Complex partial seizures with orofacial involvement in 35 cats: MRI changes,cerebrospinal fluid analysis,voltage-gated potassium channel antibodies and survival

Abstract

Objectives

The objective of this study was to evaluate the occurrence of voltage-gated potassium channel (VGKC) antibodies and the pattern of MRI changes in cats with complex partial seizures with orofacial involvement (CPSOFI), as well as to investigate whether there are factors influencing survival that could be used as prognostic markers in those cats.

Methods

Cats with CPSOFI were identified retrospectively. The following data were retrieved from the hospital database: signalment, age at first seizure and presentation, the presence of antibodies against VGKC (leucine-rich glioma inactivating factor 1 (LGI1), contactin-associated protein 2 (CASPR2)) and cerebrospinal fluid (CSF) analysis findings. Brain MRI scans were evaluated for T2 and/or fluid-attenuated inversion recovery (FLAIR) hippocampal hyperintensity and hippocampal contrast enhancement by a radiologist. The correlation between being positive for VGKC antibodies and MRI hippocampal hyperintensity was investigated.

Results

Serum antibodies against VGKC were found in 31.4% of cats, with 10/11 positive cats having antibodies against LGI1 and 1/11 against CASPR2. MRI T2/FLAIR hippocampal hyperintensities and T1 contrast enhancement were seen in 57.1% and 27.1% of cats, respectively. There was a weak correlation between those antibodies and MRI changes (phi coefficient 0.12). The following factors did not influence survival on multivariable regression analysis either for all cats together or when the cats that died were evaluated separately: age at first seizure, breed, sex, T2/FLAIR hippocampal hyperintensity, being positive for VGKC antibodies and prednisolone treatment.

Conclusions and relevance

There are no consistent findings on MRI or on testing for VGKC antibodies in all cats with CPSOFI. None of the factors investigated can be used to predict outcomes.

Plain language summary

Seizure with facial twitches in 35 cats: findings on brain imaging, brain fluid analysis, autoantibodies directed against the brain and survival

Cats are sometimes affected by seizures that present with facial twitches and salivation. The aim of this study was to describe the results of brain imaging and the frequency of antibodies produced by the cat’s own immune system that were directed against the brain, as well as factors influencing the survival of those cats. Affected cats were identified retrospectively and information describing the cat, the way seizures look, results of a brain fluid analysis and detection of antibodies directed against two specific receptors in the brain were collected. Pictures taken of the brain of those cats using MRI were reviewed. Connections between imaging findings and the existence of antibodies were investigated. Antibodies directed against the brain were found in 31.4% of cats. On brain imaging, diffuse fluid accumulation was found in a specific structure deep inside the brain, the hippocampus, in 57.1% of cats. Only a few of those cats with antibodies had the described changes on brain imaging. No factor was identified that influenced the survival time of affected cats. There are no consistent test findings across all cats affected by this specific seizure type. As a result, a potential causative relationship between brain-directed antibodies and imaging findings remains to be clarified.

Introduction

Complex partial seizures with orofacial involvement (CPSOFI) have been described in cats since 2011.^1,2 However, this was more the introduction of a new subclassification of a known seizure phenomenology rather than the description of an entirely new seizure type, since clinical signs of CPSOFI had been described before.^{3 –5}

During the diagnostic work-up of cats with CPSOFI, hippocampal hyperintensity is a common finding and is often interpreted as hippocampal necrosis, even though definitive proof of this is lacking in surviving cases.^5,6

Hippocampal necrosis in cats with seizures was first described in 1977 by Fankhauser and Fatzer, ⁷ who speculated that these changes might be caused by toxins that were yet to be identified. This phenomenon did not attract further attention until a case series describing the clinical and pathological findings in 38 cats with necrosis of the hippocampus and piriform lobe was published in 2000. ³ The authors concluded that these changes may be the cause of seizures rather than being a change secondary to epileptic seizures. In the following years, several other reports mention similar findings in the hippocampus and piriform lobe.^{2,4,5,8 –11} Such reports further fuelled the debate regarding hippocampal and piriform lobe necrosis and whether these changes are the cause or the consequence of epileptic seizures in cats, especially when two cats were described in 2015 as having hippocampal and piriform lobe necrosis presumably caused by severe cluster seizures. ¹²

However, it has become increasingly accepted that hippocampal necrosis, independent from the underlying pathology, might be causing seizures rather than being the consequence of them.^8,13

