Ethical approval for this study was granted by the Animal Ethics Committee of Can Tho University, Vietnam (CTU-AEC26001). The study involved client-owned dogs presented to the Veterinary Teaching Clinic, Can Tho University, Vietnam, for routine clinical evaluation and diagnostic investigation. No experimental infection, unnecessary invasive intervention, or non-therapeutic procedure was performed during the study period. All diagnostic procedures, including abdominal ultrasonography, blood collection, hematologic analysis, serum biochemical testing, and PCR-based screening, were conducted in accordance with standard veterinary clinical practice guidelines to minimize stress and discomfort in the animals.

Written informed consent was obtained from all dog owners before enrollment. Owners were informed about the objectives of the study, sample collection procedures, the use of anonymized clinical data, and the publication of ultrasonographic images for scientific purposes. Participation was voluntary, and all collected data were handled confidentially in accordance with institutional ethical standards.

This prospective, hospital-based observational study was conducted between January and November 2025 at the Veterinary Teaching Clinic, College of Agriculture, Can Tho University, Vietnam. The clinic functions as a tertiary referral center receiving both first-opinion and referred canine cases from urban and peri-urban areas within the Mekong Delta region.

All client-owned dogs presented to the Veterinary Teaching Clinic during the study period (n = 1,278 presentations) were screened for eligibility. Dogs were classified as having suspected hepatobiliary disease if they fulfilled at least one clinical criterion, including vomiting, diarrhea, anorexia, lethargy, icterus, cranial abdominal pain, abdominal distension/ascites, or suspected hepatomegaly, together with at least one clinicopathologic abnormality, including increased alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), hyperbilirubinemia, or hypoalbuminemia. Neonatal puppies aged <2 months and dogs lacking essential clinical, ultrasonographic, or laboratory data were excluded from the study.

Sample size was determined based on all eligible cases presented during the study period, and no a priori power calculation was performed because of the exploratory and clinic-based nature of the investigation. This study design provided a pragmatic framework for imaging-based phenotypic classification in settings where histopathology is infrequently available and where accessible imaging and molecular diagnostic tools are more clinically feasible.

Abdominal ultrasonography was performed using an ultrasound system (Mindray DP10VET, Mindray Bio-Medical Electronics Co., Ltd., Shenzhen, China) by a single clinician with >5 years of experience in small animal diagnostic imaging to minimize interobserver variability. Dogs were fasted for at least 8 h before examination whenever clinically feasible.

A microconvex probe (5–10 MHz) was used with standardized gain and depth settings. Dogs were examined in dorsal and lateral recumbency. The liver was evaluated for size, echogenicity, echotexture, hepatic margins, and vascular structures. Heterogeneous echotexture was defined as nonuniform parenchymal echogenicity, and blunt margins were defined as rounded hepatic edges. Hepatic hypoechogenicity/hyperechogenicity was assessed relative to the renal cortex.

Gallbladder wall thickening was defined as >2.0 mm, and gallbladder volume (VGB) was estimated using the ellipsoid formula:

The calculated value was normalized to body weight (BW; mL/kg) [6, 7]. Ultrasonographic findings were recorded as binary variables (present/absent) for phenotyping analyses (Table 1) [6, 7].

Dogs with ultrasonographic evidence of liver disease were classified into three imaging-based phenotypes (imaging-supported rather than biopsy-confirmed etiologic diagnoses). The hepatitis phenotype was characterized by diffuse parenchymal changes dominated by hypoechogenicity and heterogeneous echotexture, frequently accompanied by hepatomegaly and blunt hepatic margins, with or without hepatic venous dilation; concurrent gallbladder abnormalities were also possible [6, 7]. The hepatic lipidosis phenotype was defined by diffuse hepatic echogenicity, increased liver-to-renal cortical echogenicity, and reduced conspicuity of intrahepatic and portal vasculature, often accompanied by hepatomegaly. Severity grading (grades 1–3) was performed according to published ultrasonographic criteria [7, 8].

The cirrhosis phenotype was characterized by diffuse, heterogeneous echotexture, an atrophic liver, and irregular hepatic margins, frequently associated with peritoneal effusion and hepatic venous dilation [6, 7, 9].

These imaging-based classifications were interpreted as clinical phenotypes rather than definitive diagnoses, as histopathological confirmation was not obtained. Liver biopsy was not conducted because of clinical constraints, including cost, owner consent, and the invasive nature of the procedure in a routine clinical setting. Table 2 summarizes the ultrasonographic phenotypes identified in this study [7–9].

