Deleterious effects of formalin-fixation and delays to fixation on RNA and miRNA-Seq profiles.
Author(s): Jones W, Greytak S, Odeh H, Guan P, Powers J, Bavarva J, Moore HM
Publication: Sci Rep, 2019, Vol. 9, Page 6980
PubMed ID: 31061401 PubMed Review Paper? No
Purpose of Paper
This paper investigated the effects of formalin fixation and delay to fixation (DTF) on RNA and microRNA (miRNA) profiles determined by next-generation sequencing.
Conclusion of Paper
Compared to frozen specimens, formalin-fixed paraffin-embedded (FFPE) specimens had more reads that corresponded to the intronic/exonic/untranslated regions (55-65% versus 15-25%). The majority of the differences in expression pattern were attributable to tissue type, followed by preservation method and contributing institution with little effect of the duration of DTF. Although the overall gene expression patterns showed little effect of preservation method, 82% of specimens had higher expression of select genes in the formalin-fixed specimens than the matched frozen specimens. The formalin-fixation gene signature was less common in colon than ovarian and kidney specimens but was not affected by the delay to fixation. The relative density of material in the coding and UTR regions was more strongly correlated between FFPE and frozen specimen than the relative density of material in the intronic regions, thus complicating the estimation of splice variation in FFPE specimens.
Principle component analysis showed no bias in the miRNA expression profile due to preservation protocol in the top five principal components. miRNA detection levels were negatively correlated with DTF and variability increased with increasing DTF, resulting in less power to detect differential expression with increased DTF.
Studies
-
Study Purpose
This study investigated the effects of formalin-fixation and DTF on RNA and miRNA profiles as determined by next-generation sequencing. As part of the National Cancer Institute’s Biospecimen Pre-analytical Variables (BPV) program, 30 tumor specimens including 10 renal clear cell carcinomas (kidney), 10 serous ovarian carcinomas (ovary), and 10 colon adenocarcinomas (colon) were collected at four medical centers. Each specimen was divided into six samples. One specimen was snap-frozen in liquid nitrogen vapor within 1 h of resection. The remaining samples were stored in a humidified chamber at room temperature for 1, 2, 3, or 12 h; fixed in 10% neutral buffered formalin (NBF) for 12 hours; and then paraffin-embedded. FFPE blocks were stored for 3 to 18 months before sectioning. RNA was extracted from three 10 µm sections using the QIAsymphony kits. Sequencing libraries were constructed using the Illumina TruSeq Stranded Total RNA Sample Kit and the Illumina's TruSeq Small RNA Sample Preparation Kit and sequenced on a HiSeq 2500.
Summary of Findings:
FFPE specimens had more reads that corresponded to the intronic/exonic/untranslated regions (55-65% versus 15-25%). This was observed for all genes and all specimens. Overall, the gene expression patterns showed little effect of DTF or preservation method as the lists of differentially expressed genes between tumor types displayed more than 70% overlap between different preservation methods and delays. A closer examination found that 82% of specimens had higher expression of specific genes in the formalin-fixed specimens than the matched frozen specimens. Interestingly, these genes did not represent genes associated with stress and hypoxia but instead with chromatin packaging and nucleosome organization. The formalin-fixation gene signature was less common in colon than ovarian and kidney specimens but was not affected by DTF. Use of linear models found that 34.1-41.6% of genes were affected by tissue type, 17.2-20.6% by the fixation method (snap-frozen versus FFPE), 2.5-2.6% by the institution that did collection. and only 0.1-0.2% by the duration of DTF. The relative density of material in the coding and UTR regions was very strongly correlated between FFPE and frozen specimen (r=0.97), but only a strong correlation between the relative density of material in the intronic region (r=0.74), thus complicating the estimation of splice variation in FFPE specimens.
All 150 specimens were of sufficient quality for miRNA analysis. Principle component analysis showed no clear bias due to preservation protocol in the top five principal components or clustered expression profiles based on tissue type. miRNA detection levels were negatively correlated with DTF with a decrease in counts of the upper quartile by approximately 15% observed after a 12 h DTF compared to frozen specimen. A corresponding increase in library material greater than 25 bases in length indicated that the decreased counts may be due to dilution of the library with degradation products. Additionally, the variability of miRNA expression increased by 10-30% with increased DTF; decreasing the power to detect differential expression. As expected, there was >70% overlap in the tissue dependent differential miRNA expression between snap-frozen and DTF specimens (1 or 12 h), but fewer miRNA were detected in the 12 h DTF specimens.
Biospecimens
Preservative Types
- Frozen
- Formalin
Diagnoses:
- Neoplastic - Carcinoma
Platform:
Analyte Technology Platform RNA Next generation sequencing Pre-analytical Factors:
Classification Pre-analytical Factor Value(s) Biospecimen Acquisition Cold ischemia time 1 h
2 h
3 h
12 h
Biospecimen Preservation Type of fixation/preservation Formalin (buffered)
Frozen