*2.3. Immunohistochemistry*

Immunohistochemical labeling for E-cadherin, CD44, p16 and 14-3-3σ was performed on 4-μm-thick sections mounted on positively charged slides (SuperFrost Plus; Menzel Gläser, Braunschweig, Germany). Antigen retrieval, labeling, and counterstaining were performed on a Bond-Max Autostainer (Leica Biosystems, Newcastle-upon-Tyne, UK) using the Bond Polymer Refine detection system (Leica Biosystems). Primary antibodies and retrieval conditions were as follows: p16 (PA0016, 1:100,Novocastra, Newcastle Upon Tyne, UK); pH 6.0 buffer (ER1, Leica Biosystems) for 20 min, 14-3-3σ (SC-100638, 1:40, Santa Cruz Biotechnology, Heidelberg, Germany); pH 6.0 buffer (ER1) for 20 min, Ecadherin (NCH-38, 1:100, Agilent, Stockport, UK); pH 6.0 buffer (ER1) for 20 min, and CD44 (ab157107, 1:50 Abcam, Cambridge, UK); pH 9.0 buffer (ER2, Leica Biosystems) for 20 min. Internal positive tissue controls for E-cadherin and CD44 were available on each section.

#### *2.4. Evaluation of E-Cadherin, CD44, p16 and 14-3-3 σ Immunolabeling*

Analysis of immunolabeled samples was performed by 3 veterinary pathologists (A.H., A.S.-B. and S.L.P.), discrepancies were discussed by use of a multi-headed microscope, and a consensus was reached. In all cases, areas of ulceration and/or necrosis were avoided for interpretation.

The distribution and intensity of E-cadherin and CD44 labeling were analyzed for each case using a similar semiquantitative scoring system. The percentages of tumor cells, neoplastic acini and intravascular tumor emboli that expressed CD44 and E-cadherin were assessed. Labeling for CD44 was scored as follows: 0, no labeling; 1, 1–9% positive tumor cells; 2, 10–49% cells; and 3, 50–99% cells. Labeling for E-cadherin was scored as follows: 0, no labeling; 1, 1–9% positive tumor cells; 2, 10–49% cells; 3, 50–79% cells; and 4, 80–100% cells. The intensity of E-cadherin and CD44 labeling (0—negative, 1—mild, 2—moderate, and 3—strong) and whether labeling was membranous and/or cytoplasmic was also recorded [39–41]. A total immunohistochemical score (TIS) [39] for E-cadherin (ranging from 0 to 12) and TIS CD44 (ranging from 0 to 9) was calculated as the product of the distribution and intensity scores. Normal gastric surface epithelium served as a positive internal control for E-cadherin and lymphoid follicles, satellite glial cells and macrophages for CD44.

For p16, neoplastic cells with cytoplasmic and/or nuclear antigen expression were interpreted as immunopositive. An immunohistochemical score was assigned based on the intensity of staining (0—none, 1—weak, 2—moderate, and 3—strong) and percentage of positive tumor cells (0, 0% cells; 1, <25% cells; 2, 25–50% cells; 3, 51–75% cells; 4, >75% cells) [42]. The product of the intensity and distribution scores gave a TIS p16 ranging from 0 to 12. As a positive control, normal skin was used where epithelial cells within the basal layer of the epidermis exhibited weak immunoreactivity.

14-3-3σ labeling was assessed using a previously published semiquantitative scoring system [43]. The scores for the percentage (1, ≤10% cells; 2, 11–50% cells; and 3, >50% cells) and staining intensity (0, negative; 1, weak; 2, moderate; and 3, intense) of positive cells were recorded and a TIS 14-3-3σ (ranging from 0 to 9) was calculated as the product of these two parameters for each of the studied cases. Normal canine urinary bladder, which showed cytoplasmic immunolabeling of the urothelium, was used as a positive control.

#### *2.5. Statistical Analysis*

Results were analyzed for any significant relationship between the following: signalment (i.e., age, sex and breed); biopsy type (endoscopic vs. full-thickness biopsies); and histopathological features (tumor classification, chronic inflammation score, helicobacter identification, quantification, and localization). Statistical comparisons were performed

using GraphPad Prism version 8.0 for Windows (GraphPad Software, San Diego, CA, USA). Bar graphs were constructed to depict the mean and standard error of the parameter assessed for each group. To test for the significance of a relationship between two categorical variables, a Fisher's test was used to analyze the significance of contingency tables such as endoscopic vs. full-thickness biopsies for finding helicobacter. An unpaired *t*-test was used to compare the mean inflammation scores of different groups. For continuous data, the Student's *t*-test, analysis of variance (ANOVA), or Mann–Whitney U analysis test was used, depending on whether the data were normally distributed and whether two, or more than two groups, were compared. The significance level for all statistical tests was set at *p* < 0.05. The expression of p16, 14-3-3<sup>σ</sup>, E-cadherin and CD44 were compared with histological subtype and features of malignancy.
