Characterization of clinical presentation, histological features, ultrasonographic findings, and survival in 29 dogs with granulomatous hepatitis

2.1 Case selection

Ours was a retrospective observational study. The electronic medical record database at the Cummings School of Veterinary Medicine at Tufts University was searched for the diagnosis of GH between the years of 2010 and 2021. Inclusion criteria included a hepatic biopsy sample judged to be adequate for diagnosis, a morphologic diagnosis of GH based on World Small Animal Veterinary Association (WSAVA) published guidelines by a single board-certified pathologist (NMP) in consultation with a board-certified small animal internist with expertise in hepatic disease (CRLW).¹ Each dog also needed to have a medical record available for review.

Nineteen of 29 archived hepatic biopsy samples had adequate tissue to be graded and staged. A single board-certified anatomic pathologist (NMP) reviewed the histopathologic slides of all 19 cases and formulated pathological scores for grade (inflammation, necrosis) and stage (fibrosis), based on established criteria.^{1, 11} Ultrasound images and video clips from dogs were reviewed by a board-certified veterinary radiologist (LAS) using the American College of Veterinary Radiology ultrasound consensus statement as a guideline.¹² In brief, the liver was evaluated for size, margination, echogenicity and the presence of focal parenchymal lesions. Notations were made when these lesions were located adjacent to vasculature and if they deformed vascular margins. Liver parenchymal echogenicity was described as either homogenous or heterogeneous. The presence or absence of abdominal effusion and its localization (diffuse vs localized between liver lobes) as well as the echogenicity of the effusion (anechoic vs hyperechoic) and quantity (mild, moderate, severe) were noted. In dogs in which the hepatic lymph node was identified, its size and echogenicity were noted. The gallbladder was evaluated for thickness and character of the gallbladder wall. Luminal bile echogenicity and the presence of a gallbladder mucocele or biliary sludge were noted. The presence of the following focal hepatic lesions was recorded and tabulated: hypoechoic nodules, hyperechoic nodules, hyperechoic bands of liver parenchyma surrounding vasculature, hyperechoic nodules with a peripheral hypoechoic rim, and cystic changes in liver. The location of the lesions in the left, right or central liver was assessed. These lesions also were described as having either a peripheral or central location within the liver lobes. Notations were made when these lesions were located adjacent to vasculature and if they deformed vascular margins.

Dogs were excluded if they had positive diagnostic tests for infectious diseases known to be associated with granulomatous inflammation (eg, leishmaniasis, mycobacteriosis, bartonellosis, leptospirosis, toxoplasmosis, neosporosis, systemic fungal, or rickettsial disease). Not all dogs, however, had an extensive diagnostic evaluation for infectious diseases likely because of multiple factors including owner financial constraints, the low incidence of certain infectious diseases in the area, and the knowledge that previous publications in both dogs and humans have shown that extensive testing for infectious disease was of low yield in GH.^{3, 13-15} All dogs with clinical, biochemical, or imaging findings suggestive of multi-systemic granulomatous disease in organ systems other than the liver were excluded. Other exclusion criteria were a diagnosis of cancer, recent exposure (within 1 month) to hepatotoxic drugs or supplements known to elicit a granulomatous response, or hepatic copper concentrations >1000 μg/g dry weight with documented response to hepatic copper chelation treatment. Dogs with increased hepatic copper concentrations that failed to respond to adequate chelation treatment, defined as persistently increased liver transaminase activities, were included in the study because these cases were at least in part considered to have clinically relevant GH in conjunction with copper hepatopathy.^{16, 17}

2.2 Data collection

The following data were extracted and summarized from the medical record: signalment; history; clinical signs; physical examination findings; CBC and serum biochemistry test results; hepatic histopathology reports, including semi-quantitative (rhodamine staining) and quantitative copper assessment; and, associated ancillary testing for infectious organisms, liver and gallbladder aerobic and anaerobic bacterial culture results, and ultrasonographic findings. Treatment information was recorded and categorized as use of hepatoprotectants, antibiotics, or immunosuppressive drugs.

A clinical severity score was calculated for each dog at the time of biopsy, using a classification system adapted from the Child-Pugh score used in humans and previously applied to dogs with chronic hepatitis (see Data S1^{17, 18}).

Clinical remission was defined as the lack of any clinical signs as reported by the owner. Biochemical remission was defined as normalization of serum alanine transaminase (ALT) activity. Because of the possible effect of corticosteroid administration on serum ALT activity, we also evaluated normalization of serum aspartate aminotransferase (AST) activity as an indication of response to treatment.¹⁹ We defined short-term and long-term survival as survival of < or >30 days, respectively.

