Neuroinflammatory diseases of the central nervous system of dogs: A retrospective study of 207 cases (2008–2019)

Abstract

In this study we describe 207 cases of neuroinflammatory diseases of the central nervous system (CNS) in dogs autopsied at the Athens Veterinary Diagnostic Laboratory (University of Georgia, United States) from 2008 to 2019. Idiopathic and infectious diseases were diagnosed in 111 cases (53.6%) and 96 cases (46.4%), respectively. Idiopathic diseases consisted of granulomatous meningoencephalomyelitis (n = 42; 37.8% of idiopathic cases), nonspecific lymphoplasmacytic meningoencephalomyelitis (n = 39; 35.1%), necrotizing meningoencephalomyelitis (n = 22; 19.8%), presumed steroid-responsive meningitis-arteritis (n = 6; 5.4%), and necrotizing leukoencephalitis (n = 2; 1.8%). Infectious diseases consisted of bacterial infections (n = 49; 51% of infectious cases), viral infections (n = 39; 40.6%), fungal infections (n = 5; 5.2%), and parasitic infections (n = 3; 3.1%). Our study provides an overview of the most frequent neuroinflammatory diseases of the CNS of dogs in our diagnostic routine; indicates that a comprehensive diagnostic approach, including a thorough evaluation of the pathology findings and ancillary laboratory testing results, is important for an adequate diagnosis of neurologic diseases in dogs; and underscores the problems associated with the variability in tissue sample collection methods among cases. The great number of nonspecific lymphoplasmacytic meningoencephalitis also highlights the need for development of molecular laboratory tests to identify potential infectious agents in these cases.

Introduction

According to the 2019–2020 American Pet Products Association’s National Pet Owners Survey (1), nearly 63 million households in the United States own a dog. The closely shared environment, as well as the genomic similarities and the parallel biologic behavior of many naturally occurring cancers in pet dogs and humans have led to important advances in comparative oncology, strengthening the dog as a candidate animal model for a wide variety of oncology studies (2). Similarly, many infectious diseases that affect dogs can be transmitted to humans (3), with a special interest for those that affect the central nervous system (CNS) (3,4). These disorders typically cause inflammation of the nervous tissue with progressive clinical signs that culminate in spontaneous death or euthanasia of affected dogs, especially in cases in which rabies is suspected (3,4).

A search in our web-based laboratory data system revealed that 17 291 dogs were admitted to the University of Georgia Veterinary Teaching Hospital in 2019. Of those, 1809 (10.5%) had neurologic signs and were seen by the neurology service. It is important, therefore, that veterinary practitioners, neurologists, and pathologists are aware of the most common neurologic diseases affecting dogs and are able to prioritize and rule in or out various diseases in each case. However, data focusing on the frequency of neuroinflammatory diseases of the CNS in dogs are scarce in the veterinary literature (4–8). Here we describe 207 cases of neuroinflammatory diseases of the CNS in dogs submitted for autopsy at the Athens Veterinary Diagnostic Laboratory (AVDL), University of Georgia, Georgia, United States, from 2008 to 2019.

Materials and methods

We searched the electronic web-based archive of the AVDL for cases of meningoencephalitis and meningomyelitis in dogs between 2008 and 2019 using the following search terms: meningitis, encephalitis, meningoencephalitis, myelitis, meningomyelitis, meningoencephalomyelitis, and vasculitis. Submission forms and autopsy reports from retrieved cases were reviewed, and cases were included in the study if a final diagnosis related to an inflammatory lesion in the brain or spinal cord was documented. Archived hematoxylin and eosin (H&E)–stained slides were reviewed, and neuroinflammatory tissue changes were categorized as non-infectious (idiopathic) or infectious, according to the original diagnoses and the fulfillment of a minimum set of diagnostic criteria for each disease based on previous publications (Supplementary Table 1). Only cases in which representative areas of the brain (cerebral cortex and subcortical white matter, basal nuclei, thalamus, brainstem, and cerebellum) and affected spinal cord were available were examined. The signalment, clinical signs, pathologic changes, and test methods were reviewed for all cases. All ancillary tests were conducted by the original pathologist reviewing each case and included bacterial or fungal culture and fluorescent antibody test (FAT) that were performed on fresh or frozen samples following autopsy, as well as special stains, immunohistochemistry (IHC), and polymerase chain reaction (PCR).

Table 1.

