Association Between Macroscopic Appearance of Liver Lesions and Liver Histology in Dogs With Splenic Hemangiosarcoma: 79 Cases (2004–2009)

Introduction

Canine hemangiosarcoma (HSA) is a highly malignant tumor of vascular endothelial origin characterized by early, aggressive metastasis and, in most cases, a poor prognosis. Although HSA may arise from any site in the body, the most common primary sites of visceral HSA include the spleen, right atrium, auricular appendage, and liver. The most common sites of metastasis (liver, omentum, mesentery, lungs) occur either hematogenously or through transabdominal transplantation via seeding after tumor rupture. Although the liver can be a primary site for visceral HSA, it most often serves as a metastatic site from a primary splenic tumor.^1–5

The spleen is the primary site for 35–62% of all primary HSAs.^6–8 Although clinical staging appears to be an inconsistent predictor of outcome, it is often performed prior to initiation of treatment. According to a modified staging system, stage I splenic HSAs are tumors confined to the spleen with no evidence of rupture/metastatic disease and stage II HSAs are tumors that have ruptured with hemoperitoneum but without evidence of detectable metastasis. Finally, stage III HSA has gross evidence of metastasis present.^2,3,6,9–11

The liver is a frequent site of metastasis in dogs with splenic HSA. The presence of liver metastasis is, therefore, a common reason for assigning stage III disease to a dog with splenic HSA. Liver biopsy obtained during presurgical imaging with abdominal ultrasound can assist with the staging process, and recent studies have discovered improved detection of metastatic hepatic HSA via contrast harmonic imaging. In diagnostic ultrasound, contrast harmonic imaging is used to evaluate tissue perfusion and tumor vascularity. Contrast agents in the form of microbubbles are injected IV into the systemic circulation. The beam of a special ultrasound machine transmits a harmonic signal that is received and returned by the soft tissue as a result of the IV administered microbubbles passing through capillary beds. Those microbubbles have a high degree of echogenicity (i.e., the ability to reflect ultrasound waves) and are viewed very easily due to the significant contrast compared with the surrounding soft tissues. That technique enables detection of various characteristic vascular patterns of abnormal lesions.^12–15 MRI also provides a useful modality for abdominal imaging and has a 94% accuracy in differentiating between benign and malignant lesions in the liver and spleen.¹⁶ Although the liver is routinely evaluated for detectable lesions during splenectomy, biopsies are often performed only when gross changes are apparent to the surgeon.

Clinically normal livers may not be biopsied; therefore, the likelihood of detecting micrometastases through a randomly selected biopsy is not known. A previous study reported that only 17 of 45 liver biopsies (37%) in dogs with splenic HSA had histologically confirmed HSA, and it was noted that most dogs with metastatic disease in the liver had gross liver changes at the time of surgery.⁵ There are a number of surgeons that routinely biopsy the liver in suspected splenic HSA cases regardless of whether there are gross lesions present. Nonetheless, the usefulness of performing liver biopsies in dogs with grossly normal livers is yet to be determined.

The goals of this study were to examine the association between the gross appearance of the liver during surgery with histologic findings in dogs with splenic HSA, determine the usefulness of liver biopsy in cases of canine splenic HSA when the liver is grossly normal, and determine whether the gross appearance of the liver can predict the likelihood of liver metastasis.

Materials and Methods

Results

Seventy-nine dogs (3 females, 8 males, 36 spayed females, and 32 castrated males) met the inclusion criteria. Two dogs diagnosed with HSA were excluded from this study because a liver biopsy was not obtained. Mean patient age was 11.5 yr (range, 6–15 yr; median, 10 yr). The most prevalent breeds were mixed-breed dogs (n = 22), Labrador retrievers (n = 7), golden retrievers (n = 4), and German shepherd dogs (n = 4). Twenty-six other breeds were represented in smaller numbers.

