Introduction

Occurrence and clinical features of splenic neoplasia in dogs have been reported in several studies involving a variety of populations. Specifically, studies consisting of general pathology laboratory submissions, dogs with both neoplastic and non-neoplastic splenic disease, and populations of dogs presenting with hemoabdomen with or without an abdominal mass have been reported.^1–4 The results of these studies, as well as those involving dogs with hemoperitoneum, currently provide information regarding the occurrence of neoplasia and prognosis in dogs.^5–8 Previous studies have found that breeds such as the German shepherd dog, Labrador retriever, miniature schnauzer, and golden retriever are all overrepresented for splenic hemangiosarcoma (HSA).^1,2,4,7–11 Additionally, both splenic hematoma and HSA are reported to be more common in older, medium, and large breed dogs.^8,12 To date, no studies have statistically evaluated the occurrence of canine malignant splenic neoplasia and relevant clinicopathologic associations in relation to body weight.

Previous studies reported that 44–50% of all splenic samples obtained from surgical and general necropsy specimens were malignant, of which HSA accounted for 44–48% of those malignancies.^1–3 These results suggest that approximately 50% of splenic masses are expected to be malignant, with HSA representing approximately 50% of these malignancies.^1–3 Another study by Johnson et al. reported that in dogs presenting with splenomegaly, histopathology samples obtained at surgery or necropsy indicated that 66% of splenic lesions were malignant and 65% of those malignancies were HSA.⁴ In these studies, all causes of splenomegaly, regardless of the absence of a mass lesion, were included for analysis, and body weight was either not reported or not statistically analyzed.^1–4

Several studies have evaluated various clinicopathologic features of splenic malignancy. Preoperative decrease in plasma total protein and thrombocytopenia have been reported to be associated with a diagnosis of splenic HSA, and the presence of anemia has been found to be associated with a diagnosis of splenic malignancy.^4,7 Extrapolation of the occurrence of clinicopathologic findings in relation to malignant splenic disease and HSA from these reports may not be reliable in dogs of smaller body weight.

To date, no studies have specifically determined the occurrence of HSA or other forms of splenic malignancy, nor established a correlation of preoperative clinical features with a histopathologic diagnosis in dogs subjectively considered to be small in terms of body weight. The purpose of this study was to evaluate all dogs undergoing splenectomy for treatment of a splenic mass or nodule at two university teaching hospitals in order to assess whether size of dog, based on the body weight, was associated with diagnosis of malignancy or HSA. A second objective was to determine if there was a correlation between diagnosis and clinicopathologic factors such as preoperative anemia, hypoproteinemia, thrombocytopenia, presence of hemoabdomen, requirement for preoperative whole blood or packed red blood cell transfusion, and development of postoperative arrhythmias. Our hypothesis was that smaller dogs would have a lower occurrence of malignancy and HSA as compared to larger dogs, and that preoperative anemia, hypoproteinemia, thrombocytopenia, administration of a blood transfusion, and hemoabdomen would be associated with a diagnosis of HSA in both size groups.

Materials and Methods

Medical record reviews of all dogs undergoing splenectomy at the Kansas State University Veterinary Health Center and The Purdue University Veterinary Teaching Hospital were performed for the periods of January 2004 to July 2013 and July 2003 to July 2013, respectively. Medical records of dogs were included for analysis if splenectomy was performed as treatment for a splenic mass or nodule. Dogs diagnosed with splenic torsion, splenomegaly, or masses smaller than one centimeter were excluded from further analysis. Inclusion criteria were a complete medical record including signalment, body weight in kilograms, confirmation of a mass or nodule measuring at least 1 cm or greater at the time of surgery, performance of a splenectomy, and histopathology of the spleen performed by a veterinary pathologist. Other data collected included preoperative packed cell volume (PCV), total protein, and platelet counts; largest dimension of the mass recorded either from the ultrasound, surgery, or gross pathology reports; presence of hemoabdomen as determined from the surgery report or abdominocentesis performed prior to surgery; administration of either a packed red blood cell or whole blood transfusion; or development of postoperative arrhythmias. Data for preoperative PCV, total protein, and platelet counts were collected from either the most immediate pre-surgical or anesthetic induction periods. For the purpose of this study, anemia was defined as PCV < 37%, hypoproteinemia as <5.4 g/dL, and thrombocytopenia as platelets < 164,000/uL, based upon the lowest reference values for the submitting institutions' clinical pathology laboratories. For the purposes of statistical comparison, histopathology results were then used to establish groups for HSA, non-HSA splenic malignancy, and benign lesions. Non-HSA splenic malignancy included other sarcomas, lymphoma, and mast cell tumors. Occurrence and tumor type for the subcategory “other sarcomas” are reported in Table 1. Benign lesions included splenic hematoma, lymphoid nodular hyperplasia, fibrohistiocytic nodules, extramedullary hematopoiesis, myelolipoma, lipoma, or abscess. For purposes of statistical analysis, lesions diagnosed with lymphoid nodular hyperplasia and fibrohistiocytic nodules and those with myelolipomas and lipomas were combined into two respective groups.

