Cyclooxygenase-2 as a biomarker with diagnostic, therapeutic, prognostic, and predictive relevance in small animal oncology

Neoplasms are the second leading cause of death in the world after cardiovascular diseases. According to the World Health Organisation, 8.8 million people died of tumours in 2015 (http://www.who.int/cancer/en/index.html) Polish National Cancer Registry data show that the prevalence of malignant tumours has more than doubled in the last 30 years. In 2016 there were 164,140 cancer sufferers in Poland (427.1 cases/ 100,000 people) (http://onkologia.org.pl)

In canine and feline populations, the number of neoplasm cases also continues to increase around the world (60). According to Bronson (10) the death rate from tumours is the highest in older animals and reaches 45% in dogs older than 10 years. In cats, tumours are also most prevalent in older animals with an estimated death rate of 32% in cats older than 10 years. This can be attributed to the fact that companion animals live in the human environment and their predisposing risk factors are the same as humans. Advanced diagnostic techniques have also increased the detection rate for tumours in veterinary medicine (60).

In research centres, attempts are being made to identify new biomarkers that speed up and improve the quality of oncological diagnostics and have prognostic and predictive relevance in humans and animals with tumours (36). Rapid advances in knowledge of biochemical and especially molecular mechanisms responsible for oncogenesis and the progression of various types of tumours facilitate the development of new, more effective diagnostic methods and therapeutic protocols. The existing methods should be integrated, and the results of clinical evaluations, post-mortem examinations, histopathological, immunohistochemical, cytological, and molecular analyses should be used to deepen our understanding of tumour development and to improve diagnosis and treatment. The development of reliable parameters with prognostic and predictive applicability poses one of the greatest challenges in the treatment of tumours. Various biological, clinical, histological, immunohistochemical, and molecular parameters have been evaluated to date, but the choice of the most effective markers remains an open issue (28, 36, 46).

Cyclooxygenase-2 (COX-2) is an important diagnostic and prognostic biomarker to which human oncology has increasing recourse, while veterinary oncology does so to a lesser extent (50). COX-2 is not yet used as a biomarker in routine cancer screening in human medicine (36). This enzyme is overexpressed in humans presenting with various types of tumours and in selected types of tumours in animals, particularly in dogs. These findings indicate that COX-2 could be a potentially effective biomarker with diagnostic, therapeutic, prognostic, and predictive relevance in oncology. However, further research is required, particularly in veterinary medicine, because researchers are divided in their opinions on the expression of COX-2 in different types of tumours (7, 41).

Cyclooxygenase (COX) is an enzyme belonging to the myeloperoxidase family, which catalyses conversion of the arachidonic acid to prostanoids. These comprise prostaglandins, prostacyclin and thromboxane, which are bioactive proteins which regulate various physiological processes in human and animal organisms (14). COX-2 as an inducible isoform is found at low levels in mammalian cells and it is not generally detected in physiological conditions, although the presence of COX-2 was observed in the central nervous system, alimentary tract, heart, kidney, eye, and reproductive organs. In the beginning COX-2 was connected with the response to stress, but its overexpression is also found during fever, pain, and inflammation (50). In several studies, increased COX-2 expression was also demonstrated in neoplastic tissues, which suggests participation of this enzyme in carcinogenesis. COX-2 overexpression causes the cells’ phenotype to change from benign to malignant, which is connected with disruption of their growth and proliferation and an increase in cells’ ability to evade apoptosis and immune response, promote new blood vessels, and raise their invasive potential (50). More details of the function of COX-2 in the organism and the significance of COX-2 in oncogenesis were presented in our previous paper (57).

The mechanisms of COX-2 participation in oncogenesis are very complicated and weakly understood, particularly in animals. They comprise interactions between tumour cells and the surrounding microenvironment to create the best conditions for their growth, proliferation, and dissemination (22). A strict connection between chronic inflammatory processes and carcinogenesis was observed, when overexpression of COX-2 can contribute to the conversion of inflammation into cancer. Gandhi et al. (21) summarised the latest scientific achievements highlighting the significance of COX-2 and its downstream signalling effectors’ role in life-cycle events of the gammaherpesviruses – Epstein-Barr virus (EBV) and Kaposi’s sarcoma-associated herpesvirus enabling cancer progression. Among various inflammatory mediators including pathological processes leading to cancer, COX-2 and its effector molecules are of greater significance. They generate a microenvironment highly favourable for cancer development, progression, and metastasis. Recent studies showed a link between upregulated COX-2 levels and induction of lytic reactivation in gammaherpesvirus-infected cells. There are several cases on record of patients with chronic inflammatory processes and with COX-2 overexpression showing high incidences of EBV-associated malignancies, indicating the role of increased COX-2 levels in virus-mediated tumourigenesis.