Several cases of CPSOFI have been traced back to hippocampal necrosis, but notably, the severity of MRI changes was not related to clinical outcome. ¹ However, the MRI scores for hippocampal changes were higher in cats with orofacial involvement than those without. ⁶ More recently, hippocampal changes were linked to limbic encephalitis in cats with antibodies against voltage-gated potassium channels (VGKC), specifically against contactin-associated protein-like 2 (CASPR2) and leucine-rich glioma inactivated 1 (LGI1).^{14 –17} However, VGKC antibodies were detected in only 5/14 cats with suspected limbic encephalitis. ¹⁴

To date, it is still up for debate whether the hippocampal changes on MRI are linked to the existence of VGKC antibodies as an indicator of limbic encephalitis in cats with epileptic seizures. Therefore, the aims of this study in cats with CPSOFI were as follows: (1) to determine the prevalence of hippocampal changes on MRI and antibodies against VGKC; and (2) to investigate whether one or a combination of the following parameters can serve as prognostic indicators in cats with CPSOFI: patient signalment, age at first seizure, T2 and/or fluid-attenuated inversion recovery (FLAIR) hippocampal hyperintensity, existence of antibodies against VGKC, findings of cerebrospinal fluid (CSF) analysis and treatment with prednisolone.

Materials and methods

The hospital database was screened retrospectively by two authors (LPS and JEK) for cats with CPSOFI presented between January 2014 and October 2022 that were tested for VGKC antibodies. Cats were considered for inclusion if the following data were available: patient signalment, epileptic seizure description (CPSOFI alone or in combination with generalised tonic–clonic seizures), age at first seizure, brain MRI findings, results of antibody testing against VGKC (LGI1 or CASPR2), results of CSF analysis, data on medical treatment and minimum follow-up time of 3 months.

MRI was performed under general anaesthesia in sternal recumbency using a 3 T MRI (Ingenia 3.0T; Philips Healthcare) and eight-channel or 16-channel Small Extremity Coil (dStream Small Extremity 8ch coil; Philips Healthcare or dStream Small Extremity 16ch coil; Philips Healthcare). In all cats, the imaging protocol consisted of two-dimensional T2-weighted turbo spin-echo sequences (repetition time [TR] 2795–4751 ms, echo time [TE] 90–100 ms, slice thickness 1.8–2.0 mm, no gap), two-dimensional FLAIR (TR 11000 ms, TE 125–128 ms, inversion time [TI] 2800 ms, slice thickness 1.7–2.0, no gap) and pre- and postcontrast (0.1 mmol/kg IV, Dotarem; gadoteric acid 0.5 mmol/ml; Guerbet) two-dimensional T1-weighted fast field echo sequences (TR 223–499 ms, TE 2.5–20 ms, slice thickness 1.7–2.5 mm, no gap) in the transverse plane.

MRI scans were re-evaluated by a radiologist in training (MF). The observer was asked to determine if there was any T2 and/or FLAIR hyperintensity of the hippocampus and if there was any contrast enhancement of the hippocampus on T1-weighed images, regardless of symmetry. Both criteria were subjectively graded as present, maybe present or not present. For both hyperintensity and contrast enhancement, the observer had to determine whether that finding was unilateral or bilateral. The observer was not aware of the interpretation of the images in the initial report. In addition, the observer was asked to report any other regions of hyperintensity on T2 or FLAIR images.

CSF analysis was performed after suboccipital puncture using the same anaesthesia as for MRI. A nucleated cell count below 6 cells/µl and a total protein concentration below 0.25 g/l were considered normal. The nucleated cell count was corrected in cases of more than 500 red blood cells/µl using the following formula: reducing the measured nucleated cell count by one for every 500 red blood cells. LGI1 and CASPR2 IgG antibodies were measured in serum by a commercial laboratory using immunofluorescence tests (Klinisch-immunologisches Labor). The commercial Anti-VGKC-ass.-Proteine-Mosaik 1 (EURIOMMUN Medizinische Labordiagnostik AG) with LGI1 and CASPR2 transinfected cells was used, but the anti-human IgG fluorescein isothiocyanate (FITC) was replaced by an anti-feline IgG FITC. Titres of 1:10 or higher were considered to be positive.

Statistical analysis was performed using the software SPSS Statistics version 27 (IBM). Continuous data were tested for normal distribution using the Shapiro–Wilk test. Data are reported as mean or median with the SD, range or 95% confidence interval (CI), depending on their normality. The correlation between two categorical variables was tested using the phi coefficient. The influence of the following factors on survival time was tested by simple as well as multivariable linear regression analysis: breed, age at first seizure, sex, T2/FLAIR hippocampal hyperintensity, being positive for VGKC antibodies and prednisolone treatment. A value of P ⩽0.05 was considered significant for all analyses.