Gross and histologic lesions in the brain and spinal cord of 207 cases of neuroinflammatory diseases of dogs.

Disease category	Total cases	Gross neuropathology	Neurohistopathology
Non-infectious (idiopathic)	111
Granulomatous meningoencephalomyelitis	42	Random, pinpoint, dark red (n = 5) or translucent (n = 3) foci in the white matter (mainly midbrain and brainstem); cerebral swelling (n = 3); cerebellar herniation (n = 2); pale yellow nodule in the left cerebellar peduncles, medulla oblongata, and cervical spinal cord (n = 1).	Perivascular and neuroparenchymal clusters of epithelioid macrophages with lymphocytes and plasma cells associated with microgliosis, astrocytosis, necrosis, and hemorrhage in the gray and white matter (mainly midbrain, brainstem, cerebellum, and spinal cord).
Nonspecific lymphoplasmacytic meningoencephalomyelitis	39	Cerebral swelling (n = 2); cerebellar herniation (n = 1).	Lymphocytic and plasmacytic perivascular infiltration of the leptomeninges and less commonly neuroparenchyma with microglial nodules and astrocytosis in the gray and white matter.
Necrotizing meningoencephalomyelitis	22	Cerebral swelling (n = 4); cerebellar herniation (n = 3); neuroparenchymal dark red foci with collapse of the telencephalon (n = 2); secondary hydrocephalus (n = 2).	Extensive necrosis of the telencephalic cortex and subcortical white matter, infiltration by foamy macrophages, microgliosis, astrocytosis, and lymphoplasmacytic perivascular inflammation.
Presumed steroid-responsive meningitis-arteritis	6	Brain stem swelling (n = 2); leptomeningeal hemorrhage (n = 2); right temporal/parietal hematoma (n = 1).	Neutrophilic arteritis with fibrinoid change predominantly in the leptomeninges but also extending into the cortical gray matter, fibrin and hemorrhage around affected blood vessels, neuronal necrosis.
Necrotizing leukoencephalitis	2	None reported.	Extensive necrosis of the white matter around lateral ventricles with adjacent axonal spheroids, microgliosis, astrocytosis, and lymphocytic perivascular inflammation.
Infectious	96
Bacterial	49	Neuroparenchymal necrosis and hemorrhage (n = 20); leptomeningeal hemorrhages (n = 7); suppurative exudate (n = 5); cerebral swelling (n = 3); cerebellar herniation (n = 2); secondary hydrocephalus (n = 2).	Perivascular accumulation of fibrin, hemorrhage, and neutrophils in the leptomeninges and cortical gray matter and less often white matter; neutrophilic vasculitis with fibrinoid change, vascular thrombosis, neuronal and neuroparenchymal necrosis with foamy macrophages, microgliosis, astrocytosis, bacterial clusters.
Viral	39
Canine herpesvirus-1	24	None reported.	Minimal to mild perivascular lymphoplasmacytic inflammation in the leptomeninges and cortical gray matter with vasculitis and hemorrhage, microglial nodules, neuronal necrosis, and rare intranuclear viral inclusions in endothelial cells.
Canine distemper	14	None reported.	Perivascular lymphoplasmacytic inflammation in the neuroparenchyma (mainly white matter) and leptomeninges, white matter vacuolation (mainly brainstem and cerebellum), neuroaxonal necrosis, astrogliosis and astrocytosis, and neuronal or astrocytic nuclear and/or cytoplasmic viral inclusions.
Rabies	1	None reported.	Perivascular lymphoplasmacytic inflammation in the gray and less often white matter and leptomeninges, neuronal necrosis, and intracytoplasmic viral inclusions in Purkinje neurons.
Fungal	5
Cryptococcosis	2	Cloudy leptomeninges (n = 2).	Perivascular lymphoplasmacytic inflammation in the leptomeninges; fungal yeasts.
Phaeohyphomycosis	2	Cerebral swelling (n = 2).	Necrogranulomas in the gray and white matter with intralesional pigmented fungal hyphae, microgliosis, astrocytosis, and perivascular lymphoplasmacytic inflammation.
Candidiasis	1	None reported.	Extensive perivascular fibrin and neutrophilic accumulation in the leptomeninges (mainly near the pons).
Parasitic	3
Neosporosis	1	Right lateral hydrocephalus (n = 1).	Random gray matter necrogranulomas with microgliosis and tachyzoites.
Toxoplasmosis	1	Cerebral swelling (n = 1).	Random gray matter necrogranulomas with microgliosis and tachyzoites.
Dirofilariasis	1	None reported.	Perivascular eosinophilic and lymphoplasmacytic inflammation in the leptomeninges.