Higher COX-2 expression was found in tumours of various organs in humans (17, 28) and dogs (19, 39), and to a lesser extent in cats (7, 39), which is often connected with a higher histological grade of tumour malignancy, worse prognosis, and shortening of overall survival (OS) (13, 23). In Poland previous studies evaluating COX-2 expression in animal tumours are relatively limited, comprising mainly canine mammary (44) and mast cell tumours (27).

Studies of COX-2 expression in animal populations in various countries have demonstrated that it is overexpressed in various canine epithelial tumours, including adenocarcinomas and carcinomas of the mammary gland, ovarian and prostate tumours, transitional cell (urothelial) carcinoma (TCC), colorectal and small intestine tumours, squamous cell carcinoma (SCC) of the skin and oral cavity, osteosarcoma, and melanoma (19, 20, 39, 46). The enzyme is minimally expressed or is not expressed in canine lymphoma, sarcoma, fibrosarcoma, glioma, and mastocytoma (41).

Research conducted on small populations of feline patients has revealed that unlike in dogs, COX-2 is expressed more weakly and far less frequently in cats. The enzyme was found in one study by Millanta et al. (39) to be overexpressed in 96% (45/47) of cats with invasive mammary carcinoma and in selected cases of urothelial TCC, skin and oral SCC, and pancreatic adenocarcinoma, but not in feline patients with intestinal, lung or mammary gland carcinomas, lymphomas of the nasal cavity and the intestines or vaccine-induced fibrosarcomas (7, 19). These differences have been attributed to variations in the size of analytical samples, sample preparation methods, lower COX-2 levels in cats with tumours, and the absence of interspecies reactions with human antigen antibodies (7).

Several research works on COX-2 expression in canine and feline neoplasms have been written in recent years. In most of them, overexpression of COX-2 in tumours occurring in dogs and cats was found. Millanta et al. (38) observed COX-2 positivity in 83% (30/36) of canine and 82% (41/50) of feline mammary carcinomas. The frequency of COX-2 expression was significantly higher in canine carcinomas than in non-neoplastic tissues (18%, 4/22) or adenomas (20%, 2/10), which supports the existence of a role of the COX-2/prostaglandin E₂ (PGE₂) pathway in the pathogenesis of these tumours. Their results are very similar to earlier results obtained by Sayasith et al. (53), who noticed COX-2 expression in 88% (35/40 cases) of feline mammary carcinomas at low (50%, 20/40), intermediate (33%, 13/40) and high (5%, 2/40) levels. Investigations of the relationship between the expression of COX-2 mRNA level and malignancy degree in canine malignant mammary tumours were conducted by Anadol et al. (1). They found that the expression of COX-2 mRNA was significantly higher in both benign and malignant mammary tumours than in adjacent non-neoplastic mammary gland tissue. COX-2 mRNA levels were related to the histological grade of malignancy, being higher in grade 3 malignant mammary tumours than in grade 2 tumours and higher in these than in grade 1 tumours.

Some studies concerned cutaneous and oral SCC in dogs and cats. Millanta et al. (37) noticed COX-2 overexpression in 53% (8/15) of canine and 61% (22/36) of feline SCC, which was higher in cutaneous (88%, 7/8 with overexpression in canines; 70%, 19/27 in felines) than in non-cutaneous lesions (14%, 1/7 with overexpression in canines; 33%, 3/9 in felines). In both species, expression of COX-2 correlated with the progression of disease but not with lymphatic invasion, tumour grading, or tumour classification in the cutaneous tumours. Sparger et al. (56) revealed positive but weak COX-2 immunostaining of neoplastic epithelium and stroma in 75% (9/12) cases of feline oral SCC. In a similar study, Nasry et al. (42) observed that tumour cells were more likely to express COX-2 (51%, 22/43) than stroma (19%, 8/43) and adjacent oral epithelium (29%, 9/31). These results give confirmation of former results obtained by Bardagi et al. (6), who found COX-2 immunoreactivity in all of their 27 feline and 9 canine cases of SCC.

