Introduction

Cancer is the leading cause of death in adult dogs, accounting for 20% of deaths in 79% of breeds evaluated.¹ Among cancers in dogs, hemangiosarcoma, a tumor of vascular endothelial origin, accounts for 12% to 21% of all mesenchymal tumors, approximately 66% of all malignant splenic masses, and is overrepresented in large-breed dogs such as Golden Retrievers, Labrador Retrievers, and German Shepherd Dogs.² The prognosis for splenic hemangiosarcoma is poor, with a mortality rate approaching 100% in a matter of months despite intensive multimodal treatment.²

Although overall prognosis with hemangiosarcoma is poor, dogs with less advanced stage disease have improved survival rates, compared with dogs with more advanced stage disease.³ However, in most dogs with hemangiosarcoma, the diagnosis is made late in their disease course, with many dogs presenting with tumor rupture, massive hemorrhage into the abdomen, and micrometastasis or gross metastasis to other organs because of the lack of screening tests to allow for early identification of affected patients.^3,4 Splenic hematoma is another common cause of splenic masses that can also be a life-threatening disease in dogs as a result of sudden blood loss following rupture.^5,6,7 As with splenic hemangiosarcoma, the recommended treatment for splenic hematoma is splenectomy, which is curative. Currently, no blood tests can screen dogs for splenic masses, and other diagnostic tests are unable to differentiate the etiology of splenic masses prior to surgical removal. Although the presence of hemoperitoneum is more common in dogs with hemangiosarcoma, this clinical finding is neither a sensitive nor a specific indicator of hemangiosarcoma, as other splenic diseases can also result in hemoperitoneum.^5,6,7 Fine-needle aspiration is also unrewarding because of blood contamination and poor exfoliation of the tumor, in addition to the risk of rupture and seeding of the tumor.⁸ Various other methods have been evaluated for diagnosis of hemangiosarcoma, including comparison of mass to splenic volume and splenic weight to body weight, contrast harmonic ultrasonography, CT, MRI, and blood testing including multiparameter flow cytometry and measurement of plasma vascular endothelial growth factor concentrations, but none have proven sensitive or sufficiently specific for definitive preoperative diagnosis.^{8,9,10,11,12,13} A need exists to identify biomarkers to allow for identification of hemangiosarcoma earlier than is currently possible. Earlier identification will allow for diagnosis at less advanced disease stages, leading to increased survival times.

A class of potential biomarkers includes miRNAs, which are 22- to 26-nucleotide-long RNAs that are known to regulate cellular responses and host pathways including malignancy.^14,15 MicroRNAs affect posttranscriptional regulation via binding of mRNAs, either preventing their translation into proteins or causing their degradation. Substantial alteration of miRNA expression exists in malignant cells, compared with their nonmalignant counterparts, with cancer tissues exhibiting a specific miRNA signature reflecting their malignant state and features of that particular cancer and its progression.^16,17,18 In addition to having altered tissue expression levels, miRNAs also have altered expression levels in the circulation of patients with cancer, compared with patients without cancer.^19,20,21 Dysregulation of host processes has been shown to contribute to the pathogenesis of canine hemangiosarcoma, with several genes being implicated, including PTEN and those involved in the retinoblastoma protein pathway and expression of vascular endothelial growth factor.^22,23,24 A previous study²⁵ found alteration of miRNA expression in the splenic tissue among dogs with histologically normal spleens, dogs with hemangiosarcoma, and those with splenic nodular hyperplasia.

The objective of the present study was to identify miRNAs that differ between dogs with histologically normal spleens, dogs with splenic hemangiosarcoma, and dogs with splenic hematoma and to evaluate both tissue and serum miRNA expression. The hypothesis was that miRNA expression profiles in both tissue and serum would differentiate dogs with hemangiosarcoma, splenic hematoma, and histologically normal spleens.