The hospital's consent-to-treat form, signed by all owners, permitted the use of archived hepatic biopsy tissue and information in each animal's electronic medical information for research purposes.

2.3 Outcome

The outcome for each dog was determined by medical record review in combination with follow-up with the owner and referring veterinarian, as needed, via telephone. Outcomes were categorized as alive, lost to follow-up, death, or euthanized because of hepatic or non-hepatic disease. Euthanasia because of progressive clinical signs (eg, ascites, lethargy, anorexia), coagulopathies, or hepatic encephalopathy was categorized as euthanasia secondary to hepatic disease. Overall median survival time was recorded for all dogs in the study. Dogs still alive at the time of data analysis, and those lost to follow-up, were censored at the last date known to be alive.

2.4 Statistical analysis

Data distribution was evaluated by the construction of histograms and determination of skewness and kurtosis. Data with a non-normal distribution are presented as median and range. Data with a normal distribution are presented as mean and SD. Categorical data are reported as proportions.

The association of clinical score and histological stage and grade were evaluated using Spearman rank correlation. All analyses were performed using commercial software. P values <.05 were considered significant.

3.1 Patient population

Fifty cases of GH were identified. Twenty-one dogs were excluded for the following reasons: failure to meet WSAVA criteria for GH (10/21), evidence of extrahepatic granulomatous disease (6/21), hepatobiliary neoplasia (3/21), and documented systemic infectious disease (2/21). Twenty-nine dogs were included in the study. The median age was 7 years old (range, 8 months to 12 years). Of the 29 dogs, 12/29 (41%) were spayed females, 11/29 (38%) were neutered males, 3/29 (10%) were intact females, and 3/29 (10%) were intact males. Twenty of 29 (69%) dogs were listed as purebred, with 9/29 (31%) listed as mixed breeds. Nineteen breeds were represented, with the most common being Labrador retrievers (8/29, 27%) and Samoyeds (2/29, 7%). There was 1 of each of the following breeds: Beagle, English springer spaniel, Border terrier, Great Dane, Nova Scotia duck tolling retriever, Spinone Italiano, English bulldog, Jack Russell terrier, and Golden retriever.

3.2 Drug history

Five dogs had a history of administration of potentially hepatotoxic drugs, but all of these cases were included in the study because expert opinion by 1 of the authors (CRLW) indicated that a drug-induced liver injury was unlikely. Three of the 5 dogs received doxycycline and 1 of the 5 received carprofen at the time of discovery of increased serum liver enzyme activity, but not before. One of the 5 dogs had been receiving phenobarbital for control of seizures for 12 months. This dog was not excluded because the onset of increased serum liver enzyme activity was acute, and GH has not been associated with phenobarbital toxicity in previous reports.^20-23

3.3 Comorbidities

Known comorbidities documented at the time of initial presentation were arthritis (7/29, 24%), history of cruciate ligament disease (4/29, 14%), skin allergies (3/29, 10%), hypothyroidism (2/29, 7%), laryngeal paralysis (2/29, 7%), benign splenic disease (2/29, 7%), facial paralysis (1/29, 3%), idiopathic epilepsy (1/29, 3%), and urinary incontinence (1/29, 3%).

3.4 Clinical signs

The most common clinical signs reported included decreased appetite (19/29, 66%), lethargy (16/29, 55%), and vomiting (11/29, 38%). Persistent fever, defined as body temperature > 39.2 C for at least 24 hours, was present in 13/29 (45%) dogs. Polyuria and polydipsia were recorded in 7/29 (24%) dogs, icterus in 7/29 (24%), and ascites in 3/29 (10%). One dog (3%) had diarrhea. Two dogs (7%) had no clinical signs at presentation.

3.5 Clinical pathology

A summary of selected clinical pathology results is presented in Table 1. All dogs had increased serum activity of ALT and AST. Serum alkaline phosphatase (ALP) activity was increased in 28/29 (97%) dogs, and serum gamma-glutamyl transferase (GGT) activity was increased in 6/27 (22%) cases. The mean increases in serum ALT, AST, and ALP activity were 6.9 ± 3.5 times upper limit of normal (ULN), 4.1 ± 3.1 times ULN, and 4.4 ± 3.4 times ULN, respectively. In most dogs (23/29, 79%), the magnitude of increase in serum ALT activity was higher than that of serum ALP activity. Total bilirubin concentration was increased in 17/29 (58%) dogs. Hypoalbuminemia was present in 9/29 (31%) dogs.

