Pediatric Anti-NMDAR Encephalitis in Southern China: Analysis of 107 Cases

Background: Research on pediatric anti-N-methyl-D-aspartate receptor (NMDAR) encephalitis is limited. We studied clinical features of children with anti-NMDAR encephalitis in southern China. Methods: Pediatric anti-NMDAR encephalitis between 2014 and 2019 from one tertiary medical center was analyzed. The Modied Rankin Scale (mRS) score was adopted to evaluate outcomes. Results: 107 children (M/F=50/57; mean onset age=6.3 y) with anti-NMDAR encephalitis were involved. Boys signicantly showed earlier onset age, higher ratio of fever, longer hospital day, more courses of steroid treatment, higher mRS score after treatment. 48.6% of patients had prodromal events with infectious events most common. The most common symptom at onset and during whole course was seizures and psychiatric symptoms respectively. Initial median white cell count of cerebral spinal uid (CSF) was 22.0x10 6 /L. 12.1% of patients had positive herpes simplex virus DNA of CSF. All patients had CSF positive NMDAR antibodies, although 86.9% of patients had serologically positive. 9.3% of patients had overlapped other neuronal antibodies. Electroencephalograph showed abnormalities with slow wave (95.3%) and epileptic activity (41.5%). Delta brush was revealed in 3.8% of patients. 36.4% of patients had lesions in brain MRI, with local lesions most common. No tumor was found in all. 84.9% of patients responded well to the rst line therapy (steroid plus immunoglobulin), while 12 patients accepted second-line therapy (Rituximab) with 91.7% of effective rate. 12.1% of patients relapsed. 2 male patients died. The median length of hospital stay was 28 days. The mRS score was signicantly improved after treatment. 51.0% of patients had a full recovery. 76.6% of patients had mild neurological disability (mRS ≤ 2). Male, speech disorder, initial score of mRS and administration of second-line therapy were independent risk factors associated with the outcome.

resemble those in NMDAR hypofunction caused by NMDAR antagonists. NMDAR hypofunction caused by antibody mediated NMDAR internalized is a popular hypothesis for the mechanism of anti-NMDAR encephalitis [1,6]. But this hypothesis is insu cient to explain some core symptoms like seizures [1,6].
Besides, the clinical features of anti-NMDAR encephalitis vary in different ages, ethnicities and regions. Reginal studies of anti-NMDAR encephalitis may promote the research on its mechanism and enrich the clinical phenotypes. However, the regional studies of local spectrum especially focusing on the pediatric population with large sample size is still considered limited.
As our hospital is the reginal tertiary center for children's medicine in south China, we reported 107 pediatric anti-NMDAR encephalitis patients from southern China, focusing on the clinical features of pediatric anti-NMDAR encephalitis including demographics features, prodromal events, clinical manifestations, laboratory investigations, neuroelectrophysiological and neuroimaging examinations, treatments, therapy outcomes and prognosis.

Patients
This was a retrospective study of children diagnosed with anti-NMDAR encephalitis from October, 2014 to October, 2019 in the department of neurology of Guangzhou Women and Children's Medical Center. This study was approved by the Ethics Committee of Guangzhou Women and Children Medical Center (2019052419364384).

Disease Severity Evaluation
Neurological disability was assessed by modi ed Rankin Scale (mRS) [8,7,9] at the time before immunothrapy di ned as initial mRS, after immunotherapy and when discharged and at the end of follow-up. No obvious or mild neurological disability was de ned as mRS score ≤ 2. Severe neurological disability was di ned as mRS score ≥ 4 and ≤ 6. A poor response was de ned as no improvement in the mRS score or as mRS score ≥ 4 for 4 weeks. Clinical improvement was de ned as a decrease in mRS score ≥ 1 compared with the previous visit. Long-term good prognosis was de ned as mRS score ≤ 2 and poor prognosis was de ned as mRS score > 2.

