This paper compared the yield and integrity of RNA extracted from formalin-fixed paraffin-embedded (FFPE) tumor blocks stored for 1.6-4.9 years using three different extraction kits. Next generation sequencing (NGS) metrics and expression were compared between FFPE specimens subjected to exome enrichment using three different methods and case-matched frozen specimens from the same tumor.
Functional comparison of exome capture-based methods for transcriptomic profiling of formalin-fixed paraffin-embedded tumors.
Shohdy KS, Bareja R, Sigouros M, Wilkes DC, Dorsaint P, Manohar J, Bockelman D, Xiang JZ, Kim R, Ohara K, Eng K, Mosquera JM, Elemento O, Sboner A, Alonso A, Faltas BM
Purpose
Conclusion
The authors report comparable RNA yields and RNA integrity numbers (RINs) among the three extraction methods, although DV200 values were lower in some specimens after extraction with the RNeasy kit. DV200 was modestly correlated with block age but not correlated with number of mapped reads or the percentage of uniquely mapped reads. Themean number of reads and percentage of genes with zero expression were comparable among the three exome capture methods evaluated. While the highest percentage of uniquely mapped reads and the lowest percentage of multi-mapped reads were obtained when exome capture was with SureSelect, exome capture with TWIST resulted in higher fragments per kilobase of exon model per million reads mapped (FPKM). The global gene expression patterns and junction read count were strongly correlated among the exome capture methods and co-clustered for 10 of 11 patients.
FFPE specimens had a lower percentage of uniquely mapped reads than frozen specimens, but percentages of multi-mapped reads and zero expressed genes were comparable. Importantly the global expression profiles and FPKM values for 73 immuno-related genes were strongly correlated between case-matched FFPE and frozen specimens, and gene expression clustered by patient in 8 of 11 sample sets. There was 100% concordance in detection of expression outliers (1.5 times the interquartile range, z-score ≥ 2 and FPKM ≥ 20) between case-matched frozen and FFPE specimens.
Subtype classification of the five urothelial tumors evaluated based on miRNA expression showed 80%, 50% and 60% agreement between frozen and FFPE specimens when exome capture was with Agilent, TWIST and IDT, respectively. Seven of the eight fusions identified using frozen specimens were detected using each of the three exome capture methods evaluated.
Studies (1)
Purpose
This study compared the yield and integrity of RNA extracted from FFPE tumor blocks stored for 1.6-4.9 years using three different extraction kits. NGS metrics and expression were compared between FFPE specimens subjected to exome enrichment using three different methods and case-matched frozen specimens from the same tumor. The study included banked surgical FFPE tumor specimens and case-matched frozen specimens from 11 patients and included urothelial cancer, gastroesophageal junction (GEJ) adenocarcinoma, oligodendroglioma, cancer of unknown primary (CUP), leiomyosarcoma, papillary thyroid cancer, and colorectal cancers. Specimen procurement and processing details were not included, but specimens were stored 1.6-4.9 years. RNA was extracted from ten 10 µm-thick FFPE sections and from frozen specimens using the Maxwell 16 LEV simplyRNA Tissue Kit. To compare the effects of extraction method, RNA was extracted from FFPE sections of three tumors using the Roche High Pure FFPET, Qiagen RNeasy FFPE kits and the MaxWell kit. RNA yield was quantified by spectrophotometer and Qubit Fluorometer and integrity was assessed by a bioanalyzer. Exome capture was performed with Sure SelectXT Human All exon v6+UTR, Twist Human Core Exome, or IDT xGen Exome Research Panel v1.0 and libraries were constructed using the SureSelect RNA Direct kit (SureSelect capture) and the NEBNext Ultra II Directional kit (Twist and IDT). Sequencing libraries were constructed using RNA from frozen specimens with TruSeq RNA Library Prep kit v2. All libraries were sequenced using a HiSeq4000 machine.
