• Open Access

Cerebrospinal Fluid Eosinophilia in Dogs


Corresponding author: Dr B.K. Sturges, DVM, MS, Department of Surgical and Radiological Sciences, University of California, Davis, CA 95616-8747; e-mail: bksturges@ucdavis.edu.


Background: Marked eosinophilic meningitis or meningoencephalomyelitis (EME) is rarely reported in dogs and the cause is usually undetermined. Long-term prognosis for dogs with cerebrospinal fluid (CSF) eosinophilia is variable.

Animals: Twenty-three client-owned dogs.

Methods: Retrospective case series. Dogs with eosinophilic CSF, defined as total nucleated cell count (TNCC) >3 cells/μL with >20% eosinophils, were identified by a computerized search of all dogs having cisternal and/or lumbar CSF analyzed as part of the diagnostic workup between 1992 and 2007.

Results: TNCC in CSF ranged from 4 to 4,740 cells/μL (median 84 cells/μL, reference range ≤3 cells/μL), with 22 to 95% (median 78%) eosinophils in the differential count. An infectious agent was identified on necropsy in 4 of 23 (17%) dogs (Cryptococcus neoformans [n = 2], Neospora caninum [n = 1], and Baylisascaris procyonis [n = 1]). Each of these dogs had progressive neurologic deterioration. Sixteen dogs had idiopathic EME. Magnetic resonance imaging (MRI) findings were abnormal in 7 of 13 dogs with EME; 2 dogs had focal lesions and 5 dogs had multifocal lesions. Clinical signs in 12 of 16 (75%) dogs with idiopathic EME resolved with prednisone treatment. Three dogs with acute intervertebral disc herniations recovered after decompressive surgery alone.

Conclusions: Idiopathic EME is a common cause of eosinophilic pleocytosis in dogs. MRI findings are variable. Infectious causes of EME were less common and had a poor prognosis.

Eosinophils are rarely identified in the cerebrospinal fluid (CSF) of dogs and other species. When the percentage of eosinophils in the CSF is only mildly increased (<5%), this finding is considered nonspecific and is associated with a variety of diseases, including infectious and noninfectious inflammatory disease as well as neoplasia, infarction, shunt placement, and spinal cord compression.1–4 It is recommended that an eosinophilia of 10% of the total white cells in the CSF should be used as a minimum criterion for the diagnosis of eosinophilic meningitis in people.2

CSF eosinophilia in animals is classically associated with migrating helminth infection in dogs, sheep, horses, llamas, goats, foxes, and primates.5–13 Diseases with a predominantly eosinophilic CNS infiltrate in the nervous tissue include Dirofilaria immitis migration in a cat, Cuterebral myiasis in a cat, Toxoplasma gondii infection in horses, amebic infection in a sheep, and Hypoderma bovis migration in a horse.14–18 Eosinophilic meningitis occurs in acute lead poisoning in calves and salt poisoning in swine and idiopathic eosinophilic meningoencephalitis is described in 2 cows and a cat.19–22

Marked CSF eosinophilia associated with an infectious etiology is rarely reported in dogs. Angiostrongylus infection is a cause of eosinophilic meningomyelitis in Australian dogs but is rarely reported in other countries.12,13 Eosinophilic CSF occurs in association with Neospora sp., Prototheca sp., and Cryptococcus sp. infection in dogs.1,12,13,23,24 Eosinophils occur in the CSF of dogs with canine distemper virus (CDV), rabies virus, bacterial encephalitis, and toxoplasmosis, although not in a high enough percentage to be considered as eosinophilic pleocytosis.1 Eosinophilic meningitis/meningoencephalitis/meningoencephalomyelitis (EME) of undetermined etiology (idiopathic EME) is reported in 11 dogs.23,25,26 The purpose of this study was to describe the presentation, CSF analysis, magnetic resonance imaging (MRI), and clinicopathologic findings in 23 dogs with eosinophilic pleocytosis.

