2.1. Patient Selection and Sample Collection

The study included a total of 91 female dogs. Of these, 61 dogs were prospectively enrolled and diagnosed with simple mammary carcinomas (groups G3 and G4), regardless of breed and age. The remaining 30 samples (G1 and G2) were retrieved from the archival database of the Veterinary Pathology Department at the School of Agricultural and Veterinary Sciences at UNESP and consisted of 19 simple adenomas and 11 normal mammary tissues. All included dogs were entire females (not spayed).

These patients were examined at the Veterinary Obstetrics and Oncology Services of the Veterinary Hospital at the School of Agricultural and Veterinary Sciences, UNESP, Jaboticabal, Brazil. Clinical staging was conducted based on physical examinations, laboratory analyses, and imaging studies. The animals underwent complete clinical evaluation, including palpation of the mammary chain and regional lymph nodes, cardiopulmonary auscultation, abdominal palpation, and mucosal assessment. In addition, complementary imaging exams (three-view thoracic radiography and abdominal ultrasound) and laboratory analyses such as complete blood count (CBC), serum biochemistry, and urinalysis were performed.

Tumor samples were collected during unilateral total mastectomy, following the standard surgical protocol at the Veterinary Hospital. Regional lymph nodes (inguinal and axillary) were also removed for histopathological evaluation. Only one tumor per dog was analyzed, focusing on the most prognostically significant lesion when multiple nodules were present, following the criteria proposed by Cassali et al. [34] and Rasotto et al. [5]. Special subtype carcinomas, carcinomas in mixed tumors, and mammary sarcomas were excluded from the study [34].

Histopathological evaluation was employed to classify the samples into various experimental groups. Lymph node metastasis was ruled out based on histopathological examination, assessing for the presence of neoplastic epithelial cells [34].

All tissue samples were fixed in 10% buffered formalin for 24–48 h and then preserved in 70% ethanol until processing. Subsequently, the samples were paraffin-embedded and sectioned to a thickness of 3 μm for histopathological and immunohistochemical analyses. Hematoxylin and eosin (HE) staining was used for histopathological evaluation.

Dogs diagnosed with distant metastases at initial presentation were excluded, but those that developed metastases during follow-up remained in the study.

2.2. Survival Analysis

Overall survival time (OST) was defined as the duration from diagnosis to death. For animals that remained alive at the end of the 3-year follow-up, survival data were censored. Updates on patient status were collected through telephone contact with owners and documented in medical files.

2.3. Validation of Antibodies Through Western Blot

The antibodies employed in this study were previously validated for canine mammary tissues using the Western blot technique, in accordance with the protocol established by Senhorello et al. [31].

2.4. Immunohistochemistry

Immunohistochemical analyses followed established protocols [31, 35] to evaluate the expression of H3K9Ac and H4K12Ac, along with the HDACs 1, 2, and 6.

Paraffin-embedded tissue sections (3-μm thick) were mounted on silanized slides, deparaffinized, and rehydrated using standard procedures [36]. Antigen retrieval was performed in a microwave oven for three cycles of 4 min each, followed by incubation with primary antibodies (detailed in Table 1). A commercial polymer-based detection system (Novolink DS, Leica Biosystems) was used, and reactions were counterstained with Harris hematoxylin and dehydrated in graded concentrations of alcohol and xylene before mounting.

Human mammary tissues, generously provided by the VETPAT laboratory (Campinas, São Paulo), served as reaction controls, with adjacent healthy mammary tissue used as positive controls for the neoplastic samples. Negative controls were processed by replacing primary antibodies with antibody diluents to verify specificity.

Immunohistochemistry was chosen for its ability to spatially localize protein expression in tissue architecture, which is essential for distinguishing nuclear from cytoplasmic patterns, particularly for HDAC isoforms. Additionally, it allows for semiquantitative analysis while preserving morphological context [37].

