Detection of neural autoantibodies in dogs with non-infectious inflammatory central nervous system disorders

Abstract

Background

Easily obtainable biomarkers for non-infectious, inflammatory central nervous system (CNS) disorders in dogs are lacking.

Hypothesis/Objectives

Dogs with non-infectious inflammatory CNS disorders might have specific neural autoantibodies in serum or cerebrospinal fluid (CSF), or both, discoverable by commercially available indirect immunofluorescence (IIF) assays.

Animals

Dogs (n = 23) with non-infectious inflammatory CNS disorders, including meningoencephalomyelitis of unknown origin, were recruited from the University Hospital for Companion Animals (Copenhagen, Denmark).

Methods

Using an exploratory, prospective study design, serum and CSF from dogs were investigated for specific neural autoantibodies with cell- and tissue-based IIF assays used to identify neural autoantibodies in humans.

Results

Astrocytic autoantibodies against glial fibrillary acidic protein were present in 2 dogs, neuronal targeted autoantibodies against N-methyl-D-aspartate receptors (NMDAR) in 2 dogs, and tripartite motif-containing protein 46 in 1 dog. Additionally, autoantibodies toward myelin were suspected in 6 dogs and myelin-oligodendrocyte glycoprotein in 4 dogs. Autoantibodies against a neuronal surface structure were suspected in 1 dog but could not be further specified.

Conclusions and clinical importance

This study confirmed the presence of neural autoantibodies in several dogs with non-infectious inflammatory CNS disorders. The importance of both tissue- and cell-based assays (CBA) is emphasized by the fact that all specific autoantibodies apart from those targeting NMDAR were detected by tissue-based assays and then confirmed on cell-based assays (CBA).

Introduction

Several non-infectious inflammatory central nervous system (CNS) disorders exist in dogs,^1,2 including meningoencephalomyelitis of unknown origin (MUO), eosinophilic meningoencephalitis (EME) and idiopathic generalized tremor syndrome (IGTS).^1,3,4 For MUO, subdivision is based on a histopathological diagnosis of granulomatous meningoencephalitis (GME), necrotizing leucoencephalitis (NLE), or necrotizing meningoencephalitis (NME).¹ Non-infectious inflammatory CNS disorders might cause debilitating or lethal neurological disease,^2,5,6 with signs of neurological disease responding to immunosuppression, supporting an immune-mediated etiology.^2,7 However, pathogenesis remains partly unsolved. As some dogs affected by IGTS and MUO have normal standard cerebrospinal fluid (CSF) analysis and MRI findings,^2,8 reaching an ante-mortem diagnosis in these dogs is especially challenging.⁸ The presence of neural autoantibodies is infrequently reported in dogs, including autoantibodies against N-methyl-D-aspartate receptors (NMDAR) in 3 out of 19 dogs with MUO,⁹ and γ-aminobutyric acid-A receptors (GABA_AR) in 1 dog.¹⁰ Astrocytic-targeted autoantibodies against glial fibrillary acidic protein (GFAP) are reported primarily in Pugs affected by NME.^11-17 In dogs with IGTS, autoantibodies against metabotropic glutamate receptor 1 (mGluR1) have been found in 3 out of 33 dogs.⁴ In humans, neuronal autoantibodies against NMDAR, GABA_AR, and mGluR1 are found in autoimmune encephalitis (AE),^18-22 and autoantibodies against GFAP are associated with autoimmune meningoencephalomyelitis.^12,23 When autoantibodies are present in humans, disease can be mediated by pathogenic autoantibodies targeting neural extracellular antigens, causing dysfunction without necessarily causing damage to the tissue.²⁴ Alternatively, cytotoxic T cells mediate disease by targeting neural intracellular antigens causing tissue lesions, often related to underlying neoplasia.²² Regardless of antigen placement, neural autoantibodies might be found in serum or CSF or both.²⁴ Clinically, patients with AE present acutely with progression of neurological signs, potentially over days to weeks,^25-27 equivalent to what is seen in many dogs with non-infectious inflammatory CNS disorders.^2,5 Immunosuppression is an important part of treatment in both dogs and humans.^19,28 The similarities between dogs with non-infectious inflammatory CNS disorders and humans with autoantibody-positive CNS conditions are intriguing, as identification of specific neural autoantibodies is not only used to reach a diagnosis, but also holds important information regarding prognosis, underlying causes, including neoplasia, and optimal treatment.^22,29,30