Leukocytosis was documented in 12/27 (44%) dogs. Twenty-four dogs had a leukocyte differential count performed, and 12/24 (50%) had mature neutrophilia. Anemia was reported in 6/27 (22%) dogs. Thrombocytopenia was found in 8/29 (27%). Prolongation of prothrombin time (median, 1.15 times ULN; range, 1.03-2.1 times ULN) was found in 6/24 (25%) dogs, whereas prolongation in partial thromboplastin time (median, 1.18 times ULN; range, 1.02-1.26 times ULN) was documented in 5/24 (21%) cases.

Urinalysis was available in 20 dogs. Isosthenuria was present in 8/20 (40%) dogs. Mild proteinuria (urine protein : creatinine ratio >0.5, but <1.0) was found in 3/6 (50%) cases.

3.6 Ultrasound imaging

Twenty-two of the 29 dogs had ultrasound examination findings available for review. Twenty-one of these dogs had ultrasound imaging performed at the time of, or within a week of, liver biopsy. Fourteen dogs had both still images and video clips available for review, and 6 dogs had only still images available for review. Twenty of 22 (91%) dogs had >/1 hepatic abnormalities on imaging (Table 2). The most common findings were changes in echogenicity (82%), hepatomegaly (50%), irregular margination (50%), presence of nodules (41%) and presence of hyperechoic parenchymal bands (36%).

Hepatic nodules ranged in size from 7 to 22 mm. Six of 22 (27%) dogs had multiple hypoechoic nodules. Hyperechoic nodules with a peripheral hypoechoic rim were identified in 2 (9%) dogs (Figure 1C,D). One dog had 4 rim lesions noted whereas the second dog had a single rim lesion.

One-third of the dogs had hyperechoic bands surrounding the periportal vasculature (Figure 1A,B). The size of these bands ranged from 7 to 19 mm. Some of the thicker band lesions had a lobular appearance along the vasculature. None of these lesions deformed the vascular margins.

Anechoic abdominal effusion was identified in 7 (32%) dogs. Six dogs had mild and 1 dog had marked effusion. Hepatic lymph nodes were identified in 4 dogs, and were variable in echogenicity, with 2 dogs having hypoechoic lymph nodes, 1 dog having a heterogeneous lymph node, and 1 dog having a normally echogenic lymph node.

Gallbladder wall thickness was normal in 21/22 (95%) dogs. One dog had a thickened gallbladder wall measuring 4.5 mm. A hypoechoic rim on the gallbladder wall (double wall) was identified in 2/22 (9%) dogs, consistent with edema. Gallbladder sludge was noted in 9/22 (41%) dogs. An immature gallbladder mucocele was present in 1 dog (5%).

3.7 Hepatic histopathology

Hepatic biopsies were performed by ultrasound-guided percutaneous needle biopsy in 17/29 (58%) dogs (using a 16G needle in 8/17 dogs, an 18G needle in 7/17 dogs, and needle gauge was not recorded in 2/17 cases), laparoscopy in 7/29 (24%) cases, or laparotomy in 4/29 (13%) cases. One dog's biopsy sample was obtained using computed tomographic guidance.

Granulomatous hepatitis was identified in all 29 dogs. Biopsy material was available for grading and staging in 19/29 dogs. The overall scores for grade and stage were 6.3 ± 2.5 and 2.1 ± 0.80, respectively. A breakdown of the grade and stage scoring is presented in Table 3. Overall, 84% of dogs had moderate to severe necroinflammation, and 21% had moderate to severe fibrosis.

Hepatic copper was evaluated by semiquantitative rhodamine staining in 28/29 (97%) cases and by quantitative analysis using atomic absorption spectroscopy in 15/29 (52%) cases. On rhodamine staining, 8/28 (32%) biopsy samples had increased (defined as >2/5) copper scores. Nine of 15 (60%) dogs had hepatic copper concentrations >400 μg/g dry weight. Five dogs with concentrations ranging from 476 to 974 μg/g dry weight range were judged to have secondary copper accumulation by virtue of the morphologic distribution pattern of copper (ie, not centrilobular, but random or associated with periportal areas of inflammation). In the other 4 dogs, hepatic copper concentrations were >1000 μg/g, and copper accumulation could not initially be ruled out as the primary driving force for the inflammatory process. All 4 dogs failed to respond to appropriate treatment aimed at decreasing hepatic copper concentrations and subsequently were treated using immunosuppressive medications. Only 1 dog had no assessment of copper status. The dog had eosinophilic and granulomatous inflammation, a pattern not reported before with copper hepatopathy, and thus the dog was retained in the study.