2.5. Evaluation of Immunohistochemical Staining

Nuclear staining of H3K9Ac, H4K12Ac, HDAC1, and HDAC2 was verified by the presence of diffuse brown staining within the nuclei of cells from various experimental groups, consistent with patterns observed in positive controls. Immunohistochemical assessment was conducted using a staining score protocol. Five random fields per sample were photographed at 400x magnification with a Novel microscope (BM2100) using a Bioptika camera (CMOS-HD). Manual counting of 100 cells per field was performed with ImageJ software (v. 1.44p) and the “Cell Counter” tool, followed by the calculation of arithmetic means in accordance with the methodology applied in previous studies in canine tissues [29, 31, 35]. Fields were categorized based on the percentage of positive cells (0 = negative; 1 = 1–25%; 2 = 26–50%; 3 = 51–75%; 4 = 76–100%) and staining intensity (0 = negative; 1 = weak; 2 = moderate; 3 = strong), independently assessed by two blinded observers. The final staining score was obtained by multiplying these values, and the results were utilized for statistical analyses [31, 35, 38].

For HDAC6, which shows cytoplasmic expression, a different quantification method was utilized. Positive staining was indicated by brown cytoplasmic coloring, with the total percentage of the positive area being calculated. Five random fields (400x) were analyzed, as previously described [31, 35, 39]. The total photographed area was measured manually, excluding noncellular regions. The positively stained area was then selected, thresholded, and quantified. The percentage of the positive area was calculated for each image and averaged across the five images from each sample.

2.6. Statistical Analysis

Comparisons of histone (H4K12Ac, H3K9Ac) and HDAC (HDAC1, HDAC2) expression among experimental groups (G1, G2, G3, and G4) were carried out using the Kruskal–Wallis test, followed by Dunn’s post hoc test. For HDAC6, intergroup comparisons were conducted only, as it was measured as a percentage of marked cells instead of analyzed with histones within groups.

Multivariate statistical methods were applied to explore associations between histones and HDACs within and between groups. Hierarchical cluster analysis was used to identify subgroups within each experimental group, with identified subgroups randomly assigned as A or B. Principal component analysis (PCA) was conducted to determine interactions between subgroups and latent variables (PC1, PC2). Exploratory factor analysis was then used to explain data variance, with factors (F1, F2) identified based on eigenvalues and Varimax rotation. Variables with absolute loadings > 0.5 were considered relevant. Differences in factors between Subgroups A and B were evaluated using the Wilcoxon–Mann–Whitney test.

To differentiate the groups, H3K9Ac, HDAC1, HDAC2, and HDAC6 were chosen based on the highest factor loadings. ROC curves were generated using F1 and F2 values to evaluate specificity, sensitivity, and cutoff points for distinguishing between groups (G2, G3, and G4) and G1. Survival analysis was performed using Kaplan–Meier estimates, with group comparisons conducted using the log-rank test.

Statistical analyses were carried out using R and Statistica software, with a significance threshold set at 0.05.

3.1. Immunoexpression of Acetylated Histones and Deacetylase Isoenzymes: Univariate Analysis

Antibody staining was observed in all analyzed groups. When comparing the staining patterns with the positive control, similarities were noted as H3K9Ac, H4K12Ac, HDAC1, and HDAC2 exhibited predominantly nuclear staining, while HDAC6 showed predominantly cytoplasmic staining (Figure 1).

The Kruskal–Wallis test revealed significant differences (p < 0.05) in antibody scores for HDAC1, HDAC2, and H3K9Ac among the groups, as well as in the percentage of HDAC6 isoenzyme staining. However, no significant difference was noted for H4K12Ac antibody scores between the groups.

Statistical differences in H3K9Ac expression were specifically observed between groups G1 and G3 (p = 0.022), G1 and G4 (p = 0.002), and G2 and G4 (p = 0.028), showing lower expression in malignant neoplasms. In terms of HDAC1 expression, differences were noted among groups G1 and G2 (p = 0.002), G1 and G4 (p < 0.001), and G3 and G4 (p = 0.010), with higher expression in normal mammary tissue compared to simple adenomas and metastatic carcinomas. Moreover, significant differences in semiquantitative HDAC2 expression were identified between groups G1 and G3 (p = 0.007) and G1 and G4 (p = 0.035), indicating lower HDAC2 expression in mammary carcinomas relative to normal tissue.