Whereas indirect immunofluorescence (IIF) cell-based assays (CBA) have identified neural autoantibodies against NMDAR and GABA_AR in dogs,^9,10 each CBA is limited to expressing 1 antigen. Therefore, inclusion of IIF tissue-based assays (TBA) could allow for a broader screening and discovering novel autoantibodies, as has previously been done in IGTS⁴ and pug dogs with NME.¹⁷ This dual approach, combining TBA with a confirmatory test, is recommended in order to reduce the risk of both false positives and negatives.³¹ The purpose of the current study was therefore to investigate serum and CSF from dogs with non-infectious inflammatory CNS disorders for the presence of neural autoantibodies using IIF CBAs and TBAs to elucidate pathophysiology and investigate a translational potential.

Materials and methods

Animals

Client-owned dogs were recruited from the Neurology Clinic at the University Hospital for Companion Animals (UHCA), Frederiksberg, Denmark from the January 1, 2022-April 1, 2025. The study was approved by the local ethical committee (approval number 2021-38). Dogs with a suspected diagnosis of non-infectious inflammatory CNS disease were eligible for inclusion with the owner’s written consent. Dogs were excluded if an infectious cause was suspected or confirmed, or if serum or CSF was not available. When histopathology was not performed, advanced imaging was needed for inclusion, preferably MRI; however, if only CT was available, CSF findings and a positive response to immunosuppressive were used for inclusion. Magnetic resonance images were obtained on a 1.5 Tesla Siemens Magnetom Altea, and CT images were obtained on a Siemens Somatom Emotion single-slice helical CT-scanner. All advanced imaging was performed under generalized anesthesia. Standard hematology and biochemistry were available in all cases, and additional blood tests to rule out extracranial causes and infectious diseases were performed at the responsible veterinarian’s discretion.

Dogs were assigned to 5 subgroups depending on the level of diagnostic confidence to ensure transparency regarding the certainty of the diagnosis. The subgroups were; (1) G1: A definite histopathological diagnosis of NLE, NME, or GME, (2) G2: A clinical diagnosis of MUO supported by characteristic imaging findings on MRI defined as focal or multifocal T2-weighted/FLAIR hyperintense lesion(s) with variable contrast enhancement³² and a positive CSF analysis defined by a mixed or mononuclear pleocytosis (total nucleated cell count >5 cells/μL) with or without elevated protein concentration,³³ (3) G3: A probable diagnosis of MUO supported by either CSF analysis or imaging findings as described for G2, (4) G4: A possible diagnosis of MUO supported by a positive response to immunosuppressive treatment, defined as resolution of neurological signs within 1 month from treatment initiation, and normal CSF analysis and MRI, (5) G5: Dogs with either IGTS as defined in Phillipps et al.² or EME, defined with imaging findings as for MUO,³² combined with an eosinophilic pleocytosis and exclusion of infectious causes.³⁴

Each dog was given an ID-number indicating group followed by their unique number within the group (GX-X). Subgroup 4 (G4) was included to not exclude dogs similar to some forms of human AE with normal MRI and CSF.³⁵

Serum was collected in Vacuette® tube CAT serum separator clot activator (3.5 mL) and CSF was collected in Eppendorf tubes for diagnostic purposes. Excess material was stored in aliquots of 50-200 μL at −80 °C to avoid multiple thaw–freeze cycles until IIF examinations.

A small mixed breed dog (7.5 kg), male entire, 1.5 years old, was included as a negative control in all IIF examinations. The dog presented for elective euthanasia for non-health related reasons at the UHCA and was donated for research. The dog had no history or clinical signs of systemic or neurological disease, including behavioral. Serum and EDTA blood were collected from the venous catheter placed in relation to euthanasia, and CSF was collected from the cerebello-medullary cistern within 10 min after death. Hematology, biochemistry, and standard CSF parameters, including total nucleated cell count and protein, were within reference ranges. Histopathological examination of hematoxylin and eosin-stained sections of all major brain regions (basal nuclei, thalamus, hippocampus, midbrain, cerebellum-pontine area, and medulla oblongata³⁶) was performed to disclose subclinical neuropathology. Serum and CSF were handled as for dogs included in the disease groups.