Hematologic analysis was performed using an automated hematology analyzer (Dymind DF50, Dymind Biotechnology Co., Ltd., Shenzhen, China), and serum biochemical analysis was conducted using a biochemical analyzer (MNCHIP Pointcare M4, MNCHIP Technologies Co., Ltd., Tianjin, China) according to manufacturer-validated protocols for canine samples. Reference intervals were based on manufacturer guidelines and previously published canine reference values [10].

Blood samples were processed within 2 h after collection to maintain analytical stability. Internal quality-control procedures were routinely performed in accordance with standard laboratory protocols.

For dogs exhibiting a hepatitis ultrasonographic phenotype, ethylenediaminetetraacetic acid-anticoagulated blood samples were subjected to conventional PCR targeting selected infectious agents, including vector-borne pathogens (Ehrlichia canis, Anaplasma platys, Babesia spp., and Hepatozoon canis), Leptospira spp., and CAdV-1 (Table 3) [11–16].

Genomic deoxyribonucleic acid was extracted using a commercial extraction kit (Isolate II Genomic DNA Kit, Meridian Bioscience, London, United Kingdom) according to the manufacturer’s instructions. DNA concentration and purity were evaluated before amplification. PCR amplification was performed using MyTaq™ HS DNA Polymerase (Bioline Reagents Ltd., London, United Kingdom) in a final reaction volume of 25 µL according to previously validated protocols [11–16]. All assays were conducted as singleplex reactions.

Each PCR run included a positive control, a no-template negative control, and an internal amplification control. Amplicons were separated by 1.5% agarose gel electrophoresis and interpreted based on expected fragment sizes. Blood-based PCR was selected as a noninvasive screening approach; however, its sensitivity may be reduced in cases of low pathogen load, inappropriate sampling timing, prior antimicrobial therapy, or infections localized primarily within hepatic tissue.

Hierarchical agglomerative clustering was performed using a binary matrix of predefined ultrasonographic variables coded as present or absent. Jaccard distance was selected to quantify pairwise dissimilarity because it is appropriate for asymmetric binary datasets and emphasizes shared positive features. Clustering was performed using the average linkage method.

Dogs with >20% missing ultrasonographic variables were excluded to minimize bias. For the remaining cases, missing binary values were imputed with the modal value to preserve sample size, although this approach may introduce bias and should therefore be interpreted with caution. As a sensitivity assessment, clustering results were visually compared with complete-case patterns and were found to be broadly consistent.

The dendrogram was predefined to obtain three clusters corresponding to the hepatitis, hepatic lipidosis, and cirrhosis phenotypes. Alternative cluster solutions ranging from two to five clusters were explored; however, these showed lower clinical interpretability and did not meaningfully improve phenotype separation.

Cluster stability and structural consistency were assessed qualitatively through dendrogram inspection and feature-grouping consistency. Clustering findings were interpreted as exploratory and hypothesis-generating rather than confirmatory. Clinicopathologic variables were subsequently summarized across clusters to facilitate interpretation of phenotypic patterns.

All clustering analyses were performed using R software version 4.0.1 (R Foundation for Statistical Computing, Vienna, Austria). Hierarchical agglomerative clustering was used to explore co-occurrence patterns of clinical signs and ultrasonographic findings. Binary variables were coded as presence (1) or absence (0). Pairwise dissimilarities between dogs were calculated using the Jaccard distance implemented in the vegdist function of the vegan package (method = “jaccard”, binary = TRUE). Clustering was then conducted using the hclust function with the average-linkage method (method = “average”).

Associations between host-associated factors and ultrasonographic positivity were evaluated using univariable logistic regression and reported as odds ratios (OR) with 95% confidence intervals (CI). Variables were initially screened using univariable analysis; however, multivariable modeling was not performed because of the limited sample size and sparse subgroup data. Therefore, all estimates should be interpreted as exploratory.

Body condition score was assessed using a standardized 9-point scale by the attending clinician. Breeds were categorized as exotic (purebred) or indigenous (local or mixed breed) for statistical analysis.

Proportions were calculated as the number of positive cases divided by the total number of dogs within the relevant category and were expressed as percentages. The 95% CI for proportions was estimated using the Wald approximation for binomial data and computed in R version 4.0.1. Because of the relatively small sample sizes in certain subgroups, some CI values were wide and should be interpreted with caution.

