Metagenomic analysis reveals altered gut virome and diagnostic potential in pancreatic cancer.

ABSTRACT

Pancreatic cancer (PC) is a highly aggressive malignancy with a poor prognosis, making early diagnosis crucial for improving patient outcomes. While the gut microbiome, including bacteria and viruses, is believed to be essential in cancer pathogenicity, the potential contribution of the gut virome to PC remains largely unexplored. In this study, we conducted a comparative analysis of the gut viral compositional and functional profiles between PC patients and healthy controls, based on fecal metagenomes from two publicly available data sets comprising a total of 101 patients and 82 healthy controls. Our results revealed a decreasing trend in the gut virome diversity of PC patients with disease severity. We identified significant alterations in the overall viral structure of PC patients, with a meta-analysis revealing 219 viral operational taxonomic units (vOTUs) showing significant differences in relative abundance between patients and healthy controls. Among these, 65 vOTUs were enriched in PC patients, and 154 were reduced. Host prediction revealed that PC-enriched vOTUs preferentially infected bacterial members of Veillonellaceae, Enterobacteriaceae, Fusobacteriaceae, and Streptococcaceae, while PC-reduced vOTUs were more likely to infect Ruminococcaceae, Lachnospiraceae, Clostridiaceae, Oscillospiraceae, and Peptostreptococcaceae. Furthermore, we constructed random forest models based on the PC-associated vOTUs, achieving an optimal average area under the curve (AUC) of up to 0.879 for distinguishing patients from controls. Through additional 10 public cohorts, we demonstrated the reproducibility and high specificity of these viral signatures. Our study suggests that the gut virome may play a role in PC development and could serve as a promising target for PC diagnosis and therapeutic intervention. Future studies should further explore the underlying mechanisms of gut virus-bacteria interactions and validate the diagnostic models in larger and more diverse populations.