Inclusion Criteria
Patients aged younger than 18 years old, diagosed with anti-NMDAR encephalitis according to diagnostic criteria proposed by Graus et al [5] and excluding primary schizophrenia and mental behavior abnormalities secondary to intracranial infection, drugs, poisoning, brain trauma, genetic and metabolic diseases and psychological diseases. Relapse was de ned as the new onset or worsening of symptoms occurring after at least two months of improvement or stabilization [10].
Antibodies Test NMDAR IgG, and gamma-aminobutyric acid type B IgG in both serum and cerebrospinal uid (CSF) were determined by cell-based assay (EUROIMMUN, Lübeck, Germany). CSF glial brillary acidic protein (GFAP) IgG and myelin oligodendrocyte glycoprotein (MOG) IgG and aquaporin-4 (AQP4) IgG in serum were determined by cell-based assay. These assay methods have been reported in detail in our previous study [11].

Peripheral blood test results
All patients underwent blood routine test, blood gas and electrolyte examination, liver and renal function test, erythrocyte sedimentation rate and blood C-reactive protein (CRP) test when they were rst admitted to our hospital. The results showed the median white blood cells count was 9.4 × 10 9 /L (IQR7.3-12.7 ×

Neuroelectrophysiological Examination
Nearly all patients underwent EEG, visual evoked potential (VEP) and brainstem auditory evoked potential (BAEP) examination. 106 patients underwent 167 times EEG examination during acute stage phrase.
95.3% (101/106) of patients had slow wave and 41.5% (44/106) had epileptic activity in EEG. Among 167 times EEG examination, normal background was seen in 7.8% (13/167), epilepsy discharge was seen in 32.3% (54/167) including 16 episodes of clinical seizures (10 episodes of focal origin, 2 episodes of focal status epilepticus, and 2 episodes of generalized tonic-clonic status epilepticus), 4 episodes of electric seizures (2 out of 4 episodes with non-convulsive status epilepticus) and 34 episodes of interictal epileptiform discharges and Delta brush was seen in 3.8% (4/106). 13 cases (12.1%, 13/107) showed the bilateral latency of P100 wave of VEP was prolonged and 13 cases (12.1%, 13/107) showed the latency of III and V wave of BAEP in unilateral or bilateral was prolonged at acute phase. The abnormal outcomes either from VEP or from BAEP were restored in the following recovery period.

Imaging Examination
All patients underwent brain MRI examination in acute phase and non-speci c changes was found. 36.4% (39/107) of patients had lesions in brain MRI. 43.6% (17/39) of patients had local lesions in brain MRI, while 30.8% (12/39) of patients had multiple lesions including 7 cases of multiple demyelination lesions in white matter. And 5 cases among these 7 cases had MOG antibody positive. Patients having MOG antibody positive were reported in our previous study [11]. 10 Table 2 and Table 3. On univariate analysis, the factors associated with poor outcome were aged 3 years or younger, male, with prodromal event, consciousness disturbance, speech disorder, autonomic dysfunction, brain MRI changes, duration from onset to diagnosis longer than 14 days, initial mRS higher than 3 and RTX treatment ( Table 2). On multivariate analysis, male, speech disorder, RTX treatment and initial mRS were independent risk factors associated with outcome (Table 3). While consciousness disturbance, anti-NMDAR antibody titer and relapse were not associated with outcome (Table 2).

Clinical Comparisons Between Different Genders Of Patients
Besides onset age of boys was younger than girls mentioned above, we found differences of clinical features between them (summarized in Table 1). The ratio of fever in boys was higher than girls (58.0%, 29/50 vs 38.6%, 22/57, P = 0.045, Table 1). The median mRS after treatment of boys was 2 (IQR 1-2, ranging from 1 to 6, Table 1) which was higher than that of girls (2; IQR 1-2, ranging from 1 to 4, P = 0.036, Table 1). 82.0% (41/50) of boys needed 1 course of steroid treatment and 18.0% (9/50) needed 2 courses of steroid treatment. While for girls, 94.7% (54/57) of girls needed 1 course of steroid treatment and 5.3% (3/57) needed 2 courses of steroid treatment. Boys were more likely to need more course of steroid treatment (P = 0.037, Table 1). And the length of hospital day in boys was 28 days (IQR 22-39 days, ranging from 2 to 89 days) which was slightly longer than that of girls (26 days) (IQR 22-31 days, ranging from 12 to 59 days, P = 0.036, Table 1). However, there was no statistical difference of ratio of symptoms including consciousness disturbance, movement disorder, limb paralysis, ataxia, seizures, speech disorder, psychiatric symptoms, autonomic dysfunction, sleep disorder and memory disorder, laboratory ndings including CSF pleocytosis, protein increase and anti-NDMAR titer, abnormal outcomes of MRI, EEG with epilepsy discharge, initial mRS, RTX treatment, IVIG course and relapse between boys and girls (Table 1).
In total, boys presented different clinical features partially. Boys were more likely to occur onset at younger age and have fever, longer hospital day, higher mRS score after treatment and needed more course of steroid treatment compared with girls.