Other canine and feline neoplasms were also the object of studies. Gregorio et al. (23) evaluated 50 cases of canine mast cell tumours (MCT) immunohistochemically for the expression of several biomarkers including COX-2. They found an association between COX-2 expression and higher grades of malignancy on the Patnaik and Kiupel grading scales. COX-2 expression was also associated with higher cell proliferative antigen (Ki-67) scores, higher mitotic index, and higher microvascularisation density, suggesting an active role of COX-2 in MCT oncogenesis mainly through proliferation and angiogenesis stimulation. Therefore in the authors’ opinion, COX-2 is a potentially relevant clinical prognostic marker and therapeutic target. Samarini et al. (51) investigated COX-2 expression in 15 cases of feline meningioma for any possible association between COX-2 immunoreactivity and tumour grade. They noticed that all tumour cases were immunoreactive to COX-2. No significant correlation between COX-2 expression and tumour grade was found, but some was found between COX-2 expression and necrosis. The results of these studies indicate COX-2 expression in feline meningiomas, but without any difference between low- and high-grade tumours, however the association between COX-2 expression and the presence of necrosis indicates the possibility for therapy with selective COX-2 inhibitors. The very recent study of Santelices Iglesias et al. (52) to determine COX-2 expression in 117 cases of feline injection site sarcoma (FISS) showed that COX-2 immunolabelling was positive in 56.4% (66/117) of FISS cases. There was a significant association between COX-2 expression by neoplastic cells and a higher degree of inflammation, but COX-2 expression was lower in tumours with a higher degree of anaplasia. The authors conclude that these findings may be useful in predicting the sensitivity of FISS to treatment with COX-2 inhibitors, but their potential therapeutic use could be restricted to tumours with a lower degree of anaplasia. These results were in accordance with those of the investigations of Magi et al. (35) and Carneiro et al. (11), who found COX-2 expression in 97.0% (30/31) and 61.9% (13/21) of FISS cases, respectively, but in contrast to those of Beam et al. (7), who did not find COX-2 expression in any FISS cases.

The recent very considerable progress in molecular techniques for tumour diagnosis, including the emergence of DNA and RNA sequencing tools, single-nucleotide-polymorphism–based genotyping, and evaluation of mRNA, ncRNA, and miRNA profiles, as well as the achievements in proteomics, metabolomics, and bioinformatics have expanded our understanding of the molecular mechanisms responsible for tumour initiation, progression, and response to treatment, in particular in combination with histopathological and clinical evaluations (28, 36, 50). The results of intensive large-population studies are rapidly implemented in tumour treatment, albeit with varying degrees of success. However, they are less often used to develop sensitive and specific markers with diagnostic, prognostic, and predictive relevance in cancer medicine. Despite the subordinacy of this goal, the number of identified biomarkers with potential use in cancer diagnosis and treatment as well as prognostic and predictive biomarkers continues to increase (36).

Prognostic biomarkers such as β-tubulin, carbohydrate antigen 19-9 (CA19-9), cell-surface antigen CD44 (CD44), carcinoembryonic antigen, ColoPrint, circulating tumour cell, cyclin D1, E-cadherin, epidermal growth factor receptor (EGFR), inhibitor of growth protein 3 (ING3), Ki-67, matrix metalloproteinase-2, p-21, retinoblastoma gene, ribonucleotide reductase M1, and vascular endothelial growth factor (VEGF) are used to evaluate the malignant potential of tumours and measure the patients’ OS and progression-free survival (PFS) without treatment and after conventional treatment. These biomarkers are applied to qualify patients for treatment, but they do not support the prediction of treatment outcomes (Table 1) (36).

Predictive biomarkers such as epidermal growth factor receptor 1 (EGFR1), excision repair cross-complementation group 1, O(6)-methylguanine-DNA methyltransferase, thymidyne phosphorylase, and phosphatase and tensin homolog support objective identification of individuals who are more likely to benefit from a given treatment or aid the evaluation of differences in the outcomes of two or more treatment procedures in view of their toxicity (Table 2) (36).

Some biomarkers, including breast cancer gene (BRCA1), carbonic anhydrase IX (CAIX), oestrogen receptor (ER), progesterone receptor (PR), tumour suppressor protein (p53), human epidermal growth factor receptor 2 (HER2/neu), and Kirsten rat sarcoma oncogene, have both prognostic and predictive relevance (36). Multigene panel tests are also used to identify groups of up to several dozen genes, mainly in the diagnosis of breast cancer, for which application MammaPrint (59) or Mammostrat (49) are examples of available assays.