Results

During the studied period, 4311 dogs were autopsied at the AVDL. Of these, 207 (4.8%) met the inclusion criteria and were included in our study (Table 1). Non-infectious, idiopathic diseases (Supplementary Table 2) were the most common diagnostic category with 111 cases (53.6%). Infectious diseases (Supplementary Table 3) comprised 96 cases (46.4%).

Granulomatous meningoencephalomyelitis (GME) was the most common idiopathic disease diagnosed during the studied period (42/111 cases; 37.8% of non-infectious cases), followed by nonspecific lymphoplasmacytic meningoencephalomyelitis (NLM) (39/111 cases; 35.1%), necrotizing meningoencephalomyelitis (NME) (22/111 cases; 19.8%), presumed steroid-responsive meningitis-arteritis (SRMA) (6/111 cases; 5.4%), and necrotizing leukoencephalitis (NLE) (2/111 cases; 1.8%).

The age of dogs affected by GME varied from 2 mo to 15 y (mean: 5.1 y). Females were affected in 30 cases (71.4%) and males in 12 cases (28.6%). The most affected breeds included mixed breed dogs (6 cases), Chihuahua (4 cases), Shi Tzu (4 cases), Dachshund (3 cases), and Maltese (3 cases). Gross pathology changes (Figure 1 a) were reported in 13 cases and consisted of random areas of hemorrhage (5 cases) or softening (3 cases) in the brainstem, cerebral swelling characterized by flattening of gyri (3 cases), and cerebellar herniation through the foramen magnum (2 cases). Histologic changes consisted of clusters of epithelioid macrophages with lymphocytes and plasma cells that expanded the neuroparenchymal and leptomeningeal perivascular spaces (Figure 1 b). Similar inflammatory infiltrates were also observed within the neuroparenchyma and were often associated with extensive areas of gliosis and necrosis (16 cases; Figure 1 c). Changes were more pronounced in the midbrain, brainstem, and cerebellum. The spinal cord had similar lesions in 13 of 14 cases examined.

Figure 1.

a — Granulomatous meningoencephalomyelitis (Case 24). Cross-section of the brain at the level of the cerebellum and cerebellar peduncles. There are multiple dark red areas of hemorrhage throughout the cerebellar peduncles and cerebellar white and gray matter.

b — Granulomatous meningoencephalomyelitis (Case 24). Numerous inflammatory cells expand the perivascular spaces in the leptomeninges and superficial telencephalic cortex. Hematoxylin and eosin (H&E); bar = 1 mm.

c — Granulomatous meningoencephalomyelitis (Case 24). Perivascular clusters of epithelioid macrophages and fewer lymphocytes and plasma cells admixed with areas of hemorrhage invade the adjacent neuroparenchyma. H&E; bar = 300 μm.

d — Nonspecific lymphoplasmacytic meningoencephalomyelitis (Case 60). Perivascular spaces are expanded by numerous lymphocytes and plasma cells (center) with adjacent areas of microgliosis (bottom). H&E; bar = 300 μm.

e — Necrotizing meningoencephalomyelitis (Case 94). Cross-section of the brain at the level of the hippocampus and thalamus. There is unilateral collapse of the right parietal telencephalon with yellow subcortical areas of necrosis (asterisk) and secondary hydrocephalus.

f — Necrotizing meningoencephalomyelitis (Case 94). Extensive areas of necrosis form a neuroparenchymal cleft within the telencephalic cortex (asterisks). H&E; bar = 1 mm.

g — Necrotizing meningoencephalomyelitis (Case 94). Necrotic areas in the telencephalic cortex are characterized by vacuolated neuroparenchyma with foamy macrophages and swollen endothelial cells. H&E; bar = 300 μm.

h — Steroid responsive meningitis-arteritis (Case 106). Two superficial cortical venules have undergone fibrinoid change and are surrounded by hemorrhage. H&E; bar = 300 μm.

i — Necrotizing leukoencephalitis (Case 110). Periventricular area of necrosis characterized by widespread infiltration with foamy macrophages that occasionally cluster around blood vessels (asterisk). H&E; bar = 50 μm.