Materials and Methods

A computerized medical record search from 1992 to 2007 was performed for dogs presenting for neurologic signs to the University of California Davis, Veterinary Medical Teaching Hospital (UCD VMTH) with eosinophilic CSF. Dogs were included if they had a CSF pleocytosis (>3 nucleated cells/μL, UCD VMTH Clinical Pathology Laboratory reference range ≤3 cells/μL), with ≥20% eosinophils in the differential count. Given that 10% CSF eosinophils is recommended as a minimum criterion for eosinophilic meningitis, we reasoned that ≥20% eosinophils in CSF was unequivocal evidence of specific recruitment of eosinophils into the CNS and constituted marked CSF eosinophilia and eosinophilic meningitis.2

Signalment, clinical history, physical, and neurologic examination findings were recorded for all dogs. Results of diagnostic tests, including CBC, serum biochemical panel, urinalysis, thoracic radiographs, and abdominal ultrasound, were recorded. The total nucleated cell count (TNCC), percentage of eosinophils in the differential count, and protein concentration for the initial CSF analysis and all subsequent CSF analyses were recorded for all dogs. Results of advanced imaging, including myelography, computed tomography (CT), and MRI, were also reviewed by the authors (RCW, BKS). All CT and MR images were acquired with settings considered optimal for imaging the CNS. Infectious disease testing on serum, CSF, or both included Cryptococcus latex agglutination Cryptococcal antigen test (n = 14; 7 CSF, 5 serum, and 2 CSF and serum), T. gondii IgG and IgM antibody (n = 10; 3 serum, 6 CSF, and 1 CSF and serum), Neospora caninum immunofluorescent antibody test (IFA) (n = 8; 4 serum, 2 CSF, and 2 CSF and serum), Coccidiodomycosis immunodiffusion (n = 8; 6 serum, 1 CSF, 1 CSF and serum), PCR panel for CDV, West Nile virus, N. caninum and hughesi, Rickettsia rickettsia, Ehrlichia canis, Borrelia burgdorferi, and T. gondii (n = 7), D. immitis antigen (n = 7), E. canis/E. equi antibody (n = 7 serum), CDV IgG and IgM antibody (n = 4; 2 CSF, 1 serum, and 1 serum and CSF), Babesia canis antibody (n = 2 serum), and Brucella canis rapid slide agglutination test (n = 2). CSF culture and fecal culture and sedimentation were performed in 2 dogs each.

Reported response to treatment and overall outcome was determined. Necropsy findings were examined if available.


Twenty-three dogs met the inclusion criteria. An infectious cause for marked CSF eosinophilia was evident in 4 dogs; Cryptococcus neoformans (2 dogs), N. caninum (1 dog), and Baylisascaris procyonis migration (1 dog). Three dogs had intervertebral disc (IVD) extrusions. Infectious disease titers were not submitted in these dogs and all 3 recovered completely with surgical decompression without concurrent steroid administration. In 16 dogs, an underlying etiology was not identified and idiopathic EME was diagnosed.


There were 15 large breed (>25 kg) dogs representing 12 breeds and 8 small breed dogs representing 7 breeds. There were 11 females (2 intact, 9 spayed) and 12 males (5 intact, 7 castrated), with a median age of 3.5 years (mean 4 years, range 8 months to 13 years).

Presenting Complaints

Nineteen of the 23 dogs presented solely for their neurologic abnormalities and had no other clinical or physical exam findings suggestive of systemic illness. Three dogs with idiopathic EME showed systemic signs, including diarrhea/fecal incontinence, vomiting/abdominal pain, and nausea/inappetance in 1 dog each. One dog with B. procyonis migration also presented for vomiting and anorexia. Neurologic abnormalities reported by the owners included generalized ataxia (n = 8), seizures (n = 8), cervical and/or spinal pain (n = 6), mentation or behavior changes (n = 5), paraparesis (n = 4), lameness (n = 2), vestibular signs (n = 2), blindness (n = 1), and anisocoria (n = 1).