Twenty-one dogs had a grossly normal liver and 58 dogs had gross liver lesions. All of the 21 dogs with grossly normal livers had either histologically normal livers or benign liver changes (7 normal, 14 benign abnormal changes) with no evidence of metastasis. Benign abnormal changes (in order of frequency) included nodular hyperplasia, hepatic vacuolar degeneration, lipogranulomas, extramedullary hematopoiesis, hemosiderosis, cholestasis, thrombosis/necrosis, steroid hepatopathy, chronic lymphoplasmacytic hepatitis, central venous arteriolarization, hepatocellular glycogenesis, and periacinar coagulative necrosis with neutrophilic hepatitis. Of the 58 dogs with liver lesions, 29 dogs (50%) had benign histologies (one or more of the above described changes) and 29 dogs (50%) had malignant histologies, including 28 HSAs and 1 spindle cell sarcoma.

An overall significant association between liver histology and gross clinical characteristics was observed (P = .0001). All HSA-positive liver biopsies were taken from grossly abnormal livers. However, 59% of benign liver biopsies were also obtained from grossly abnormal livers.

Table 1 summarizes the association between the gross clinical appearance of the liver and liver histology. There were 47 dogs with multiple liver nodules and 32 dogs without multiple liver nodules (10 single liver nodules, one mottled liver, and 21 grossly normal). One dog with multiple liver nodules on gross examination had hepatic spindle cell sarcoma. A significant association was found between liver histology and either the presence or absence of multiple liver nodules, and dogs with multiple liver lesions were more likely to have HSA. The majority of dogs (93%) with biopsy-confirmed liver HSA had multiple liver nodules. Additionally, dogs with multiple gross liver nodules were statistically more likely to have HSA on liver histopathology (P = .0001).

TABLE 1 Association Between Gross Appearance of Liver Lesions and Liver Histology in Dogs With Splenic HSA

View Fullscreen

Pertaining to color, there were 14 dogs with liver nodules in group 1 and 14 dogs with liver nodules in group 2. Thirty dogs did not have the color of the liver lesions noted. A significant association was also found between liver histology and the color of the liver nodules (P = .0006). The majority of dogs (80%) in group 1 had biopsy-confirmed HSA. Only four dogs were noted to have actively bleeding nodules. All of those four dogs had biopsy-confirmed hepatic HSA, which was also statistically significant, (P = .0415).

Table 2 reflects the sensitivity, specificity, PPV, and negative predictive values of gross changes in the liver associated with metastatic lesions on histology. The presence of active bleeding was associated with a 100% specificity and PPV of 100%. However, few dogs with HSA in the liver had evidence of active bleeding (n = 4); therefore, the sensitivity in detecting HSA is low. Group 1 (i.e., dark lesions) was associated with a high sensitivity (85%), specificity (85%), and PPV (86%). Overall, the presence of any gross abnormalities of the liver was associated with a high sensitivity (100%), but the specificity and PPV were relatively low at 41% and 48%, respectively.

TABLE 2 Sensitivity, Specificity, and PPV and Negative Predictive Values of Gross Changes in the Liver Associated With Metastatic Lesions on Histology

View Fullscreen

Thirteen dogs had metastasis in other locations other than the liver, including omentum (n = 7), lymph nodes (n = 2), kidneys (n = 1), retroperitoneal space (n = 1), right atrium (n = 1), and cutaneous (n = 1). Of those 13 dogs, only 5 had liver metastasis (4 in the omentum, 1 in the kidney). The remaining eight dogs had histopathologically benign livers.

Only 20 dogs had multiple biopsies of the liver. Of those 20 cases, 12 dogs had two biopsies taken, 7 dogs had three biopsies taken, and 1 dog had multiple biopsies taken from the liver. Unfortunately, due to the retrospective nature of this study, information was neither available nor consistent in regard to documentation of what liver lobes were biopsied, the gross appearance of every sample obtained, or the correlation of histopathology with the gross appearance of each liver sample submitted for biopsy.

