Patients and clinical data
Archived formalin-fixed, paraffin-embedded tissues were obtained from 396 patients with gastric carcinoma. Among them, 146 had early gastric cancer and 250 had advanced gastric cancer. Standard resection with curative intent was performed at Korea University Guro Hospital from 2002 to 2005. No preoperative treatment was performed. According to tumor stage, 96 % (62 % of all the patients studied) of the indicated patients received postoperative 5-fluorouracil-based adjuvant chemotherapy. The mean follow-up period was 50.3 months (median 53.9 months; range 0–84.5 months). Clinicopathologic data were obtained from medical records and the histopathologic features of all patients were reviewed by pathologists. Tumors were classified according to the seventh American Joint Committee on Cancer (AJCC) TNM cancer classification system, the World Health Organization classification, and the Japanese classification [12–14]. The clinicopathologic characteristics of the patients are described in Table 1. The institutional review board of our institution approved this study.
Tissue microarray preparation and immunohistochemistry
Sections of formalin-fixed, paraffin-embedded tissues were prepared and stained with hematoxylin and eosin. Under the microscope, representative tumor areas were chosen at the deep invasive front with peritumoral stroma and prepared so as to create a tissue microarray. One or two different regions per case were punched out from donor blocks of 3 mm in diameter. Immunohistochemistry was performed on 4-μm tissue sections with a Ventana Bench Mark XT Autostainer (Ventana Medical Systems, Tucson, AZ, USA) with NOVA1 (dilution 1:500; Abcam, Cambridge, UK), cytokeratin (clone AE1/AE3; dilution 1:600; DAKO, Glostrup, Denmark), CD68 (dilution 1:150, clone PG-M1; DAKO), CD163 (dilution 1:100; clone MRQ-26; Cell Marque, Rocklin, CA, USA), CD3 (dilution 1:200; LabVision, Fremont, CA, USA), CD4 (dilution 1:200; Cell Marque), myeloperoxidase (MPO; dilution 1:2000, DAKO), and forkhead box P3 (FOXP3; dilution 1:100; Abcam) primary antibodies.
NOVA1 expression was analyzed according to the previously described semiquantitative approach [11, 15]. Staining intensity (1, no staining to weak intensity; 2, moderate intensity; 3, strong intensity) was multiplied by the percentage of positive cell nuclei (1, 0–9 %; 2, 10–19 %; 3, 20–29 %; 4, 30–39 %; 5, 40–49 %; 6, 50–59 %; 7, 60–69 %; 8, 70–79 %; 9, 80–89 %; 10, 90–100 %). NOVA1 expression was scored from 1 to 30 and classified as strong (score 21–30), moderate (score 11–20), or weak (score 1–10). The expression pattern of NOVA1 was evaluated in tumor cells, stromal spindle cells (fibroblasts, support cells, and endothelial cells), and immune cells (Fig. 1A). Regarding stromal spindle cells and immunes cells, NOVA1 expression was evaluated within the tumor cell nests and invasive tumor front area. Considering that our previous study had found that NOVA1 was expressed only in T cells and not in B cells or macrophages (Fig. S1) , we focused on the NOVA1 expression in T cells by comparing the CD3+ cells. For indicated cases that showed noncohesive, poorly differentiated histologic features, tumor cells were identified by comparing cytokeratin-positive cells.
Although all tissue cores were confirmed to contain appropriate tumor tissue occupying at least 50 % of the core area, the ratio of epithelial (tumor) area to stromal area differed among tissue cores. Therefore we counted tumor-infiltrating immune cells within the tumor cell nests and invasive tumor front area . Stained slides were scanned with a ScanScope CS system (Aperio Technologies, Vista, CA, USA), and the images were viewed with use of the Aperio ImageScope program (version 184.108.40.2062; Aperio Technologies, Vista, CA, USA). The morphology and number of macrophages, T lymphocytes, and neutrophils were assessed by a hematopathologist (S.O.Y.). CD68+ macrophages, CD163+ M2 macrophages, CD3+ T lymphocytes, MPO+ neutrophils, and FOXP3+ regulatory T cells (Tregs) that infiltrated within the tumor microenvironment were evaluated as follows: The five most representative ×400 magnification high-power fields were selected from the digital immunohistochemistry slides for CD68, CD163, CD3, MPO, and FOXP3. Preserved intact nuclei were counted manually, and the cell numbers were averaged. When the cell density was above the mean value of the overall cases, the sample was defined as high density for that inflammatory cell type (Fig. 1B).
Pearson correlation, one-way ANOVA, t, and χ
2 tests were used as appropriate to analyze differences between the variables examined. Survival rates were analyzed with the Kaplan–Meier method, and differences were compared with the log-rank test. Overall survival was measured from the date of diagnosis to that of gastric cancer-associated death or the last follow-up visit. Univariate and multivariate analyses were performed with the Cox proportional hazards model. Parameters showing statistical significance (P < 0.05) on univariate analysis were included in the multivariate analysis, which was performed with a backward stepwise method. Two-sided P values less than 0.05 were considered statistically significant. Statistical analyses were performed with IBM SPSS Statistics for Windows version 22.0 (IBM, New York, NY, USA).