Detection of viral pathogens in high grade gliomas from unmapped next-generation sequencing data

• Unmapped next-generation sequencing reads from human cancer can detect viral sequences.
• High grade gliomas showed the presence of Epstein–Barr virus (EBV) and Roseolovirus.
• EBV is not transcriptionally active in these cases.
• Cytomegalovirus was not detected in cases of glioblastoma.

Viral pathogens have been implicated in the development of certain cancers including human papillomavirus (HPV) in squamous cell carcinoma and Epstein–Barr virus (EBV) in Burkitt's lymphoma. The significance of viral pathogens in brain tumors is controversial, and human cytomegalovirus (HCMV) has been associated with glioblastoma (GBM) in some but not all studies, making the role of HCMV unclear. In this study we sought to determine if viral pathogen sequences could be identified in an unbiased manner from previously discarded, unmapped, non-human, next-generation sequencing (NGS) reads obtained from targeted oncology, panel-based sequencing of high grade gliomas (HGGs), including GBMs. Twenty one sequential HGG cases were analyzed by a targeted NGS clinical oncology panel containing 151 genes using DNA obtained from formalin-fixed, paraffin-embedded (FFPE) tissue. Sequencing reads that did not map to the human genome (average of 38,000 non-human reads/case (1.9%)) were filtered and low quality reads removed. Extracted high quality reads were then sequentially aligned to the National Center for Biotechnology Information (NCBI) non-redundant nucleotide (nt and nr) databases. Aligned reads were classified based on NCBI taxonomy database and all eukaryotic viral sequences were further classified into viral families. Two viral sequences (both herpesviruses), EBV and Roseolovirus were detected in 5/21 (24%) cases and in 1/21 (5%) cases, respectively. None of the cases had detectable HCMV. Of the five HGG cases with detectable EBV DNA, four had additional material for EBV in situ hybridization (ISH), all of which were negative for expressed viral sequence. Overall, a similar discovery approach using unmapped non-human NGS reads could be used to discover viral sequences in other cancer types.