Neurologic Exam Findings

All dogs presented with signs of neurologic disease. Abnormal neurologic exam findings included postural reaction deficits (n = 18), generalized ataxia (n = 14), menace response deficit (n = 8), mentation changes (n = 7), vestibular signs (n = 6), pupillary changes (miosis, mydriasis, or anisocoria, n = 5; reduced responsiveness to light, n = 2), circling (n = 4), paraparesis (n = 3), cervical pain (n = 3), reduced thoracic limb reflexes (n = 3), cerebellar signs (hypermetria, truncal sway, tremors, n = 2), and diffuse spinal pain (n = 1). The neuroanatomic localization was multifocal CNS (12 dogs), cerebrum/thalamus (5 dogs), and cerebellum/brainstem (1 dog). Five dogs had signs of myelopathy only, which localized as a C1-C5 myelopathy (1 dog), C6-T2 myelopathy (1 dog), T3-L3 myelopathy (2 dogs), and T3-S3 myelopathy (1 dog).

Diagnostic Evaluation

Diagnostics to rule out extracranial disease were performed in all dogs and included CBC (n = 23), biochemistry panel (n = 23), urinalysis (n = 22), thoracic radiographs (n = 21), abdominal ultrasound (n = 21), coagulation panel (n = 1), and pre- and postprandial bile acids (n = 1). Peripheral eosinophilia was reported in only the dog with B. procyonis migration (2,829 eosinophils/μL; reference range 0–1,500 eosinophils/μL). Serum biochemistry abnormalities included mild increases in the activity of ALT and ALP (n = 4), CK and AST (n = 2), and hypoalbuminemia and hypocholesterolemia (n = 1). Urinalysis and thoracic radiographs were unremarkable in all dogs. Abdominal ultrasound in the dog with Baylisascaris sp. migration revealed sublumbar lymphadenopathy.

CSF Analysis

CSF samples were obtained from the cerebellomedullary cistern in 17 dogs, the lumbar cistern in 3 dogs, and both in 3 dogs. The TNCC ranged from 4 to 4,740 cells/μL (median 84 cells/μL, reference range ≤3 cells/μL). The percentage of eosinophils in the CSF samples ranged from 22 to 95% (median 78%, reference range = 0%). The RBC count ranged from 0 to 3,300 cells/μL (median 40 cells/μL, reference range = 0 cells/μL). Protein concentration ranged from 15 to 803 mg/dL (median 41 mg/dL, reference range <35 mg/dL for lumbar, <25 mg/dL for cisternal).

The TNCC in 4 dogs with infectious EME ranged from 62 to 4,740 cells/μL (median 875 cells/μL), with 30–95% eosinophils. The TNCC in 16 dogs with idiopathic EME ranged from 4 to 3,880 cells/μL (median 99 cells/μL), with 22–95% eosinophils. The TNCC in 3 dogs with IVD extrusions ranged from 23 to 84 cells/μL, with 22–78% eosinophils. Five dogs had TNCC > 1,000 cells/μL, including 3 dogs with idiopathic EME, 1 dog with cryptococcosis, and 1 dog with B. procyonis migration. Seventeen dogs had >50% eosinophils. Four dogs had >90% eosinophils and included the dog with B. procyonis migration and 3 dogs with idiopathic EME.


MRI of the brain was performed in 16 dogs, including 3 dogs with infectious diseases and 13 dogs with idiopathic EME. MRI was abnormal in all dogs with infectious etiologies. MRI of the dog with B. procyonis migration indicated marked loss of gray/white matter distinction throughout the cerebral cortex, T2 hyperintensity within the associated internal capsule and corona radiata, and cerebellar herniation. MRI of the dog with neosporosis indicated T2 hyperintensity of the vermis and the left cerebellar hemisphere and subtle regions of T1 postcontrast enhancement in the left cerebellum. MRI of 1 dog with cryptococcosis indicated diffuse T1 postcontrast meningeal enhancement and multiple focal areas of ill-defined T1 postcontrast enhancement throughout the brain parenchyma.

