All experimental procedures were reviewed and approved by the Institutional Animal Care and Use Committee (Approval No. MUVS-2020-09-41). Written informed consent was obtained from all dog owners before surgery and sample collection. Gingival mass excision was performed as part of routine clinical management at the Small Animal Hospital, Faculty of Veterinary Science, Mahidol University. All dogs received standard anesthesia, analgesia, and postoperative monitoring in accordance with institutional clinical protocols. Animals were humanely handled throughout the study to minimize pain, discomfort, and distress. Fine-needle aspiration (FNA) of excised gingival masses was performed immediately after surgical removal under sterile conditions. All procedures complied with institutional animal welfare policies and followed the ethical principles outlined in the Animal Research: Reporting of In Vivo Experiments 2.0 guidelines.

This study was conducted over a 15-month period from December 2020 to February 2022. Clinical sampling took place at the Small Animal Hospital, Faculty of Veterinary Science, Mahidol University, Nakhon Pathom, Thailand. Cytological processing and evaluation were performed at the Pathology Laboratory Unit, Center for Veterinary Diagnosis, Faculty of Veterinary Science, Mahidol University, Thailand.

This prospective diagnostic accuracy study enrolled consecutive canine patients presenting with gingival masses, using histopathology as the reference standard. Dogs of any age, sex, or breed were eligible for inclusion provided they were free from systemic illness, had not received prior treatment for oral lesions, and were clinically fit for general anesthesia based on physical examination and routine preoperative assessment.

Cytological samples were collected from gingival masses measuring at least 2 cm in diameter, which aligns with the conventional clinical staging threshold for canine oral tumors [22]. Lesions on both the maxillary and mandibular gingiva were included. Exclusion criteria included poor cytological smear quality and the absence of a definitive histopathological diagnosis.

Immediately after surgical excision, FNA was performed on each gingival mass to mimic preoperative cytological sampling. Three smears per lesion were prepared using standardized techniques and assigned to the following analyses: Romanowsky staining for cytological classification and cytomorphometric analysis, silver staining of AgNORs for proliferative activity, and acridine orange staining for micronuclei testing. The study workflow, including screening, eligibility assessment, enrollment, and analysis, is shown in Figure 2.

Demographic data, including breed, sex, and age, were recorded for each enrolled dog and used as descriptive covariates in the analysis.

A total of 46 eligible cases were included during the study period. For comparisons among the three diagnostic groups, this sample size provided about 80% statistical power (α = 0.05) to detect medium-to-large effect sizes (Cohen’s f ≈ 0.35). In correlation analyses, the Fisher z-standard error (≈ 0.152) resulted in a 95% confidence interval half-width of roughly 0.29 for Pearson’s correlation coefficient, indicating acceptable precision for correlation estimates.

Romanowsky-type staining was performed using a modified Wright–Giemsa protocol as previously described [23]. Cytological evaluation was conducted under a light microscope (ACCU-SCOPE Inc., Commack, NY, USA), starting with initial low-magnification assessment and followed by detailed examination at 400× magnification to assess cellular morphology. Gingival masses were cytologically classified into three categories, epithelial hyperplasia, benign neoplasms, and malignant neoplasms, based on established cytological interpretation and malignancy criteria guidelines [24, 25].

Histopathologic evaluation was performed independently by a veterinary pathologist certified by the Thai Board of Veterinary Pathology (DTBVP), who was blinded to the cytological findings. Lesions were classified based on tissue structure, cellular atypia, invasiveness, and mitotic activity. Initial cytological assessments were carried out by a licensed veterinarian experienced in diagnostic cytology, and all final cytological diagnoses were confirmed by a DTBVP-certified veterinary pathologist.

Only cytological smears with optimal preservation, minimal background debris, clear nuclear detail, and limited cell overlap (<30%) were included. One representative smear per staining technique was analyzed for each case. A minimum of 500 well-preserved epithelial cells per smear was required for inclusion. Systematic, non-overlapping high-power fields were selected for evaluation.

All cytological assessments, including cytomorphometric measurements, AgNOR counts, and micronuclei scoring, were conducted by a single licensed veterinarian who was blinded to the histopathological diagnosis and group allocation. Each slide was evaluated twice in separate sessions, and the average values were used for analysis. To assess repeatability, 10% of slides were re-evaluated, with results remaining within predefined quality control limits (≤15% variation for continuous variables and ≥90% concordance for categorical variables). All measurements and data were subsequently reviewed and validated by a DTBVP-certified veterinary pathologist.

