Serum anti‐GM2 and anti‐GalNAc‐GD1a ganglioside IgG antibodies are biomarkers for immune‐mediated polyneuropathies in cats

Abstract Recent work identified anti‐GM2 and anti‐GalNAc‐GD1a IgG ganglioside antibodies as biomarkers in dogs clinically diagnosed with acute canine polyradiculoneuritis, in turn considered a canine equivalent of Guillain‐Barré syndrome. This study aims to investigate the serum prevalence of similar antibodies in cats clinically diagnosed with immune‐mediated polyneuropathies. The sera from 41 cats clinically diagnosed with immune‐mediated polyneuropathies (IPN), 9 cats with other neurological or neuromuscular disorders (ONM) and 46 neurologically normal cats (CTRL) were examined for the presence of IgG antibodies against glycolipids GM1, GM2, GD1a, GD1b, GalNAc‐GD1a, GA1, SGPG, LM1, galactocerebroside and sulphatide. A total of 29/41 IPN‐cats had either anti‐GM2 or anti‐GalNAc‐GD1a IgG antibodies, with 24/29 cats having both. Direct comparison of anti‐GM2 (sensitivity: 70.7%; specificity: 78.2%) and anti‐GalNAc‐GD1a (sensitivity: 70.7%; specificity: 70.9%) antibodies narrowly showed anti‐GM2 IgG antibodies to be the better marker for identifying IPN‐cats when compared to the combined ONM and CTRL groups (P = .049). Anti‐GA1 and/or anti‐sulphatide IgG antibodies were ubiquitously present across all sample groups, whereas antibodies against GM1, GD1a, GD1b, SGPG, LM1 and galactocerebroside were overall only rarely observed. Anti‐GM2 and anti‐GalNAc‐GD1a IgG antibodies may serve as serum biomarkers for immune‐mediated polyneuropathies in cats, as previously observed in dogs and humans.

Anti-GM2 and anti-GalNAc-GD1a IgG antibodies may serve as serum biomarkers for immune-mediated polyneuropathies in cats, as previously observed in dogs and humans. Poly(radiculo)neuropathies in cats considered to be of immunemediated aetiology accounted for nearly 60% of feline nerve biopsy material submitted for histological examination in a previous study. 4 These conditions may present clinically in acute or chronic patterns, may affect specific anatomical regions, such as bilateral brachial plexus neuritis, and may be more prevalent in certain breeds, such as Bengal Cat Polyneuropathy. 2 Bengal Cat Polyneuropathy typically affects young cats, may manifest with multiple episodes and a relapsing/ remitting course, and a full or partial recovery is achieved in around 90% of cases. Electrophysiological and histological features comprise a demyelinating and distal denervating phenotype. 5 Over recent years, a polyneuropathy with similar clinical and biopsy features to Bengal Cat Polyneuropathy, termed "Heterogenous Motor Polyneuropathy in Young Cats" has been observed in many other breeds, including Domestic Short-or Longhaired, Siamese and Persian cats. 6 Birman and British Shorthair cats are also considered to potentially be predisposed to immune-mediated polyradiculoneuropathies. 4 In acute and chronic immune-mediated peripheral neuropathies in man, including Guillain-Barré syndrome (GBS), many anti-glycolipid antibodies (AGAbs) have been identified as serological markers of disease. 7,8 Whilst anti-GM2 and anti-GalNAc-GD1a antibodies are not the most common AGAb biomarkers in man, both are well-described in acute and chronic auto-immune neuropathy syndromes. [9][10][11][12][13] Knowledge of human disease biomarkers in GBS led us to first investigate dogs clinically diagnosed with acute canine polyradiculoneuritis (ACP), a canine equivalent to GBS, for similar AGAbs. In ACP, initially in a pilot study and latterly in a larger international cohort, we observed a high prevalence of serum anti-GM2 and anti-GalNAc-GD1a antibodies. 14,15 Herein, we investigated cats clinically diagnosed with immune-mediated polyneuropathies (IPN) in comparison to neurological and non-neurological control groups for the prevalence of similar serum AGAbs.

| Samples
Cat serum samples were acquired over 8 years (2015-2022), following a national (UK) and international call for diseased and control serum samples sent out initially to board-certified veterinary neurologists.
The study was ethically approved by the University of Cambridge (CR101) and University of Glasgow (Ref14a/16).
The presumptive diagnosis of immune-mediated polyneuropathy (IPN) was based on signalment (commonly young Bengal or other purebred cat, but also other cat breed, of any sex, with initial presentation typically at less than 1 year of age), and a clinical history of progressive (typically over 1-2 weeks) or relapsing/remitting para-or tetraparesis, decreased or absent withdrawal reflexes in all limbs, and occasional hyperaesthesia and cranial nerve involvement (typical bilateral facial nerve paresis). The diagnosis was supported by ancillary investigations including an absence of biochemical abnormalities to explain the presenting clinical signs, negative infectious screens (eg, for Toxoplasma gondii, feline leukaemia virus (FeLV), feline immunodeficiency virus (FIV) and feline coronavirus (FCoV)), supportive electrophysiological changes, such as abnormal spontaneous myofibre activity on electromyography (EMG), decreased motor nerve conduction velocity and compound muscle action potential (CMAP) amplitudes, and variable conduction block, 5,6,16  and ancillary investigations carried out.
All samples were shipped with cooling agents, were immediately blinded by coding upon receipt and stored at À80 C until use.