3.8 Exclusion of infectious etiology

Bacterial cultures of the liver were performed in 14/29 (48%) dogs and were negative in all cases. Bile cultures were performed in 11/29 (38%) cases, and all were negative. Fluorescence in situ hybridization for bacteria was performed on liver tissue in 3/29 (10%) dogs and was negative in all 3 cases. Six dogs were tested for Bartonella spp. by PCR of hepatic tissue, and all dogs were negative. Three dogs were negative for Histoplasma capsulatum antigen (by enzyme immunoassay on urine samples). Toxoplasma gondii titers were performed in 2 dogs, both of which were negative. Serum from 14 dogs was evaluated by leptospiral microscopic agglutination test (MAT; targeting serovars Bratislava, Canicola, Grippotyphosa, Hardjo, Icterohaemorrhagiae, and Pomona), and all were negative.

3.9 Clinical score

Each dog was assigned a clinical score that ranged from 0 to 14 based on established criteria.^{17, 18} The mean ± SD clinical score was 3.75 ± 2.75. Clinical scores were significantly negatively correlated with overall survival (rho = 0.49, P = .01) and histological stage (rho = −0.47, P = .04), but not with overall histological grade (rho = −0.29, P = .35).

3.10 Treatment

Treatment protocols were highly variable. Twenty-three of 29 (79%) dogs received antibiotics early in the course of their illness. Antibiotic regimens differed, but 17/23 (74%) dogs received a fluoroquinolone, 16/23 (69%) an aminopenicillin, and 13/23 (57%) metronidazole. Twenty of 22 (90%) dogs received a combination of these antibiotics. Other antibiotics used included doxycycline in 3 dogs, azithromycin in 2 dogs, and clindamycin in 1 dog. Nineteen of 29 (66%) dogs received immunosuppressive drugs. In 5/19 (26%) dogs, immunosuppressive drugs were first-line treatments. In the other 14/19 (74%) dogs, immunosuppressive drugs were started after failure of clinical or biochemical response to antibiotic treatment or both. Immunosuppressive drugs included cyclosporine (11/19, 58%), corticosteroids (15/19, 78%), azathioprine (1/19, 5%) or mycophenolate (3/19, 16%). Seven of 19 (36%) dogs were treated solely with corticosteroids, and 4/19 (21%) with only cyclosporine. In the remaining 8 dogs, corticosteroids were used as initial treatment and adjunctive immunosuppressive drugs were used for maintenance, including cyclosporine (7/8), azathioprine (1/8) and mycophenolate (3/8). Eight dogs never received immunosuppressive drugs but were treated with variable combinations of antibiotics. One dog received neither antibiotic nor immunosuppressive treatment. Twenty-five of 29 (86%) dogs received treatment with ≥1 hepatoprotective agents (S-adenosylmethionine, ursodiol, vitamin E). Prescription diets formulated for liver disease were recommended in 9/29 (31%) dogs.

Long-term follow-up was available for 26/29 (90%) dogs. Resolution of clinical signs occurred in 18/26 (69%) dogs, and normalization of serum ALT or serum AST activity was documented in 15/26 (58%) and 18/26 (69%) dogs, respectively. All dogs that were not in clinical remission had persistent increases in serum ALT and AST activity. Three dogs in clinical remission still had increased serum ALT activity and, of these, 1 dog had a concurrent increase in serum AST activity. The median survival time for all dogs was 635 days (range, 1-2482 days). Twenty of 26 (77%) dogs survived for at least 30 days. All dogs that did not survive >30 days post-diagnosis were euthanized because of liver disease. Median long-term survival in the remaining 20 dogs was 865 days (range, 200 to 2482 days). Of the 10 dogs that were non-survivors in the long-term follow-up cohort, only 2/10 (20%) were euthanized because of liver disease. The remaining 10 dogs were still alive at the time of manuscript preparation.

Fourteen dogs were managed initially with antibiotics, and when they failed to show clinical or biochemical improvement or both, they were switched to immunosuppressant drugs. These 14 dogs had a median survival of 980 days (range, 135-2482 days) with minimal short-term 30-day mortality (2/14, 14%). In contrast, 4/8 (50%) dogs treated with antibiotics alone died within 30 days. However, the other 4 dogs treated with antibiotics alone had a median survival time of 865 days (range, 557-1526 days). Five dogs treated with immunosuppressive drugs alone, with no prior antibiotics, were long-term survivors (median, 2108 days; range, 200-2438 days).