The increased expression of COX-2 in various types of tumours, in particular in dogs, but also in cats, suggests possibilities for its utilisation in practice. Its introduction may be feasible into routine evaluation as a diagnostic, therapeutic, prognostic, and predictive biomarker in small-animal veterinary oncology especially, in like manner to how it is exploited to a certain extent in human oncology (5).

In human medicine, COX-2 overexpression in tumour patients is often associated with poor prognosis and reduced OS and/or PFS (30). The applicability of COX-2 in the diagnosis of canine tumours requires further research because the results of studies evaluating these associations and another between the overexpression and response to treatment are contradictory (19, 20). Correlations with poor prognosis and reduced OS have been observed in canine mammary gland carcinoma (47), whereas no such relationships have been reported in canine prostatic carcinoma (55). Queiroga et al. (47) evaluated COX-2 expression in canine mammary tumours to assess its prognostic significance and any connection with clinical and pathological parameters. They examined 129 mammary tumour samples from 57 bitches of various breeds aged 6–14 years, including 22 from dysplastic lesions, 40 from benign and 57 from malignant tumours, and 10 from inflammatory carcinomas. Thirteen samples from normal tissues were examined for comparison. COX-2 expression was found in all samples, but with various intensities – the lowest in normal tissues and the highest in inflammatory carcinomas. COX-2 expression increased with tumour malignancy. Directly proportional relationships were found between COX-2 expression and various clinico-pathological parameters – tumour size, skin ulceration, adherence to the tissues and skin, histological type, time of metastases and relapses, worse prognosis, and shorter PFS and OS, especially in inflammatory carcinomas. In a similar study, Millanta et al. (39) determined COX-2 expression in invasive mammary carcinomas in 47 queens aged 8.8 ± 2.5 years and 28 bitches aged 10.9 ± 2.7 years and also measured the expression of this enzyme in normal tissues to assess the relationship to clinico-pathological features and explore its prognostic aptitude. COX-2 expression was evaluated in relation to age, tumour size, histological type, blood vessel density, expression of ER and PR receptors, expression of Ki-67, HER-2 and VEGF, and OS. In both animal species COX-2 expression was not observed in normal tissues, but it was found in tumour cell cytoplasm in 100% (28/28) of bitches and 96% (45/47) of queens; in 79% of bitches and 81% of queens the COX-2 expression was rated as average to strong. In bitches, an increase in COX-2 expression was significantly correlated with overexpression of HER-2 and weak differentiation of tumour cells, but in queens with the status ER(−) and PR(+) and increased VEGF expression, it indicated higher malignancy. Heller et al. (25) selected 50 bitches and performed an evaluation of the interaction of COX-2 expression and various histological types of canine mammary carcinomas – adenocarcinoma, solid carcinoma, and anaplastic carcinoma. Expression of COX-2 was found in 56% (28/50) of cases, including 47% (17/37) of adenocarcinomas and 100% (11/11) of anaplastic carcinomas, while there was no expression (0/2) in solid carcinomas. The intensity of COX-2 expression (the immunohistochemical score – IHS) varied within the range 1–3 (average IHS 1.0) in adenocarcinomas and 2–12 (average IHS 5.1) in anaplastic carcinomas. These studies demonstrated that weakly differentiated tumours show stronger COX-2 expression than well-differentiated ones, e.g. adenocarcinomas. Queiroga et al. (48) examined the prognostic value of COX-2 expression in malignant canine mammary tumours in 27 bitches also through evaluation of correlation with clinico-pathological parameters such as tumour size, histological type, presence of necrosis, metastases to lymph nodes, and PFS and OS. COX-2 expression was found in all cases over a broad range of IHS scores (3, 4, 5, 6, 7, 8, 9, 10, 11, 12), but was predominantly noted at scores for stronger expression intensity (average IHS 8.8). Expression of COX-2 was significantly higher in tumours with metastases to lymph nodes, but a correlation between COX-2 expression and tumour size, histological type, or presence of necrosis was not noticed. However, the study showed a statistically significant correlation between strong COX-2 expression and shorter PFS and OS and worse prognosis.