MRI was unremarkable in 5 dogs with idiopathic EME and revealed an extra-axial caudal cerebellar cystic lesion and bilateral hydrocephalus in 1 dog, which was deemed incidental. In the remaining 7 dogs with idiopathic EME, MRI abnormalities were focal in 2 dogs and multifocal in 5 dogs. Three dogs showed patchy regions of T2 hyperintensity and contrast enhancement in the cerebral cortex (Fig 1). Two mass lesions were noted in 1 dog and solitary mass lesions were identified in the frontal sinus (biopsied as resolving hematoma) and thalamus in 1 dog each. One dog had extensive T1 postcontrast meningeal enhancement, focal T2 hyperintensity in the cerebellopontine angle, and enlarged T1 postcontrast enhancing V, VII, and VIII cranial nerves (Fig 2a and b). One dog had diffuse T1 postcontrast meningeal enhancement throughout the brain, with a focal T2 hyperintensity centrally within the spinal cord at the level of C3.

Figure 1.

 T2 weighted transverse MR image of an adult Labrador Retriever with idiopathic eosinophilic meningoencephalitis indicating patchy regions of hyperintensity in the cerebral cortex and in the thalamus. Clinical signs in this dog resolved with prednisone.

Figure 2.

 (a and b): Thirteen-year-old German Shepherd dog cross. Comparison of precontrast T1 weighted (W) (a) and postcontrast T1W (b) transverse MR images reveals bilateral contrast enhancement of cranial nerve V and meningeal enhancement (more severe on the right). The dog improved clinically with a tapering dose of prednisone and clinical signs did not recur (10-month follow-up).

Other Diagnostic Imaging

Myelograms performed in 3 dogs presenting for paraparesis were consistent with IVD extrusion. One of these dogs had been surgically treated previously for an intracranial intra-arachnoid cyst before presenting for an acute IVD extrusion at T13-L1, and a spinal arachnoid cyst at C2-3 was also identified in this dog. Surgical treatment for the intra-arachnoid cyst in that dog has been previously reported.27 Myelogram/CT performed in 1 dog with idiopathic EME presenting for severe cervical pain revealed a concurrent caudal cervical spondylomyelopathy.

Infectious Disease Titers

Titers to various infectious diseases were submitted in 19 dogs. In 3 dogs with IVD extrusion and 1 dog with B. procyonis infection, infectious disease titers were not submitted. In 2 dogs, crypotococcosis was diagnosed via LCAT. In another dog, N. caninum was diagnosed via serum (1 : 20,480) and CSF (1 : 640) IFA titers. Infectious disease testing was negative in the other 16 dogs.

Treatment and Outcome

All dogs with infectious EME died or were euthanized. The dog with B. procyonis migration was euthanized 1 day after diagnosis because of rapid neurologic deterioration. Necropsy revealed severe multifocal subacute eosinophilic, histiocytic, lymphoplasmacytic perivascular meningoencephalitis, and vasculitis in the cerebrocortex, brainstem, and spinal cord. Focal nematode granulomas were identified in the lungs and in the stomach.

The dog with neosporosis initially improved after treatment with trimethoprim-sulfonamide (15 mg/kg PO q12h), pyrimethamine (0.78 mg/kg PO q12h), ponazuril (5 mg/kg PO q24h), and prednisone (0.5 mg/kg PO q12h), but was euthanized 5 months later because of recurrence of signs. Necropsy revealed severe locally extensive necrotizing granulomatous encephalitis with intralesional protozoa consistent with Neospora spp.

Both dogs with cryptococcosis declined neurologically despite treatment. One dog was treated with amphotericin (0.5 mg/kg IV q72h), flucytosine (50 mg/kg PO q8h), fluconazole (5 mg/kg PO q12h), and prednisone (0.33 mg/kg PO q24h) before euthanasia 9 days after diagnosis. The other dog was treated with dexamethasone sodium phosphate (0.1 mg/kg IV once) and fluconazole (8 mg/kg PO q12h) before euthanasia 5 days after diagnosis. Necropsy revealed disseminated cryptococcosis in both dogs. Severe diffuse histiocytic and lymphoplasmacytic inflammation with cryptococcal organisms was identified in the meninges, brain, and caudal nasal turbinates of 1 dog and severe multifocal granulomatous meningoencephalomyelitis (GME) with intralesional yeast was identified in the cerebrum, brain stem, cerebellum, and spinal cord of the other dog.