Materials and Methods

Results

Samples were obtained from 5 healthy control dogs (paired tissue and serum samples, n = 5), 10 dogs with splenic hemangiosarcoma (paired tissue and serum samples, 5; serum samples only, 5), and 5 dogs with splenic hematoma (serum samples only, 5). Healthy control dogs had a median age of 4.4 years (range, 4.4 to 6.4 years), and all 5 were sexually intact female mixed-breed dogs. Dogs with splenic hemangiosarcoma had a median age of 10.9 years (range, 6.8 to 12.1 years) and were castrated male (n = 8) or spayed female (2) dogs. Types and breeds of dogs with splenic hemangiosarcoma were as follows: mixed breed (n = 4), Golden Retriever (3), Labrador Retriever (2), and Australian Cattle Dog (1). Dogs with splenic hematoma had a median age of 8.0 years (range, 6.1 to 10.5 years) and were castrated male (n = 2), spayed female (2), or sexually intact male (1) dogs. Types and breeds of dogs with splenic hematoma were 1 each of American Pit Bull Terrier, German Shepherd Dog, Labrador Retriever, mixed breed, and Siberian Husky.

For the 10 dogs with hemangiosarcoma, 6 dogs had hemoperitoneum and 4 did not. Seven dogs had a single mass within the spleen, and 3 dogs had multiple masses. The median largest mass dimension was 9 cm (range, 1.8 to 20 cm). Five dogs had no evidence of metastatic disease. One dog had omental metastasis, and another had metastasis to the liver. One dog had suspected metastasis to a local lymph node, and 1 dog had a single nodule identified on thoracic radio-graphs. One dog had a concurrent hemangiosarcoma of the third eyelid. Two dogs had no known comorbidities. Comorbidities in the 8 remaining dogs included orthopedic disease (n = 3), dermatologic disease (2), undefined heart murmur (2), myxomatous mitral valve dysplasia (1), epilepsy (1), and cystolithiasis (1) with some dogs having > 1 comorbidity.

For the 5 dogs with hematoma, 1 had hemoperitoneum and 4 did not. Four dogs had a single mass within the spleen, and 1 dog had multiple masses. The median largest mass dimension was 10 cm (range, 8 to 20 cm). No dogs with hematoma had evidence of metastasis. Three dogs had no known comorbidities, and the remaining 2 dogs had a testicular seminoma (n = 1) or pulmonary hypertension (1).

Of 25 miRNAs evaluated, 21 miRNAs were successfully amplified (Table 1), 4 were not identified in the samples (ie, miR-134-5p, miR-136-5p, miR-758-3p, and miR-876-5p), and all 3 reference miRNAs (miR-103a-3p, miR-16-5p, and miR-93-5p) were successfully amplified (Supplementary Table S1, available at: avmajournals.avma.org/doi/suppl/10.2460/javma.82.8.659). Control tissue and serum samples had similar expression patterns (Table 2). Hemangiosarcoma tissue and serum samples had varying expression values for 14 of 21 miRNAs. Of these, 13 were more highly expressed in serum, compared with tissue, and 1 was more highly expressed in tissue, compared with serum. Control tissue and hemangiosarcoma tissue samples had varying expression values for 12 of 21 miRNAs. Ten were more highly expressed in control tissue than hemangiosarcoma tissue, and 2 were more highly expressed in hemangiosarcoma tissue than control tissue.

Five miRNAs had significantly different expression in serum among the 3 groups (Table 3), with all 5 having lower expression in control serum than serum from dogs with either hemangiosarcoma or hematoma.

Discussion

Circulating miRNAs have been successfully used in human medicine to differentiate healthy from disease states,^19,20,21 and the objective of the present study was to determine whether similar methods could be used to differentiate dogs with histologically normal spleens, dogs with splenic hemangiosarcoma, and dogs with splenic hematoma. Five circulating miRNAs (miR-214-3p, miR-452, miR-494-3p, miR-497-5p, and miR-543) were identified that distinguished healthy control dogs from dogs with splenic masses. These 5 miRNAs had higher expression in dogs with either hemangiosarcoma or hematoma than healthy control dogs, indicating upregulation of these circulating miRNAs occurs in these splenic disease states. Of the 5 miRNAs that had differential expression between control dogs and dogs with hemangiosarcoma or hematoma, 2 have been studied in canine hemangiosarcoma previously.