Indirect immunofluorescence examinations

To investigate for the presence of neural autoantibodies, serum and CSF were examined using IIF assays (EUROIMMUN Medizinische Labordiagnostika AG, Lübeck, Germany). Both TBAs and CBAs were included initially (Table 1) and supplemented with additional examinations based on findings (Table 2). The specific CBAs included were chosen based on previous reported findings in dogs,^9,10,16 as well as clinical similarities to specific forms of human AE. A CBA for anti-myelin-oligodendrocyte glycoprotein (MOG)-autoantibodies was performed when serum was available after performing initial and additional assays (n = 18).

Table 1.

Overview of assays performed on serum and CSF. All assays and positive controls, including canine ANA positive serum, produced by EUROIMMUN Medizinische Labordiagnostika AG, Seekamp 31, 23560 Lübeck, Germany.

Assay	Target	Assay type	Controls
EUROIMMUN IIFT Mosaic: Cerebellum (Rat)/ Hippocampus (Rat)	Cerebellum (rat), Hippocampus (rat)	Tissue-based assay (rat)	Positive: Kit-provided positive human sample (NMDAR), canine ANA-positive sample (S 1:10) Negative: Healthy dog sample (S 1:10, CSF undiluted)
EUROIMMUN IIFT: Autoimmune Encephalitis Mosaic 6	Glutamate receptor type N-methyl-D-aspartate (NMDAR), glutamate receptor type α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) 1/2, contactin-associated protein 2 (CASPR2), leucine-rich glioma-inactivated protein 1 (LGI1), dipeptidyl aminopeptidase-like protein 6 (DPPX), γ-Aminobutyric acid type B-receptor (GABAB)	Cell-based assay: Transfected cells (EU90)	Positive: Kit-provided positive human sample (NMDAR), canine ANA-positive sample (S 1:10) Negative: Healthy dog sample (S 1:10, CSF undiluted)
EUROIMMUN Anti γ-Aminobutyric acid type A-receptor (GABA A ) IIFTa	Aminobutyric acid type A-receptor (GABAA)	Cell-based assay: Transfected and control-transfected (control) cells (EU90)	Positive: Kit-provided positive human sample (GABAA) Negative: Healthy dog sample (S 1:10, CSF undiluted)
EUROIMMUN Anti-Glutamic Acid Decarboxylase (65 kDa) (GAD65) IIFT	Glutamic acid decarboxylase 65 kDa (GAD65)	Cell-based assay: Transfected and control-transfected (control) cells (EU90)	Positive: Kit-provided positive human sample (GAD65), canine ANA-positive sample (S 1:10) Negative: Healthy dog sample (S 1:10, CSF undiluted)
EUROIMMUN Anti-Myelin-oligodendrocyte glycoprotein (MOG) IIFT	Myelin-oligodendrocyte glycoprotein (MOG)	Cell-based assay: Transfected and control-transfected cells (EU90)	Positive: Kit-provided positive human sample (MOGNegative: Healthy dog sample (S 1:10)

^aResearch-use only.

Abbreviations: ANA = anti-nuclear antibodies; CSF = cerebrospinal fluid; EU90 = specific cell-line used; IGTS = idiopathic generalized tremor syndrome; IIFT = indirect immunofluorescence test; S = serum.

Table 2.

Overview of additional assays performed on serum and CSF based on findings on assays reported in Table 1.

