We reviewed the medical records of 399 pediatric patients (≤18 years old) with encephalitis who were admitted to Samsung Medical Center (Seoul, Korea) from December 1998 to March 2011. This study included patients who diagnosed with ANE based on the criteria proposed by Mizuguchi¹⁾, acute encephalopathy following viral disease with rapid deterioration of consciousness, absence of cerebrospinal fluid (CSF) pleocytosis, increase in CSF protein, neuroimaging findings of symmetric brain lesions involving the bilateral thalami, elevated serum aminotrasferase to a variable degree, no increased blood ammonia, and clinical absence of other diseases resembling ANE. We exclude other resembling diseases, suggested by that Mizuguchi¹⁾, such as overwhelming bacterial and viral infections, fulminant hepatitis, Reye syndrome, other toxin induced disease, Leigh encephalopathy and related mitochondrial cytopathies, Wernicke encephalopathy, acute disseminated encephalomyelitis or other types of encephalitis, vasculitis, cerebral infarction, and so on.

In each episode, clinical data were collected based on the retrospective chart review including sex, age, initial clinical manifestations, clinical course, initial laboratory findings, electroencephalography (EEG) findings, causative organism of associated febrile illness, treatment modalities, and outcome.

Neuroimaging was reviewed by a neuroradiologist. The location and extent of signal abnormalities were checked by brain computed tomography (CT) or MRI with fluid-attenuated inversion recovery (FLAIR), T1-, and T2-weighted images. The presence of contrast enhancement in the lesions was also identified. For differentiation of cytotoxic and vasogenic edema of the lesions, diffusion-weighted imaging (DWI) and apparent diffusion coefficient (ADC) map images were used. Vasogenic edema (interstitial edema) usually shows high-signal intensity on DWI with an increased ADC map. However, cytotoxic edema (cellular swelling) expresses high-signal intensity on DWI and low-signal intensity on ADC maps. Detection of hemorrhage in the lesion was aided using T2- and T1-weighted images and gradient-echo imaging in patients with MRI. Neuroimaging findings were analyzed by the distribution, pattern and presence of hemorrhage. We paid particular attention to the analysis of the extent and signal intensity of brain stem lesions, the midbrain, pons, and medulla oblongata, for the identification of prognostic factors. The findings of brain stem involvement were as follows: diffuse broad hypointensity on CT, broad hyperintensity on T2-weighted/FLAIR MRI, diffusion restriction (cytotoxic edema) on DWI/ADC, and hemorrhage. The brain stem lesions were divided into three groups (none, mild, and definite) based on extent and signal-intensity as followed: none defined as absent involvement, mild defined as extent of involvement in below 2 lesions among midbrain, pons, and medulla oblongata and subtle signal change without diffusion restriction and hemorrhage, and definite defined as broad extent more than three or marked signal change with diffusion restriction or hemorrhage.

Ethical approval for this retrospective study was provided by the Institutional Review Board of Samsung Seoul Medical Center in Seoul, Korea (SMC 2011-04-104-002).

Data are presented as median and range. Statistical analysis was performed to identify different factors between death and survival groups with the Fisher exact test. And the predictors of mortality in ANE was estimated using multivariate Cox regression and Mann-Whitney test with PASW Statistics 18.0 (SPSS Inc., Chicago, IL, USA). The statistical significance threshold was set at P<0.05 corrected.

Of the 399 pediatric patients, we identified ten patients (male:female=6:4) with ANE, including a patient with two episodes (Table 1). The median age of onset was 1.5 years (0.4-8.4 years). Family history of ANE was unremarkable in all patients. None had any recent immunization.

The clinical characteristics of the patients and the clinical presentations of viral illness are summarized in Table 1. Prior to the onset of ANE, nine patients showed fever, and all children had symptoms of viral illness; upper respiratory tract infection (n=6), acute gastroenteritis (n=7) and acute pharyngo-tonsillitis (n=2). The median body temperature was 38.5℃ (37.5℃-40℃) during the antecedent febrile illness. The first sign of encephalopathy appeared after a median of four days (1-9 days) from the onset of antecedent infection (Table 2). Altered mental status was observed in all cases: drowsiness (n=10, including recurrent case) and confusion (n=1). Nine patients (90%) had seizures (generalized seizure, n=5; complex partial seizure, n=4). They were managed with various anticonvulsants (phenytoin, n=8; phenobarbital, n=5; topiramate, n=1; carbamazepine, n=1; levetiracetam, n=1). The seizures occurred in the initial phase when the deterioration of consciousness appeared. The seizures were well controlled with antiepileptic drugs without recurrence during hospitalization. Progression of encephalopathy to maximum level of mental status deterioration occurred between two to twelve days after the initial vial illness. The lowest level of consciousness ranged from drowsy to coma (Table 2).