The ages of 39 dogs affected by NLM varied from 1 mo to 15 y (mean: 4 y). Males were affected in 20 cases (51.3%) and females in 19 cases (48.7%). Gross pathology changes were observed in 3 cases and consisted of cerebral swelling (2 cases) and cerebellar herniation (1 case). Histologically, random perivascular infiltrates of lymphocytes and plasma cells occurred mainly in the leptomeninges but also in the neuroparenchyma (Figure 1 d); areas of astrocytosis, microglial nodules (4 cases), and rare neuronal necrosis (1 case) were also reported. The spinal cord had similar changes in 3 of 5 cases examined.

The age of the 22 dogs affected by NME varied from 6 mo to 12 y (mean: 4.3 y) and the disease occurred in 13 females (59%) and 9 males (41%). Pug dogs were the breed most afflicted with NME (11 cases or 50%). Gross lesions were reported in 10 cases and were characterized by cerebral swelling (4 cases), cerebellar herniation (3 cases), secondary hydrocephalus (2 cases), and telencephalic collapse (2 cases) with subcortical yellow areas of necrosis or dark red areas of hemorrhage (Figure 1 e). Histologic findings consisted of extensive areas of necrosis (Figure 1 f) with foamy macrophages (Figure 1 g) that frequently affected the telencephalic cortex and the subcortical white matter. These lesions expanded the adjacent neuroparenchyma, which exhibited extensive areas of rarefaction, astrogliosis, neuronal necrosis, and perivascular accumulations of lymphocytes and plasma cells. The spinal cord was not examined in these cases.

The mean age of 6 dogs affected by presumed SRMA was 3.6 y, and the disease affected 4 females and 2 males. Gross changes in 4 cases were characterized by swelling of the brainstem (2 cases), leptomeningeal hemorrhage (2 cases), and a telencephalic hematoma (1 case). Histologically, perivascular spaces and vascular walls throughout the brain were expanded with neutrophils and fewer lymphocytes and plasma cells. Affected vessel walls had often undergone fibrinoid change, with perivascular exudation of fibrin and hemorrhage (Figure 1 h). The adjacent neuroparenchyma exhibited areas of edema and neuronal necrosis. The spinal cord was not examined. Two cases had similar systemic vascular lesions affecting the heart (1 case) and the liver, kidney, lung, mesentery, and urinary bladder (1 case).

Necrotizing leukoencephalitis occurred in a 1-month-old and a 1-year-old male dog. No gross changes were reported. Histologically there were areas of necrosis with foamy macrophages within the telencephalic white matter, predominantly around the lateral ventricles (Figure 1 i). Neuroparenchymal vacuolation with astrogliosis and microgliosis, neuronal necrosis, axonal spheroids, and perivascular accumulations of lymphocytes and plasma cells were observed in the adjacent areas. The spinal cord was not examined.

Bacterial infections were the most common category within the infectious disease group (49/96 cases; 51% of the infectious cases). The age of affected dogs varied from 3 d to 16 y (mean: 4.6 y). Males were affected in 32 cases (65.3%) and females in 17 cases (34.7%). Lesions in the CNS were associated with systemic bacterial infection in 40 cases (81.6%), were primary in 5 cases (10.2%), and were associated with adjacent lesions (rhinitis, sinusitis, and inner ear infection) in 4 cases (8.2%). Gross changes were reported in 30 cases and consisted mainly of random areas of neuroparenchymal hemorrhage and softening (20 cases; Figure 2 a), leptomeningeal hemorrhage (7 cases), suppurative exudate (5 cases), cerebral swelling (3 cases), cerebellar herniation (2 cases), and secondary hydrocephalus (2 cases). Histologically, there were widespread perivascular accumulations of fibrin and hemorrhage with numerous neutrophils within the leptomeninges and neuroparenchyma (Figure 2 b). The wall of affected blood vessels was often infiltrated by neutrophils or had undergone fibrinoid change (Figure 2 c). Other vessels were occluded by fibrin thrombi and surrounded by hemorrhage. The associated neuroparenchyma often exhibited areas of neuronal necrosis with astrogliosis and microgliosis or areas of necrosis with foamy macrophages. Lesions in the spinal cord were reported in 4 cases. Intralesional bacteria were occasionally reported in the brain (5 cases) or extraneural tissues (18 cases), including left atrioventricular valve (10 cases), lung (5 cases), spleen (2 cases), and myocardium, aortic valve, kidney, and liver (1 case each).

