We retrospectively enrolled 67 children diagnosed with ANE at Wuhan Children’s Hospital between January 2014 and June 2022. All patients met the diagnostic criteria for ANE revised by Mizuguchi et al. (10), including: (1) Clinical features: Acute encephalopathy occurring 1–3 days after febrile illness, characterized by rapidly worsening consciousness, (2) Cerebrospinal fluid examination: No significant increase in cell count, elevated protein levels, (3) Imaging findings: Head CT/MRI shows symmetrical, multifocal lesions, necessarily involving bilateral thalami; MRI characteristic manifestations include T2/FLAIR hyperintensity, diffusion restriction on DWI, sometimes accompanied by hemorrhage, (4) Laboratory tests: Serum transaminases elevated to varying degrees, normal blood ammonia. Patients with infectious, metabolic, toxic, or autoimmune encephalopathies were excluded. For death cases, those whose death was unrelated to ANE (e.g., accidental trauma) and cases lost to follow-up were excluded. Clinical data were retrospectively analyzed, including sex, age, season of onset, personal, past medical and family history. Information on prodromal infection symptoms and pathogens, the interval between prodromal infection and encephalopathy onset, and clinical manifestations during the acute stage was collected. We collected laboratory examinations at the first admission, including blood routine tests, liver and renal function, electrolytes, cardiac enzymes, coagulation profile, procalcitonin (PCT), cytokines, and cerebrospinal fluid (CSF) analysis (routine and biochemistry). Glasgow Coma Scale (GCS) is based on the score at the first admission. Neuroimaging data were obtained from cranial MRI, and treatment information was also recorded.

To investigate prognostic factors, follow-up was conducted until death for non-survivors and up to 12 months for survivors. Follow-up assessments included the Extended Glasgow Outcome Scale (EGOS) score, residual clinical symptoms, and cranial MRI findings. Outcomes within 3 months were defined as short-term prognosis, whereas outcomes at 12 months were defined as long-term prognosis. Cranial MRI scans were collected at admission and at 1, 3, 6, and 12 months after disease onset. The study flowchart is presented in Figure 1.

Neuroimaging features were defined as follows: hyperintensity on T1-weighted imaging (T1WI) was considered hemorrhage; hypointensity on T1WI, hyperintensity on T2WI, and hypointensity on FLAIR were defined as cystic degeneration; brain atrophy was defined as parenchymal volume loss with or without gliosis and widening of sulci. MRI severity was evaluated using the scoring system proposed by Wong et al. (11), ranging from 0 to 4 points, based on the presence of hemorrhage, cystic changes, brainstem involvement, and white matter involvement (cerebral, cerebellar, or both). Disease severity was further assessed using the ANE Severity Score (ANE-SS) proposed by Yamamoto et al. (12), with a maximum of 9 points, categorized into low risk (0–1), moderate risk (2–4), and high risk (5–9). Prognostic outcomes were assessed using EGOS (13). For survivors, EGOS scores were recorded at 3, 6, and 12 months of follow-up.

All statistical analyses were conducted using SPSS version 26.0 (IBM, Armonk, New York, United States), GraphPad Prism version 9.5.1 (GraphPad Software, San Diego, California, United States), and Origin 2021 (OriginLab Corporation, Northampton, MA, United States). The statistical data of clinical variables are presented in the form of proportions and interquartile ranges. Pearson chi square test, Fisher’s exact test, Mann Whitney U test, and correlation analysis were used to compare the statistical significance between groups. Logistic regression analysis was performed to determine predictive factors for ANE mortality and long-term prognosis. A Sankey plot of risk factors associated with mortality and long-term outcomes was generated using Origin software. Use Kaplan Meier method to construct survival curves and compare mortality rates among different variable groups. We calculated the odds ratio (OR) and their respective 95% confidence intervals (CI). Statistical tests were two-tailed, and significance was set at p < 0.05.

Participants (n = 67) with ANE were included in this study. Baseline characteristics are summarized in Table 1. The male-to-female ratio was 1.4:1, with an age range of 1–142 months. Most patients (62.7%) were under 4 years old, and 68.7% had disease onset during the winter or spring. 62 patients (92.5%) had no significant past medical or family history. One patient had an elder brother who developed encephalitis and died at the age of 3 following a similar infection, and genetic testing confirmed a RANBP2 mutation in this patient. Additionally, one patient had a history of motor developmental delay, one had recurrent pneumonia, one had febrile seizures, and one had purulent meningitis.

