Peripheral blood eosinophilia in dogs: Prevalence and associated diseases

Abstract Background Canine eosinophilia has not been evaluated over the last two decades. As in human local differences, changes in the prevalence and associated diseases over time can be expected. Objective This study aims to determine the prevalence and causes of marked blood eosinophilia in dogs. Methods Retrospective study. A total of 317 clinical histories of dogs with an eosinophil concentration > 1.5 × 109/L (marked eosinophilia) between 2013 and 2017 were evaluated. Patients were allocated to 10 groups according to their major clinical findings. Results Eosinophilia was present in 1,592 of 10,829 dogs (14.7%); it was mild (0.8–1.49 × 109/L) in 78.4%, moderate (1.5 – 4.9 × 109/L) in 20.5% and severe (> 5 × 109/L) in 1.1% of cases. Rottweilers were overrepresented (16.1%). Of 317 cases with marked eosinophilia, 19.6% had neoplasia, 19.1% gastrointestinal disorders, 13.6% health check, 10.4% endoparasites, 6% respiratory, 5.4% neurologic, 5.4% dermatologic, 4.8% urogenital, 3.2% endocrine disorders and 12.6% miscellaneous. Lymphomas (29%) and mast cell tumours (12.9%) were the most frequent tumours in the neoplasia group. A total of 72.6% of tumour‐bearing dogs were older than 8 years, while 63.6% of dogs had endoparasites, and 86% of apparently healthy dogs were younger than 5 years. Eosinophilia was significantly higher in patients with respiratory disorders (p < 0.0146). Leukocytosis was found in 50.2% of cases. Conclusion Malignancy was the most common cause of marked blood eosinophilia in older dogs and endoparasitism in younger dogs. Eosinophilia was common in apparently healthy young dogs and may be related to undiagnosed parasitic infestations.


INTRODUCTION
Peripheral blood eosinophilia is related to a wide variety of causes, ranging from hypersensitivity and parasitic infestations to neoplasia.
As in people, it seems that local differences in the incidence also exist in animals (Lilliehöök et al., 2000;O'Connell & Nutman, 2015). While in resource-limited countries, parasitic infestations represent the most common cause of eosinophilia in people, hypersensitivity reactions should be considered a primary differential diagnosis in developed countries. Likewise, it can be expected that climatic changes affect the distribution of intermediate hosts necessary for the replication of helminths or the spread of infections transmitted by ticks, which finally leads to a change in the distribution of parasites that could be reflected in a local change in causes of eosinophilia (Skuce et al., 2013). Even if the causes of eosinophilia are well known, studies in local dog populations are scarce, and recent data are missing-the last one was published about two decades ago (Lilliehöök et al., 2000).
Moreover, changes in the prevalence of eosinophilia and associated diseases in Central Europe can be expected as a result of changes within the dog population (e.g., spectrum of breeds, age) or the increase in certain immunologic or neoplastic diseases. The latter is associated with increased life expectancy and better veterinary care for geriatric pet animals. The aim of this retrospective study was to determine the prevalence, grade and associated diseases in dogs with marked peripheral blood eosinophilia presented to the Small Animal Clinic of the University of Veterinary Medicine, Vienna over a 4-year period.
TA B L E 1 Dog breeds with a high prevalence (more than 10 dogs) of marked eosinophilia (≥1.5 × 10 9 /L, n = 314) in relation to the total of canine complete blood cells counts for the same breeds during the study period Immunophenotype was available in 13 dogs with lymphoma; eight dogs presented a B-cell lymphoma, and five dogs presented a T-cell lymphoma. In five dogs, the immunophenotype was not determined.
Eosinophilia was moderate in all but one dog in this group (median 1.86, range 1.5-20.03 × 10 9 /L). Senior dogs were overrepresented in this group ( Figure 1); forty-five dogs (72.6%) with neoplasia were older than 8 years (median 9.9, range 0.7-18.6 years); only one young dog was in this category. This 7-month-old dog was diagnosed with acute myeloid leukaemia and showed the highest eosinophil and leukocyte counts in the present study (20.03 × 10 9 /L and 143.1 × 10 9 /L, respectively). in all but one dog in this group (median 1.95; range 1.5-6.2 × 10 9 /L).
Only one 2-month-old dog with severe acute gastroenteritis showed severe eosinophilia. Additionally, this dog showed a leukaemoid reaction (103 × 10 9 /L), and endoparasites and parvovirosis tested negative several times.
Forty-three (13.6%) dogs were presented for health checks. These dogs were apparently healthy and included blood donors or dogs that were clinically evaluated before elective surgery (castration, minor surgical procedures). Thirty-seven (86%) dogs in this group were younger than 5 years (five young and 32 young adults; median 1.7, range 0.8-12.8 years). Severe eosinophilia was found in only one dog (median 1.97; range 1.51-5.87 × 10 9 /L). This dog, a Greyhound from a Spanish shelter, was presented for entropium surgery. This patient was dewormed and tested negative for leishmaniosis.