Because limited veterinary literature is available regarding prognostic biomarkers for canine renal cell carcinoma (CRCC), Carvalho et al. (13) retrospectively evaluated COX-2 expression and histological and clinical features associated with the prognosis of CRCC in 64 cases in which nephrectomy had been required. COX-2 expression was significantly associated with overall median survival time (MST) – 420 days if the COX-2 score was > 3 versus 1176 days if it was < 3. The authors concluded that the addition of COX-2 immunostaining to standard histopathological evaluations would help to predict outcomes in CRCC patients. De Campos et al. (15) investigated several prognostic factors including COX-2 in feline mammary gland neoplasms, correlating them with OS. Immunoreactivity for COX-2 was higher in metastases than in primary tumours and was directly correlated with OS. The authors suggest that COX-2 inhibition may represent a therapeutic option for malignant feline mammary gland neoplasms. COX-2 scores should be analysed in primary tumours and metastases for a better understanding of disease outcome in patient conditions characterised by a poor prognosis. The recent study of Gregorio et al. (23) revealed a strict connection of COX-2 overexpression to OS in canine MCT. Confirmation of these findings could also be recognised in the results obtained by Carvalho et al. (12) on 109 cases of canine mammary tumours. High COX-2 expression was associated with more serious grades of malignancy, lymph node metastasis, and shorter OS. In a similar study, during investigation of several biomarkers including COX-2, Araujo et al. (2) found a concordance of COX-2 expression, worse prognosis and shorter OS in canine mammary primary carcinomas with lymph node metastasis. Nobrega et al. (43) evaluated the five biomarkers - factor VIII (FVIII), COX-2, vascular endothelial growth factor (VEGF), proliferating cell nuclear antigen (PCNA), and caspase-3 (casp-3) in relation to OS in 60 cases of canine cutaneous haemangiosarcoma (cHSA). Marker expression was positive in 80–100 % of samples, with weak to moderate labelling intensity for FVIII, COX-2, and VEGF and strong for PCNA and casp-3, but without any relationship to OS. The authors concluded that although expression of COX-2 and VEGF is frequent in canine cHSA, these possible therapeutic targets need further investigations for greater clarity about their potential in treatment.

Experimental, clinical, and epidemiological studies have demonstrated that non-steroidal anti-inflammatory drugs (NSAIDs), in particular selective COX-2 inhibitors (coxibs), effectively inhibit tumour progression and improve chemotherapy outcomes in human patients (16). Specific COX-2 inhibitors, including celecoxib and rofecoxib, have been developed to minimise some mainly gastrointestinal side-effects of NSAIDs. In addition to anti-inflammatory, analgesic and antipyretic effects, these compounds also deliver anticarcinogenic effects by inhibiting the production of prostanoids. In some cases, however, anticarcinogenic effects were observed independently of COX-2 inhibition. Tamura et al. (58) examined the antitumour effects of celecoxib in an AZACB canine mammary tumour cell line and utilised a cell line expressing low levels of COX-2 to minimise its effect on celecoxib’s activity. They revealed that celecoxib downregulated COX-2 expression, induced cell apoptosis and inhibited cell proliferation mainly via COX-2-independent mechanisms. The results of these studies suggest that celecoxib might be used in the treatment of canine mammary tumours regardless of COX-2 expression, also in combination with other antitumour agents. This discovery has led to the development of structural analogs such as dimethyl-celecoxib (DMC), which effectively inhibits cell proliferation and induces apoptosis through the downregulation of survivin and cyclins A and B and the ensuing loss of cyclin-dependent kinase activity. DMC does not provoke the side-effects associated with COX-2 inhibition; however, further research on it and the compounds of its type is required (29).

The discovery that coxibs possess anticarcinogenic properties laid the groundwork for clinical research in human oncology, which initially focused on coxibs’ chemopreventive and subsequently on its chemotherapeutic effects. Initial studies demonstrated that coxibs are effective in the treatment of familial adenomatous polyposis (FAP), but subsequent large-population research programmes revealed that coxibs have significant cardiovascular side-effects. Due to safety concerns, rofecoxib has been withdrawn from the pharmaceutical market, and celecoxib is presently prescribed only as a chemopreventive agent for FAP (3). However, a review of 72 research programmes carried out by Harris (24) did not confirm those concerns and found that coxibs caused side-effects only in patients with a higher risk of cardiovascular diseases.

The therapeutic effects of NSAIDs in cancer treatment have been confirmed by numerous studies which investigated the combined application of NSAIDs, radiotherapy, and chemotherapy in human patients (34). Overexpression of COX-2 has also been observed in some canine and feline tumours, and research findings indicate that this enzyme could be more widely used as a biomarker in veterinary medicine, in the diagnosis and treatment of cancer with the use of COX-2 inhibitors (39). This biomarker could be applied to identify patients where the use of non-selective and, in particular, selective COX-2 inhibitors could reduce COX-2 overexpression, limit tumour progression and increase survival rates (16, 34).