Assay name	Target	Assay type	Controls
EUROIMMUN IIFT Neurology Mosaic 8	Cerebellum (monkey), peripheral nerve (monkey), intestinal tissue (monkey), pancreas (monkey)	Tissue-based assay	Positive: Kit-provided positive human samples (Hu, GAD65), ANA-positive dog sample (S 1:10) Negative: Healthy dog sample (S 1:10, CSF undiluted)
EUROIMMUN IIFT Purkinje Cell Mosaic 1a,b	Carbonic anhydrase-related protein VIII (CARP) Yo/cerebellar degeneration related protein 2 (CDR2) Inositol 1,4,5-Trisphosphate Receptor Type 1 (ITPR1) Delta/notch-like epidermal growth factor-related receptor (Tr/DNER)	Cell-based assay: Transfected cells (EU90)	Positive: Positive human sample (Yo, S 1:100) Negative: Healthy dog sample (S 1:10, 1:100, CSF 1:2, 1:10)
EUROIMMUN IIFT Neurology Mosaic 11aa,b	Glutamate receptor type α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) type 1/2, contactin-associated protein 2 (CASPR2), dipeptidyl aminopeptidase-like protein 6 (DPPX), γ-Aminobutyric acid type A/B-receptor (GABAA/B), glycine receptor subunit α1β (GLRA1b), leucine-rich glioma-inactivated protein 1 (LGI1), metabotropic glutamate receptor 1 (mGluR1), metabotropic glutamate receptor 5 (mGluR5), glutamate receptor type N-methyl-D-aspartate (NMDAR)	Cell-based assay: Transfected cells (EU90)	Positive: Positive human sample (mGluR1, S 1:10) Negative: Healthy dog sample (S 1:10, 1:100)
EUROIMMUN IIFT Neurology Mosaic 1Ca,b	IgLON family member 5 (IgLON5), Metabotropic glutamate receptor 5 (mGluR5), Glycine receptor subunit α1β (GLRA1b)	Cell-based assay: Transfected and control-transfected (control) cells (EU90)	Positive: Positive human sample (GLRA1b, S 1:10) Negative: Healthy dog sample (S 1:10, 1:100, CSF 1:2, 1:10)
EUROIMMUN Anti-Amphiphysin IIFTa,b	Amphiphysin	Cell-based assay: Transfected and control-transfected (control) cells (EU90)	Positive: Positive human sample (amphiphysin, S 1:1000) Negative: Healthy dog sample (S 1:10, 1:100, CSF 1:2, 1:10)
EUROIMMUN Anti-Aquaporin-4 (AQP4) IIFT	Aquaporin-4 (AQP4)	Cell-based assay: Transfected and control-transfected (control) cells (EU90)	Positive: Kit-provided positive human sample (AQP4Negative: Healthy dog sample (S 1:10, 1:100, CSF 1:2, 1:10)
EUROIMMUN Anti-CV2/collapsing response mediator protein 5 (CRMP5) IIFTa,b	CV2/collapsing response mediator protein 5 (CRMP5)	Cell-based assay: Transfected and control-transfected (control) cells (EU90)	Positive: Positive human sample (CV2/CRMP5, S 1:320) Negative: Healthy dog sample (S 1:10, 1:100, CSF 1:2, 1:10)
EUROIMMUN Anti-Flotillin-1/2 IIFTa,b	Flotillin-1/2	Cell-based assay: Transfected and control-transfected (control) cells (EU90)	Positive: Positive human sample (Flotilin-1/2, S 1:100) Negative: Healthy dog sample (S 1:10, 1:100, CSF 1:2, 1:10)
EUROIMMUN Anti-Glial fibrillary acidic protein (GFAP) IIFTa,b	Glial fibrillary acidic protein (GFAP), astrocytes	Cell-based assay: Transfected and control-transfected (control) cells (EU90)	Positive: Positive human sample (GFAP, S 1:100) Negative: Healthy dog sample (S 1:10, 1:100, CSF 1:2, 1:10)
EUROIMMUN Anti-Tripartite motif-containing protein 46 (TRIM46) IIFTa,b	Tripartite motif-containing protein 46 (TRIM46)	Cell-based assay: Transfected and control-transfected (control) cells (EU90)	Positive: Positive human sample (TRIM46, S 1:10) Negative: Healthy dog sample (S 1:10, 1:100)

^aPerformed at EUROIMMUN Medizinische Labordiagnostika AG, Seekamp 31, 23560 Lübeck, Germany. ANA-positive sample not included.

^bResearch-use only.

All assays and positive controls, including ANA positive dog serum, produced by EUROIMMUN Medizinische Labordiagnostika AG, Seekamp 31, 23560 Lübeck, Germany. Abbreviations: ANA = anti-nuclear antibodies; EU90 = name of human cell line; IIFT = indirect immunofluorescence test.