Discussion
Anti-NMDAR encephalitis is the most common type of autoimmune encephalitis [3].Some clinical features are different between children and adult with anti-NMDAR encephalitis. And they may vary from regions and racial. Besides most published papers focused on adult patients. Our study focused on children with anti-NMDAR encephalitis and had relatively large cases in southern China to provide more practical information and guidance.
There were some differences in demographic data of patients comparing with the previous studies.
According to previous studies, pediatric anti-NMDAR encephalitis in China and other countries was more common in children aged around 8-12 years [3,[12][13][14][15][16]. While in our study, the mean age at onset was 6.3 ± 3.1 years old which was slightly younger (See Additional le 1: Table S1). In addition, the onset age of boys in our study was younger than that of girls, which was different from the outcomes of UK and Thailand studies [13,12]. These differences suggested that the demographics feature of age might range from regions and nations. The gender ratio in our study was 50:57 (male: female), which was similar to the outcomes of previous studies in Asia including studies from different regions of China in which the ratio of female ranged from 50-70% (See Additional le 1: Table S1) [17-19, 13, 15, 16, 14, 20]. While in studies from western country female was moderately predominate (higher than 70%) [10]. And the gender ratio was also various in different age, with the age increase, proportion of female patient increase [14,20,15]. One reason for the predominance of female is teratomas which are more common in female more than 14 years old [10]. Therefore, we presumed the dissimilarity of gender ratio in our patients was likely to racial, region and age differences.
The initial symptoms were various in the different age. Previous studies found that in adult and children older than 12 years old, psychiatric symptoms were the most common initial presentation [17,10,15,14].
However, in children younger than 12 years old, the proportion of psychiatric symptoms at initial symptoms was nearly equal to that of seizures, but more common than that of movement disorder [10,15,21]. It remained unclear why there existed age differences on the initial symptoms of anti-NMDAR encephalitis. In general speaking, the manifestation of anti-NMDAR encephalitis associated with NMDAR dysfunction in neurobiology. One hypothesis for explaining the anti-NMDAR clinical features is that psychiatric symptom which is more common in adults is caused by NMDAR hypofunction resulted from antibody mediated NMDAR internalized, while neurological symptoms like seizures and movement disorders as initial symptoms are caused by agonist effect of NMDAR, extrasynaptic NMDAR hyperfunction or neuronal network imbalance with impaired intraneuronal activity [6,1]. Extrasynaptic NMDAR is more commonly in GluN1/2B, while synaptic NMDAR is more commonly in GluN1/2A [6]. And during development from juvenile to adult, NMDAR subunit switch from GluN2B to GluN2A [22]. In future, more researches about the NMDAR function network in physiology, disease during development need to do in order to elucidate the pathological mechanism of anti-NMDAR encephalitis.
The EEG abnormalities were not speci c, and the proportion of EEG abnormalities was similar to other studies [23,24,8,17]. Delta brush was relatively rare among the EEG examinations in current study, if compared to the EEG outcomes of Zhang et al study from north of China [16] (See Additional le 1: Table  S1), which might be related to the time of EEG monitoring for patients. Though the MRI change of anti-NMDAR was non-speci c, we found some abnormalities suggesting an association with other autoantibodies, so called "antibody overlapping syndrome". The outcomes of MRI scan in 7 cases in our previous studies showed the abnormal of multiple demyelination lesions, among which 5 cases having MOG antibody positive [11]. And among other 3 cases with MRI abnormalities of leptomeninges enhancement, one case had GFAP antibody positive.