Figure 2.

a — Bacterial meningoencephalitis (Case 147). Extensive leptomeningeal hemorrhage in the telencephalic leptomeninges.

b — Bacterial meningoencephalitis (Case 123). Leptomeningeal vessels are surrounded and infiltrated by a great number of neutrophils and fibrin that also extend into the superficial telencephalic cortex (asterisk). H&E; bar = 1 mm.

c — Bacterial meningoencephalitis (Case 123). Clusters of neutrophils and hemorrhage surround a cortical telencephalic venule (center). H&E; bar = 300 μm.

d — Herpesvirus meningoencephalitis, dog 171. Inflammatory cells expand a leptomeningeal vessel wall in the telencephalon. Rare endothelial cells contain intranuclear viral inclusions (arrow). H&E; bar = 100 μm.

e — Herpesvirus meningoencephalitis (Case 172). A microglial nodule surrounding an area of necrosis in the telencephalic cortex. H&E; bar = 100 μm.

f — Canine distemper (Case 190). The cerebellar white matter is vacuolated and a gemistocyte (arrow) is present. An astrocyte contains an intranuclear viral inclusion (arrowhead). H&E; bar = 300 μm.

g — Rabies (Case 199). An intracytoplasmic viral inclusion (arrowhead) in the cytoplasm of a Purkinje neuron. H&E; bar = 100 μm.

h — Cryptococcosis (Case 203). Leptomeningeal perivascular spaces are surrounded by fungal yeasts morphologically consistent with Cryptococcus spp. and a small number of lymphocytes. H&E; bar = 100 μm.

i — Phaeohyphomycosis, Case 200. Neuroparenchymal necrogranulomas with intralesional pigmented fungal hyphae. H&E; bar = 300 μm.

Bacterial culture on different sets of fresh or frozen tissues was completed in 39 cases, and polymerase chain reaction (PCR) was performed in 2 cases. Frequent isolates included Escherichia coli (8 cases), Streptococcus canis (6 cases), Staphylococcus intermedius (3 cases), Staphylococcus pseudintermedius (3 cases), Pseudomonas aeruginosa (2 cases), and Klebsiella pneumoniae (2 cases). Viral infections comprised 39 cases (41% of the infectious cases) and consisted of canine herpesvirus 1 (24 cases), canine distemper virus (14 cases), and rabies virus (1 case) infection. Most dogs with viral diseases (37 cases) were less than 1 y old (mean: 1 mo). Males were affected in 28 cases (71.8%) and females in 11 cases (28.2%). No specific gross changes were reported in the brain.

Histologically, canine herpesviral infection was characterized by subtle perivascular lymphoplasmacytic inflammation in the leptomeninges (Figure 2 d) and neuroparenchyma with occasional vasculitis and hemorrhage, glial nodules associated with neuronal necrosis, and rare intranuclear viral inclusions (Figure 2 e) in endothelial cells in the brain (5 cases) and extraneural tissues (21 cases). Canine herpesviral infection was confirmed by fluorescent antibody testing (FAT) in 22 cases.

Canine distemper was characterized by perivascular lymphoplasmacytic inflammation within the neuroparenchyma and leptomeninges, with extensive white matter vacuolation that was more pronounced in the brainstem and cerebellum. Affected areas had astrogliosis with neuroaxonal necrosis and neuronal or astrocytic nuclear or cytoplasmic viral inclusions (10 cases; Figure 2 f). Canine distemper viral inclusions were also detected in extraneural tissues (8 cases). Canine distemper infection was confirmed by FAT (8 cases), IHC (6 cases), or PCR (4 cases).

Rabies was characterized by a low number of perivascular lymphocytes and plasma cells within leptomeninges and neuroparenchyma with occasional neuronal necrosis and intracytoplasmic viral inclusions (Figure 2 g). Rabies virus infection was confirmed by FAT and IHC.

Fungal infections were diagnosed in 5 cases and occurred in 4 males and 1 female (mean: 2.2 y). Cryptococcosis was characterized by clusters of fungal yeasts and small perivascular lymphocytes within the leptomeninges (Figure 2 h). Phaeohyphomycosis was associated with multiple necrogranulomas with intralesional pigmented fungal hyphae within the neuroparenchyma (Figure 2 i). Candidiasis was characterized by fibrinous leptomeningitis with numerous neutrophils, macrophages, and fewer lymphocytes and plasma cells. The lesion was an extension from a primary infection in the tympanic bulla. Fungal culture yielded positive results in 2 cases of fungal infection.