Microscopic imaging was conducted using an Accu-Scope 3000-LED light microscope and a Nikon Eclipse 80i fluorescence microscope equipped with a Nikon DS-Fi3 digital camera. Fluorescence imaging used excitation wavelengths of 485–492 nm and emission filters of 520–550 nm. Image-analysis software (CaptaVision+ version 2.3, ACCUSCOPE, Inc., USA) was calibrated with a 10-μm stage micrometer.

Cytomorphometric evaluation was performed following the criteria described by Bhavasar et al. [20]. From modified Wright–Giemsa–stained smears, the first 100 non-overlapping, well-preserved tumor cells were analyzed at 400× magnification. Nuclear and cytoplasmic boundaries were manually outlined using CaptaVision+ software (version 2.3, ACCU-SCOPE Inc., Commack, NY, USA). Anucleate, degenerated, and overlapping cells were excluded. Measured parameters included NA, CA, ND, and CD. The nuclear-to-cytoplasmic (N:C) ratio was calculated for each cell and averaged for each case.

The micronuclei assay was conducted using acridine orange staining as previously described [26]. A 0.1% (w/v) acridine orange stock solution was prepared and stored at 4 °C, with a freshly diluted working solution at 1:15 in phosphate buffer (pH 6.8–7.2). Slides were stained for 3 minutes in a dark chamber at 22°C–25°C, rinsed three times with phosphate buffer, air-dried, and examined under a fluorescence microscope (Nikon Corporation, Tokyo, Japan). Batch-to-batch quality control involved preparing fresh working solutions and verifying fluorescence uniformity.

Micronuclei were identified and scored using established criteria [27, 28]. They were defined as round-to-oval extranuclear chromatin bodies located in the same focal plane and measuring about one-third to one-sixth the diameter of the main nucleus. A total of 500 well-preserved mononucleated epithelial cells per slide were evaluated at 400× magnification. Binucleated cells, apoptotic or karyorrhectic nuclei, and cells with cytoplasmic fragmentation were excluded. All micronuclei within a single cell were counted, and overlapping or out-of-focus structures were ignored to reduce artifacts.

AgNOR staining was conducted using a modified version of a previously established protocol [29]. The working solution included a mixture of gelatin–formic acid (solution A) and freshly prepared silver nitrate (solution B) for each staining batch. Slides were incubated in a dark, humid chamber at 22°C–25°C for 40 min, rinsed twice with distilled water, and air-dried. Quality control was ensured by including a standard reference smear in each staining run and by verifying staining consistency.

AgNOR dots appeared as black intranuclear granules and were evaluated at 400× magnification in 200 non-overlapping nuclei per slide. Parameters assessed included the mean number of AgNOR dots per nucleus, the shape of the dots (round, oval, irregular), and their intranuclear distribution (central, border, or diffuse). Discrete dots were counted individually, while confluent clusters were counted as a single structure when separation was unclear. Mean values were calculated for each case.

All statistical analyses were conducted using SPSS version 21.0 (IBM Corp., NY, USA), with a significance level set at p < 0.05. Data normality and homogeneity of variances were evaluated using the Shapiro–Wilk and Levene’s tests, respectively. Group differences were examined with one-way analysis of variance (ANOVA) within the general linear model, followed by Duncan’s test and least significant difference post hoc comparisons. Effect sizes were calculated using eta squared (η²). Relationships between continuous variables were analyzed using Pearson’s correlation coefficient, while associations involving categorical variables were assessed with Chi-square or Spearman’s rank correlation tests, as appropriate. Each dog was considered an independent statistical unit for all analyses.

A total of 46 dogs with gingival masses participated in this study, including 22 males and 24 females, with an average age of 9.9 ± 3.8 years (range: 3–17 years). The most common breed was mixed-breed dogs (52.2%), followed by Labrador Retrievers (13.0%), Shih Tzus (8.7%), and Pomeranians (6.5%). Histopathological examination verified all cytological diagnoses.

Based on combined cytological and histopathologic evaluation, 11 cases were classified as epithelial hyperplasia, 14 as benign neoplasms, and 21 as malignant neoplasms. The mean ages of dogs in the hyperplastic, benign, and malignant groups were 9.3, 8.6, and 11.3 years, respectively. Statistical analysis showed no significant differences among groups for age (p = 0.261) or sex distribution (p = 0.445). Detailed demographic data across diagnostic categories are summarized in Table 1.