Summary of Findings
The authors report comparable RNA yields and RINs among the three extraction methods, but DV200 values were lower in some specimens when extraction was with the RNeasy kit. DV200 values were modestly correlated with block age (ρ=-0.45, P=0.002) and RIN (ρ=0.54, P=0.001), but were not correlated with number of mapped reads or the percentage of uniquely mapped reads. The mean number of reads were comparable among the three exome capture methods for FFPE specimens. The percentage of uniquely mapped reads was higher using SureSelect than TWIST (94% versus 89%, P=0.036), but was comparable between IDT and SureSelect and TWIST exosome capture methods. SureSelect had a lower percentage of multi-mapped reads than IDT (2% versus 5%, P=0.0001) or Twist (2% versus 3%, P<0.0001). Exome capture with TWIST resulted in higher fragments per kilobase of exon model per million reads mapped (FPKM) than SureSelect or IDT (P<0.0001, both). The percentage of genes with zero expression was comparable among the exome selection methods. The majority of genes detected were observed across all three selection methods (17801 genes), but additional genes were only detected with two of the three methods (832 genes with IDT and TWIST, 549 genes with TWIST and SureSelect, and 226 genes with IDT and SureSelect) and an additional 1880, 360 and 218 genes were detected only using SureSelect, TWIST, and IDT, respectively. The global gene expression patterns were strongly correlated among the exome capture methods (r=0.86-0.95) and co-clustered for 10 of 11 patients.
The average number of total and uniquely mapped reads were higher for frozen specimens than for FFPE specimens (44.4 million versus 38.6 million and 39 million versus 35 million), but percentages of multi-mapped reads and zero expressed genes were comparable. While gene expression was bimodal for both FFPE and frozen specimens, the second density peak was at 1000 for FFPE specimens versus 100 for frozen specimens. Importantly, global expression profiles were strongly correlated between case-matched FFPE and frozen specimens (r=0.79-0.90) and clustered by patient in 8 of 11 sample sets. Importantly, there was 100% concordance in detection of expression outliers (1.5 times the interquartile range, z-score ≥ 2 and FPKM ≥ 20) between case-matched frozen and FFPE specimens. FPKM values for 73 immuno-related genes were strongly correlated between the three exome capture methods and frozen specimens, with higher overall concordance with case-matched frozen specimens when exome capture from FFPE was conducted with TWIST rather than IDT (P=0.003) or SureSelect (P=0.001). Subtype classification of five urothelial tumors based on miRNA expression showed 80%, 50% and 60% agreement between frozen and FFPE specimens when exome capture was with Agilent, TWIST, and IDT, respectively, with LumNS misclassified in all FFPE specimens. Of the eight fusions identified using frozen specimens, all were detected using each of the three exome capture methods except MKRN2-PPARG in one specimen after capture using TWIST. The junction read count and expression levels were strongly correlated for all exome capture methods (ρ=0.99, P<0.0001 for SureSelect; ρ=0.92, P=0.001 for TWIST; and ρ=0.85, P=0.003 for IDT) and each were comparable to those in frozen specimens.
Biospecimens
TissueThyroid GlandTissueColorectalTissueEsophagusTissueBladderTissueNerveTissueOtherPreservation Types
FormalinFrozenDiagnoses
NeoplasticCarcinomaNeoplasticSarcomaPlatforms
RNASpectrophotometryRNAFluorometryRNANext generation sequencingRNAAutomated electrophoresis/BioanalyzerPre-analytical Factors
Classification Pre-Analytical Factor Value(s) Biospecimen Preservation Type of fixation/preservation Formalin (buffered)FrozenStorage Storage duration 1.6-4.9 yearsAnalyte Extraction and Purification Analyte isolation method Maxwell 16 LEV RNA FFPE Purification KitQiagen RNeasy FFPE KitRoche High Pure FFPET KitNext generation sequencing Specific Targeted nucleic acid Exome enrichment with Agilent SureSelect V6Exome enrichment with IDT XGen Exome Research PanelExome enrichment with TWIST NGS Exome