Parasitic infections were identified in only 3 cases and occurred in 2 males and 1 female dog (mean: 3.3 y). Protozoal infections were characterized by foci of necrosis with epithelioid macrophages associated with parasitic forms morphologically consistent with Toxoplasma or Neospora spp. One case of eosinophilic leptomeningitis was attributed to Dirofilaria sp. infection given the presence of nematodes in the heart and lungs.

Discussion

Neuroinflammatory CNS diseases comprised 4.8% of the canine autopsy diagnoses in our study. The prevalence of noninfectious diseases (53.6%) was slightly greater than infectious diseases (46.4%), but those numbers need to be interpreted carefully, given that many idiopathic cases were diagnosed as such based on the absence of infectious organisms on routine histology or other ancillary tests at the time of diagnosis. Differences in the frequency of specific disease groups (non-infectious versus infectious) or particular disease types may be explained by environmental predisposing factors (9), institutional variations, methods employed to diagnose specific diseases (6), and the population of canine breeds in a given geographic area (9–11). Given the retrospective nature of our study, we were unable to retest cases for specific infectious pathogens using molecular techniques such as next-generation metagenomic sequencing (12). For this reason, 50 of the 111 (45%) idiopathic diagnoses, for which ancillary diagnostics were not completed, may represent infectious diseases that were undiagnosed because of the lack of appropriate tests at the time of diagnosis (12).

Most neuroinflammatory diseases in our study consisted of non-infectious (idiopathic) meningoencephalomyelitis, also referred to as meningoencephalomyelitis of unknown origin or MUO (10,12,13). This category encompasses diseases such as GME, NME, SRMA, and NLE, among others (14). The cause and pathogenesis of these diseases remains widely unknown (9,10,12,13). However, multifactorial mechanisms associated with immune dysregulation (15–23), genetic predisposition (24), and environmental factors (25) have been suspected. Treatment with immunosuppressive drugs together with other palliative care medications has been frequently used to stabilize or improve the quality of life in patients affected by MUO (10). The partial efficacy of such treatments has been advocated as supportive of the proposed underlying immune dysregulation in these cases (10). Although CNS infection has also been proposed as a possible cause of MUO, investigations using different laboratory techniques have overwhelmingly failed to identify pathogens in the tissues from affected dogs (6,12,16,25–27).

As evidenced by our study, the 3 main types of histologically recognizable MUO include GME, NME, and NLE, which affect mainly, but not exclusively, small breeds of dogs (10,12,24,28–30). The distinction among these disorders in a diagnostic setting is usually achieved based on the distribution of the lesions, the nature of the inflammatory infiltrates, and the exclusion of infectious agents commonly implicated in canine CNS disease (10,12,24,28–30). Gross lesions of GME were uncommonly reported in our study and consisted mainly of red to dark brown, random areas of hemorrhage in the white matter (13,19). Although not consistently present in cases of GME, these foci reflect areas of perivascular hemorrhage (as observed histologically) and should not be surprising given the angiocentric nature of GME lesions (13,19). A mass in the left cerebellar peduncles, medulla oblongata, and cervical spinal cord was reported in 1 case, consistent with the localized form of GME (10,12,24,28–30). Histologic lesions of GME were primarily angiocentric and localized mainly in the white matter and leptomeninges of the brainstem, cerebellum, and less often telencephalon and spinal cord (24,31–33). In our study, 13 of 14 spinal cords had histologic changes of GME, suggesting that lesions may extend to the spinal cord more often than previously thought (32). Any segment of the spinal cord can be affected, but lesions tend to be more frequent in the cervical portion (32), as seen in 5 of our cases. Although optic neuritis was described in 2 of our cases, the number of cases in which optic nerves were examined remains unknown based on the examination of autopsy reports. Ancillary testing was reported in 27 of the 42 GME cases in our dataset, and consisted mainly of FAT for rabies and canine distemper or special stains of tissue sections to rule out bacterial or fungal infection.

Necrotizing meningoencephalomyelitis and NLE are similar disorders that were first described in pugs and Yorkshire terriers, respectively, but have now been reported in many other toy breeds and may represent different lesion patterns of a single entity (24,28–30,34–36). The main distinctive feature between the 2 conditions is the localization of the lesions (24). Necrotizing meningoencephalomyelitis affects predominantly the leptomeninges, telencephalic cortex, and subcortical white matter (29,35,37); NLE targets the periventricular white matter and less often the brainstem (24). As in our cases, the main lesions associated with NME and NLE are extensive areas of necrosis with cavitation and foamy macrophages (29,35,37). Similar to reports for GME, ancillary testing for NME and NLE was negative for infectious agents.