Discussion

This is the first study to describe the association between gross appearance and histopathology of the liver in dogs with splenic HSA. In 79 dogs diagnosed with splenic HSA, 21 dogs (27%) had grossly normal livers, and the remaining 58 dogs (73%) had grossly abnormal livers. All of the biopsy samples obtained from dogs with a normal liver were histopathologically benign, reflecting a low yield associated with a randomly selected biopsy site in this subset of the population. The overall incidence of gross liver lesions appears to be quite high. In this study, as many as 58 of the 79 dogs were noted to have gross abnormalities according to the surgery reports. However, only 29 of those 58 dogs (50%) had metastatic disease on histological examination. All others were benign, with the exception of one dog with spindle cell sarcoma present in the liver. When classifying the gross lesions based on color, presence of bleeding, and multiple lesions, a significant association between specific gross characteristics and presence of HSA liver metastasis was found. As depicted in Table 2, the sensitivity of having multiple nodules, color group 1, and presence of any gross abnormalities were 93, 84, and 100%, respectively. In contrast, the presence of active bleeding was associated with a low sensitivity (14%) but a high specificity of 100%. In a clinical setting, however, it is the PPV and negative predictive value of a test that will assist in clinical decision making. The PPV reflects the conditional probability that an individual that tests positive has the disease. Patients with liver lesions categorized in color group 1 had an 86% probability of having HSA metastases. Similarly, the presence of active bleeding was associated with a PPV of 100%. Therefore, when encountering a patient with either dark-colored lesions or actively bleeding lesions during exploratory laparotomy in a dog with suspected splenic HSA, there is a high chance that they may represent metastatic liver lesions. A previous study showed a significantly improved outcome in dogs with HSA when all gross and suspicious lesions were removed during splenectomy and received follow-up doxorubicin-based chemotherapy. That study did not state either the location or number of metastatic lesions resected.¹⁷ Many other studies have not demonstrated a significant difference in outcome between stage II and III splenic HSA.^2,3,7,18 However, many of those studies were underpowered and may have had the outcome affected by type II errors. Furthermore, the staging procedures were not specified (i.e., whether all suspicious lesions were biopsied); therefore, dogs might have been understaged or even overstaged if stage was assigned based on the presence of gross liver lesions.

Fifty percent of the dogs in this study had liver lesions that were benign on histopathological exam. That was consistent with another study that reported the incidence of benign hepatic nodular hyperplasia and metastatic lesions in dogs with splenic HSAs.¹⁹

Abdominal ultrasound can be helpful in detecting subcapsular nodules that are deep within the parenchyma.^12–15 Although imaging may direct where the surgical biopsy is obtained, it is extremely difficult for the surgeon to obtain a representative biopsy sample when the surface of the liver is grossly normal. Although intraoperative ultrasound may provide better guidance with surgical biopsy, it is not readily used in most practices due to availability. Preoperative imaging with either ultrasound or MRI may also influence which dogs go to surgery (i.e., fewer dogs may proceed to surgery if diagnosed with advanced stage disease), which can affect the sample population.^12–16

None of the grossly normal livers were found to have liver metastasis according to the histopathology results. Those results suggest that performing routine biopsies for staging purposes in dogs with clinically normal livers is a low-yield procedure. It is important to note, however, that a negative biopsy result does not necessarily mean that the liver is free of micrometastasis. It may only reflect that a randomly selected biopsy site is unlikely to detect it.

Limitations of the study were mainly due to its retrospective nature. Neither the size and location of the random liver biopsies obtained in grossly normal livers were noted nor were the techniques used to obtain a representative sample. Livers with either subcapsular nodules or masses deep within the parenchyma appear normal to a surgeon on gross examination; therefore, pathology may be missed with a random biopsy. The apical portions of the liver lobes are most likely to be sampled in a grossly normal liver, mainly due to accessibility, rather than the midbody or deep parenchymal core samples, which may yield undetected disease. Also, in a grossly normal liver, usually only one sample is obtained as opposed to multiple samples from multiple liver lobes, which may influence the detection of metastatic neoplasia. Multiple liver biopsies were not obtained in all dogs, only random dogs with multiple liver nodules. Descriptive characteristics of the liver on gross examination were incomplete in many surgery reports throughout the sample population, and the color of the nodules was not always noted. Active bleeding of the nodules was also another descriptor that may not have been documented in all cases.

Conclusion

The data suggest that obtaining biopsies of grossly normal livers during surgery for suspected splenic HSA is unnecessary. The appearance of multiple, dark-colored, or bleeding nodules in the liver is highly associated with metastatic disease. Although it may be tempting to use this information to facilitate clinical decisions during surgery, the authors caution that the information should be verified in a prospective clinical study.