| Data analysis and statistics
FIU values were converted to binary data with the application of an optimal cut-off threshold calculated for each unique glycolipid target to discriminate between IPN and combined ONM and CTRL groups by plotting the ROC curve (MedCalc Statistical software version 19.7, Ostend, Belgium) and utilising the Youden index (J) method 17 when giving equal weight to sensitivity and specificity. Both neurological/ neuromuscular (ONM) and non-neurological (CTRL) control groups were combined to allow for comparison of IPN to both neurological controls and to ageÀ/breed-specific control sera at the same time, bearing the distinct age-and breed-specific picture (BCP) for many IPN-cases in mind. The DeLong method 18 was employed for the comparison of paired ROC curves. Array data for each cat serum was graphically displayed as heat maps using the rainbow scale to broadly distinguish intensity values across different glycolipid targets (MultiExperiment Viewer software; version 4.9.0; TM4 software suite).
Gaussian data were analysed using a two-tailed Student's t test.
Proportion data were analysed using Fisher's exact test for counts. org/). 19 3 | RESULTS

| Anti-glycolipid antibody serology
Other AGAbs such as anti-GA1 and/or -sulphatide IgG antibodies were present in a high proportion of IPN samples. However, these were deemed to be non-specific for IPN-cats, due to their ubiquitous presence across all sample groups (Figure 2A). Interestingly, this antibody pattern contrasts with the absence or low frequency of IgG AGAbs detected against glycolipids GM1, GD1a, GD1b, SGPG, LM1 and GalC ( Figure 2B), other than when antibodies against such targets formed a part of heteromeric complexes with glycolipids GM2 or GalNAc-GD1a (see above).
The clinical diagnosis for the single ONM-cat exhibiting both anti-GM2 and anti-GalNAc-GD1a IgG AGAbs was myasthenia gravis. This diagnosis was reached following electrophysiological investigations, but the anti-nicotinic acetylcholine receptor antibody titre was within normal limits. A positive clinical response to corticosteroid therapy was reported in this case.