All statistical analyses, including logistic regression and clustering procedures, were performed using R software version 4.0.1. A p < 0.05 was considered statistically significant.

During January–November 2025, 1,278 dogs were presented to the Veterinary Teaching Clinic, Can Tho University, Vietnam. Of these, 111/1,278 dogs (8.69%; 95% CI: 7.27–10.31) were suspected of hepatobiliary disease and subsequently underwent abdominal ultrasonography. Liver abnormalities compatible with hepatobiliary disease were detected in 63/111 suspected dogs (56.76%; 95% CI: 47.53–65.55), corresponding to an overall detection proportion of 63/1,278 dogs (4.93%; 95% CI: 3.85–6.29). Table 4 summarizes the host-associated risk factor screening results for ultrasonographically confirmed liver disease among clinically suspected dogs.

No significant associations were identified for sex (p = 0.873) or housing system (p = 0.194). In contrast, breed, age, and body condition were associated with ultrasonographic positivity. Exotic breeds had higher odds of ultrasonographic positivity than indigenous breeds (OR, 2.29; 95% CI: 1.03–5.07). Compared with dogs aged ≤1 year, the odds of ultrasonographic positivity increased progressively with age, with the strongest associations observed in dogs aged >8 years (OR, 13.93; 95% CI: 2.71–71.55) and dogs aged 4–8 years (OR, 6.92; 95% CI: 1.29–37.05). Relative to dogs with normal body condition, obesity (OR, 4.78; 95% CI: 1.72–13.30) and underweight status (OR, 3.38; 95% CI: 1.15–9.87) were associated with increased odds of ultrasonographic positivity (p = 0.007).

Among dogs with ultrasonographic evidence of liver disease (n = 63), hepatitis was the most common imaging phenotype (36/63; 57.14%; 95% CI: 44.90–69.34%), followed by hepatic lipidosis (19/63; 30.16%; 95% CI: 18.86–41.46%) and cirrhosis (8/63; 12.70%; 95% CI: 4.50–20.90%) (Table 5).

Clinical signs in dogs diagnosed with hepatitis were dominated by nonspecific gastrointestinal manifestations (Figures 1–2). Anorexia was the most frequent clinical sign (34/36; 94.44%), followed by vomiting (30/36; 83.33%) and diarrhea (18/36; 50.00%). Icterus was observed in 13 of 36 dogs (36.11%). Abdominal pain was recorded in 9/36 dogs (25.00%), ascites in 7/36 dogs (19.44%), whereas muscle tremors were uncommon (2/36; 5.55%).

Ultrasonography revealed a consistent hepatobiliary pattern among dogs with hepatitis (Figures 1 and 3). Diffuse hepatic parenchymal hypoechogenicity was the predominant abnormality (32/36; 88.89%) and was frequently accompanied by hepatomegaly (19/36; 52.77%). Dilated hepatic veins and blunt hepatic margins were each detected in 18/36 dogs (50.00%). Gallbladder-associated abnormalities were common, including increased VGB (15/36; 41.70%) and gallbladder wall thickening (11/36; 30.56%). Biliary sludge or cholelithiasis was identified in 7/36 dogs (19.44%). Representative ultrasonographic images demonstrated hepatitis with concurrent biliary abnormalities, including heterogeneous hepatic echotexture with gallbladder wall thickening (4.98 mm) and gallbladder distension (VGB = 19.94 mL) (Figure 4), as well as hepatic venous dilation with gallbladder distension (Figure 5).

Hematologic and biochemical classifications are summarized in Table 6. Thrombocytopenia and increased hepatobiliary enzyme activities were frequently observed in dogs with hepatitis.

PCR etiologic screening results for dogs with a hepatitis phenotype are summarized in Table 7. An infectious etiology was identified in 20/36 dogs (55.56%), including vector-borne pathogens (15/36; 41.67%), Leptospira spp. (3/36; 8.33%), and CAdV-1 (2/36; 5.56%). Dogs with positive PCR findings were excluded from the PCR-negative subgroup used for descriptive comparison. The PCR-negative subgroup comprised 16/36 dogs (44.44%) that tested negative for all PCR targets.