Results

A total of 35 cats were included, with the following breeds represented: domestic shorthair cat (n = 26), British Shorthair cat (n = 5), Norwegian Forest cat (n = 1), Persian cat (n = 1) and mixed-breed cat (n = 2). The sex distribution was as follows: intact male (n = 5), castrated male (n = 18), intact female (n = 4) and spayed female (n = 8). The mean age at onset of seizures was 4.7 years (range 3 months to 12.3 years). In 20/35 (57.1%) cats, CPSOFI was the only type of seizure seen, whereas 11 (31.4%) cats experienced CPSOFI as well as generalised tonic–clonic seizures. In 4/35 (11.4%) cats, the seizure semiology described in the patient’s file did not allow definite differentiation between seizure types.

MRI was performed at a median of 18 h (range 0.2–264) after the last observed seizure. Out of the 35 cats, 29 (82.8%) presented with MRI abnormalities, including those cases categorised with ‘maybe’ abnormal findings. MRI hyperintensity of the hippocampus was seen on T2 and/or FLAIR images in 20/35 (57.1%) cats, whereas the hippocampus was subjectively judged to be ‘maybe’ hyperintense in a additional nine (25.7%) cats. No hyperintensity was seen in 6/35 (17.1%) cats. Hippocampal hyperintensity was bilateral in 18/20 (90.0%) cats. Contrast enhancement of the hippocampus was seen in 13/35 (37.1%) cats, with unilateral enhancement seen in 1/13 (7.7%). All cats showing hippocampal contrast enhancement also exhibited T2/FLAIR hyperintensities in the hippocampus. Additional T2 and/or FLAIR hyperintensities were noted in the piriform lobe (n = 3), frontal lobe (n = 1) and olfactory bulb (n = 1), and were observed in 14.3% of the total cohort. The results of the MRI findings are summarised in Table 1.

Table 1

Numbers of cats with hippocampal hyperintensity on MRI and detection of voltage-gated potassium channel (VGKC) in 31 cats with different seizure semiologies

Type of seizures	Total number	Hippocampal T2 hyperintensity	Bilateral T2 hippocampal hyperintensity	Hippocampal contrast enhancement	Antibodies againstVGKC
					CASPR2	LGl1
CPSOFI only	20	11	11	5	1	6
CPSOFI plus generalised tonic–clonic seizures	11	7	6	6	0	2

CASPR2 = contactin-associated protein 2; CPSOFI = complex partial seizures with orofacial involvement; LGI1 = leucine-rich glioma inactivated 1

The nucleated cell count in CSF was abnormal in 2/35 (5.7%) cats, with cell counts of 9 and 212/µl. Total protein concentration was elevated in only one cat (0.45 g/l), with an associated nucleated cell count of 212/µl.

Serum antibodies against VGKC were found in 11/35 (31.4%) cats. In total, 10 cats tested positive for LGI1 IgG antibodies, whereas one cat tested positive for CASPR2 IgG antibodies. No cat was positive for both antibodies. Both cats with an elevated nucleated cell count in CSF were negative for LGI1 and CASPR2 antibodies but did have hippocampal hyperintensities on T2 and/or FLAIR images. Of these two cats, one also had bilateral hippocampal contrast enhancement. Of the 11 cats that tested positive for antibodies against at least one VGKC, four (36.4%) exhibited hippocampal hyperintensity. When limiting the analysis to cats in which hippocampal hyperintensity was definitively classified as present or absent (n = 22), there was a weak correlation between T2 and/or FLAIR hyperintensity and VGKC antibody positivity (phi coefficient 0.12). The results of VGKC antibody testing are summarised in Table 1.

Cats were discharged with the following anticonvulsive treatments: phenobarbital only (7/35), levetiracetam only (1/35), phenobarbital plus levetiracetam (11/35), phenobarbital plus diazepam (8/35) and phenobarbital plus levetiracetam plus diazepam (8/35). Of the 35 cats, 26 received additional prednisolone at a mean initial dose of 1.27 mg/kg body weight daily.

In total, 33 cats survived to discharge, whereas two cats were euthanased during hospitalisation, one because of a coexisting hepatopathy and the other because of a cardiac arrest. Of the 33 cats that were discharged, three died or were euthanased within the first month for different reasons. The mean survival time of the remaining 30 cats was 771 ± 646 days (range 79–2372), with 19/30 (63.6%) cats alive at the time of writing the manuscript; five cats were lost to follow-up. The mean survival time of those cats that survived the first months after diagnosis but were deceased at the time of writing the manuscript was 485 days, whereas the mean survival time of all cats still alive was 868 ± 671 days (range 184–2372). The survival times of all cats are summarised in Figure 1. Prednisolone treatment was the only factor influencing survival time on simple linear regression analysis (P = 0.14). However, multivariable regression analysis revealed that none of the following factors influenced survival – age at first seizure, breed, sex, T2/FLAIR hyperintensity or VGKC antibody positivity – either across all cats or when analysing only those that were deceased.

