Differences and diversity of autoimmune-mediated encephalitis in 77 cases from a single tertiary care center

Background The incidence of autoimmune encephalitis has risen globally. There are two general categories of disease-associated antibodies that can be tested for: neuronal surface and intracellular. However, testing both groups of autoantibodies are costly. This study aims to identify differences between groups by comparing clinical presentations, radiological findings and CSF profile of patients, and determine if any parameters are indicative of one group of autoantibodies over another. Additionally, we aim to report the local incidence of less common groups of disease-associated antibodies as well. Methods 77 records of autoimmune encephalitis/encephalomyelitis patients admitted to between October 2010 and February 2017 were reviewed. Patients with infections or those with classic central nervous system demyelinating features were excluded. Results Of 77 patients, 40% presented with neuronal surface antibodies and 33% had intracellular antibodies. The most common autoantibody detected in each group was anti-NMDAr antibody (25/31, 81%) and anti-Ri antibody (7/25, 28%) respectively. In the neuronal surface antibody group, behavioral change was the most common complaint (45%), followed by seizures (39%) and abnormal movements (29%). In the latter group, seizure was the most common presenting symptom (32%), followed by motor weakness (20%), behavioural change (16%) and abnormal movements (16%). Patients with neuronal surface antibodies were younger (35 vs 48 years old, p=0.04) and more likely to present with behavioral change (45% vs 16%, p=0.02). Mortality rate was higher in the intracellular group (16% vs 3.2%, p=0.09). No differences were detected in magnetic resonance imaging (MRI) and CSF profile. Conclusions In the early stages of the disease, both groups have comparable clinical outcomes. Although there were significant differences in age and percentage of patients with behavioral change, both groups of autoimmune encephalitis still shared many clinical

features and could not be distinguished based on MRI and CSF profiles. Therefore, we recommend that patients with features of autoimmune encephalitis should be screened for both the neuronal surface and intracellular antibodies regardless of clinical presentation.

Background
Autoimmunity is a major cause of encephalitis. It is now as common as infectious etiology.
Antibodies responsible for autoimmune encephalitis are broadly classified into two categories: neuronal surface and intracellular antibodies (16). The neuronal surface group comprises of antibodies to surface receptors and protein complexes such as NMDAr, AMPAr, CASPR2, LGI-1 and GABAr (17,18). On the other hand, the intracellular group (classic paraneoplastic antibodies), consists of antibodies against intracellular antigens (e.g., Hu, Ri, Yo and CV2) (19).
Several methods can be used to identify these autoantibodies in the serum and cerebrospinal fluid (CSF). In autoimmune encephalitis, techniques such as immunoblotting, immunohistochemistry, and immunocytochemistry with immunofluorescence assays are most commonly used (20). However, these assays are time-consuming and costly especially if several panels of autoantibodies are tested.
Therefore, by comparing the clinical data, radiological results and CSF profiles of patients, we hope to identify key differences that will determine which set of autoantibodies patients are likely to have in order to eliminate unnecessary tests. This would free up time and resources that can be spent on other aspects of healthcare. Additionally, the results could be applicable to other developing countries who could benefit from this cost-saving strategy. Furthermore, this study also aims to report the local incidence and clinical presentations of other less common disease-associated antibodies groups.

Methods
Participants 77 records of patients with autoimmune encephalitis/encephalomyelitis admitted to KCMH between October 2010 and February 2017 were reviewed.

Selection Criteria
Patients with infectious encephalitis and those with classical demyelinating features were excluded from this study. Patients were included if they met the criteria for encephalitis, which is one or more of the following presentations (21): cognitive disturbance, behavioral change, focal neurological abnormalities and seizures plus the detection of well-defined antibodies against cell-surface, synaptic, or intracellular proteins in serum or CSF.
According to the criteria proposed by Graus et al., antibody negative autoimmune encephalitis was diagnosed in patients who had rapid progression of working memory deficits, altered mental status, or psychiatric symptoms and absence of well-characterized autoantibodies in serum and CSF (22). The diagnosis of neuro-psychiatric lupus erythematosus (NPLE) was initially determined by clinical assessment, followed by positive results of either ANA, anti-ribosomal P, anti-cardiolipin or lupus anticoagulant in serum (23). The diagnosis of Hashimoto's encephalitis comprised of encephalopathy with seizures, myoclonus, hallucinations or stroke-like episodes and subclinical or mild overt thyroid disease with positive serum thyroid antibodies (22). We used the descriptions of psychopathological features according to Al-Diwani et al. (24) in our study. "Behavior" included agitation, aggression, disorganisation, incoherent speech, violence, incongruent laughter/crying, disinhibition, hyper-religiosity, impulsivity and talking to self. "Catatonia" referred to mutism, stupor, verbigeration and waxy flexibility.