For quantitative assessment, a total of 23,000 epithelial cells were evaluated in the micronuclei assay (500 cells per case), 9,200 nuclei were analyzed for AgNOR staining (200 nuclei per case), and 4,600 cells were included in cytomorphometric analysis (100 cells per case). Specifically, 5,500, 7,000, and 10,500 cells from epithelial hyperplasia, benign neoplasms, and malignant neoplasms, respectively, were assessed for micronuclei frequency. For AgNOR evaluation, 2,200, 2,800, and 4,200 nuclei were examined, while 1,100, 1,400, and 2,100 cells from the same respective groups were analyzed for cytomorphometric parameters (Table 2).

Representative Romanowsky-stained smears, AgNOR preparations, and micronucleus images for each diagnostic category are shown in Figure 3.

Cytomorphometric analysis revealed clear differences among diagnostic groups. The epithelial hyperplasia group showed the highest mean ND (10.92 ± 1.80 μm), followed by malignant (10.42 ± 2.16 μm) and benign neoplasms (9.98 ± 2.22 μm). Likewise, the mean CD was largest in epithelial hyperplasia (30.49 ± 9.82 μm), compared to malignant (19.22 ± 9.20 μm) and benign lesions (18.65 ± 8.55 μm).

The highest NA and CA were also observed in epithelial hyperplasia (94.36 ± 31.85 μm² and 766.89 ± 444.52 μm², respectively), followed by malignant and benign neoplasms. In contrast, the nuclear-to-cytoplasmic (N:C) ratio was highest in benign neoplasms (0.44 ± 0.23), slightly lower in malignant neoplasms (0.41 ± 0.19), and lowest in epithelial hyperplasia (0.17 ± 0.14).

Statistical analysis showed significant differences between groups for ND, NA, CA, and N:C ratio (F(2,43)=62.4–710.7, P < 0.01, η² = 0.03–0.24). Of these variables, the N:C ratio had the largest effect size (η² = 0.24), followed by CA (η² ≈ 0.22). A gradual morphological change was clear, with decreasing nuclear and cytoplasmic sizes and increasing N:C ratio from epithelial hyperplasia through benign to malignant lesions. Partial overlap of N:C ratios between benign and malignant groups indicated a continuum of cytological change rather than a strict categorical distinction. These results are summarized in Table 3.

Micronuclei were clearly identified under fluorescence microscopy as round-to-oval extranuclear chromatin bodies with fluorescence intensity similar to the main nucleus (Figure 4). The malignant group showed the highest average micronuclei count (7.76 ± 2.10; 95% CI: 6.93–8.59), followed by benign neoplasms (4.93 ± 2.73; 95% CI: 4.00–5.86) and epithelial hyperplasia (3.73 ± 1.34; 95% CI: 3.24–4.22).

One-way ANOVA revealed significant differences in micronuclei frequency across the three diagnostic groups (F (2,43) = 14.40, p < 0.01, η² = 0.40), showing a strong influence of lesion type on genomic instability. A steady increase in micronuclei frequency was observed from epithelial hyperplasia to benign and malignant neoplasms, aligning with increased chromosomal damage during neoplastic progression. These findings are summarized in Table 4.

AgNOR dots appeared as black intranuclear granules under light microscopy (Figure 5). The epithelial hyperplasia group had the lowest average AgNOR count (1.60 ± 0.75; 95% CI: 1.33–1.87), with most dots being round (96.91%) and located centrally within the nucleus (89.73%). Rare oval and irregular shapes, and minimal diffuse distribution, were observed.

In benign neoplasms, the average AgNOR count rose to 2.43 ± 1.17 (95% CI: 2.02–2.84). AgNOR dots mostly remained round (97.71%), with a more varied intranuclear distribution, including increased peripheral (21.32%) and diffuse (14.93%) localization.

Malignant neoplasms showed the highest average AgNOR count (4.04 ± 2.81; 95% CI: 3.12–4.96), along with a higher proportion of irregularly shaped dots (5.07%) and a significant shift toward diffuse intranuclear distribution (33.86%). Central localization decreased to 46.74%, indicating increased nuclear disorganization. Overall, AgNOR number, shape irregularity, and diffuse distribution progressively increased from epithelial hyperplasia to malignant lesions, reflecting increased proliferative activity and cytological atypia.