DISCUSSION
The aim of the present study was to determine the prevalence of blood eosinophilia in dogs and to determine the cause of marked eosinophilia in the Vienna City area, Austria. Causes of peripheral blood eosinophilia are well-established, but as in humans, regional differences should be expected (Barrett et al., 2017). Only one paper has reviewed the causes of blood eosinophilia in dogs within the last 20 years (Lilliehöök et al., 2000). In that study, the prevalence for eosinophilia was 10%, given a cut-off of 1.25 × 10 9 /L. When we consider our laboratory's reference value of 0.8 × 10 9 /L, the prevalence of eosinophilia in the present study was 14.7% and 5.2% when using the same cut-off as the previous Swedish study and 3.17% for marked eosinophilia (≥ 1.5 × 10 9 /L). As in Sweden, we also found differences among breeds; Rottweilers showed more frequent eosinophilia and exhibited significantly higher eosinophil counts, compared with other breeds (Lilliehöök et al., 2000). This could be related to a predisposition of Rottweilers for eosinophilic disorders such as eosinophilic gastrointestinal (GI) diseases and meningoencephalitis as well as the rarely with haematologic malignancies such as lymphomas and several solid tumours (Gotlib, 2017). In the present study, lymphomas represent 29% by far the largest group among neoplasia-associated eosinophilia.
In contrast to previous studies, in our cohort, B-cell lymphomas were more frequently associated with eosinophilia (eight out of 13) than T-cell lymphomas (Marchetti et al., 2005;Ozaki et al., 2006). However, B-cell lymphomas are only occasionally reported to be associated with eosinophilia (Tomiyasu et al., 2010). Mast cell tumours were the second most frequently diagnosed neoplasia, while eosinophilic tissue infiltration is known to be a common finding in this type of tumour, and peripheral eosinophilia is only rarely documented (O'Keefe et al., 1987;Skor et al., 2017;Takahashi et al., 2000;Tomiyasu et al., 2010).
So far, only one report has described a case of a dog with severe eosinophilia (24 × 10 9 /L) having a cutaneous low-grade mast cell tumour. Eosinophils returned to normal concentrations after starting therapy with prednisone, increasing again after recurrence and lymph node metastasis (Musser et al., 2018). Nevertheless, in a smaller number of dogs, we found that eosinophilia was present in a wide variety of other neoplasms. However, the association with non-haematologic tumours, especially carcinomas, is well known (Losco, 1986;Samoszuk, 1997). The role of eosinophils in neoplasia is not clear; possible mechanisms are related to remodelling of the connective tissue after tissue damage by the growing tumour, but cytotoxic effects of eosinophils on tumour cells have also been suggested (Davis & Rothenberg, 2014).
The release of protein material through necrotic changes as well as cytokines such as interleukin-5 (interleukin [IL]-5), IL-3 and eotaxin-1 produced by tumour or host inflammatory cells attract eosinophils to neoplastic tissue. On the other hand, blood eosinophilia results from the stimulation of the bone marrow by colony-stimulating factors produced by some specific neoplastic cells (Samoszuk, 1997). The significance of eosinophilia in tumour behaviour is not clear, but blood eosinophilia seems to be related to widespread metastases and poor prognosis. The fact that eosinophils return to normal values after treatment and increase again with recurrence or metastasis in patients with neoplasia can help to predict the course of the disease (Musser et al., 2018).
GI disorders (19.1%) represented the second most common cause of eosinophilia in the dogs of the present study, comparable to the Swedish survey, where 25% of cases were in this group. Even if peripheral eosinophilia is an inconstant finding in patients with GI disorders (Talley et al., 1990), it has been related to acute and chronic GI diseases (Lilliehöök et al., 2000;Mehta & Furuta, 2015), and less frequently with other inflammatory conditions such as pancreatitis, especially when severe tissue injury to adjacent organs occurs (Tokoo et al., 1992). Eosinophils seem to play an important role in the GI tract not only in the sense of defence against parasites but also in the case of bacterial and viral infections, preventing uncontrolled bacterial invasion after damage to intestinal epithelial cells, as seen in dogs with parvovirosis (Goddard & Leisewitz, 2010;Yousefi et al., 2008).
In contrast to the most recent study in Sweden (Lilliehöök et al., 2000), where only one dog with a hookworm infestation was documented, endoparasitism was more frequently associated with eosinophilia in the present study (10.4%). Helminthiases have a reputation to produce a marked peripheral eosinophilia. The degree and duration of eosinophilia are determined by the development, migration and distribution of the parasite and the immune response of the host, being high during tissue migration and lower with intraluminal parasites or in those that produce cysts unless these become disrupted (Nutman, 2007). Other endoparasites, such as Giardia spp. and intestinal coccidial infections have been related to a lesser extent to eosinophilic tissue infiltration (Aloisio et al., 2006) and peripheral blood eosinophilia (Center et al., 1990). Some authors suggest that Giardia spp. may produce some allergen, which could reach a deeper layer of intestinal mucosa during infection, resulting in eosinophilia (Dos Santos & Vituri, 1996).
The association between eosinophilia and parasite infestation is hampered by the fact that it can take months for eosinophils to return to normal values after antiparasitic treatment (Leder & Weller, 2000).
