EBV is linked to a variety of lymphoid and epithelial malignancies. In transplant recipients EBV is associated with the development of B cell lymphomas in post-transplant lymphoproliferative disorder (PTLD). We performed an integrative, multi-cohort analysis of EBV-positive and negative tumor samples to identify shared gene-signatures associated with EBV oncogenesis.

We selected three gene expression data sets (Gastric Cancer, PTLD, and Hodgkin's Lymphoma) that compared EBV-positive to EBV-negative tumors (n =170, collected from Gene Expression Omnibus). For each data set, we plotted the geometric mean of each gene probe to check for proper normalization and batch effects. The Hedges' g effect size, a measure of magnitude, was calculated for each gene in each dataset. To study the absolute change in expression of the three data sets, a meta-effect size was created by combining the gene effect sizes from each data set. Using a False Discovery Rate of < 0.05 and an Effect Size of 0.8, we utilized two meta-analysis methods, the random-effects model and the Fischer's sum-of-logs method. Leave-one-out validation was used to prevent bias from a large dataset.

We identified 30 human genes that were significantly up-regulated and five human genes that were significantly down-regulated in EBV-positive tumors. Of the top 15 most significant genes, nine have proposed roles in oncogenesis; for example, PMAIP1 is up-regulated in adult T cell leukemia and SESN2 contributes to p53 signaling. Only FGR was previously shown to be up-regulated in EBV-associated tumors. Overall, of the 35 genes, 17 have known or proposed mechanisms in oncogenesis, six have unknown functions, eight are immune regulators, and four are cellular homeostasis enzymes. Of note, CD38 and TLR7, known targets in cancer therapy, were up-regulated in EBV-positive tumors indicating that this approach can identify viable therapeutics.

Receiver operating characteristic curves, violin plots, and forest plots demonstrated that the 35 genes were able to significantly distinguish EBV-positive from EBV-negative tumors in each dataset. Importantly, this data was validated with an independent Burkitt's Lymphoma cell line dataset.

These results suggest common underlying mechanisms of viral transformation in EBV-positive tumors and may identify new opportunities for drug targeting or predictive diagnostics.

CITATION INFORMATION: Maloney E, Bongen E, Vallania F, Kotecha N, Khatri P, Martinez O. Integrative, Multi-Cohort Analysis of Epstein-Barr Virus (EBV)-Positive and Negative Tumor Samples to Identify Gene-Signatures Associated with EBV Oncogenesis. Am J Transplant. 2016;16 (suppl 3).

« Back to 2016 American Transplant Congress