Some patients with anti-NMDAR encephalitis have prodromal events. 48.6% of anti-NMDAR encephalitis patients in our study had prodromal events. 92.9% of these patients with prodromal events of intracranial infection was HSV encephalitis. Anti-NMDAR encephalitis is a major constitution of autoimmune encephalitis after HSV encephalitis and patients aged 4 years or younger are more likely to develop to autoimmune encephalitis after HSV encephalitis and develop movement disorder and worse condition, whereas older children develop behavioral and psychiatric manifestations predominated [9,25,15,10]. The pathophysiology of anti-NMDAR encephalitis after HSV encephalitis is still unclear. In vitro model of the blood-brain barrier, HSV can reduce the cell-cell barrier resistance lead to a viability loss of the infected endothelial cells [26]. Blood-brain barrier disruption provides a convenient for entry of complement and other pro-in ammatory molecules. And complement may be involved in the pathological process of anti-NMDAR encephalitis and cause neuronal degeneration which is different from the other anti-NMDAR encephalitis [1,27]. That's a hypothesis for why young children with anti-NMDAR encephalitis after HSV encephalitis are worse than others with classical encephalitis. Analysis of the prodromal events is helpful for further mechanism study of anti-NMDAR encephalitis.
The relapse rate of anti-NMDAR encephalitis ranges from 0.0-25% [15,10,20,16,14] (See Additional le 1: Table S1). And it was 12.1% in our study. The relapse rate may be associated with duration of followup. The fatality rate of anti-NMDAR encephalitis ranges from 0-12.5% [10,28,29,20,16,14,15] (See Additional le 1: Table S1). While it was 1.8% in current study, that was similar to studies from other region of China [20,16,14,15], which ranged from 0.0-2.3% (See Additional le 1: Table S1), being relatively low. A systematic review of reported cases which involved studies of pediatric and adult anti-NMDAR encephalitis showed that patients who died had an older onset age than patients who survived, and pediatric patients had a better outcome than the adult [17]. This might be an explanation for the lower fatality rate in our study and studies form other regions of China.
The ratio of patients with good prognosis in our study was 76.6% which similar to other study [30,16,10,14,12]. The predictors of poor outcome in previous study including decrease consciousness, speech disorder, longer hospital day, higher initial mRS, and treatment with RTX were consistent with the outcomes of our study [30,16,10,14,12,31]. Byrne et al found the patients who recovered completely had shorter interval (15 days) between onset and initiation of treatment than that of those who did not recovered completely (21 days) [32]. Similarly, we found a longer duration (more than 14 days) served as a predictor of poor outcome, in consistent with the general value of "early treatment with good outcome".
Male, lesions in brain MRI and autonomic dysfunction served as predictors of poor outcome in our study, although previous study did not [30,16,10,14,12,31]. We found patients aged 3 years or younger was a risk factor for outcome. The previous study involved both pediatric and adult anti-NMDAR encephalitis, found the younger age was a predictor of poor outcome [20], although the study only concentrating on pediatric patients did not found [16,14,12]. In our study, we found prodromal event was also associated with poor outcome that is rarely analyzed in previous studies. Prodromal events could play an important role in epigenetic change in patients. For now, epigenetic mechanisms for NMDAR receptor hypofunction in schizophrenia provided new insights into understanding of its neuropathophysiology and early interventions [22].
Besides boys occurred onset at younger age, we also found boys were more likely to have fever, longer hospital day, higher mRS score after treatment and needed more course of steroid treatment compared with girls. And male was an independent risk factor of poor outcome.
While no difference of initial mRS, level of CSF anti-NDMAR titer, ratio of abnormal MRI, initial mRS, RTX treatment and relapse between boys and girls was found. And these difference between boys and girls in our study might suggest though no difference of severity at initial onset, boys might more likely recover slower and have poor outcomes than girls. One study from UK which involved 31 pediatric patients with anti-NMDAR encephalitis also found partial phenotypes were different between girls and boys and boys were more likely to present predominant movement [13]. But in contrast to one study from Thailand which only involved 14 pediatric patients with anti-NMDAR encephalitis found that there was no difference of duration hospitalization between boys and girls and the ratio of fever in boys was tend to be higher than girls but this difference didn't have statistical signi cance and as a result, male was not a risk factor of poor outcome [12]. The difference of results between ours and other study might be caused by the sample size. And the sample size in our study was relatively large.  Distribution of mRS scores at initial, after treatment and at last follow-up.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.