Risk factors and outcome in dogs with recurrent massive hepatocellular carcinoma: A Veterinary Society of Surgical Oncology case–control study

Abstract Local recurrence after surgical excision of canine massive hepatocellular carcinoma (HCC) has been poorly studied in veterinary medicine with scant information published regarding risk factors for and outcome following recurrence. The aim of this case–control study was to describe the time to recurrence, evaluate potential risk factors for recurrence, and report the outcome in dogs with massive HCC. Medical records for 75 dogs who developed recurrence and 113 dogs who did not develop recurrence were reviewed. Statistical analyses were performed to determine risk factors for recurrence as well as the median time to develop recurrence and overall survival time (OS). None of the risk factors evaluated were significant for the development of recurrence. The median time to develop recurrence was 367 days (range 32–2096 days). There was no significant difference in median OS for dogs who developed recurrence vs. those who did not (851 vs. 970 days). For dogs with recurrent HCC, treatment at recurrence trended toward prolonged OS but was not significantly different from dogs not undergoing treatment at recurrence. There was no significant difference in median OS for dogs with histologically complete vs. incomplete tumour excision (990 vs. 903 days). Although specific risk factors for recurrence were not identified, elevations in liver values were noted in patients with recurrent disease and could act as a noninvasive surveillance tool. Recurrence was noted earlier in dogs who had routine post‐operative surveillance (228 vs. 367 days). Routine surveillance for recurrence is recommended especially in dogs where further intervention is possible and should extend beyond 1 year. Patients with massive HCC have a good long‐term prognosis regardless of incomplete excision, pulmonary metastasis, or recurrent local disease.

recurrence were reviewed. Statistical analyses were performed to determine risk factors for recurrence as well as the median time to develop recurrence and overall survival time (OS). None of the risk factors evaluated were significant for the development of recurrence. The median time to develop recurrence was 367 days (range 32-2096 days). There was no significant difference in median OS for dogs who developed recurrence vs. those who did not (851 vs. 970 days). For dogs with recurrent HCC, treatment at recurrence trended toward prolonged OS but was not significantly different from dogs not undergoing treatment at recurrence. There was no significant difference in median OS for dogs with histologically complete vs. incomplete tumour excision (990 vs. 903 days). Although specific risk factors for recurrence were not identified, elevations in liver values were noted in patients with recurrent disease and could act as a noninvasive surveillance tool. Recurrence was noted earlier in dogs who had routine post-operative surveillance (228 vs. 367 days).
Routine surveillance for recurrence is recommended especially in dogs where further intervention is possible and should extend beyond 1 year. Patients with massive HCC have a good long-term prognosis regardless of incomplete excision, pulmonary metastasis, or recurrent local disease. No sex predisposition or specific risk factors have been recognized, although male dogs may be overrepresented. 4 Retrospective studies have identified an increased risk of HCC development in Welsh Corgis and beagles, Scottish Terriers with vacuolar hepatopathy, and dogs with hyperadrenocorticism. 5,6 Three morphologic subtypes of HCC are described: massive, nodular, and diffuse. Nodular and diffuse morphologies present as multifocal or diffuse infiltrating tumours and historically carry a poorer prognosis. 2 The massive morphology is defined as a large tumour affecting a single liver lobe and represents 53%-83% of all canine HCC. 2 Massive HCC is most often reported in the left division, is typically slow-growing, and reported to have a favourable prognosis with prolonged survival times following surgical excision. Survival greater than 1400 days irrespective of surgical margins and 765 days with incomplete margins is reported. 1,7 Factors reported to be associated with a poorer prognosis include lack of surgical treatment, right-sided location, or disease in the quadrate lobe. 8 Similarly, elevated serum alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP) to AST ratio, blood urea nitrogen (BUN), potassium, and gamma-glutamyl transferase (GGT) have been identified as potential poor prognostic indicators. 1,8 The distant metastatic rate for massive HCC is variable ranging between 0% and 37%. 2 Information regarding local recurrence rates following complete or incomplete resection, risk factors for local recurrence, and outcome following treatment of recurrent lesions are not well described. One study reported progressive local disease in 3 of 25 dogs with complete excision and 7 of 12 dogs with incomplete excision and included dogs whose recurrence was multifocal. 7 Another study reported no local recurrence in 42 dogs with 4 having undergone incomplete resection. 1 A study evaluating long term outcomes in dogs undergoing liver lobectomy for any liver tumour found the most common cause of tumour related death to be recurrence which occurred in 14 of 19 dogs that died. 9 No study reported time to recurrence, potential risk factors for recurrence, survival following recurrence, or any further treatment of recurrent cases. The objective of this retrospective case-control study was to describe the time to local recurrence, evaluate the potential risk factors for local recurrence, and report the outcome in dogs with massive HCC in a large cohort of dogs. intraoperative and postoperative complications encountered), and histopathology following primary surgery (margins, degree of overall differentiation, nuclear pleomorphism, amount of necrosis, mitotic count). For cases, the date of recurrence and reason for reevaluation leading to recognition of recurrence, workup performed at the time of recurrence (bloodwork, imaging, FNA cytology), and type of management for recurrence (surgery vs. medical) was collected. Local recurrence was defined as the presence of a mass in the same liver lobe (when the lobe was not completely resected at initial surgery), an immediately adjacent lobe, or the same or immediately adjacent liver division as the initial mass. If surgical treatment was performed, the location of recurrence, intraoperative and postoperative complications and histopathology findings of the recurrent mass were recorded.