On the other hand, eosinophilia may be present in parasitised dogs with a negative coproscopic result. Nevertheless, the prevalence of intestinal parasitosis in dogs in developed European countries reaches values between 30% and 50% for pets living in cities and metropolitan areas, respectively, being most common in dogs younger than 1 year (Barutzki & Schaper, 2011;Zanzani et al., 2014) (Clercx et al., 2000). In a recent study, eosinophilic lung disease was classified into three categories, eosinophilic bronchitis (EB), granuloma (EG) and bronchopenumopathy, based on radiography, bronchoscopy, BALF and haematologic findings; in patients with EB, peripheral eosinophilia was rare, percentages of eosinophils in BAL fluid were low, and only minimal bronchoscopic changes were present, while changes, including bronchiectasis, were more severe in the last two groups, suggesting more marked airway inflammation with consecutive tissue damage. Recognising peripheral severe eosinophilia can help in the classification and in determining the prognosis of these patients, reported as guarded in patients with EG (Johnson et al., 2019). Determining the cause of eosinophilia in these patients may be challenging since many different factors may lead to pulmonary hypersensitivity, such as exposure to certain microorganisms (parasites, fungal and bacterial infections) and chemical substances (toxic products, medications), among others. In many cases, the cause cannot be elucidated, but knowing the prevalence of some infectious diseases, such as parasite infections, is essential for diagnostic workup. This is of great importance in the case of helminthiasis, for which prevalence rates might differ significantly among countries (Giannelli et al., 2017).
In some cases, diagnosis can be easily overlooked, since larvae are rarely present in fecal or transtracheal wash samples, and antiparasitic therapy should be considered in these patients even when faecal examinations are negative (Shaw et al., 1996). Negative results are expected in the early phases of infection and remain negative for up to 8 weeks after the onset of pulmonary signs (Campos, 2009).
Similar to endoparasites, differences were found in dogs with skin disorders when compared with the study in Sweden (Lilliehöök et al., 2000), where 7% of dogs in the Swedish study had a Sarcoptes scabiei infection, and none was detected in our study. In that sense, flea allergy dermatitis was the most common cause of peripheral eosinophilia in cats (20%) in another study (Center et al., 1990), compared to only two dogs (0.63%) in the present study. Peripheral eosinophilia is a common but inconstant finding in eosinophilic dermatoses, even in patients with severe eosinophilic infiltration (Mauldin et al., 2006), and as already reported, it was related, in addition to parasitic infestations, to allergic reactions and nonspecific dermatopathies, including bacterial dermatitis and autoimmune disorders, such as pemphigus foliaceus (Center et al., 1990;Lilliehöök et al., 2000).
Hypoadrenocorticism was diagnosed in all but one dog with an endocrine disorder. Eosinophilia is one of the hallmarks of these patients, being present in 20% of cases, and hypoadrenocorticism should always be considered in chronically ill patients with eosinophilia and a concurrent sodium:potassium ratio lower than 27:1 (Peterson et al., 1996). Additional calculation of neutrophil to eosinophil as well as to lymphocyte ratio might help in the diagnosis of hypoadrenocorticism when adrenocorticotropic hormone (ACTH) stimulation test or measurement of ACTH concentrations are not available since these differ from dogs with other diseases (Zeugswetter & Schwendenwein, 2014).
Hypereosinophilia refers to marked blood eosinophilia (≥1.5 × 10 9 /L) for at least 6 months with or without tissue eosinophilia and is further classified as HES when damage to end organs is present together with no apparent aetiology or evidence of clonality (Davis & Rothenberg, 2014). This is a very rare disease in humans, in which diagnosis is usually made per exclusion. In many cases, HES has been re-classified as primary clonal eosinophilia after confirmation of cell clonality (Brito-Babapulle, 2003). In veterinary medicine, a tentative diagnosis of HES is commonly made if no other relevant causes are detected after deworming of patients with chronic hypereosinophilia.
In conclusion, as reported in humans, we found a clear difference in the distribution of causes of marked blood eosinophilia in dogs when compared to previous studies, and, surprisingly, the most common cause of marked eosinophilia in the current study was malignancy. This may be due to a higher incidence of diseases such as cancer, the age of the population under investigation, an observational bias due to the local oncologic referral centre and absence of infectious diseases such as heartworms that are more common in southern countries. However, the high number of apparently healthy young patients with eosinophilia in our study raises suspicion of undetected parasitic infestation. Therefore, parasitic infestation should be considered in puppies and young adults with unknown eosinophilia, while neoplasia should become part of differential diagnoses in older patients.

CONFLICT OF INTEREST
The authors declare no conflict of interest.

ETHICS STATEMENT
This study is a retrospective study and did not require Ethics Committee approval.

AUTHOR CONTRIBUTIONS
Design, data analysis, supervision of project, writing-original draft and review and editing: Guija-de-Arespacochaga. Data collection, data analysis, writing-original draft: Kremer. Writing-review & editing: Künzel.
Design, data analysis and processing, writing: review and editing: Schwendenwein.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the corresponding author.

PEER REVIEW
The peer review history for this article is available at https://publons.