All assays were performed according to the manufacturers recommendations with the following protocol modifications³⁷; (1) to include a positive dog control verifying binding by the fluorescence-marked anti-dog-IgG serum, serum from a dog positive for antinuclear antibodies (EUROIMMUN) in assays where nuclear antigen was exposed, (2) the healthy control dog was included instead of a human negative control, (3) fluorescence-marked anti-dog-IgG (EUROIMMUN) was used as the conjugate for dog samples, (4) incubation of transfected cells/tissue and dog material were prolonged to 120 min to ensure optimal binding between assay and autoantibodies, (5) serum samples were diluted to 1:10 and 1:50, and (6) for CSF examinations, 25 μL were used undiluted. If neural autoantibodies were suspected to be present on TBAs, or when clinical findings seemed suspicious for specific neural autoantibodies because of similarities to a type of human AE, CBAs were performed to identify and confirm the presence of autoantibodies (Table 2).³¹ For included research-use-only CBAs, the manufacturers’ recommendations regarding incubation time (30 min) and serum dilutions (1:10, 1:100) were followed, and CSF was tested in 1:2 and 1:10 dilutions.

Results of IIF were interpretated by trained personnel at the Department of Clinical Immunology, Odense University Hospital (Odense, Denmark) or at the Institute for Experimental Immunology, affiliated to EUROIMMUN Medizinische Labordiagnostika AG (Lübeck, Germany). Fluorescence staining was evaluated quantitatively by comparing reactivity to the included negative control at the cut-off dilution (1:10 for serum samples and undiluted for CSF) for both TBAs and CBAs. If a specific fluorescence pattern was present when evaluating TBAs, relevant CBAs were used to confirm a suspected autoantibody. If a fluorescence pattern was present on TBAs, but no specific autoantibodies were suspected, additional assays, either CBAs or TBAs, were performed.

When evaluating CBAs, the number of fluorescent cells (an expression of the transfection rate) was compared to the positive human control and needed to be similar, whereas the included plasmid-transfected control cells had to be negative, for a sample to be deemed positive.

Results

Twenty-three dogs were included as follows: Subgroup 1 n = 3 (G1-1 and G1-2 confirmed GME, G1-3 confirmed NME), subgroup 2 n = 3, subgroup 3 n = 10, subgroup 4 n = 4, and subgroup 5 n = 3. Subgroup 5 included 2 dogs with IGTS (G5-1, G5-2), and 1 dog with EME (G5-3).

An overview of all included dogs, including neuro-localization, imaging and CSF findings, and performed IIF assays and results can be found in Table S1, summarized in Table S2. Neural autoantibodies were identified in 4 out of 24 dogs, with 1 dog (G2-3) having autoantibodies against both GFAP and NMDAR. Autoantibodies were found against GFAP (n = 2), NMDAR (n = 2), and tripartite motif-containing protein 46 (n = 1). The latter is described in detail in Heidemann et al.³⁸ and is not addressed further here. Autoantibodies were suspected for 7 dogs, ie, a positive fluorescence pattern was present on TBAs, but specific autoantibodies could either not be confirmed with available CBAs, or results were inconclusive. In 2 of these dogs, other specific autoantibodies were identified (G2-3, G5-3). All IIF assays for the included healthy dog control were negative.

Dog G1-3, a 6-year-old female neutered Shih Tzu with NME was presented due to an acute onset of seizures 2 days prior to sample collection, and dog G2-3 was a 4.5-year-old female neutered Chihuahua with several previous relapses of presumed MUO. Both dogs had an initial TBA pattern suspicious of anti-GFAP-autoantibodies in CSF (Figure 1), subsequently confirmed in both serum and CSF with a specific CBA (Figure 2). Dog G2-3, harboring anti-GFAP-autoantibodies, and dog G5-3, a Bernese Mountain dog affected by EME, were both positive for anti-NMDAR-autoantibodies on the specific CBA in serum (Figure 3).

Figure 1.

Distinct fluorescent astrocyte-related pattern on “EUROIMMUN Mosaic: Cerebellum (rat)/Hippocampus (rat)” for dog G1-3 and G2-3. (A) Hippocampus and (B) cerebellum showing a distinct fluorescent pattern from dog G1-3 (CSF, 1:5). (C) The same distinct fluorescent pattern on hippocampus (rat) and (D) cerebellum (rat) for dog G2-3 (CSF, undiluted). (E) Negative assay for the healthy control dog (serum, 1:10) on hippocampus and (F) cerebellum (rat) for comparison. Abbreviations: CSF = cerebrospinal fluid; GFAP = glial fibrillary acidic protein.

