Fig 1.
Copy number of EBV-DNA in EBV-associated gastric carcinoma.
Estimation of cancer cell ratio (CCR) was calculated for the adjustment of qPCR data. An EBV-associated gastric carcinoma (EBVaGC) case (A, hematoxylin and eosin staining) was immunostained with the monoclonal antibody AE1/AE3 (B). Using immunohistochemistry images, all the nuclei (blue) were automatically counted by the Tissue Studio software in the whole and cytokeratin immunostained regions (C). The CCR in EBVaGC cases (n = 43) distributes 4.1%–49% with a median of 22% (D). EBV-CN was obtained from qPCR data divided by the CCR (qPCR/CCR). Copy numbers of EBV-DNA per genome (EBV-CN) in EBVaGC cases (n = 43) are in the range between 1.2 and 185 with a median of 9.9 (E). A–C: Scale bar represents 100 μm. B: Inset is negative control, in which staining was performed without primary antibody. D, E: The bars represent the median and quartiles.
Fig 2.
Fluorescent in situ hybridization of EBV and its correlation with qPCR/CCR data in EBV-associated gastric carcinoma.
Fluorescent in situ hybridization (FISH) of EBV-DNA was performed with a specific DNA probe against the full-length EBV genome. Each red signal represents a single copy of the EBV genome. Each nucleus of EBV-infected cell line Namalwa contains one or two signals in each cell (A). Each nucleus of NCC-24 contains a lot of red signals in some cells, while no red signal in others, resulting in the average 3.9/nucleus (B). SNU-719 shows a lot of red signals in each nucleus, 34 on average. There are 7 nuclei in the figure (C). In two EBV-associated gastric carcinoma (EBVaGC) cases, the one shows 9 signals per nucleus (D) and the other 26 signals per nucleus (E). In both figures, several nuclei are aggregated. The graph presents correlation of FISH analysis with qPCR/CCR data (F). The y-axis represents the average number of EBV per nucleus by FISH. The x-axis represents the EBV copy number per nucleus, which was twice the corresponding value of the EBV-CN, assuming no amplification or deletion of GAPDH in the cancer cell nucleus. Dots and bars represent the mean and standard deviation, respectively. For the cases in which 2 × EBV-CN is less than 40, the number of EBV signals in FISH is proportional to 2 × EBV-CN; it gradually comes close to a plateau of 44 when 2 × EBV-CN exceeds 40 (R2 = 0.75 on the Hill curve fitting). A-E: Scale bar is 5 μm. Yellow arrowheads point to a single red signal of the EBV genome in each figure.
Fig 3.
Association of PD-L1 expression with EBV-CN or EBV-FISH.
Comparison of EBV-CN between PD-L1-positive and -negative tumors revealed significantly higher EBV-CN in PD-L1-positive tumors (P = 0.005, Mann-Whitney U test) (A). Representative images of hematoxylin and eosin staining, PD-L1 immunohistochemistry, and EBV-FISH of an EBVaGC case with 41 EBV signals per nucleus (B, D, and F, respectively) and an EBVaGC case with 9 EBV signals per nucleus (C, E, and G, respectively). B, C: Hematoxylin and eosin staining; D, E: PD-L1 immunohistochemistry; F, G: EBV-FISH.
Table 1.
EBV-CN and clinicopathological characteristics of EBVaGC.
Fig 4.
Prognostic significance of viral burden in cancer cells in EBV-associated gastric carcinoma.
Disease-specific survival of patients with EBVaGC comparing low and high EBV-CN groups. The median of EBV-CN was used as a threshold. Note the shorter survival in patients with high EBV-CN (P = 0.041).