Figure 1

Kaplan–Meier survival curve of cats with complex partial seizures with orofacial involvement for which follow-up information was available (30/35). In total, 19/35 cats were censored as they were still alive at the time of writing the manuscript

Discussion

Serum antibodies were identified in 31.4% of cats with CPSOFI, nearly the same as the 35.7% described previously.^14,16 In those preceding reports, all positive cats had antibodies against LGI1, and none was positive for antibodies against CASPR2.^14,16 Similarly, in the study presented here, only one cat was positive for antibodies against CASPR2, with the remainder having LGI1 antibodies only.

In another study including 32 cats with VGKC antibodies, 81% were positive for LGI1 but none for CASPR2 antibodies. ¹⁸

In another report describing a single cat with presumed limbic encephalitis, only antibodies against LGI1 were detected. ¹⁷

It has been shown that antibodies binding to VGKC may lead to lymphocytic infiltration, glial cell activation and complement-associated neuronal damage, resulting in hippocampal sclerosis. ¹⁹ In a different report, a cat with histologically confirmed limbic encephalitis LGI1 antibodies but was demonstrated to have antibodies against the netrin-1 receptor. ²⁰ This raises the question of whether there are more unidentified antigens against which autoantibodies may initiate the same pathophysiological process leading to limbic encephalitis in cats.

Only two cats exhibited an increased nucleated cell count on CSF analysis, which is somewhat surprising in an inflammatory brain disease. However, normal median nucleated cell counts have also been found in humans with LGI1- and CASPR2-associated encephalitis. ²¹ Furthermore, normal cell counts have been described in canine immune-mediated encephalitides, such as granulomatous meningoencephalitis and necrotising leucencephalitis.^{21 –23} Still, the question remains unanswered as to whether all cats included here would have inflammatory changes on histopathological examination or whether the normal nucleated cell count correctly reflects a lack of significant inflammation, at least at this time point of the disease. The question can be extended further by asking if we are truly looking at the same disease entity in all these cats.

In addition, T2 and/or FLAIR hyperintensities were found in other brain areas outside the hippocampus in 5/35 cats. These changes could potentially reflect other pathologies, such as postictal oedema or infectious encephalitis. The question about the true nature of these lesions is nearly impossible to answer without regular histopathological examination or by identifying another reliable biomarker that can be found in most cats with immune-mediated limbic encephalitis.

Hyperintensity on T2-weighted and/or FLAIR MRI, not including those with brain imaging categorised as ‘maybe hyperintense’, was seen in 57.1% of cats, with no hyperintensity seen in 17.1% in the study presented here. Another study identified T2 hippocampal hyperintensities in all six cats with complex partial seizures with orofacial involvement, a term used before VGKC antibodies were detected as a cause of limbic encephalitis in cats. ¹ This is in contrast to the findings demonstrated in another study of cats with suspected limbic encephalitis, where the percentage found to have normal MRI scans was 80%. ¹⁸ The same study found similar results on MRI in cats with and without antibodies against LGI1, whereas the study presented here identified a weak correlation between being positive for VGKC antibodies and T2/FLAIR hippocampal hyperintensity (phi coefficient 0.12), which is the more expected finding. The authors of the referenced feline study speculate on the relatively low incidence of hippocampal hyperintensity, noting that in humans with limbic encephalitis, mesotemporal MRI changes are seen in 41% of cases that are LGI1 positive.^1,21 They argue that hippocampal inflammation in the small feline brain may be too subtle to detect using 1.5 T MRI. In the present study, 3 T MRI was employed, which, following that line of reasoning, might be expected to detect more subtle hippocampal changes. Nevertheless, only 40% of cats with VGKC antibodies showed T2 and/or FLAIR hyperintensities on MRI, which is lower than expected. Although a similar percentage has been reported in humans, one might expect a higher percentage if VGKC antibodies truly indicate the presence of limbic encephalitis. ²¹ The reason why this is not consistently reflected on MRI can only be speculated. It is possible that observed T2 and/or FLAIR hippocampal hyperintensities reflect only the more severe or chronic manifestations of the inflammatory process.

