Page 76 - Read Online

Page 10 of 15 Miliotis et al. J Cancer Metastasis Treat 2020;6:13 I http://dx.doi.org/10.20517/2394-4722.2020.12

activated, such as through the recognition of neoantigens or viral antigens presented by cancer cells, they
produce inflammatory cytokines like tumor necrosis factor alpha (TNFα) and interferon gamma (IFNγ) [97,98] .
[99]
IFNγ has been shown to induce PD-L1 expression in various cancers . In GC, PD-L1 induction by IFNγ
appears to occur mainly through the JAK/STAT/Interferon regulatory factor 1 (IRF1) signaling axis [82,100-102] .
The extent to which different cancer cells are responsive to IFNγ and the downstream effects of IFNγ
exposure vary among cancer types and molecular subgroups within a cancer type.

In TCGA and other GC cohorts, when EBVaGC samples are compared to EBV-negative ones, they
demonstrate elevated IFNγ signatures, indicated by higher expression of IFNγ, JAK/STAT signaling
components, and several IFNγ-induced genes [4,103] . In addition, in vitro studies have shown that EBVaGC
cell lines induce PD-L1 expression in response to IFNγ to a much higher extent than EBV-negative GC
cell lines [93,104] . PD-L1 induction in response to IFNγ has also been shown to be significantly elevated
in other EBV-associated epithelial malignancies, such as NPC [105] . In NPC, the viral protein LMP1
acts synergistically with IFNγ to induce PD-L1 expression through the activation of the JAK3/STAT3,
NFκB, and AP-1 signaling pathways. In GC, EBNA1 has been shown to promote IFNγ-induced PD-L1
[93]
overexpression. Moon et al. showed that EBNA1 knockdown in SNU719, an EBVaGC cell line, resulted
in the transcriptional downregulation of JAK2. EBNA1 knockdown also resulted in a small but significant
reduction in constitutive and IFNγ-induced PD-L1 levels. However, ectopic EBNA1 expression in the
EBV-negative GC cell line AGS did not affect constitutive or IFNγ-induced PD-L1 expression. In contrast,
Su et al. [106] reported that AGS-EBV cells are more sensitive to IFNγ/TNFα treatment, showing higher
downstream PD-L1 upregulation, than the uninfected parental cell line. This suggests that other viral
factors, in addition to EBNA1, might be necessary for increasing IFNγ-induced PD-L1 expression in EBV-
infected GC cells. Further studies are required to determine how EBNA1 and other viral or host factors
promote increased IFNγ sensitivity and PD-L1 expression in EBVaGC.

Nakayama et al. [107] analyzed 43 EBVaGC samples and showed that the number of EBV genomes per cancer
[31]
cell (EBV copy number) correlates positively with PD-L1 expression. Similarly, Strong et al. performed
cellular gene expression analysis on 12 EBV-positive and 20 EBV-negative TCGA GC samples and showed
that, following hierarchical clustering, the 4 EBV-positive samples that had a much higher EBV coverage
depth (in RNAseq data) than the rest of the EBV-positive samples formed a well-defined gene expression
cluster. When they compared expression between “high” and “low” EBV GC samples, a large proportion
of the genes that were upregulated in the “high” EBVaGC group were immune-related, including IFNγ,
[31]
STAT1, IRF1, and multiple IFNγ-induced genes . As the authors state, EBERs, which have been shown to
induce IFNγ and TNFα production in peripheral blood mononuclear cells in vitro, could be contributing to
the elevated IFNγ signature found in “high” compared to “low” EBV GC samples [108] .

CONCLUSION
Even though GC is declining in the United States, it still has one of the lowest 5-year survival rates of
[2]
any cancer type . This highlights the need for new therapeutic strategies, especially for metastatic cases
that have the poorest prognosis and account for most of the new diagnoses every year. The molecular
heterogeneity of GC correlates with the response rate to different therapies, indicating that different
approaches should be considered for different molecular subgroups. The FDA recently approved
pembrolizumab as second-line therapy for patients with advanced MSI-H tumors of any type, including
GC [109] . In the last few years, several molecular and clinical studies present EBVaGC as another subgroup of
GC that could benefit from early-line treatment with immune checkpoint inhibitors [6,107] .

In EBVaGC, high immune activation in the tumor microenvironment likely acts as a driving force for
the selection of immune escape mechanisms such as PD-L1 overexpression. Different mechanisms act
independently or synergistically to induce PD-L1 expression. These include somatic genomic modifications,