Table 1 Distribution of Splenic Histopathology Diagnoses

View Fullscreen

Results

Initial medical records review yielded 396 dogs undergoing splenectomy, with 187 and 209 dogs from Kansas State University Veterinary Health Center and The Purdue University Veterinary Teaching Hospital, respectively. Of these, 325 met inclusion criteria, with the median weight of 26.7 kg (range 2.4–73.0 kg) and the median age of 10.7 yr (range 3–20 yr).

The cutoff value for body weight based on the Youden index was identified at 27.8 kg and was used to dichotomize the dogs into small (body weight ≤27.8 kg) and large (body weight >27.8 kg) groups (sensitivity 0.57; specificity 0.59). Distribution of histopatholgic diagnoses by dog size is reported in Table 1. Among study dogs, 175 (54%) were classified as small dogs, and 150 (46%) were large dogs. The occurrence of HSA, non-HSA malignancy, and benign splenic disease was significantly different (P = .019) between large and small dogs. Among all splenic masses, small dogs (25%) were less likely to be diagnosed with HSA as compared to large dogs (39%) (Table 1). Logistic regression analysis showed a weak positive relationship between weight and odds of HSA, with an increase in body weight having increased odds for HSA (P = .062). The overall occurrence of malignant and benign splenic diseases was not different between size groups (P = .291). However, when considering only malignant diagnoses, HSA was significantly more likely to occur in large dogs (65%) than in small dogs (46%) (P = .009). Breeds of dogs diagnosed with HSA and represented more than twice were the following: mixed breed, 25 (24%); Labrador retriever, 14 (14%); German shepherd dog, nine (9%); golden retriever, eight (8%); miniature schnauzer, six (6%); Australian shepherd, three (3%); boxer, three (3%); and cocker spaniel, three (3%). Breeds of dogs with a diagnosis of non-HSA malignancy and represented more than twice were as follows: mixed breed, 21 (25%); golden retriever, eight (10%); Labrador retriever, eight (10%), rottweiler, seven (8%); beagle, six (7%); border collie, three (4%); and bichon frise, three (4%).

Overall percentage of occurrence of clinicopathologic variables associated with a diagnosis of HSA, non-HSA malignancy, or benign lesions in small and large dogs are presented in Table 2. Odds ratios, 95% confidence intervals, and P values for the development of specific clinicopathologic variables in relation to histopathologic diagnosis are reported in Table 3. Valid platelet counts were reported in 181 dogs (58%), with the remainder having either no value reported or a result considered invalid due to the presence of clumping. As compared to non-HSA malignancy, small dogs with HSA were 4.9 times more likely to have anemia (P = .001), 8.4 times as likely to have thrombocytopenia (P = .001), and 7.8 times as likely to have hemoabdomen (P < .001). Within this same comparison for large dogs, odds ratios were greater than five only for development of hemoabdomen (9.0, P < .001). Odds ratios for development of anemia, thrombocytopenia, and hemoabdomen with HSA compared to benign disease were 5.4 (P = .001), 9.0 (P < .001), and 9.8 (P < .001), respectively, for small dogs and 4.9 (P < .001), 12.0 (P < .001), and 5.4 (P < .001), respectively, for large dogs. Overall, of all dogs with HSA, 58/103 (56%) dogs were reported to have hemoabdomen. Postoperative arrhythmias developed in 15% of small and 33% of large dogs, with no significant difference in odds ratios of this finding for various histopathologic categories (Table 3).

