Taking advantage of the copious amounts of data generated from The Cancer Genome Atlas (TCGA), researchers have created a comprehensive map of the viruses present in a wide range of cancer types. The result provides a new tool to better understand how viral infections play a role in tumor development and clues on how viral oncogenes function. The study is published in Nature Communications.
Researchers at the Institute of Biomedicine and the University of Gothenburg Sweden used RNA-sequencing data from more than 4,000 tumor samples from 19 different cancer types available through the TCGA. Viral DNA was detected in 178 tumor samples.
“We didn't look specifically for human viruses. We were open to the possibility of finding anything—more than 3,000 viral genomes were considered,” said lead author Erik Larsson, PhD, of the University of Gothenburg. The viruses detected turned out to be almost exclusively known human pathogens.
The researchers did not find viruses linked to breast cancer or glioblastoma multiforme, even though researchers had speculated that a virus might influence the development of these two tumors. Of 167 glioblastoma multiforme and 810 breast cancer samples, viral DNA was below the positive detection threshold, arguing “strongly against viral etiology” for these tumor types, the authors concluded, although a rare virus involvement cannot be ruled out. “Considering that we saw no expression of viral genes in almost 1,000 tumors from these cancers it seems unlikely that viruses have a widespread role,” said Larsson, although there could be silent viruses incorporated into the genome.
“A possible explanation would be regional differences, and one should keep in mind that our study is based on RNA. In theory there could be silent viruses incorporated into vital DNA regions. However, considering that we saw no expression of viral genes in almost 1,000 tumors from these cancers it seems unlikely that viruses have a widespread role. It would be interesting to apply our methodology to materials where viruses have previously been identified.”
The presence of human papillomavirus (HPV) was confirmed in 96.6% of the cervical cancers sampled; cervical cancer is predominantly caused by infection of HPV. The hepatitis B virus (HBV) was found among 32.4% of hepatocellular cancer samples, confirming the role of this virus in a portion of this type of liver cancer. The cancer type with the next highest proportion of viral-positive DNA (14.8% of samples were positive for HPV) was head and neck cancer.
The data show that viruses may have a role in a small proportion of tumor types not generally linked to viral infection including bladder and colorectal cancers. The presence of viral RNA was found in 7.3% of bladder cancers, 9.8% of colon cancers, and 5.6% of rectal cancers.
A total of 12 HPV subtypes were identified among 84 cervical cancer tumors. No other viruses were found associated with cervical cancer. The two most frequent HPV subtypes were HPV16 (65.5% of positive cases) and HPV18 (13.1% of positive cases). HPV subtypes were also identified in several bladder urothelial carcinomas, lung squamous cell carcinoma, and uterine endometroid carcinoma. HPV18 was also found in 1.9% of colorectal tumors (5 cases).
Taking advantage of sequencing technologies to identify viral sequences in tumor tissue in an unbiased way, this study goes beyond the traditional single-focus approaches to identify the relationship of viruses to tumor initiation.
High-throughput approaches have already identified genomic sites of viral genome integration in cervical and head and neck cancers. In this current study, HPV insertions in the ERBB2 (2 cases) and RAD51B genes (4 tumors) were observed. Based on the data, the authors suggest that both the RAD51B and the ERBB2 genes are functional targets for viral integration. Tumors with viral integration in the RAD51B gene had slightly lower expression of this gene and one tumor type (of two that had high ERBB2 expression) had viral integration within this gene.
Although the presence of viral RNA is not enough evidence to suggest that the virus plays a causal role in the cancer, presence of viral RNA and expression of viral oncogenes is a strong indicator that the virus contributed to tumorigenesis, said Larsson.
Larsson and colleagues suggest that “analysis of host transcriptome perturbations caused by tumor virus proteins can facilitate identification and prioritization of cancer-causing genes and pathways,” a concept that has previously been proposed. The study identified what the authors say are a surprisingly low number of novel viruses: but that research on rare, under-studied cancer types may uncover new virus types that may have roles in cancer formation.
The researchers are continuing to analyze viral RNA in other tumor types.