Statistical analysis confirmed significant differences in AgNOR counts among groups (F (2,43) = 5.78, p < 0.01, η² = 0.21), indicating that lesion classification explained about 21% of the variance in AgNOR number. Detailed AgNOR parameters are shown in Table 5, with proportional distributions depicted in Figure 6. Figure 7 shows the concurrent upward trends of AgNOR and micronuclei counts across diagnostic categories.

Data normality was confirmed using the Shapiro–Wilk test, allowing the use of Pearson’s correlation coefficient for parametric analysis. Correlation results among cytomorphometric parameters, AgNOR counts, and micronuclei frequency are summarized in Table 6.

The N:C ratio showed moderate positive correlations with micronuclei frequency (r = 0.317, p = 0.032, 95% CI: 0.03–0.56, r² = 0.10) and AgNOR count (r = 0.319, p = 0.030, 95% CI: 0.04–0.56, r² = 0.10). A strong positive correlation was observed between AgNOR count and micronuclei frequency (r = 0.631, p < 0.01, 95% CI: 0.42–0.78, r² = 0.40), indicating a close link between proliferative activity and genotoxic damage.

These findings confirm that increases in proliferative and genotoxic indices correspond with cytomorphometric progression in canine gingival lesions. The relationships among key quantitative variables are shown in Figure 8.

In routine veterinary practice, cytological diagnosis mainly depends on subjective interpretation of cellular morphology to distinguish neoplastic from non-neoplastic cells. However, relying solely on morphology may limit diagnostic accuracy, especially in borderline or early-stage lesions. This study shows that adding cytomorphometric analysis, AgNOR staining, and the micronuclei assay to routine cytology greatly improves diagnostic detail. We found significant positive correlations among these three methods, indicating that their combined use offers an objective and biologically relevant assessment of cellular proliferation and genotoxicity. This combined tri-marker approach enables a multidimensional cytological evaluation beyond morphology alone, enhancing the ability to differentiate benign from malignant gingival lesions before histopathologic confirmation. Importantly, these quantitative measures may also give insights into tumor aggressiveness and aid in clinical decision-making for managing potentially malignant gingival lesions in dogs. Additionally, because these markers are objective, they lay the groundwork for automated or AI-assisted cytological scoring systems that can reduce observer bias.

Cytomorphometry has been extensively used in human oral oncology as a quantitative tool for detecting malignancy, with parameters such as ND, NA, CD, CA, and nuclear-to-cytoplasmic (N:C) ratio showing strong diagnostic importance [30–32]. However, studies using cytomorphometric analysis of canine gingival tumors remain limited. To address this gap, our study found significant differences in ND, NA, and CA among epithelial hyperplasia, benign neoplasms, and malignant neoplasms (p < 0.01), with epithelial hyperplasia consistently exhibiting the highest mean values.

Specifically, epithelial hyperplasia showed larger nuclear and cytoplasmic sizes than neoplastic lesions, likely indicating lower cellular pleomorphism and more consistent nuclear structure. In neoplastic cells, greater variation in nuclear and cytoplasmic shape and size may lower average morphometric values despite increased atypia. Conversely, the N:C ratio, an established marker of malignancy, was significantly higher in both benign and malignant neoplasms compared to epithelial hyperplasia, supporting its usefulness in distinguishing reactive from neoplastic processes. Besides the N:C ratio, most cytomorphometric parameters were notably higher in malignant than in benign lesions, with values rising progressively with lesion severity. These results agree with previous canine and human studies showing increased nuclear sizes and decreased cytoplasmic maturation in malignant cells, due to heightened proliferative activity and altered cellular differentiation [33–37]. The use of digital morphometric workflows further improves reproducibility and facilitates integration into automated image-analysis platforms.

Micronuclei assays are well-established indicators of genotoxic damage caused by chromosomal breakage or mis-segregation during abnormal mitosis. Persistent DNA damage can result in micronuclei formation, genomic instability, and downstream cellular effects such as apoptosis or mutagenesis, all of which contribute to carcinogenesis [38]. Consistent with this biological rationale, an increase in micronuclei frequency from normal or hyperplastic epithelium to malignant lesions has been reported as a marker of neoplastic progression [39, 40]. Despite this importance, micronuclei assays are still underused in veterinary oncology, with only limited studies in animal buccal cells [41–43].