Table 2 Percentage of Occurrence Clinicopathologic Variables Organized by Size Group

View Fullscreen

Table 3 Age-Adjusted Odds Ratios (OR) of Clinicopathologic Variables in Relation to Histopathological Diagnosis of Splenic Mass by Body Weight Size Group

View Fullscreen

Discussion

The results of this study showed that occurrence of benign and malignant splenic disease did not differ between the two sizes of dogs evaluated. However, in dogs diagnosed with a splenic mass or nodular lesions and treated by splenectomy, HSA occurred significantly less often in dogs ≤27.8 kg (25%) as compared to dogs >27.8 kg (39%). Additionally, as a percent of malignancy, HSA was significantly lower in small (46%) as compared to large (65%) dogs. The results of this study further characterize both the occurrence of and relevant clinical features associated with splenic nodular disease in a sample of dogs treated with splenectomy based upon body weight. In prior studies, no attempt was made to establish a correlation between body weight and the diagnosis of a malignant disease. The authors sought to investigate the validity of their clinical impression that smaller dogs were less likely to be diagnosed with HSA or other forms of splenic malignancy. The 58% overall occurrence of malignant splenic disease in this study is comparable to previous reports that evaluated smaller samples, had broader inclusion criteria, and found malignant splenic disease in 44–66% of dogs.^3,4 As a percentage of malignancy, HSA represented 46% of diagnoses in small dogs in our study, as compared to the 65% for dogs of all sizes as reported by Johnson et al.⁴ In other studies involving broader inclusion criteria, many forms of splenic disease were included for analysis, which led to inclusion of splenic samples without gross evidence of a mass or nodular disease and a somewhat lower percentage of HSA.^1,2

The distribution of histopathologic diagnosis (HSA, non-HSA malignancy, or benign splenic disease) was statistically different between the small and large dog groups, with HSA diagnosed in 25% and 39%, respectively. However, the ROC analysis showed an overall poor discrimination of body weight for diagnosis of HSA (Area Under the ROC Curve 0.57). The poor discriminatory ability of body weight for a diagnosis of HSA was not unexpected, considering the diversity of clinical presentations and histopathologic diagnoses for splenic neoplasia in dogs. However, this method served the purpose of establishing two groups for comparison by body weight and allowed evaluation of smaller dogs for differences in histopathologic diagnosis and clinical features of splenic neoplasia.

The diagnosis of HSA, as compared to non-HSA malignancy and benign disease, was positively associated with the development of clinicopathologic findings in the current study. In this study, small and large dogs diagnosed with HSA were 5.4 and 4.9 times more likely to be diagnosed with anemia, respectively, compared to dogs diagnosed with benign disease. These results are in contrast to recent reports of dogs in which there was no significant difference in PCV or other hematologic variables evaluated when HSA was compared to other forms of splenic disease.^5,7 An association between the diagnosis of splenic neoplasia and presence of anemia has been reported previously; however, a relationship between HSA and anemia has not been previously established.⁴ Likewise, both small and large dogs with HSA had significantly greater odds of receiving a blood transfusion compared to non-HSA malignancy or benign splenic disease. The most likely explanation for the association of anemia with a diagnosis of HSA may be due to the increased propensity of dogs with HSA to develop hemoabdomen, but other causes of anemia cannot be excluded.

Significant associations were identified for a diagnosis of HSA and thrombocytopenia in both sizes of dogs, and hypoproteinemia only in small dogs. The odds of small and large dogs having thrombocytopenia when diagnosed with HSA were 8.4 and 4.2, respectively, compared to non-HSA splenic malignancy. Additionally, the odds of small and large dogs diagnosed with HSA having thrombocytopenia were 9.0 and 12.0, respectively, compared to a diagnosis of benign splenic disease. Our study reported platelet counts for 58% of the dogs in the study, and information regarding thrombocytopenia should be interpreted with some caution. Histopathologic diagnosis had no statistical association with hypoproteinemia other than in small dogs with HSA as compared to benign disease. This increased odds ratio is most likely secondary to the presence of hemoperitoneum, but other causes of hypoproteinemia cannot be excluded. These results are similar to a previous study that established that both thrombocytopenia and hypoproteinemia were significantly more common with a diagnosis of HSA. The dogs included in that study were diagnosed with hemoperitoneum, had a mean body weight of 33 kg, and required a blood transfusion,⁷ whereas the dogs included in the present study did not have specific inclusion criteria requiring the presence of hemoperitoneum.