Investigations
Routine laboratory results were reviewed. These included complete blood count, blood urea nitrogen, creatinine, serum electrolytes, HIV antibody, chest x-ray and liver function tests. Blood cultures were performed in every febrile case. Serum and/or CSF were also After collection, the sample was diluted 10-fold and incubated for 30 minutes at room temperature on slides containing either specific antigen-expressing HEK cells or tissues that naturally expressed these antigens (e.g., cerebellum, pancreas, intestine and nerve cell). They were then incubated in secondary antibody conjugated with fit-C fluorescence and observed under an inverted fluorescence microscope. As for the immunoblot assay (EUROLINE®), specimen was diluted accordingly (1:101 for anti-neuronal) put on the test strip and incubated for 30-120 minutes. After a wash, the strip was incubated with conjugated enzyme and substrate before being evaluated by EUROLineScan®. The IIF assay for aquaporin-4 (AQP4) antibody, anti-thyroid peroxidase and anti-thyroglobulin was also conducted in some cases based on clinical suspicion.
All patients were screened for tumor by computed tomography (CT) of chest and whole abdomen including pelvis. Pelvic examination and ultrasonography were performed in female patients with anti-NMDAr and anti-Yo. Mammogram was also indicated in some cases based on clinical judgement. Modified Rankin Scale (mRS) was used to evaluate patients before discharge.

Statistics
SPSS software version 17 was used to describe and compare epidemiological data between neuronal surface and intracellular groups. Categorical data were analyzed using chisquare and Fisher's exact test. Mann-Whitney U test was performed to analyze numerical data. P-value of less than 0.05 was considered statistically significant. atypical AQP4-IgG, which presented with encephalitis syndrome (Figure 1). The prevalence and characteristics of autoantibodies from each group of autoimmune encephalitis are described in Figure 1 and Table 1. The most common abnormality in brain MRI was increased signal intensity in T2-weighted image at the temporal lobe (3 cases, 11.1%). Other findings were leptomeningeal enhancement, subcortical, basal ganglia and multifocal lesions. In the neuronal surface antibody group, three cases of tumor (9.7%) were found on admission or during the first follow-up (Table 1).

Results
14 patients (45.2%) received immunotherapy (IVIg 2g/kg/course or plasmapheresis 5 cycles) and 18 patients (58.1%) were given intravenous methyl prednisolone 1g/kg/day for 5 days. Tumors were appropriately treated upon discovery. Median length of stay in the hospital was 24 days (ranged from 4-150 days). Seven patients (22.6%) had poor outcome at discharge from hospital (mRS score 4-5) and one died from hospital acquired infection.

Intracellular Antibody Group
Anti-Ri was the most common autoantibody detected in the intracellular antibody group As seen in Table 2, seizure was the most common presenting symptom (8 cases, 32.

Systemic Autoimmune Encephalitis Group
There were nine cases in the systemic autoimmune encephalitis group with six resulted from neuropsychiatric lupus erythematosus (NPLE