The neuroimaging findings are summarized in Table 3. Eight patients were evaluated with both brain CT and MRI. However, two patients (patients 1 and 9) underwent only brain CT, because they were clinically too unstable to have brain MRI. The abnormalities in brain MRI were most well defined by T2-weighted images and FLAIR sequences (Fig. 1).

Every patient showed symmetric involvement of the thalamus, however, patient 6 showed bilateral but asymmetric (left>right) thalamic involvement (Fig. 2). She had a milder clinical course (confused mental status and intensive care unit [ICU] care for only two days), and completely recovered without neurologic sequelae and no residual lesions on follow-up brain MRI.

Cervical spinal cord lesions were observed in patient 2 (Fig. 3). He had extensive involvement of the brain including the bilateral thalami, basal ganglia, broad brain stem and cerebellum.

Seven cases (63.6%) had brain stem involvement. We analyzed the extent (the midbrain, pons, and medulla oblongata) and signal-intensity of brain stem lesions. The extent and signal-intensity of brain stem lesions were divided into three groups: absent signal change (none, n=4), subtle signal change (mild, n=3), and broad extent and marked signal change (definite, n=4). The findings of brain stem involvement were as follows: diffuse broad hypointensity on CT, broad hyperintensity on T2-weighted/FLAIR MRI, diffusion restriction (cytotoxic edema) on DWI/ADC, and hemorrhage.

All patients underwent EEG during acute encephalopathic periods. Ten cases (10/11, 91%) initially showed abnormal EEG. The EEG revealed diffuse encephalopathic findings characterized by diffuse background slow activity (n=6) or voltage suppression (n=3) on cerebral background rhythm. In addition, regional intermittent slow (n=1) and focal epileptiform activities (n=3) were observed. EEG seizure was recorded in two patients (patient 3 and 10).

Mild elevation of liver transaminases was found in ten cases. The median level of aspartate aminotransferase was 53 U/L (27-209 U/L), and the median level of alanine aminotransferasewas 31 U/L (15-242 U/L). The metabolic tests such as serum amino acid analysis, serum ammonia and lactate/pyruvate level, and urine organic acid analysis were all normal. Lumbar puncture was performed in eight patients (73%); cerebrospinal fluid (CSF) pleocytosis (>5/L) was found in four patients (9-175/µL), and increased CSF protein (>45 mg/dL) was found in five patients (88-246 mg/dL). Pathogens responsible for antecedent infections were identified in 63.6% of cases (7/11) (Table 1).

The clinical course and management during their hospital admission are summarized in Table 3. All patients were admitted to the pediatric ICU, but patient 5 was cared in the general ward for the second episode. Patient 1 received unusual long term ICU care for 230 days after the diagnosis of brain death, because the parents wanted continuous care until cardiac death. The median duration of ICU stay in nine patients except patient 1 was eight days, from 1 to 15 days. Until the patients were diagnosed as having ANE with brain imaging and were ruled out of other possible causes, we provided treatment for possible viral encephalitis or bacterial meningitis such as vancomycin, third-generation cephalosporin, or acyclovir. We also prescribed oseltamivir to patients who had positive results for Influenza A or B. Intravenous immunoglobulin (IVIG, total 2 g/kg for 2-5 days) and high-dose steroids (methylprednisolone 15-30 mg/kg/day for 3 or 5 days) were used to treat ANE: both intravenous IVIG and high-dose steroid therapy (n=6), IVIG only therapy (n=1), and high-dose steroid therapy (n=2). These patients started treatment within 24 hours after diagnosis with ANE. Two patients (patient 3 and 9) were not treated with IVIG or high-dose steroid therapy. Patient 3 was transferred to our hospital in a comatose state and was confirmed the diagnosis of brain death the next day. Therefore, we did not try either IVIG or high dose steroids on the patient. The patient died two weeks after conservative management. Patient 9 was also transferred to our hospital in a comatose state, and he was declared brain death on the third day of admission and died on the fifth hospital day.

The overall mortality rate was 40% (4/10). Three patients became brain death within three days from the day of mental status change. The median follow-up duration of six survivors was 3.3 years (range, 1.6-5.6 years). Neurological impairments such as motor deficits, developmental language delay and epilepsy were observed in three of six survivors. Definite involvement of the brain stem in brain imaging exhibited a significant statistical association with mortality (P=0.041) by multivariate Cox regression and Mann-Whitney test. As stated above, patients had no sooner started the IVIG or high-dose steroid therapy than they diagnosed with ANE. Therefore, the length of time between symptom onset and treatment was almost equal to that from symptom onset to diagnosis. There were no statistical associations between mortality and the factors studied; age, duration from symptom onset to diagnosis and treatment, causative agent of antecedent infection, brain hemorrhage and use of steroids or IVIG (Table 4).