In cell lines and clinical tissue samples of dogs with hemangiosarcoma in 1 study,³² miR-214 was found to be downregulated. Another study found that miR-214 was upregulated in microvesicles from cell lines of hemangiosarcoma and was also upregulated in the plasma of dogs with hemangiosarcoma.²⁷ This is consistent with the findings of the present study, in which miR-214 expression was significantly higher in the serum than in the tissues of dogs with hemangiosarcoma. In 1 study,³³ miR-214 was shown to be anti-angiogenic in mice through its targeting of quaking, a protein required for vascular development. It has also been shown to act as a tumor suppressor in canine hemangiosarcoma by targeting p53 through modulation of COP1 protein expression.³² Other studies have demonstrated oncogenic properties of miR-214. One group found that blockade of miR-214 reduced tumor progression and metastasis in mice.³⁴ Another demonstrated that overexpression of miR-214 led to down-regulation of PTEN in mice.³⁵ Inactivation of PTEN has been demonstrated in canine hemangiosarcoma, with loss of the PTEN protein leading to increased cell proliferation and reduced apoptosis through the phosphoinositide 3-kinase/Akt pathway.²² These conflicting roles identified for miR-214 may relate to its location, being either intra- or extracellular,²⁷ and further work needs to be done to determine the specific action of miR-214 in canine splenic masses.

An miRNA-sequencing study found that miR-452 was associated with splenic hemangiosarcoma, consistent with the findings in the present study.²⁵ Although miR-452 has not been investigated thoroughly in canine splenic masses, it has been shown to promote tumor progression in human hepatocellular carcinoma through increased expression of epidermal growth factor receptor and Akt.³⁶ Akt is a prosurvival protein and is an end product in the phosphoinositide-3-kinase pathway. Reduction of PTEN protein leads to phosphorylation of Akt and a prosurvival state. In canine hemangiosarcoma cell lines, epidermal growth factor receptor has been shown to support anchorage-independent growth, which is important for metastasis.³⁷ Additional investigation needs to be performed to determine whether miR-452 acts through these same pathways in canine splenic masses.

In humans, miR-497-5p is associated with malignancy in angiosarcoma and has been shown to directly target KCa3.1, a calcium-activated potassium channel.³⁸ Such potassium channels have been previously shown to regulate activity of cancer cells, increasing cell proliferation and metastasis. Downregulation of miR-497-5p was found in human angiosarcoma tissue samples, compared with capillary hemangioma samples.³⁸ Similarly, expression of this miRNA in the present study was lower in hemangiosarcoma tissues, compared with control tissues; however, miR-497-5p had higher expression in the serum of dogs with hemangiosarcoma, compared with tissue samples, indicating that diseased tissue expression and serum levels of miRNAs may not always correspond.

Although miR-494-3p and miR-543 have not been previously associated with hemangiosarcoma in dogs or angiosarcoma in humans, they have been associated with other cancer types. In human glioma and prostate cancer, miR-494-3p is overexpressed.^20,39 In human non–small cell lung cancer and colorectal cancer, miR-543 has been shown to promote angio-genesis and tumorigenesis.^40,41 Although their specific mechanism of action in canine splenic masses is unknown, both miR-494 and miR-543 have been shown to target PTEN directly, which is known to be downregulated in canine hemangiosarcoma.^22,41,42 The results of the present study indicated that the up-regulation of these miRNAs is also present in dogs with splenic masses.

These 5 miRNAs were able to differentiate splenic hemangiosarcoma and hematoma from histologically normal spleens with high specificity and sensitivity. None of the miRNAs were differentially expressed between the serum of animals with splenic hemangiosarcoma and hematoma. These are 2 of the most common diagnoses for splenic masses, and although prognosis is substantially different between them, the initial treatment of splenectomy is the same. These miRNAs may be a useful tool for screening dogs to identify splenic masses at an earlier stage of disease, thereby decreasing morbidity and mortality rates by preventing splenic rupture and hemoperitoneum for both conditions and improving treatment outcomes for dogs with hemangiosarcoma with earlier diagnosis.