All dogs with IVD extrusions recovered after surgical decompression. The CSF was not rechecked in these dogs because of their complete recovery.

Clinical signs resolved in 12 of 16 (75%) dogs with idiopathic EME. Two dogs recovered without treatment and no further CSF analyses were performed. In 1 dog presenting for seizures, T2 hyperintensity was noted in the rostral part of the frontal sinus and T1 postcontrast enhancement was noted in the adjacent meninges lining the left forebrain and, to a lesser degree, the temporal/parietal region. CT revealed bony production and destruction of the rostral aspect of the inner table of the frontal bone consistent with osteomyelitis. A transfrontal craniotomy was performed to biopsy the frontal sinus mass, which was diagnosed as a resolving hematoma. That dog was treated postoperatively with prednisone (0.43 mg/kg PO q12h), amoxicillin-clavulanic acid (14 mg/kg PO q12h), and metronidazole (7 mg/kg PO q12h). Clinical signs resolved within 2 days. Recheck CSF analysis 2 weeks after diagnosis had TNCC = 4 cells/μL, with 0% eosinophils. Prednisone was tapered over 3 months and there was no recurrence of clinical signs at 2-year follow-up.

One dog, presenting for an acute onset of marked cervical pain, was treated with prednisone (0.3 mg/kg PO q12h), which was tapered and discontinued after 10 months of treatment. This dog also had a short-strided thoracic limb gait and ataxic pelvic limb gait, suggestive of a caudal cervical myelopathy and evidence of cervical stenotic myelopathy on myelogram/CT. Although the gait never became normal with prednisone treatment, the cervical pain completely resolved. Recheck CSF analysis 1 month after prednisone was discontinued, indicating recurrence of a marked eosinophilic pleocytosis. At that time the dog showed no evidence of cervical pain but had a persistent mild gait abnormality. The dog was treated with another tapering course of prednisone 0.48 mg/kg PO q12h and cervical pain did not recur (2-year follow-up). Because the owners' primary concern had been the cervical pain, they opted not to pursue surgery for the abnormal gait and declined repeat CSF analyses.

In 8 dogs with idiopathic EME, clinical signs resolved with prednisone treatment ranging from 0.33 to 1 mg/kg PO q12h (median 0.51 mg/kg, mean 0.54 mg/kg). All dogs were eventually tapered off prednisone without recurrence of signs after a median of 105 days (mean 109 days, range 60–165 days). Recheck CSF analysis was permitted in 5 dogs with idiopathic EME and all showed complete resolution of the eosinophilia at the first recheck (range 2 weeks to 3 months). The TNCC returned to normal at first recheck in 4 dogs and at second recheck in 1 dog.

Four dogs with idiopathic EME died or were euthanized because of poor response to corticosteroid therapy. One dog treated with prednisone (0.54 mg/kg PO q12h) died at home 2 days after diagnosis. Necropsy performed in 1 dog, euthanized 3 days after diagnosis and treatment with dexamethasone sodium phosphate (0.25 mg/kg IV q12h), indicated severe multifocal histiocytic meningoencephalitis with multifocal necrosis and malacia in the cerebrocortex and brain stem. Another dog improved with prednisone (0.67 mg/kg PO q12h) and clindamycin (5 mg/kg PO q8h) but began seizuring 6 weeks after diagnosis and was euthanized 3 months after initiating treatment. The last dog improved with prednisone (0.9 mg/kg PO q12h) for 3–4 months but was euthanized when clinical signs recurred after prednisone was discontinued (5.5 months after diagnosis).