Clinical and ultrasonographic findings in dogs with hepatic lipidosis are summarized in Figure 6. The most frequent clinical manifestations were abdominal distension/discomfort (17/19; 89.47%), followed by inappetence (10/19; 52.63%) and diarrhea (10/19; 52.63%). Vomiting (7/19; 36.84%) and lethargy (6/19; 31.57%) occurred at intermediate frequencies, whereas jaundice was less common (4/19; 21.05%).

Ultrasonographically, hepatic lipidosis was characterized predominantly by diffuse parenchymal abnormalities, including increased hepatic echogenicity and hepatomegaly (each 13/19; 68.42%), reduced conspicuity of intrahepatic and portal vasculature (13/19; 68.42%), and increased liver-to-renal cortical echogenicity (12/19; 63.16%). Perihepatic and biliary-associated abnormalities were identified in subsets of dogs, including pericholecystic fat accumulation (10/19; 52.63%) and blunt hepatic margins with increased parenchymal brightness (8/19; 42.11%). In contrast, overt gallbladder abnormalities were relatively uncommon, with gallbladder wall thickening observed in 2/19 dogs (10.53%) and gallbladder enlargement in 1/19 dog (5.26%) (Figure 6).

Representative ultrasonographic images supported these phenotypes (Figures 7–8). Grade 3 hepatic lipidosis was demonstrated by diffuse hepatic hyperechogenicity, reduced visualization of portal venous structures, and a hyperechoic layer adjacent to the gallbladder wall, consistent with pericholecystic fat (Figure 7A–B). A dog with concurrent cholecystitis and grade 1 hepatic lipidosis exhibited gallbladder wall thickening with increased echogenicity and heterogeneous hepatic parenchyma containing focal hyperechoic areas (Figure 8). Hematologic and biochemical classifications are summarized in Table 8. Hematologic and biochemical classifications are summarized in Table 8.

Hierarchical clustering analysis of dogs with cirrhosis identified two principal phenotypic patterns (Figure 9). One low-feature cluster exhibited relatively few concurrent abnormalities, including 1–2 clinical and ultrasonographic features, whereas the remaining dogs formed a higher-feature cluster characterized by multiple co-occurring clinical signs (4–5 features) and concordant ultrasonographic abnormalities (4–5 features), consistent with more advanced structural and portal hypertensive changes.

Ascites was the most frequent clinical manifestation (7/8; 87.50%), followed by jaundice (6/8; 75.00%). Vomiting and anorexia were each observed in 4/8 dogs (50.00%). Muscle tremors occurred in 3/8 dogs (37.50%), whereas chronic weight loss and poor body condition were each recorded in 2/8 dogs (25.00%) (Figure 9). Representative clinical appearances are presented in Figure 10.

Ultrasonographically, diffuse heterogeneous hepatic echotexture was identified in all dogs with cirrhosis (8/8; 100%). Peritoneal effusion occurred in 7/8 dogs (87.50%). A small or atrophic liver and irregular hepatic margins were each observed in 6/8 dogs (75.00%), whereas dilated hepatic veins were identified in 5/8 dogs (62.50%) (Figure 9). Representative ultrasonographic images demonstrated increased hepatic edge echogenicity with abdominal effusion and color Doppler evidence of hepatic venous dilation (Figures 11–12). Hematologic and biochemical classifications are summarized in Table 9.

Among clinically suspected dogs, ultrasonographic positivity increased with age and was associated with breed and extremes of body condition, whereas sex and housing system were not associated. The age-related increase, particularly in dogs aged >8 years, is consistent with the established epidemiology of chronic hepatopathies in small animal practice [17]. The association observed in exotic breeds may reflect referral bias or a potential genetic predisposition; however, this finding should be interpreted with caution because population-level denominators were unavailable. The observed association with both obesity and underweight status highlights the importance of metabolic imbalance and systemic disease in the clinical context of hepatobiliary disorders [17, 18].

An important limitation of this study is the reliance on ultrasonographic phenotyping without histopatho-logical confirmation. Consequently, overlap among imaging findings and potential misclassification are likely and should be carefully considered when interpreting the findings. Chronic hepatitis may be particularly difficult to distinguish from early-stage hepatic fibrosis or reactive hepatopathy because these conditions frequently share overlapping ultrasonographic characteristics, including heterogeneous echotexture, altered echogenicity, and variable liver size. Early fibrotic remodeling may not yet produce the architectural distortion or nodular surface irregularity characteristic of advanced cirrhosis, thereby limiting the diagnostic specificity of ultrasonography. Therefore, differentiation between inflammatory and early fibrotic liver disease remains inherently constrained when based solely on imaging findings, and histopathology remains the gold standard for definitive diagnosis and disease staging [17–21]. These considerations indicate that the phenotypic classifications described in this study should be interpreted as clinically pragmatic rather than diagnostically definitive.

