaldehyde within 22-41 min of collection.  FFPE tissues were fixed at room temperature for 23-97 h, dehydrated, cleared, and paraffin impregnated using an automated Tissue-Tek processer, embedded in paraffin and stored at room temperature for 20-42 d. Total RNA was extracted from fresh and frozen tissue with the RNeasy Fibrous Tissue Mini Kit that included on-column DNase treatment. Total RNA was extracted from two 20 µm-thick FFPE sections per block with the RNeasy FFPE Kit and on-column DNase treatment after deparaffinization with xylene. RNA extractions were performed in triplicate for tissue from 9 of the 10 patients evaluated; a single RNA extraction was performed for each time point for one individual due to the procurement of a limited amount of tissue.  RNA was quantified by Qubit and RNA quality was assessed by RIN and DV200 using a 2100 Bioanalyzer System.  Whole transcriptome libraries were prepared with the SMARTer Stranded Total RNA-seq Kit v3 -Pico Input Mammalian Kit and 10 ng of total RNA; RNA from fresh and frozen specimens, but not FFPE, was fragmented. cDNA library quality was assessed using a 2100 Bioanalyzer with the Bioanalyzer High Sensitivity DNA Assay and quantity was determined with the KAPA Library Quantification Kits-Complete Kit and a 7500 Real-time PCR machine.  Samples were pair-end sequenced on a NovaSeq 6000 instrument. To determine concordance between RNA and DNA variants of differentially preserved specimens, DNA was extracted from frozen tissue using the DNeasy Blood & Tissue Kit and SNP typing was performed with 400 ng of DNA and the Illumina Infinium Omni5-4 BeadChip Kit.

   Summary of Findings:

   RNA isolated from FFPE tissue was highly degraded (median RIN = 2.5, median DV200 = 48%) and generated a higher proportion of intronic and intergenic reads compared to RNA from fresh (median RIN = 8.1, median DV200 = 97%) or frozen tissues (median RIN =8.0, median DV200=96%) counterparts. Gene expression patterns were strongly correlated among replicates (RNA extraction performed in triplicate) for each preservation method: fresh (ρ=0.89), frozen (-80°C; ρ=0.87), and FFPE (ρ=0.83).

   The proportion of sequenced reads was consistent among patients and similar between RNA isolated from fresh and frozen cardiac tissue specimens for reads that aligned to regions of active transcription (gene-annotated reads; fresh: mean 53%, frozen: mean 44%) and multi-mapping reads (fresh= mean 31%, frozen= mean 35%).  Conversely, RNA isolated from FFPE tissue generated comparatively fewer gene-annotated (range: 10-40%) and multi-mapping reads, a higher proportion of reads that were unmapped (because they were too short or aligned to intronic/intergenic regions), and more inter-individual variability compared to what was observed among fresh and frozen counterparts.  Scatterplot analysis of the percent of gene-annotated reads relative to the total reads sequenced for individual FFPE tissues did not reveal an effect of fixation time (23-97 h) or FFPE block storage duration (20-42 d) on the percentage of gene-annotated reads.

   The GC content of aligned reads was modestly lower for RNA isolated from FFPE tissue than case-matched fresh and frozen specimens. There was a positive association between the proportion of gene-annotated reads and both RIN and DV200 scores in fresh and frozen specimens, as specimens with higher RIN and DV200 scores tended to have a higher proportion gene-annotated reads; the relationship was less apparent for RIN in FFPE specimens.  While protein-coding genes accounted for the majority of gene-annotated reads for all preservation methods (fresh=76%, frozen =73%, FFPE=48%), FFPE specimens yielded higher proportions of mtRNA and lncRNA relative to the other specimen types.  MDS plots for protein-coding RNAs, mtRNAs, and lncRNAs, clearly separated specimens by preservation method and not by individual, suggesting that the differences associated with preservation method are larger than those between individuals.

   Expression levels of protein-coding transcripts were strongly correlated among RNA from fresh, frozen, and FFPE specimens (ρ<0.94), but lncRNA expression profiles were less correlated among the three preservation methods (ρ>0.53); much of the variation that was present for both protein-coding and lncRNA was attributed to genes with a low level of expression. mtRNA gene expression profiles were strongly correlated among RNA isolated from fresh and frozen specimens (ρ<0.98), but RNA from FFPE tissue had false overexpression of a subset of mtRNA transcripts relative to RNA from fresh and frozen counterparts that resulted in a FFPE cluster on multidimensional scaling plots. Attempts to filter out genes with a low level of expression did not entirely abolish clustering by preservation method, but it did reduce overall variation associated with it and allow separation by individual patients.

   Data from WTS analysis of exon regions of 31 genes associated with cardiac disease demonstrated high concordance with DNA variants represented on a SNP array for fresh, frozen, and FFPE tissue specimens, although RNA isolated from FFPE tissue did have a significantly higher number of discordant variant calls (p<0.05) than fresh and frozen tissue and a larger proportion of variants with an insufficient read depth (<75).

   Biospecimens

   • Tissue - Heart

   Preservative Types

   • Formalin
   • None (Fresh)
   • Frozen

   Diagnoses:

   • Not specified

   Platform:

   Analyte	Technology Platform
   RNA	Fluorometry
   RNA	Next generation sequencing
   DNA	SNP assay
   RNA	Real-time qRT-PCR
   RNA	Automated electrophoresis/Bioanalyzer

   Pre-analytical Factors:

   Classification	Pre-analytical Factor	Value(s)
   Biospecimen Preservation	Type of fixation/preservation	None (fresh) Frozen Formalin (buffered)
   Biospecimen Preservation	Time in fixative	23-97 h
   Storage	Storage duration	20-42 d
   Automated electrophoresis/Bioanalyzer Specific	Quality metrics	RIN DV200

   View more