Abstract 3181: Massively parallel sequencing of RNA and DNA isolated from FFPE tissue can be used for clinical studies

Formalin-fixed, paraffin-embedded (FFPE) patient tissue samples stored globally in pathology archives represent an invaluable biobank for clinical research. Unfortunately, nucleic acids isolated from FFPE tissues tend to be fragmented and chemically modified, interfering with many classical molecular analyses. Since massively parallel sequencing (MPS) technologies rely on randomly fragmented nucleic acids, we have focused a systematic study on the potential use of FFPE samples in MPS-based clinical studies. Available FFPE extraction kits were evaluated with respect to purification of RNA and DNA from different tissue types. Although the quality of the extracted nucleic acids was found to be highly dependent on the purification method used, it was possible to isolate high molecular DNA and RNA from recent FFPE specimen (fixed within one year). Extracted DNA and RNA from matching cryopreserved and FFPE specimens were successfully used for the preparation of targeted genomic (Exome-Seq) and whole transcriptome (RNA-Seq) sequencing libraries. Illumina's TruSeq exome preparation protocol was used for the preparation of Exome-Seq libraries and sequence analysis revealed that between 95.0 and 98.8 % of the reads mapped to the human genome with mismatch rates from 0.29 to 0.78 %. The preparation of RNA-Seq libraries from FFPE RNA using classical methods based on poly(dA) selection combined with oligo(dT) and random priming of cDNA synthesis is problematic due to the RNA degradation and results in 3′ biases and uneven sequence coverage. In this study, multiplexed RNA-Seq libraries were prepared using the ScriptSeq kit in combination with Ribo-Zero technology (both from Epicentre). In brief, ribosomal RNA was depleted from fragmented total RNA using Ribo-Zero followed by tagged random primed cDNA synthesis, terminal tagged second-strand synthesis, multiplexed amplification and purification. This approach results in strand-specific and paired-end sequences of coding and non-coding RNA. Sequence analysis of RNA-Seq libraries prepared from RNA isolated from a human lung tumor revealed that 96 - 98 % of the reads could be mapped and that the directional protocol worked correctly for at least 99 % of the reads. Surprisingly, more than half of the mapped reads were found to be non-human, mapping primarily to bacterial rRNA. As bacterial contamination is a potential issue in many tissue samples, we are presently testing improved versions of Ribo-Zero, including probes to b