Lymphoplasmacytic meningoencephalomyelitis of unknown cause (also referred to as non-suppurative meningoencephalitis) was diagnosed in 39 cases and has been described in many species, including dogs, cats, and cattle (27,38). Histologic changes consist of lymphoplasmacytic perivascular inflammation throughout the brain and are typically suspected to be associated with a viral infection, which has been confirmed only rarely (27). However, further routine testing typically fails to confirm an underlying infection in these cases, as demonstrated by other investigators (27,38) and in all of our cases. It is possible that a subset of our cases may represent viral infections (West Nile virus, parvovirus, and others) that were not ruled out at the time of diagnosis (27). However, testing for additional diseases was beyond the scope of this investigation.

The lesions of presumed SRMA in our study were restricted to the brain in 4 cases and were systemic in 2 cases. Although these cases were originally diagnosed as potential SRMA, other possibilities should be considered given the lack of confirmatory tests for this condition. An immune-mediated disorder, SRMA is characterized by neutrophilic and fibrinoid arteritis that can lead to secondary ischemic changes in the neuroparenchyma (39). Previously known as beagle pain syndrome, SRMA has now been described in many other breeds of dogs (40–42). Lesions of SRMA can affect one or multiple organs, but they are rarely restricted to the CNS, as seen in 4 of our cases (41–43). One possible reason for this difference was the low number of presumed SRMA cases in our investigation, which precludes evaluation for possible lesion patterns and a comparison with other large population studies. Another possibility is that these potential SRMA cases may represent cases of systemic bacterial infections, given the similarities of the vascular changes in both categories in our study. Although increased serum acute-phase protein concentrations have been used to support a clinical diagnosis of SRMA in dogs (39,44), no such testing was performed in our cases.

Most of the infectious diseases in our study consisted of bacterial and viral infections, similar to reports by other investigators (8). The detection of intralesional organisms in tissue samples can be challenging, especially when ancillary tests are limited or not available (27). Intralesional bacteria were commonly detected in extraneural tissues in our study, especially the left atrioventricular valve, but were rarely reported in the brain. Similarly, the presence and location of viral inclusions varied according to the viral infection, but were reported mainly in extraneural tissues and less often in the brain. Bacterial infections were confirmed by bacterial culture in most cases, which revealed E. coli as the most common isolate in nearly 20% of the cases. Most of the viral infections were confirmed by visualization of viral inclusions in tissue sections and FAT on fresh or frozen tissues. In addition, IHC is a reliable tool for confirmation of canine distemper viral infection (45), as demonstrated in a subset of our cases.

Inherent limitations in our study and other retrospective investigations conducted in our laboratory include the lack of standardized CNS tissue collection and sampling during autopsy, the wide range of pathologists and pathology trainees involved in each case, the number of archived histologic tissue sections available for examination, and the lack of uniform ancillary testing among cases (38). For instance, after longitudinal sectioning of the brain, only the right or left half of the brain was kept in formalin for further gross examination and histologic evaluation in many of our cases; the other half was stored in the freezer in case ancillary tests such as bacterial culture, FAT, or PCR were needed. Although rapid and efficient, hemisectioning has the potential to preclude the diagnosis of diseases with focal or predominantly unilateral lesions, because those lesions may be located in the brain tissue that was placed in the freezer. Another inherent problem associated with sample collection is the lack of standardized tissue sections to allow for proper neurolocalization of lesions for each disease group and for a comparison among various diseases (46). In addition, inconsistencies in ancillary test requests are mainly the result of individual pathologists’ suspicions in various cases, or because of specific requests made by referring veterinarians or owners at the time of diagnosis, which do not always follow an evidence-based approach (38).

Our study indicates that a comprehensive diagnostic approach, with thorough evaluation of the pathology findings and ancillary laboratory test results, is important for the accurate diagnosis of neurologic diseases in dogs. It is also important that collection of brain and spinal cord samples follow a protocol that would allow pathologists to examine representative and standardized areas of the CNS in all cases of potential neurologic disease (46). In addition, a great number of NLM cases were present in our population set, highlighting the need for the development of new laboratory tests that might identify potential infectious agents in these cases (12,47,48).