Comparison Between Local Population and Other Studies
This study described the clinical, radiological and laboratory findings in patients with autoimmune encephalitis diagnosed and treated at a tertiary care hospital in Bangkok, Thailand. Antibodies against neuronal surface antigen were the most common followed by those against intracellular antigen (46% and 38%, respectively). The most common autoimmune encephalitis in this study was anti-NMDAr encephalitis. This is consistent with other studies, which report that anti-NMDAr encephalitis is the most prevalent immune-mediated encephalitis worldwide to date (25,26). Some reports found that anti-LGI1 was presumed to be the second most common cause of autoimmune encephalitis among the neuronal surface antibodies group (12,27). However, we only identified two cases of anti- LGI1 encephalitis among 31 cases (6%). This is likely because LGI1 antibodies were more common among elderly patients with antibody-associated central nervous system (CNS) syndrome (28,29), while the majority of patients in our study were middle-aged (mean age of 40). Both cases of anti-LGI1 encephalitis in our study had FBDS that responded to immunotherapy rather than anti-epileptic drugs, which is similarly reported by Thompson et al. (30). Additionally, these two patients were more than 50 years old and only had CNS manifestations-a finding that is consistent with Gadoth et al. (31).
Among the two cases of anti-AMPAr antibody encephalitis, one had SCLC, which was the most commonly associated tumor (32). The other patient, a 2-year-old girl with no malignancy detected, presented with a stiff-person syndrome. She eventually recovered after receiving baclofen and supportive treatment. Her serum anti-AMPAr antibody was subsequently negative at a 3-month-follow up.
There are some differences in this study and previously published literature. For instance, in the neuronal surface group, only 26% of our cohort had presented with prodromal symptoms while some studies have suggested that this figure could be as high as 61-70% (25,33,34). Additionally, CSF pleocytosis was detected in 37% of our patients compared to 80% in a previous review (33).
There was a strong association between anti-NMDAr encephalitis and ovarian tumor/teratoma (1,35,36), especially among Asian and African-American women where 50% of adult female patients were estimated to have ovarian teratoma (37). However, in our study, only two cases (9%) of ovarian tumor were found at the first immediate tumor screening despite a thorough investigation with abdominal and pelvic CT and pelvic ultrasonography. Interestingly, this is consistent with anti-NMDAr encephalitis case series in China (25) and Korea (38), which found that ovarian tumor was present in four (8%) and two (9%) patients respectively. However, since paraneoplastic neurological syndromes can present within five years of the initial diagnosis (39), annual tumor screening should be performed in the remaining cases. Furthermore, oro-facial dyskinesia was the most common movement disorder among anti-NMDAr encephalitis patients in our study, which is consistent with previous studies (26,35,40). However, we did not find dystonia as common as suggested by Varley et al.(41) and there was only one case of catatonia in our study.
There were three patients (4%) that fit the criteria of autoantibody negative autoimmune encephalitis. This raises a question of whether there is truly no antibody or if there is an unknown neuronal antibody that we could not detect with the current tests used in our study.
Finally, we discovered that some patients had non-classical paraneoplastic autoantibodies such as anti-recoverin, anti-titin and anti-SOX1, which were usually not considered as the cause of paraneoplastic CNS disease. Anti-recoverin is one of the causes of cancerassociated retinopathy (42). Here, we found three cases of anti-recoverin associated encephalitis that presented with behavioral symptoms. Anti-titin and anti-SOX1 often present with other classical paraneoplastic antibodies (anti-Hu, Ri, Yo etc.) in paraneoplastic syndrome of the nervous system (43,44). Since no other coexisting autoantibodies were identified in our cases, these antibodies may be directly responsible for the neurological symptoms seen.

Comparison Between Neuronal Surface and Intracellular Groups
Since the majority of autoimmune encephalitis patients in our study belonged to either the neuronal surface group or the intracellular group, the differences between these two groups were summarized in Table 2.
Patients in the intracellular group were significantly older than those in the neuronal surface group (48 vs. 35 years old, p = 0.04). Furthermore, behavioral change, which was the third most common symptom in the intracellular group, was significantly more common in the neuronal surface group as well (45% vs. 16%, p = 0.02). On the other hand, seizure was the most common complaint in the intracellular group, but there were no significant differences in its prevalence between the two groups (32% vs. 39%, p = 0.60). Abnormal movements and psychiatric features were more common in neuronal surface group, but similarly there were no significant differences (p = 0.24 and 0.30 respectively). Additionally, the intracellular group tended to have a longer duration of symptoms before admission to hospital (3 months vs 2 months).
As for CSF profile, the intracellular group had a large proportion of normocellular CSF (76%). Neuroimaging (MRI) data could not distinguish between these two groups of autoantibody-mediated encephalitis.
The recovery rate and morbidity at discharge from hospital between both groups were comparable. The mortality rate was noticeably higher in the intracellular group although this was not statistically significant (16% vs. 3.2%, p = 0.16). Therefore, it seems that at least in the early stages of the disease, patients with surface and intracellular antibodies have similar outcome. However, further studies with more participants and higher power are needed to confirm this result.

Conclusion
We described different groups of autoimmune encephalitis patients in Thailand. The two most prevalent groups were neuronal surface and intracellular groups, with the former being more common than the latter. However, in the earlier stages of the disease, both groups have comparable clinical outcomes. Even though the age of onset and percentage of patients with behavioral change were statistically different between these two groups, they still shared many clinical features and could not be distinguished based on radiological findings and CSF profiles. As such, we recommend that patients with suspected autoimmune encephalitis should be screened for both the neuronal surface and intracellular antibodies.

Consent for publication: Not Applicable
Availability of data and material: Data used and/or analyzed in this study are available from the corresponding author on reasonable request.
Competing interests: The authors declare that they have no competing interests.  Proportion of each groups of autoimmune encephalitis