The distribution of ultrasonographic phenotypes (hepatitis > hepatic lipidosis > cirrhosis) reflects the spectrum of cases commonly referred for imaging evaluation. Importantly, these categories should be interpreted as imaging-supported clinical phenotypes rather than etiologic diagnoses, because ultrasonography is primarily effective at identifying structural abnormalities and associated complications but has limited specificity in differentiating underlying causes. This limitation is particularly relevant when distinguishing chronic hepatitis from early fibrotic liver disease, which may demonstrate similar ultrasonographic appearances in routine clinical practice. In selected cases, advanced imaging modalities may provide complementary diagnostic information, particularly in complex hepatobiliary or biliary disease presentations [19].

Dogs classified with the hepatitis phenotype most commonly exhibited nonspecific gastrointestinal signs together with diffuse hepatic parenchymal hypoechogenicity, frequently accompanied by hepatomegaly and concurrent gallbladder abnormalities. These findings are consistent with previous descriptions of inflammatory hepatopathies, in which hepatocellular injury and cholestasis are reflected in both ultrasonographic and biochemical alterations [7, 17, 20, 22]. However, gallbladder wall thickening should be interpreted with caution because it may also occur under physiologic conditions, such as fasting, or in systemic disorders, such as hypoalbuminemia, in which edema rather than primary biliary inflammation is the underlying mechanism [23, 24].

A major strength of the present study is the incorporation of PCR-based screening for selected infectious etiologies, including blood-borne pathogens, Leptospira spp., and CAdV-1, in dogs classified with the hepatitis phenotype. This integrated approach reduces, although does not eliminate, the risk of misclassifying acute infectious hepatitis as chronic hepatopathy, particularly considering the possibility of false-negative PCR results associated with low pathogen load, inappropriate timing of sample collection, or previous antimicrobial therapy [25–27].

Dogs with hepatic lipidosis demonstrated characteristic ultrasonographic findings, including diffuse hyperechogenicity, hepatomegaly, and reduced visualization of intrahepatic vasculature, consistent with lipid infiltration altering acoustic interfaces [7, 25, 29]. Clinicopathologic findings further supported a metabolic component, with frequent alterations in hepatocellular enzyme activities and lipid metabolism [28].

In contrast, dogs with cirrhosis exhibited ultrasonographic and clinical findings consistent with advanced chronic liver disease, including ascites, heterogeneous hepatic echotexture, atrophic liver morphology, and vascular abnormalities suggestive of portal hypertension. These findings align with previous reports and reinforce the clinical relevance of ultrasonography in identifying late-stage structural alterations and associated complications [7, 20, 21, 29, 30].

The hierarchical clustering analysis provided an additional exploratory framework for assessing co-occurrence patterns among clinical and ultrasonographic abnormalities. The use of Jaccard distance for binary variables was methodologically appropriate; however, the number of clusters was predefined according to clinical phenotypes, and alternative cluster solutions were not formally optimized using cluster-validity indices. Consequently, these findings should be interpreted as descriptive and hypothesis-generating rather than confirmatory.

Several additional limitations should be acknowledged. The single-center, referral-based study design may introduce selection bias and limit generalizability to primary-care canine populations. The relatively small sample size of the cirrhosis subgroup (n = 8) reduced statistical power for subgroup comparisons. In addition, the absence of multivariable analysis limited the ability to control for potential confounding variables; therefore, the reported associations should be interpreted as exploratory. Furthermore, PCR-negative findings do not exclude infection, as detection may be influenced by pathogen load, sampling timing, prior antimicrobial treatment, and tissue-localized infections that may not be detectable in peripheral blood samples.

Despite these limitations, the present study demonstrates that an integrated diagnostic approach combining ultrasonography, clinicopathologic assessment, and targeted PCR screening can provide a practical and clinically applicable framework for the initial evaluation and triage of canine hepatobiliary disease. Future multicenter studies incorporating histopathology and expanded etiologic diagnostic testing are warranted to validate these findings, reduce potential misclassification, and improve diagnostic accuracy and disease characterization.