In the present study, micronuclei frequency increased significantly with lesion severity (p < 0.01), with mean counts of 3.73 ± 1.3 in epithelial hyperplasia, 4.93 ± 2.7 in benign neoplasms, and 7.76 ± 2.1 in malignant neoplasms. These results suggest increasing genomic instability in malignant cells and support the diagnostic value of micronuclei scoring for detecting cellular transformation. Our findings mirror observations in human oral squamous cell carcinoma, where micronuclei frequency is highest in malignant lesions and lowest in normal mucosa [44, 45]. Previous studies also show that combining micronucleus analysis with routine cytology improves the early detection of oral neoplasia in humans [46], and recent veterinary data confirm its usefulness as a biomarker of pathological cellular changes in dogs [47]. Overall, these findings support the use of micronucleus assays to assess genotoxic effects, monitor disease progression, and guide prognosis in canine oral tumors.

AgNOR staining indicates ribosomal biogenesis and protein synthesis needs in actively dividing cells, making it a sensitive marker of cellular proliferation [48]. Although AgNOR has been studied as a proliferative marker in various canine tumors [49–52], its use in cytology for canine gingival masses has been limited. In this study, AgNOR counts varied significantly among epithelial hyperplasia, benign neoplasms, and malignant neoplasms (p < 0.01), with a stepwise increase corresponding to histologic severity.

Mean AgNOR counts increased from 1.60 ± 0.75 in epithelial hyperplasia to 2.43 ± 1.17 in benign neoplasms and 4.04 ± 2.81 in malignant neoplasms. These findings align with human studies showing increased AgNOR expression in higher-grade oral lesions [53, 54]. In addition to numerical differences, AgNOR morphology and intranuclear distribution provided further diagnostic insights. Malignant lesions showed a higher proportion of irregularly shaped AgNOR dots and a significant shift toward diffuse nuclear distribution, whereas hyperplastic lesions predominantly had round, centrally located dots. These patterns match previous reports linking diffuse and irregular AgNOR configurations with increased proliferative activity and malignancy [55–57]. Therefore, the combined assessment of AgNOR count, shape, and distribution offers a useful tool for distinguishing premalignant and malignant lesions in cytological samples.

The observed link between AgNOR activity and micronuclei frequency indicates a biologically meaningful connection between cell growth and genetic instability. Cells that are dividing rapidly show increased ribosomal production and replication stress, which can overwhelm DNA repair systems and lead to chromosomal damage, as evidenced by micronuclei [58, 59]. The simultaneous increase of these markers across lesion severity supports a mechanistic connection between proliferation and DNA damage during malignant transformation. Significantly, the extent and direction of these changes in canine gingival lesions closely match those reported in human oral lesions and breast tumors [60–62], emphasizing shared biological pathways and strengthening the translational relevance of this three-part cytological method. These cross-species similarities suggest that canine gingival tumors could serve as a useful spontaneous model for certain aspects of human oral tumor biology.

A key finding of this study was the significant correlation among cytomorphometric indices, AgNOR counts, and micronuclei frequency. The N:C ratio showed a moderate correlation with both micronuclei and AgNOR counts, while a strong positive correlation was observed between AgNOR and micronuclei frequencies. To our knowledge, this is the first study to quantitatively link proliferative, genotoxic, and morphometric parameters within a single cytological dataset in canine gingival lesions. These correlations provide objective evidence that concurrent evaluation of these markers reflects lesion severity and malignant potential. The parallel increase in AgNOR and micronuclei counts across diagnostic categories (Figure 7) visually and statistically supports this bidirectional relationship.

From a diagnostic standpoint, this integrated approach reduces reliance on subjective morphology and reduces the risk of false-negative results by incorporating measurable biological indicators. The combined evaluation of nuclear morphology, proliferative activity, and genotoxic damage improves consistency, supports earlier detection of malignant transformation, and may assist in presurgical planning and post-treatment follow-up. Furthermore, the image-based and quantitative nature of these analyses naturally aligns with digital pathology and AI-assisted cytology platforms.

Despite these strengths, the moderate sample size and lack of longitudinal clinical outcome data may limit the generalizability of the findings. Future studies should validate this trimodal framework with larger, more diverse groups, examine its usefulness for other oral and skin neoplasms, and include additional proliferative markers, such as Ki-67 or proliferating cell nuclear antigen. Testing this method in liquid-based or brush cytology and linking it to long-term clinical results would also help clarify its prognostic value. Overall, such studies could establish this tri-marker cytological system as a reliable tool in veterinary diagnostics and comparative oncology.