The number of cases included in the present study did not allow for statistical analysis of these factors. In addition, these differences may simply reflect the subjectivity of judging MRI hyperintensities, since there is currently a paucity of objective criteria to assess these MRI scans. This complicating factor was highlighted by a study that included cats with and without seizures, in which the variability in assessing hippocampal intensities between different observers was demonstrated. ⁶ In addition, it has been shown that inflammatory diseases are not always identified with standard MRI. It was demonstrated, for example, that 36% of dogs with inflammatory CSF findings have a normal brain MRI examination. ²⁴ Therefore, the lack of T2 hyperintensity should not be equated with the absence of limbic encephalitis.

In the 36.4% of cats with hippocampal hyperintensities, however, adding diffusion-weighted imaging to the MRI evaluation may help establish a specific diagnosis of limbic encephalitis. A recent human study concluded that T2 and/or FLAIR hyperintensity of temporal lobes without diffusion restriction or contrast enhancement supports the diagnosis of LGI1/CASPR2 encephalitis in contrast to other differentials. ²⁵

Some of the seeming discrepancies between the results of different diagnostic tests could be explained by the period of time that has elapsed between the onset of seizures and performing diagnostics. The pathophysiological process of autoantibody-induced seizures may go through different stages, starting with the production of neutralising antibodies. Except for potential postictal oedema, histopathological changes will not be visible in that first phase. Later, acute complement activation with membrane attack complex building may lead to hippocampal necrosis without significant pleocytosis. This may be followed by antibody-dependent cytotoxicity and hippocampal lymphomonocytic pleocytosis. In a final post-necrotic stage, parenchymal atrophy, replacement oedema and gliosis may develop. Therefore, hippocampal T2 and/or FLAIR hyperintensity may not be visible in all stages, and nor does it have to coexist with the detection of VGKC antibodies, depending on the stage of the pathological process.

Equally surprising was the demonstration of cats that were negative for VGKC but exhibited hippocampal hyperintensity. This raises the question of whether those cats have postictal oedema only or whether other antibodies are potentially initiating the development of limbic encephalitides, such as the netrin-1 receptor antibodies already mentioned, or whether there are additional antibodies yet to be discovered. ²⁰

The discrepancy between antibody detection and hippocampal MRI changes may again fuel the discussion about the pathological mechanisms underlying the hippocampal MRI changes. There are cats described where the T2 and/or FLAIR hyperintensities of the hippocampus are associated with necrosis secondary to severe seizure activity, ¹² whereas some other authors speculate that seizures may develop secondary to hippocampal necrosis. ³ Others merely state that hippocampal necrosis is associated with limbic encephalitis in cats, avoiding the discussion of whether the necrosis is the cause or consequence of the seizures and how this relates to the MRI changes. ⁶ However, much of the controversy occurred before antibodies against VGKC were detected in cats. Once these antibodies were identified, the scientific opinion shifted towards hippocampal changes on MRI in cats with CPSOFI representing limbic encephalitis.^14,15,17,18 That still leaves the possibility that hippocampal hyperintensities on MRI may be a manifestation of postictal oedema in some cats, as described for dogs. Even histopathological examination of affected brains may not be able to answer that question in all cases. A small case series describes histopathological changes in four cats with seizures and hippocampal hyperintensities on MRI. ⁵ All three had findings consistent with hippocampal necrosis. However, the detailed histopathological findings were not the same for all three, with only one exhibiting a mild lymphocytic inflammation despite the hippocampal changes on MRI being similar for all three. Therefore, different underlying pathologies may result in the same MRI findings in seizuring cats.

None of the factors evaluated here, including patient signalment, age at first seizure, T2 and/or FLAIR hippocampal hyperintensity, the presence of antibodies against VGKC or treatment with prednisolone, influenced survival time on multivariable analysis. The lack of any influence of the prednisolone treatment is surprising in a presumably immune-mediated disease; indeed, simple linear regression identified an influence, but the influence is lost in combination with other factors.

In contrast to our findings, two factors – the number of seizures before presentation and the severity of MRI hippocampal changes – were associated with outcomes in another study. ¹⁹ Cats that were euthanased had a higher number of seizures and higher MRI scores, representing more severe hippocampal changes. The difference between the two studies regarding the influence of hippocampal MRI changes on the outcome may again reflect the subjectivity in assessing MRI intensities, as discussed above.

Conclusions

There is no consistent diagnostic finding in cats with CPSOFI. Serum antibodies against VGKC, usually against LGI1, were found in only 31.4%, and MRI T2/FLAIR hippocampal hyperintensities were seen in just 57.1% of cats with CPSOFI. There appears to be a weak correlation between those antibodies and MRI changes. However, the coexistence of both is not a consistent finding. In addition, none of the factors tested can be used to predict survival in cats with CPSOFI; we did not identify any factor predicting survival in these cats.