Cancer Diagnosis Program | Biorepositories and Biospecimen Research Branch

Search BRD Papers

Whole-Transcriptome Profiling on Small FFPE Samples: Which Sequencing Kit Should Be Used?

Author(s): Hilmi M, Armenoult L, Ayadi M, Nicolle R

Publication: Curr Issues Mol Biol, 2022, Vol. 44, Page 2186-2193

PubMed ID: 35678677 PubMed Review Paper? No

Purpose of Paper

Conclusion of Paper

Studies

1. Study Purpose

   The purpose of this study was to compare results of RNAseq analysis of FFPE breast tumor specimens obtained using 3’capture (Lexogen, Lexogen’s QuantSeq 3’ Kit), exome-capture (RNAaccess, TruSeq RNA exome Kit), and ribodepletion (Smarter, SMARTer Stranded Total RNA-seq Kit v3-Pico; Solovation, Ovation SoLo Kit; Sequoia, SEQuoia Complete Stranded RNA Kit) approaches to gene expression results obtained using NanoString in tumor-matched FFPE specimens and RNAseq results from case-matched frozen tumor specimens using polyA enrichment (TruSeq, TruSeq polyA enrichment kit). Twenty tumor-matched FFPE and frozen breast tumor specimens were used for all comparisons, with the exception of the RNAaccess method that used 12 specimens.  RNA was isolated from frozen specimens with the RNeasy Mini Kit and from FFPE specimens with the ALLPrep FFPE Tissue Kit.  The amount of RNA isolated from FFPE specimens used as input depended upon the library sequencing kit and assay used: Nanostring (150 ng), Truseq (400 ng), RNAaccess (400 ng), Lexogen (50, 150, 400 ng), Sequoia (2, 26 ng), Solovation (2, 5 ng), Smarter (2, 8 ng).  Samples were analyzed by Illumina short-read sequencing at a depth of 10 million single-end reads (Lexogen) or 20 million paired-end reads (all other kits).

   Summary of Findings:

   The largest number of genes quantified was from libraries constructed from frozen breast tumor specimens using the poly-A enrichment-based Truseq Sequencing Kit (35,032 genes) and from FFPE breast tumor specimens using the ribodepletion-based Smarter Sequencing Kit (34,372 genes) with  the lowest number of quantified genes was observed when FFPE breast tumor specimens were sequenced with the Lexogen Kit when 50 ng of RNA was used as input (16,764 genes) and the Sequoia Kit (18,864 genes) that used 26 ng of RNA as input.  Transcriptomic profiles from Nanostring analysis of FFPE breast tumor specimens were used to gauge the accuracy of each sequencing method evaluated; gene-wise correlation was the strongest for frozen breast tumor specimens sequenced by TruSeq (mean correlation coefficient=0.816) followed by FFPE breast tumors sequenced by the Smarter technique (mean correlation coefficient= 0.759). Statistical pairwise comparisons of correlation coefficients of Nanostring comparisons revealed that the Lexogen Kit with 50 ng (p=0.006) or 400 ng (p=0.004) of RNA as input and the Sequoia Kit (26 ng of input RNA, p=0.02) had significantly lower correlation coefficients than the RNAaccess Kit (400 ng of input RNA), whereas TruSeq (p=0.05, 400 ng input RNA from frozen specimens) and Smarter (p<0.0001, 8 ng of input RNA) Kits had significantly higher correlation coefficients. 

   Comparisons were then drawn between different RNAseq techniques based on the expression level of individual genes (high, mid, low tercile range); the correlation coefficient of case-matched frozen specimens analyzed by Truseq were compared to each of the RNAseq techniques evaluated in FFPE specimens, and statistical pairwise comparisons were made relative to results obtained from FFPE specimens with the RNAaccess Kit. For genes in the high tercile, analysis of FFPE specimens by the Smarter technique resulted in the strongest correlation coefficient with Truseq analysis of case-matched frozen specimens.  Relative to correlation coefficients obtained for FFPE specimens with the RNAaccess Kit, significantly higher correlations with Truseq results in frozen specimens occurred with the Lexogen Kit when 150 ng or 400 ng of input RNA was used, the Solovation Kit, and the Smarter Kit (p<0.0001 for all).  For genes in the mid and high terciles, correlation coefficients with Truseq results were significantly higher than the RNAaccess Kit for the Lexogen Kit when 150 ng or 400 ng of input RNA was used, the Solovation Kit, and the Smarter Kit (p<0.0001 for all); less robust but still significantly higher correlation coefficients were also observed with the Lexogen Kit when 50 ng of input RNA was used and the Sequoia Kit (p<0.0001 for all).  For all terciles, FFPE specimens with small quantities of RNA that were analyzed using the Smarter RNAseq technique were more strongly correlated to results obtained from Truseq analysis of case-matched frozen specimens with standard quantities of RNA than any other technique evaluated.

   For FFPE specimens, kit reproducibility was evaluated in duplicate samples containing an RNA input amount of 2 ng for Sequoia, Solovation, and Smarter Kits. The correlation coefficient between duplicate samples was significantly higher for the Smarter and Solovation Kits (R=0.83 for both) than the Sequoia Kit (R=0.64; p=0.029).

   Biospecimens

   • Tissue - Breast

   Preservative Types

   • Frozen
   • Formalin

   Diagnoses:

   • Neoplastic - Carcinoma

   Platform:

   Analyte	Technology Platform
   RNA	DNA sequencing

   Pre-analytical Factors:

   Classification	Pre-analytical Factor	Value(s)
   DNA sequencing Specific	Technology platform	Illumina’s TruSeq RNA exome Kit (RNAaccess) Lexogen QuantSeq FWD 3’ Kit (Lexogen) BIO-RAD’s SEQuoia complete stranded RNA library Kit (Sequoia) Takara’s SMARTer Stranded total RNA-seq Kit v3-pico (Smarter) TECAN’s Ovation SoLo RNA-Seq library preparation Kit (Solovation)
   Biospecimen Preservation	Type of fixation/preservation	Formalin (buffered) Snap frozen
   DNA sequencing Specific	Template/input amount	2 ng 5 ng 8 ng 26 ng 50 ng 140 ng 400 ng

   View more