Figure 2.

Fluorescence on “EUROIMMUN Anti-Glial fibrillary acidic protein (GFAP) IIFT” cell-based assay (EU90) confirming the presence of autoantibodies targeting GFAP in dog G1-3 and G2-3. (A) Dog G1-3 showing a strong fluorescence on transfected cells for CSF (1:10) and weak fluorescence for (B) serum (1:100). (C) Dog 2-3 showing fluorescence on transfected cells for CSF (1:2) and weak fluorescence for (D) serum (1:100). (E) Strong positive human control (serum, 1:100), and (F) negative assay for the healthy control dog (CSF, undiluted). Abbreviations: CSF = cerebrospinal fluid; GFAP = glial fibrillary acidic protein.

Figure 3.

Fluorescence on “EUROIMMUN IIFT autoimmune encephalitis mosaic 6” cell-based assay for the NMDAR biochip for dog G2-3 and G5-3. (A) Dog G2-3 (serum, 1:50), (B) dog G5-3 (serum, 1:10), (C) strong positive human control, (D) negative assay for the healthy control dog (serum, 1:10). Abbreviation: NMDAR = N-methyl-D-aspartate receptors.

On the anti-MOG CBA, for 4 dogs (G2-3, G3-2, G3-7, G3-10) fluorescence was seen on a population of transfected cells, but not on the control-transfected control cells included in the assay (ie, same cell-line but transfected with an empty plasmid instead of a plasmid containing the MOG-coding DNA sequence). However, fewer cells than for the positive human control sample showed fluorescence (Figure 4). All the dogs with a suspicious reaction towards the MOG-transfected cells had lesions related to white matter, with 3 dogs additionally having lesions in thalamus on MRI.

Figure 4.

Fluorescence on “EUROIMMUN anti-myelin-oligodendrocyte glycoprotein (MOG) indirect” immunofluorescence cell-based assays for dogs with a suspicious pattern on cells (EU90). Compared with the negative control cells, fluorescence related to cells in the same area as for the positive human control is present, but in fewer cells. (A) Dog G2-3 (serum, 1:10), (B) dog G3-2 (serum, 1:10), (C) dog G3-7 (serum, 1:10), (D) dog G3-10 (serum, 1:10), (E) strong positive human control, (F) negative assay for the healthy control dog (serum, 1:10).

In 4 dogs (G1-1, G2-2, G3-7, G5-3), a fluorescence pattern related to myelin was seen on TBA. As no specific antigen was suspected, a TBA containing peripheral nerve tissue from monkey was performed; for all 4 dogs, fluorescence related to peripheral nerve myelin was present, indicating a more general autoantibody toward myelin, rather than a target restricted to CNS.

For dog G3-1, a fluorescence pattern suspicious for extracellular targeted autoantibodies was present. Several CBAs for both extra- and intracellularly placed targets were performed, all negative.

In dog G3-9, a positive fluorescence pattern related to vessels was seen on initial TBAs. Based on the experience of the trained personnel evaluating fluorescence, it was deemed incidental without clinical relevance as it was not restricted to CNS nor related to neural tissue.

Discussion

Several dogs had either confirmed (n = 4) or suspected (n = 7) neural autoantibodies, but the role of these autoantibodies are, apart from indicating an immune-mediated etiology, unclear.

Because GFAP is an intracellular antigen, it is likely that the pathophysiology in the 2 dogs with anti-GFAP-autoantibodies was mediated by cytotoxic T cells rather than the autoantibodies themselves, as for humans.²² For dog G1-3, it seems likely that GFAP was primarily targeted by the immune system, causing NME, similar to what has been reported in humans, Pug dogs, and a Shih Tzu dog.^12,13,17 As for Pugs, the findings might indicate a heritability in the Shih Tzu breed for NME.³⁹

Dog G2-3, in contrast, had a different progression of disease than reported for anti-GFAP-positive cases of NME.^12,14,17 Whereas anti-GFAP-autoantibodies are related to meningoencephalomyelitis in humans,¹² they have also been reported secondary to traumatic brain injury where the primary astrocytic insult likely exposes intracellular GFAP and provoking an immune reaction.⁴⁰ It is not clear if the secondary immune reaction towards GFAP exacerbates inflammation, but the presence of anti-GFAP-autoantibodies does correlate negatively with outcome.⁴⁰ Dog G2-3 also harbored autoantibodies targeting NMDAR in serum and had a reaction to myelin on both rat (CNS) and monkey tissue (peripheral nerve), and to MOG-transfected cells. Taken together, this dog seemed to have several abnormal ongoing inflammatory processes, but further clarification of which process was the primary is not possible.