When comparing control tissue to hemangiosarcoma tissue, differential expression for 12 of 21 miRNAs evaluated was found, with control tissue having higher expression of 10 of 12 miRNAs. Further examination into the significance of these differences in the healthy versus diseased state may provide more insight into the role of these miRNAs in hemangiosarcoma. No differences were found in miRNA expression between control tissue and control serum, indicating that for the 21 miRNAs evaluated, expression between the spleen and the serum was stable.

For dogs with hemangiosarcoma, 14 of 21 miRNAs had differential expression between tissue and serum, with 13 being upregulated in the serum, compared with the tissues. This is interesting in that miRNAs present in serum may be regulated by more complex biological processes. Hemangiosarcoma is a tumor of the vascular endothelium and is histologically characterized by irregular vascular channels, papillary structures that bulge into these irregular vascular channels, and frequent anaplasia.⁴³ These vascular abnormalities may explain why these miRNAs are increased in the serum, as they may be more likely to be shed into circulation either from circulating tumor cells or as cell-free miRNAs.⁴⁴ Another explanation lies in the export of miRNAs from cells, either in extracellular vesicles or associated with Ago proteins.^45,46,47 Extracellular vesicles and miRNA-Ago protein complexes are exported from cells into circulation from both active cells and those undergoing apoptosis or necrosis. Extracellular vesicles have been shown to promote cancer metastasis through multiple mechanisms, including roles in angiogenesis,⁴⁵ and it is possible that the miRNAs identified here may have a therapeutic potential in future treatment of canine splenic masses.

Four miRNAs were not successfully amplified in the samples from the present study. Two of these miRNAs were found via a database search,^g and the other 2 were identified in an miRNA-sequencing study.^25,28 Despite a 100% sequence match between the human and canine miRNAs, it is possible that cross-binding or technical issues with the RT-qPCR assay occurred. Further work should be done to confirm that these sequences match between humans and dogs and to determine whether these miRNAs exist in the canine spleen.

Although the present study is limited by the small number of dogs included, 5 miRNAs were identified with higher expression in the serum of animals with splenic masses with high sensitivity and specificity. Larger sample size may lead to significant differences between expression patterns of miRNAs between hemangiosarcoma and hematoma samples. Future studies may also be used to determine whether other differences, including tumor rupture and hemoperitoneum or the receipt of a blood transfusion, affect miRNA expression. It is possible that the splenic tissue samples obtained were not reflective of hemangiosarcoma as a result of sampling location, although tissue adjacent to the samples taken for the study were submitted for histologic examination to ensure collection of hemangiosarcoma tumor cells to control for this variable. Additionally, influence of age, sex, breed, comorbidities, and other clinical factors on the results of differential miRNA expression is unknown. Other miRNAs that were not assessed in the present study may also be involved in the pathogenesis of splenic masses. Use of other methods of analysis such as microarray or RNA sequencing may be useful to identify additional miRNA that may be involved in the pathogenesis of canine splenic masses.

The 5 miRNAs identified in the present study distinguished healthy dogs from those with other splenic masses via serum evaluation, which has not been previously done. Validation testing of these 5 miRNAs to detect splenic masses is the next step in developing a minimally invasive screening test and is a future goal of this work with a larger number of dogs included. Such a minimally invasive diagnostic test is needed for the diagnosis of splenic masses in dogs. Abdominal ultra-sonography is not routinely used to screen older dogs unless symptoms are present pertaining to abdominal disease, and smaller lesions may be missed, whereas a blood test could be more easily incorporated into regular screening. Earlier intervention would be helpful to prevent tumor rupture, hemoperitoneum, and subsequent hypovolemic shock and the need for emergency surgery in dogs with hemangiosarcoma or hematoma. In the case of hemangiosarcoma, earlier intervention may also improve prognosis. With additional testing, a blood test with a panel of miRNA such as those identified here could be added to routine blood testing of older dogs to detect the presence of splenic masses earlier than may otherwise be possible.