| METHODS
When available, the date of last follow-up and date of death were recorded with patient outcome reported as alive, dead, or lost to follow up. Cause of death most likely related to HCC recurrence and necropsy findings were recorded when available.
Fine needle aspirate cytology reports provided by each institution were evaluated by a single author (JML) and categorized according to the following: carcinoma, hepatocellular carcinoma, or hepatocellular neoplasia; marked atypia, likely consistent with carcinoma or hepatocellular neoplasia; mild atypia, vacuolar hepatopathy, hepatopathy, or hyperplasia; and inconclusive or nondiagnostic, including necrosis and well differentiated hepatocytes. Histopathology reports were evaluated by a single author (JML) to extract information regarding completeness of surgical margins, the overall degree of differentiation, nuclear pleomorphism, amount of necrosis, and mitotic count. The mitotic count was reported as the number of mitosis per 10 high-powered fields. Surgical margins were reported as histologically incomplete (neoplastic cells present at the margin) or complete (no neoplastic cells present at the margin) and distance to closest surgical margin was reported in millimetres when available. Due to high variability in reporting of the degree of differentiation, nuclear pleomorphism, and amount of necrosis from each institution the following categorizations were created. The degree of differentiation was categorized as well if the terms 'well differentiated, low grade, organized, encapsulated and/or well demarcated' were listed in the histopathology report. Differentiation was categorized as moderate if the terms 'moderately differentiated, intermediate grade, moderately encapsulated and/or moderately demarcated' were listed in the histopathology report. Differentiation was categorized as poor if the terms 'poorly differentiated, high grade, disorganized, irregular, or unencapsulated' were listed in the histopathology report. Nuclear pleomorphism was categorized as mild if the histopathology report stated rare to no multinucleation, 1-2 nucleoli, well defined cell border, mild anisokaryosis, and/or mild anisocytosis. Nuclear pleomorphism was categorized as moderate if the histopathology report stated moderate multinucleation, 2-3 nucleoli, moderate anisokaryosis, and/or moderate anisocytosis. Nuclear pleomorphism was categorized as marked if the histopathology report stated frequent bi-, tri-, or multinucleation, marked anisokaryosis, and/or marked anisocytosis. The amount of necrosis was categorized as mild if the terms 'occasional, foci, scattered, some, or <25%' were listed in the histopathology report. The amount of necrosis was categorized as moderate if the terms 'regional, multiple focal regions, or <50%' were listed in the histopathology report. The amount of necrosis was categorized as marked if the terms 'expansile, extensive, large, or >50%' were listed in the histopathology report. Surgical complications were reported by each institution and categorized by a single author (JML) using the Classification of Intraoperative Complications (CLASSIC) as described in Table 1. 10

| RESULTS
Cases and controls were collected from multiple institutions, including private practice (3) and academic veterinary centres (9) in the United States (9), Canada (2), and Europe (1). A total of 92 recurrent cases were submitted. Of these, 17 cases were excluded due to incomplete medical records (7), the presence of multiple liver lobe involvement at initial surgery (9), or the presence of rapid diffuse recurrence (1) resulting in a total of 75 recurrent cases. A total of 163 control cases were submitted. Fifty controls were excluded due to incomplete medical records (7), presence of multiple liver lobe involvement at initial surgery (9), or follow-up period of less than 6 months (34) resulting in 113 control cases. One case was randomly selected for every two controls submitted by each institution, resulting in 43 recurrent cases and 86 controls evaluated for local recurrence risk factor analysis. Demographics at the time of the initial surgery are summarized in Table 2.