For both dog G2-3 and G5-3, anti-NMDAR-autoantibodies were only found in serum. A reason for the negative assay results for CSF could be a very strong affinity and binding to the CNS parenchyma resulting in no free CSF autoantibodies. However, it seems more likely that the finding is purely incidental; in previously reported cases of anti-NMDAR-autoantibodies in dogs, autoantibodies could also be found in CSF,⁹ and in humans, anti-NMDAR-autoantibodies are primarily thought to be clinically relevant if found in CSF.⁴¹ Because this autoantibody has only been reported in 3 dogs, more cases are needed to determine the clinical relevance in dogs, regardless of their presence in CSF or serum or both. The 2 dogs differed clinically; whereas dog G2-3, the Chihuahua also harboring anti-GFAP-autoantibodies, had a chronic history of several relapses, dog G5-3 had an EME with rapid disease progression. Thus, the identification of anti-NMDAR-autoantibodies could have been incidental. Another explanation could be that the anti-NMDAR-autoantibodies are not only present in a specific form of AE. For instance, in humans, anti-NMDAR-autoantibodies are found in serum from stroke patients, indicating that they could also be a more general marker for CNS disease.⁴²

For dogs with a suspicious pattern on the anti-MOG CBA (G2-3, G3-2, G3-7, G3-10), assay interpretation was inconclusive. As fluorescence was only present in the transfected cells and not in the negative control-transfected cells, there might be a reaction toward the epitope. Additionally, the dogs all had lesion distributions similar to humans with anti-MOG-autoantibodies in which encephalomyelitis can be present, and lesions primarily affect white matter and thalamus.⁴³ However, when used for diagnostic purposes in human medicine, for a CBA to be positive, the same number of cells should show fluorescence when comparing the patient sample to the positive control, which was not the case in the current study. The findings could be incidental, or they could be a result of species-specific epitope differences causing suboptimal binding between dog IgG and the MOG-expressing cells.

For dog G3-1, the presence of neuronal autoantibodies was suspected based on the TBA but none of the additional CBAs performed for this dog was positive. This could either be due to the CBAs tested not being specific for the autoantibody present, the human epitopes being too different from the canine antigen for the IgG to bind, or the autoantibody not yet having been discovered in humans. If enough material were available, immunoprecipitation and mass spectrometry could potentially have identified the autoantibodies. However, the pattern on TBA was only present for CSF, and there was not enough material for further examinations.

A main advantage of identifying neural autoantibodies in human patients is the classification into homogenous groups according to the antigen, allowing for investigations of prognosis and optimal treatment that can then be applied to patients with that specific autoantibody.^22,29 Currently, especially when dealing with the heterogenous disease group MUO, histopathology, performed mainly post-mortem, is needed to reach a final diagnosis.¹ The possibility to subdivide dogs ante-mortem into more homogenous groups would remove many of the challenges currently faced in both clinical work and research. As in humans, apart from having diagnostic value, identification of neural autoantibodies in dogs could potentially hold prognostic information and guide treatment, as well as determine the relevance of investigating for an underlying neoplasia. In human patients with confirmed autoantibody-mediated pathophysiology,²⁴ immunosuppression is combined with other treatment approaches, including removal of autoantibodies by plasma exchange.⁴⁴ Given the reserved-poor prognosis for MUO,^5,45,46 this could potentially be explored in dogs with suspected pathogenic autoantibodies.