The presence of hemoperitoneum was found to be significantly associated with a diagnosis of HSA when compared to non-HSA malignancy or benign disease in the present study. Small dogs diagnosed with HSA were 7.8 and 9.8 times more likely to have evidence of preoperative hemoabdomen compared to non-HSA malignancy and benign splenic disease, respectively. Large dogs with HSA were 9.0 times more likely than those with non-HSA malignancy to have hemoabdomen and were 5.4 times more likely than dogs with benign disease to exhibit this finding. Thus, HSA was strongly associated with development of hemoabdomen in both size groups. The presence of concurrent hemoperitoneum and splenic malignancy is reportedly 76.1–80%, with 70.4–88% diagnosed with HSA versus 29.6% diagnosed with a splenic hematoma.^5–7,12 A mean body weight was not reported in the majority of these studies, and the sample sizes were small, including only dogs with a diagnosis of hemoperitoneum.⁵ Our study evaluated a larger sample of dogs with splenic masses treated by splenectomy and supported the fact that HSA was associated with the presence of hemoabdomen, but also found a lower percentage (56%) of those with HSA diagnosed with hemoabdomen. Despite these findings, caution is warranted when considering the presence of these clinical findings for prediction of HSA.

Overall, 15% of small and 33% of large dogs developed postoperative arrhythmias with no statistically significant differences between any groups based on histopathologic diagnosis. Development of ventricular arrhythmias in dogs with splenic masses treated by splenectomy and associations with underlying diagnoses varies in the literature. Ventricular arrhythmias have occurred with a frequency of 38.5–87.5% in dogs diagnosed with HSA versus 31.3–34.1% in dogs diagnosed with benign hematoma.⁸ Other studies report an incidence of 32% for the development of ventricular arrhythmias in dogs presenting with acute non-traumatic hemoperitoneum.^6,8

Certain limitations were present in our study. By excluding dogs undergoing splenectomy for other reasons (i.e., splenomegaly, splenic torsion, nodular lesions <1 cm), we hoped to report distribution of neoplasia and overall occurrence of malignant neoplasms in a sample of dogs undergoing splenectomy. Medical record review found no dogs that died or were euthanized and had histopathology of the spleen performed. By including only dogs with a splenic mass treated by splenectomy and diagnosed by histopathology, it is conceivable that our findings may not be able to be broadly applied to the general dog population, but are representative of a sample of dogs with splenic neoplasia treated by splenectomy. The inclusion of any dog having a 1 cm or greater splenic nodule could also represent a limitation and potentially bias the results toward a lower percentage of neoplasia. Histopathologic interpretation by numerous veterinary pathologists over a 9-yr period is another potential limitation. Pathology samples were not reviewed to determine if reclassification of the splenic disease was warranted. Several studies have described difficulty in differentiating between HSA and splenic hematoma.^13–15 A previous study found that when splenic histopathology samples were reviewed in non-blinded fashion, only 3/217 samples were reclassified from hematoma to HSA.¹² Recently, immunohistochemical staining resulted in reclassification of splenic fibrohistiocytic nodule into nodular hyperplasia, lymphoma, stromal sarcoma, and histiocytic sarcoma, but not hemangiosarcoma.¹⁶ Based upon these studies, there is the potential for reclassification of a small percentage of masses in our population to a diagnosis of malignancy, had histopathologic review been conducted.

Other limitations of this study are the inherent problems of conducting a retrospective study. Data was collected from a medical records review that required reliance upon accurate and consistent documentation of all variables chosen for inclusion in statistical analysis. One example was body condition scoring, which was not consistently available at either institution and therefore was not included as a confounding variable in the data analysis. Using breed for categorizing large and small dog groups was avoided due to the large number of mixed breeds in both groups. In this study we did not intend to separate heterogeneous effects, such as genetics, within each of the two size groups and did not control for breed in the data analysis. Long-term follow-up was not available for many dogs, which prohibited accurate reporting of any survival data among groups. A prospective study would be ideal in order to evaluate both short- and long-term survival information.

Conclusion

Based on our results, we found that the occurrence of HSA in dogs undergoing splenectomy for treatment of a splenic mass was significantly lower in small (25%) versus large (39%) dogs based upon our statistically derived body weight groups. Additionally, HSA as a percent of malignancy was found to be significantly lower in small dogs (46%) as compared to large dogs (65%). There was no difference in the occurrence of malignancy between small and large dogs, with 58% of dogs in this study having a malignant diagnosis. In both body weight groups of dogs, a diagnosis of HSA was positively associated with anemia, thrombocytopenia, and hemoabdomen in comparison to non-HSA malignancy and benign disease, while decrease in total protein was only significant for small dogs with HSA. However, even with the noted associations, definitive diagnosis requires histopathology.