Correlation of expressed single nucleotide variants detected by cDNA-Capture or RNA-Seq and corresponding normalized expression values.
The research, published in the Journal of Molecular Diagnostics, describes a hybrid process for transcriptome analysis of non-fresh frozen tumor samples. While fresh-frozen (FF) tissue samples are optimal for transcriptome analysis, few clinical samples are collected as FF. Most are fixed in formalin (FFPE) to preserve tissue for pathologic examination, causing degradation of the RNA. Additionally, most clinical samples are available in only limited amounts, which yield limited DNA and RNA.
The hybrid method, called cDNA-Capture sequencing, was developed to address the challenges of transcriptome analysis of sub-optimal tissue samples (FFPE). RNA-Seq was combined with an intermediate enrichment step of exome capture, which concentrates the data yield onto the exome. This method allows detection of expressed variants from degraded RNA (formalin-damaged) and determination of gene expression levels from limited input material.
cDNA-Capture was completed in FF lung adenocarcinoma samples and compared to previously-generated RNA-Seq data from lung adenocarcinoma samples. The cDNA-Capture sequencing achieved similar coverage levels as RNA-Seq, with only 1/3 the number of reads. The method increased the representation of the lowest expressed genes in the transcriptome while minimizing the oversequencing of the most highly expressed genes. The ability of RNA-Seq and cDNA-Capture to validate the expression of single nucleotide variants (SNV) within protein-coding genes was compared, with similar results.
Gene fusions were detected in the LNCaP prostate cancer cell line by both RNA-Seq and cDNA-Capture. All of the fusions had a higher cDNA-Capture normalized fusion score compared with RNA-seq.
RNA-Seq and cDNA-Capture were compared using FFPE material from two lung adenocarcinomas. The percentage of reads aligned was nearly equivalent, but cDNA-Capture exhibited a six-fold increase in the proportion of aligned reads that mapped to a targeted region. Significant correlations were found between cDNA-Capture from FFPE and FF materials.
Not surprisingly, both lung adenocarcinomas had a greater number of expressed SNVs detected by both RNA-Seq and cDNA-Capture and a much higher percentage of reads aligned to the genome when using FF material compared to FFPE. When comparing cDNA-Capture to RNA-Seq reads, FFPE had a much larger gain in target enrichment than FF material. Regarding the percentage of mapped reads that aligned to coding regions, FF cDNA-Capture had the largest percentage (mean, 56%), followed by FFPE cDNA-Capture (32%), FF RNA-Seq (25%), and FFPE RNA-Seq (6%).
The authors suggest that cDNA-Capture enrichment may be superior to RNA-Seq at low input since it enriches for coding regions (rescuing gene-expression signals masked by noise from RNA degradation). The enrichment is sufficient to maintain the biological interpretation observed in FF material, while requiring only 1/3 of the sequencing data. cDNA-Capture costs approximately 50% less per sample than RNA-Seq, even when including the additional cost of the exome capture kit.
cDNA-Capture resulted in more even and comprehensive coverage across all expressed genes, ultimately making it possible to maximize the information provided by limited clinical samples.