Tissue-based assays (cerebellum (rat)/hippocampus (rat)) were included to do the broadest possible examination and not overlook autoantibodies in initial examinations. As TBAs are suboptimal in identifying neural autoantibodies against extracellular targets, with up to 50% false negative cases in humans with autoantibodies against NMDAR,⁴⁷ CBAs were also included. Still, the main limitation to the current study is the risk of false negative results, as neither tissue nor antigen-expressing cells were from dogs, potentially risking a mismatch between autoantibodies and antigens. Whereas rat- and monkey tissue are used routinely in humans,^48,49 there is no guarantee that the same assays are valid in dogs. However, we chose to use assays already available to allow for accessibility to the included assays for the sake of potential future studies and clinical implementation. Alternatively, TBAs based on dog tissue could have been included, although the process of developing such TBAs would be tedious and potentially pose ethical dilemmas. As the included CBAs are based on human cell lines, a similar approach could be used to produce a dog-specific CBA mosaic. However, relevant autoantigens would need to be confirmed in a larger population. We included dogs if either serum or CSF or both were available to investigate as many dogs as possible. However, some dogs for which only serum was investigated might have had intrathecally produced autoantibodies in CSF that were then missed. Another limitation is the heterogeneity in the diagnostic work-up, as this was decided solemnly by the responsible clinician due to the ethical approval granted for the study. However, a stringent study with a standardized diagnostic protocol could have ensured a more robust work-up of the population and should be considered for future studies into neural autoantibodies. Finally, as we only investigated for the presence of IgG, other isotypes could potentially have been present. Due to the often-limited amount of biological material, especially because CSF is examined undiluted, we had to prioritize what seemed most relevant based on human literature.²⁴

Conclusions

We successfully identified neural autoantibodies in a subset of dogs, but not all, with non-infectious inflammatory CNS disorders, strengthening a theory of a translational potential between dogs and humans. Based on our results, it seems reasonable to compare dogs with non-infectious inflammatory CNS disorders with neural autoantibodies to autoantibody-positive human CNS conditions, and to further explore the translational potential. Neural autoantibodies, when present, could have both diagnostic and prognostic value and could guide treatment decisions.

Abbreviations

AE

autoimmune encephalitis

ANA

antinuclear antibodies

CBA

cell-based assay

CNS

central nervous system

CSF

cerebrospinal fluid

CT

computed tomography

EME

eosinophilic meningoencephalitis

FLAIR

fluid-attenuated inversion recovery

GABA_AR

γ-aminobutyric acid-A receptors

GFAP

glial fibrillary acidic protein

GME

granulomatous meningoencephalitis

IgG

immunoglobulin G

IGTS

idiopathic generalized tremor syndrome

IIF

indirect immunofluorescence

IIFT

indirect immunofluorescence test

MUO

meningoencephalomyelitis of unknown origin

mGluR1

metabotropic glutamate receptor 1

MOG

myelin-oligodendrocyte glycoprotein

MRI

magnetic resonance imaging

NLE

necrotizing leucoencephalitis

NMDAR

N-methyl-D-aspartate receptor

NME

necrotizing meningoencephalitis

TBA

tissue-based assay

TRIM46

tripartite motif-containing protein 46

UHCA

the University Hospital for Companion Animals

Author contributions

Pernille Lindholm Heidemann (Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Writing—original draft), Christine Nilsson (Conceptualization, Formal analysis, Funding acquisition, Investigation, Supervision, Writing—review & editing), Sandra G. Schmidt (Formal analysis, Investigation, Methodology, Writing—review & editing), Erik Lattwein (Formal analysis, Funding acquisition, Investigation, Methodology, Writing—review & editing), Jørgen Steen Agerholm (Data curation, Formal analysis, Investigation, Writing—review & editing), Morten Blaabjerg (Conceptualization, Methodology, Supervision, Writing—review & editing), Hanne Birgit Gredal (Conceptualization, Data curation, Funding acquisition, Project administration, Supervision, Writing—review & editing)

Conflicts of interest

E. Lattwein is an employee of EUROIMMUN Medizinische Labordiagnostika AG. The other authors declare no interests.

Funding

The Fund for Disease Control in Companion Animals, University of Copenhagen, numbers 2021-0002 and 2022-0001.

Off-label antimicrobial declaration

Authors declare no off-label use of antimicrobials.

Institutional animal care and use committee or other approval declaration

This project underwent ethical review and was approved by the local ethical committee at the Department of Veterinary Clinical Sciences, University of Copenhagen (permission number 2021-38). Written informed consent was obtained from the owners of the dogs included in this study, as well as for the owner of the dog included as a healthy control, allowing for participation in this study.

Human ethics approval declaration

Authors declare human ethics approval was not needed.