There was no correlation between MRI findings and clinical outcome. MRI was performed in 3 of 4 dogs with idiopathic EME that did not respond to treatment and was normal in 1 dog (necropsy unavailable), indicated multifocal contrast enhancement in 1 dog (where necropsy demonstrated multifocal meningoencephalitis and necrosis), and indicated 2 extra-axial lesions in 1 dog (necropsy unavailable). MRI was unremarkable in the 2 dogs that recovered without treatment (recheck CSF analyses not performed) and in the 2 dogs that responded to prednisone. One of these dogs had cerebellar/vestibular signs, which resolved with prednisone (1 mg/kg PO q12h) for 30 days and then gradually tapered over 4 months (recheck CSF was normal 3 and 6 weeks after initiating treatment, follow-up 4 months after discontinuing prednisone). The other dog, presenting for seizures and mentation change, became clinically normal and had no further seizures after treatment with prednisone (0.5 mg/kg PO q12h) tapering over 3 months and phenobarbital (3 mg/kg PO BID) tapered after 6 months. Recheck CSF 1 month after initiating treatment was normal (follow-up time 9 months after discontinuing prednisone).


Eosinophilia is an uncommon finding in the CSF of dogs with neurologic disease. It can be associated with meningitis, encephalitis, and myelitis and occurs mainly in young to middle-aged large breed dogs. MRI findings in dogs with idiopathic EME vary considerably and focal mass lesions and cranial nerve involvement may be identified. Neurologic signs resolve in the majority of dogs with idiopathic EME most likely attributable to anti-inflammatory or immunosuppressive prednisone therapy.

Of the 23 dogs with marked CSF eosinophilia in this study, an idiopathic cause was diagnosed most frequently (13/23) with infectious and noninfectious inflammatory causes occurring less commonly. This is also the first report of intracranial nematode migration causing eosinophilic pleocytosis in the CSF. This probably reflects the rarity of the disease because B. procyonis infection is characterized by large deposits of eosinophils along necrotic migration tracks and cerebral blood vessels.28 Although there is no canine immunologic assay currently available to detect antibodies to B. procyonis, as is the mainstay of diagnosis B. procyonis in humans, fecal analyses including flotation and sedimentation techniques should be performed in all dogs with EME to look for parasites with migratory potential.

Variable CSF analyses have been reported in dogs with neosporosis and cryptococcosis, although most have a mixed, histiocytic, or neutrophilic pleocytosis.29–40 CSF eosinophilia is rarely reported in dogs with Neospora, with eosinophils typically constituting <5% of the TNCC.36,37,41 In both dogs with cryptococcosis and the dog with neosporosis in this report, few eosinophils were identified in CNS tissues collected at necropsy despite the high percentage of eosinophils in the CSF. Eosinophilic inflammation may have initially localized to the meninges in these dogs whereas granulomatous or histiocytic infiltrates predominated in the brain parenchyma.

In a majority of the dogs with eosinophilic CSF in this report, an underlying infectious agent was not identified. Idiopathic EME has been reported previously in 11 dogs; however, MRI findings were available in only 1 dog.23–26 MRI findings in the 13 dogs with idiopathic EME in this report were variable. Focal mass lesions were reported in 4 dogs, suggesting that EME, similar to GME, may present focally in some dogs. Three dogs with mass lesions responded to treatment with prednisone and had no recurrence of clinical signs when prednisone was discontinued, with a follow-up time of 2–3 years, making neoplasia unlikely. One dog with masses in the olfactory bulb and pyriform lobe died at home and necropsy was not performed. Prominence of the sulci, suggestive of cortical thinning or atrophy, was present in 1 dog, which is similar to the MRI findings in another report of a dog with idiopathic EME.26 Patchy regions of T2 hyperintensity and/or T1 postcontrast enhancement were seen in several dogs, indicating more diffuse parenchymal disease. Cranial nerve V, VII, and VIII involvement was seen clinically and on MRI in 1 dog, which has not previously been reported with EME. An underlying neoplastic process such as lymphoma infiltrating the cranial nerves was considered, but the lack of recurrence of clinical signs after corticosteroid treatment was tapered would be unusual for lymphoma. Diffuse meningeal enhancement was identified in 1 dog with an extremely high TNCC (3,880 cells/μL), which may have reflected severe meningitis. MRI indicated no intracranial abnormalities in 4 dogs, including 2 dogs with a marked CSF pleocytosis. Small or very diffuse meningeal and/or intraparenchymal lesions may be difficult to detect on MRI.