| DISCUSSION
A recent large international multicentre study described the demographics and clinical features of cats clinically and histologically F I G U R E 2 Serum IgG antibodies against glycolipids GA1, sulphatide, GM1, GD1, GD1b, SGPG, LM1 and galactocerebroside and some of their 1:1 heteromeric complexes in cats with immune-mediated polyneuropathies (IPN; n = 41), other neurological and neuro-muscular disorders (ONM; n = 9) and non-neurological control cats (CTRL; n = 46). A, Heat map showing serum IgG antibodies against glycolipids GA1 and sulphatide and some of their 1:1 heteromeric complexes in cats. Each column represents a unique glycolipid target, whilst each row represents a cat serum. The fluorescence intensity unit (FIU) intensity increases from black (0) through the rainbow scale to red (≥ 4000). IgG anti-glycolipid antibodies AGAbs are present at comparable levels across all groups and are therefore non-discriminating biomarkers in this cohort of cats. B, Heat map showing serum IgG antibodies against glycolipids GM1, GD1a, GD1b, SGPG, LM1 and galactocerebroside and some of their 1:1 heteromeric complexes in cats. IgG AGAbs are either absent or infrequently present at comparable levels across all groups and are therefore non-discriminating biomarkers in this cohort of cats.
diagnosed with presumed immune-mediated polyneuropathies. 16 The cohort examined in the current study had similar clinical features, with over-representation of young cats (typically ≤2 years old), male sex, and specific breeds comprising Bengal, Domestic and British Shorthair cats. We thus consider our cohort typically representative of the IPN diagnostic grouping.
Screening of the entire IPN-cohort for underlying infectious disease, which represents a major differential aetiology for peripheral nerve disease in young cats, did not provide any evidence of infection with FIV or Toxoplasma gondii. Experimentally, an infection with FIV has been shown to be associated with a peripheral neuropathy. 20 Toxoplasmosis, which in cats more typically causes encephalitis and myelitis, 21  form of FCoV) only rarely affects the peripheral nervous system, 24,26 and that all of the seropositive cats in this study lacked any other supportive signs of feline infectious peritonitis, makes the seropositivity likely to be coincidental and unrelated to the diagnosis of IPN.
Reports of preceding or coinciding events, such as vaccinations, gastrointestinal or respiratory disease were sparse, thereby not supporting a clinically evident association with development of disease.
However, more focused research with larger case numbers would be required to investigate whether a specific association may exist for these cases.
Interestingly, nearly two-thirds of the IPN-cats had a disease onset in the autumn (25%) and winter (40%) months, similar to observed in previous studies on immune-mediated peripheral neuropathies (ACP) in dogs. 14,27 The IPN-cats in this study possessed serum anti-GM2 and anti-GalNAc-GD1a antibodies as the most distinct AGAb marker, highly similar to our previous findings in ACP-dogs. 14 Antibodies to both these targets reached a higher sensitivity, yet slightly lower specificity, when compared to the ACP-dogs. Also as seen in ACP-dogs, many IPN-cats possessed both anti-GM2 and anti-GalNAc-GD1a AGAbs, yet with lower binding levels for the latter. This is most likely due to crossreactivity between anti-GM2 and anti-GalNAc-GD1a AGAbs, since these glycolipids contain an identical terminal trisaccharide moiety. 10,28 Co-existing anti-GM2 and anti-GalNAc-GD1a (albeit IgM) AGAbs have been reported in human patients suffering from chronic demyelinating neuropathy with sensory ataxia, 12 from pure motor chronic demyelinating neuropathy, 10 and from demyelinating neuropathy characterised by slow progression, frequent facial nerve involvement (occasionally also observed in this cohort of cats) and sensory deficits. 11 Human patients with GBS may also possess anti-GM2 and anti-GalNAc-GD1a IgG antibodies, 7-9 but these more rarely co-occur than is seen in ACP and IPN, suggesting potential differences in antibody fine specificity between cats/dogs and humans. regions. 15 Similarly, the presence of GM2 has previously been demonstrated in feline sciatic nerve, 28 and with immunohistochemical investigations has been located to the abaxonal Schwann cell membranes (data not shown). Selective Schwann cell injury has been shown to be associated with myelin fragmentation either affecting parts or the entire internode, 30 and secondary axonal degeneration is commonly observed in demyelinating neuropathies. 31  anti-GM2 antibodies, 9 but also recovery times for IPN-cats, which in recent literature range from a few days to 17 months. 16 At the same time, such a heterogeneity in targets and regions might provide a potential explanation for the relative lack of distinct AGAb-binding patterns observed between breeds, ages and phenotypes, which lies in stark contrast to more typical GBS-subtypes and their distinct AGAb-profiles, 35 and might further also explain the lack of correlation of certain IPN-breeds to histological features. 29 Whilst being unable to directly demonstrate causality, in combination with experimental work conducted on potential pathophysiological roles of AGAbs in GBS, 36,37 the results of this study strongly suggest that the anti-GM2 and anti-GalNAc-GD1a antibodies we have observed in IPN-cats are likely to be pathogenic factors, as well as suitable biomarkers. Feline IgG1 is capable of fixing complement and thus acting mechanistically in IPN to drive complement-mediated nerve injury. 38 Owing to the absence of serial samples, we were not able to assess serial titres over time and ethical considerations further preclude pathophysiological analysis of causality in modelling studies directly carried out in cats.
Finally, a number of cats in the non-neurological control population exhibited anti-GM2 and/or anti-GalNAc-GD1a antibodies above the cut-off threshold. Albeit slightly higher in frequency, this observation corresponds to our and other previous observations in neurologically normal dogs and humans, in which AGAbs are also occasionally found. 14,32 Many AGAbs exist in the normal natural antibody repertoire, notably but not restricted to anti-GA1 and -sulphatide antibodies as reported herein for cats. It is therefore imperative to carefully consider assay thresholds, choices of antigens and disease control groups when conducting studies on AGAbs.
The one ONM-cat that exhibited both anti-GM2 and anti-Gal-NAc-GD1a AGAbs was diagnosed with myasthenia gravis, but it remains possible that this case may have been affected by an alternative or concurrent IPN, particularly given the normal acetylcholine receptor antibody titre.

| CONCLUSION
This study indicates that in cats clinically diagnosed with immunemediated peripheral neuropathies, anti-GM2 and anti-GalNAc-GD1a IgG antibodies, may potentially serve as useful biomarkers for disease, with anti-GM2 AGAbs being considered the most favourable and simple antibody test to conduct.
Considering the histological and electrophysiological findings reported for cats with immune-mediated peripheral neuropathies, it seems likely that, as is the case in dogs and humans, anti-GM2 and anti-GalNAc-GD1a AGAbs are a contributing cause of AGAbmediated demyelinating neuropathy on which further pathophysiological studies are warranted.