Lastly, HDAC6 immunostaining showed significant differences, with variations in the percentage of the positive area between groups G1 and G3 (p = 0.005) and G1 and G4 (p = 0.045). This isoenzyme demonstrated higher expression in mammary carcinomas compared to normal mammary tissues. The antibody staining results for groups G1, G2, G3, and G4 are summarized in Figure 1.

Univariate analysis within the experimental groups showed no significant differences in antibody expression in G1. However, in G2, HDAC2 expression was significantly higher than that of HDAC1 (p < 0.001), and H4K12Ac was expressed at higher levels than HDAC1 (p < 0.001). In G3, significant differences were noted between H4K12Ac and H3K9Ac (p < 0.001) as well as between H4K12Ac and HDAC1 (p < 0.001), while no significant differences were observed among the other antibodies. Likewise, in G4, H4K12Ac expression was significantly greater than that of H3K9Ac (p < 0.001) and HDAC1 (p < 0.001). Overall, in groups G3 and G4, H4K12Ac demonstrated higher expression levels compared to H3K9Ac and HDAC1 (Figure 2).

3.2. Immunoexpression of Acetylated Histones and Deacetylase Enzymes: Multivariate Analysis

Exploratory factor analysis identified a positive association in Group G1 for the expression of H3K9Ac and HDAC6, while this association was negative in Groups G3 and G4. A similar association was noted between Groups G2 and G3 for H4K12Ac and H3K9Ac, as well as between Groups G3 and G4 for HDAC1 and HDAC2, both exhibiting a positive correlation. No consistent pattern of antibody expression association was observed between G1 and the other groups (Figure 3 and Table 2).

When comparing Subgroups A and B within each experimental group, Group G1 exhibited a significant difference between Subgroups A and B regarding the positively associated expression of H3K9Ac and HDAC6 (p < 0.001). In Group G2, a significant difference was noted between the subgroups for the negative association between HDAC1 and HDAC6 (p = 0.005). In the analysis of the groups with mammary carcinomas, both groups G3 and G4 showed a significant difference between subgroups concerning the positive association of HDAC1 and HDAC2 (p < 0.05). These results indicate that within the malignant neoplasia groups, there are differing expression patterns, allowing for the dichotomization of Subgroups A and B (Table 2).

After confirming the separation of samples into G3 and G4 based on the associated expression of histones, we evaluated whether this subdivision influenced prognosis. However, no significant difference in OST was observed between animals in Subgroups A and B of G3 or between Subgroups A and B of G4 (Figure 4).

Further multivariate analyses across the experimental groups identified two principal factors (Factor 1 and Factor 2), which accounted for 61.89% of the variance between groups. Factor 1 showed a positive association between HDAC1 and HDAC2 expression, while Factor 2 demonstrated a negative association between H3K9Ac and HDAC6 expression (Table 3).

ANOVA comparisons of these factors revealed significant differences among groups for Factor 1 (p = 0.002) and Factor 2 (p < 0.001). Specifically, Factor 1 showed a significant difference between G1 and all other groups, while no significant differences were noted among the neoplastic groups. For Factor 2, significant differences were identified between G1 and G3, as well as between G1 and G4, but no differences were found among the neoplastic groups. These findings indicate a distinct expression pattern for the positive association between HDAC1 and HDAC2 in normal mammary tissue compared to neoplastic tissues, while the negative association between H3K9Ac and HDAC6 separated normal from malignant mammary tissues.

To further explore these differences, receiver operating characteristic (ROC) curve analyses were conducted to evaluate the sensitivity and specificity of these factors in distinguishing normal from neoplastic tissues. Among the key findings, a high sensitivity was noted for differentiating normal mammary tissue from simple adenomas (100%) and simple carcinomas (90.9%), as well as a high specificity for distinguishing normal mammary tissue from metastatic carcinomas (90.9%) for Factor 1. For Factor 2, a high specificity was observed in distinguishing normal mammary tissue from simple carcinomas (91.3%) and a high sensitivity was noted for differentiating normal mammary tissue from metastatic carcinomas (93.3%) (Table 3 and Figure 5).