Dogs with idiopathic EME had a much better long-term prognosis than dogs with an infectious etiology. Only 4 (25%) dogs with idiopathic EME died or were euthanized. This compares to previous reports where 6 of 11 (55%) dogs with idiopathic EME died or were euthanized.23,25,26 Of those 6 previously reported dogs, 3 were euthanized without treatment and 3 died despite treatment with corticosteroids. In the 5 previously reported cases where clinical signs resolved, 4 dogs received corticosteroids.23,25 Necropsy findings in 5 previously reported cases of idiopathic EME include meningeal discoloration (n=3), spongiosis (n=2), demyelination (n=2), gliosis and astrocytosis (n=2), malacia (n=2), cortical atrophy (2), neuronal necrosis (n=1), hemorrhage (n = 1), and opaque exudate within sulci (n=1).23,25,26 The wide variety of pathologic findings described in these dogs may reflect pathologic processes of different etiologies or temporally associated factors in a continuum of the same disease.

In 3 dogs in this report, including 1 dog with 78% eosinophils in the CSF, eosinophilic CSF was associated with acute IVD extrusion. Although infectious disease testing was not performed in these dogs, all recovered without immunomodulatory treatment. It is possible that eosinophilic CSF in these dogs may have reflected an allergic or hypersensitivity reaction to IVD material or, less likely, that they had concurrent idiopathic EME that resolved without corticosteroid therapy.

It is unclear why idiopathic EME is diagnosed in a majority of the dogs with marked CSF eosinophilia whereas an infectious agent is more commonly identified in other species. It is possible that infectious causes of canine EME are not properly diagnosed. Fecal examination is the primary diagnostic tool for identifying intestinal parasites with migratory potential in dogs. If fecal floatation had been performed in all dogs in this study, it is possible that other parasitic organisms would have been identified. However, in humans, parasitic infections including Toxocara and Gnathostomiasis are commonly diagnosed with serology rather than with fecal floatation because ova from the parasites are often not identified in the feces.42–44 Therefore, it is possible that even with fecal analysis, parasitic causes of EME in dogs go undiagnosed without more thorough serologic testing. Eosinophilic meningoencephalitis has been reported in association with viral infection in humans and it is possible that some dogs with EME suffer from a transient undiagnosed viral infection.45 Two dogs in this report recovered without treatment, which may support a self-limiting viral etiology. Some cases of idiopathic EME may represent another form of pathogen-free steroid responsive inflammatory CNS disease, which is common in dogs but rare in other species. The often dramatic response to corticosteroids in many dogs with idiopathic EME supports an inflammatory process that may or may not be associated with (or preceded by) underlying allergy or infection. However, corticosteroids can also reduce blood and tissue eosinophil numbers directly via many mechanisms, including induction of apoptosis.46

Similar to other reports, a majority of the dogs in this study were young to middle aged. However, idiopathic EME was diagnosed in 2 dogs ≥10 years of age. All 4 dogs with infectious EME and a majority of the dogs with idiopathic EME (69%) were large breed dogs. Unlike GME and necrotizing encephalitis, which occur more commonly in small breed dogs and typically require life-long immunosuppressive therapy, many cases of EME in this report occurred in large breed dogs and responded to a tapering dose of prednisone, following a clinical picture more similar to steroid responsive meningitis.

Peripheral eosinophilia was identified in only 1 dog in this report (dog with B. procyonis migration). In 8 of 13 (62%) previously reported cases of EME, concurrent blood eosinophilia was noted, although there was no correlation between the magnitude of blood and CSF eosinophilia.23–26 Similarly, there is often no correlation between eosinophilia in the peripheral blood and the CSF in humans with EME.3,47

Necropsy findings in dogs with idiopathic EME are variable in distribution and severity and may suggest different underlying etiologies or the effects of temporally associated factors in a continuum of the same disease. Prognosis for infectious EME is poor and, as such, thorough infectious disease evaluation including fecal analysis and serum and CSF immunology is required to rule out known infectious causes. Pathologic examination of all euthanized dogs that fail to respond to treatment will be crucial in better characterizing the underlying etiology of idiopathic EME.