| Clinical findings at time of initial surgery
Preoperative haematology and serum biochemistry results were avail-    Table 5.  (2), chlorambucil (1), Toceranib (6), cyclophosphamide (2), doxorubicin (1), gemcitabine (1), dasatinib (1), and vinorelbine (1) with 5 patients receiving multiagent therapy. mass being deemed inoperable. This dog became hypotensive with grade 2 haemorrhage requiring a blood transfusion. The dog that underwent microwave ablation sustained grade 2 haemorrhage due to hepatic vein puncture and required a blood transfusion. One other mass was considered inoperable due to the proximity of the tumour to the caudal vena cava, and the surgery was aborted. One dog had substantial adhesions of the liver mass to surrounding structures necessitating a cholecystectomy in addition to the liver lobectomy.

| Tumour recurrence
Histologic surgical margins were reported in 60.0% (12/20) of cases that underwent surgery for treatment of local tumour recurrence and were classified as complete in 50.0% (6/12). The median histologic margin distance was 2 mm (range 2-30 mm). Further histologic information is detailed in Table 5.  Table 7.

| DISCUSSION
The present study provides information on the largest case series to date of dogs with massive HCC treated surgically, with a large subset of cases documented to have a local recurrence of HCC. In this study population, the median time to develop tumour recurrence was 367 days (range 32-2096 days). Median time to recurrence was shorter in the 40.0% of cases undergoing routine restaging (228 days, range 32-1099 days). None of the risk factors evaluated in this study appeared to be significant for the development of recurrence. In comparing dogs that developed recurrence of HCC (cases) with those that did not (controls) there was no significant difference in the median OS (851 days for cases, 970 days for controls). For cases that underwent additional treatment at the time of recurrence, survival trended toward being longer but was not significantly different from those that did not undergo additional treatment (1057 vs. 789 days). There was no significant difference in median OS for dogs with histologically complete or incomplete margins (990 vs. 903 days). Dog with suspect pulmonary metastasis at the time of recurrence had a significantly shorter survival then dogs without pulmonary metastasis (478 vs.

days).
In our study, the median time to local recurrence for dogs with HCC was around 1 year. Although previous studies have reported incidence of recurrence ranging from 0% to 27% 1,7 in dogs, none have specifically reported the timeframe to recurrence in a population of this size. Risk factors including sex, breed, age, body weight, tumour size, tumour location, FNA cytology findings, histologic findings, mitotic count, and completeness of excision were evaluated. None were found to be significantly associated with HCC recurrence in our population. Similarly, excluding age, Welsh corgi, and beagle breeds, none of these factors have previously been associated with the development of HCC. 5 In humans, HCC accounts for approximately 80% of liver tumours with a 5 years postoperative recurrence rate of 70%. 11,12 One study including 661 people with HCC found that the majority of recurrences develop within the first year following resection, with a median time to recurrence of 22 months. 12 Although our study did not identify any specific risk factors for recurrence, we did identify clinical features which may be useful in  The retrospective and descriptive nature of this study carries inherent limitations. The histopathology slides not being read by a single pathologist may have created bias due to inconsistent reporting as no standard reporting system exists for canine liver tumours. As we found that recurrence happens around 1 year following surgery, the minimum six-month follow-up for controls could have resulted in the inclusion of some controls that may have later developed recurrence. As recurrence was noted with ultrasound in the majority of patients, and ultrasound has been found to correctly localize liver masses in only 51.8%-74% of cases, the inability of imaging reports to give specific locations of recurrence is not unusual but could have allowed inclusion of some cases who had new masses instead of recurrence. 24,25 Additionally, the lack of standardized routine restaging may have led to some overestimation of the time to recurrence.
In conclusion, surgical intervention remains the mainstay of