The use of NSAIDs in the treatment of canine and feline tumours has been investigated by relatively few studies, which, nevertheless, produced interesting results. Boria et al. (9) found that cisplatin administered in combination with piroxicam induced remission in five out of nine dogs with oral SCC and in two out of eleven dogs with oral malignant melanoma. Schmidt et al. (54) observed that piroxicam administered per os at 0.3 mg/kg/day induced remission in 3 out of 17 dogs and inhibited tumour growth in 5 out of 17 dogs with oral SCC. In a similar study, which was conducted to assess COX-2 expression in feline oral SCC and the COX-2-inhibitory activity of piroxicam in carcinoma-afflicted cats, Di Bernardi et al. (18) found that piroxicam at a dose of 0.3 mg/kg b.w. daily was a potentially beneficial treatment option for cats with oral SCC and with COX-2 overexpression in cancer cells and would be a notable improvement to current therapy. In a study performed on a canine model of human invasive urinary bladder cancer, Knapp et al. (33) demonstrated that cisplatin administered intravenously at 60 mg/m² every 21 days in combination with piroxicam (0.3 mg/kg/day per os) induced remission in 10 out of 14 dogs with invasive TCC of the urinary bladder but none was observed in the animals administered cisplatin only. Piroxicam was found to reduce tumour size, induce apoptosis, and reduce angiogenesis in 12 out of 18 dogs with urothelial TCC in research by Mohammed et al. (40). Feline TCC of the urinary bladder improved clinically when treated with meloxicam in findings made by Bommer et al. (8), who saw reduction of haematuria and/or dysuria with MST of 311 days. COX-2 expression was associated with MST, which in COX-2-positive cats was 123 days and for COX-2-negative cases was 375 days. Itturiaga et al. (26), examining the influence of low-dose meloxicam (0.25 μg/mL) on CF41.Mg canine mammary carcinoma cells, noticed that cell migration and invasion were significantly reduced and suggested that meloxicam has a potential adjunctive therapeutic application useful in controlling the invasion and metastasis of canine mammary carcinoma. Similarly, Pang et al. (45) compared the in vitro action of the short-acting non-selective COX inhibitor carprofen with that of the long-acting selective COX-2 inhibitor mavacoxib on cancer cells and cancer stem cell survival. They observed that mavacoxib increases apoptosis in cancer cells and has an inhibitory effect on cell proliferation and migration, but they suggest that these anti-tumour effects of mavacoxib warrant further study. King et al. (31) evaluated the safety of NSAID COX-2 inhibitor robenacoxib in healthy young beagle dogs and found no adverse effects of this highly selective COX-2 inhibitor administered orally once daily even at a highest dose of 40 mg/kg b.w. for one month and 10 mg/kg b.w. for six months. They also found that application of robenacoxib was associated with marked inhibition of COX-2. Similar results were obtained by King et al. (32) in healthy young short-haired cats in a study concerning the safety of oral robenacoxib and inhibition of COX-2. Arenas et al. (4) researched adjuvant therapy and evaluated the disease – free survival (DFS) and OS of COX-2 inhibitor firocoxib versus those of chemotherapy with mitoxantrone in dogs with highly malignant canine mammary tumours (HM-CMTs) and those of control dogs in a case-control prospective study. They noticed that dogs receiving firocoxib treatment had statistically higher DFS and OS than control dogs. The DFS and OS of dogs medicated with mitoxantrone were, however, not statistically different from those of the controls. The authors concluded that their study supported the use of firocoxib for the treatment of HM-CMTs, but that further studies were needed to compare the efficacy of chemotherapy drugs with that of COX-2 inhibitors as adjuvant treatment in such cases in dogs.

Although there are several pieces of evidence supporting an important role of COX-2 in tumour development and progress in humans and animals, further studies are necessary to explore its significance. Subsequent investigation will elucidate the agency of COX-2 in oncogenesis, determine COX-2 expression levels in various types of canine and, in particular, feline tumours, assess the diagnostic, therapeutic, prognostic, and predictive relevance of this biomarker precisely, and evaluate the usefulness of NSAIDs in the chemoprevention and chemotherapy of canine and feline tumour patients.