All patients presented with fever, 61.2% developed respiratory symptoms, and 27% exhibited gastrointestinal symptoms. Laboratory testing confirmed influenza virus infection in 32.8% of cases. In this cohort, acute encephalopathy symptoms appeared 1–5 days after the prodromal infection, with 35.8% of children exhibiting encephalopathy symptoms within one day. Neurological manifestations included impaired consciousness (GCS ≤8) in 79.1% of patients and seizures in 82.1%, of whom 30.9% experienced status epilepticus. Multiple organ dysfunction syndrome (MODS) was observed in 64.2% of patients, while shock and disseminated intravascular coagulation (DIC) occurred in 11.9 and 7.5%, respectively.

Of the 67 patients, 18 (26.9%) died, with time to death ranging from 2 to 85 days (mean, 21 days). Thirteen of these deaths occurred within the first month. Among the 49 survivors, EGOS scores at 3, 6, and 12 months of follow-up are shown in Figure 1.

All patients underwent baseline laboratory testing on admission, including complete blood count, liver and renal function, electrolytes, cardiac enzymes, and coagulation profile. Abnormalities were observed in activated partial thromboplastin time (APTT), aspartate aminotransferase (AST), lactate dehydrogenase (LDH), procalcitonin (PCT), interleukin-6 (IL-6), and interleukin-10 (IL-10) (Table 2). Median APTT was 39.00 (29.00–46.10) s, with a maximum of 89 s; AST was 121.00 (48.00–292.00) U/L, peaking at 3732 U/L; LDH was 461.00 (320.00–654.00) U/L, with a maximum of 9,082 U/L; and PCT was 1.32 (0.33–15.13) ng/mL, reaching up to 102.21 ng/mL. Among the 42 patients tested for serum cytokines, median IL-6 was 15.97 (5.67–84.46) pg/mL, and IL-10 was 8.12 (3.58–27.92) pg/mL. The ANE Severity Score (ANE-SS) ranged from 0 to 8. 17.9% of patients were categorized as low risk, 52.2% as moderate risk, and 29.9% as high risk based on risk stratification.

All patients demonstrated thalamic involvement on acute-phase cranial MRI, with varying degrees of additional involvement in other regions (Table 1). On diffusion-weighted imaging (DWI), 82.1% showed a central hypointense zone in the bilateral thalami, surrounded by a hyperintense ring, with an additional outer hypointense rim; on apparent diffusion coefficient (ADC) maps, this corresponded to a central hyperintensity surrounded by hypointensity and an outer hyperintensity, forming a “tricolor pattern.” In 17.9% of patients, DWI revealed a central thalamic hyperintensity with peripheral hypointensity, with reversed features on ADC maps, consistent with a “bicolor pattern.”

MRI follow-up was available for 49 survivors for ≥12 months (Table 3). Serial imaging demonstrated dynamic disease evolution, with hemorrhage observed in 62.7% (42/67), cystic degeneration in 63.3% (31/49), and brain atrophy in 55.1% (27/49). Based on the MRI scoring system proposed by Wong in 2006, scores among patients with ≥12 months of follow-up were distributed as follows: 20.4% scored 1, 28.6% scored 2, 28.6% scored 3, and 22.4% scored 4.

Most patients (80.6%) were admitted to the pediatric intensive care unit (PICU), with a median PICU stay of 7 (4–12) days. Tracheal intubation was required in 35.8% of cases. Corticosteroids were administered in 94% of patients, with 31.7% receiving steroids within 24 h of onset. High-dose methylprednisolone therapy (≥20 mg/kg/day) was used in 74.6% of cases. Intravenous immunoglobulin (IVIG) was administered in 89.6%, hypothermia therapy in 40.3%, and plasma exchange in 1.5%. Among survivors, 38.8% underwent ≥6 months of standardized rehabilitation therapy at specialized hospitals (Table 1).

Analysis of the association between clinical features and mortality in children with ANE showed that influenza virus infection, an interval of ≤24 h from prodromal infection to onset of encephalopathy, GCS score <5, shock during the acute phase, and the need for tracheal intubation were significantly associated with increased risk of death. In contrast, administration of corticosteroids within 24 h of onset was associated with improved survival (Table 3). Laboratory analyses indicated that death patients were more likely to present with prolonged APTT and elevated levels of AST, LDH, PCT, IL-6, and IL-10 (Figures 2A–C).

To explore the related factors of ANE mortality risk, we used univariate logistic regression analysis. The results identified influenza virus infection, prodromal infection to encephalopathy interval ≤24 h, GCS score <5, acute-phase shock, need for tracheal intubation, higher ANE-SS score, prolonged APTT, and elevated PCT and IL-10 as significant risk factors for short-term mortality (Table 4).

Kaplan–Meier survival curves were plotted to examine mortality across different risk factors (Figure 3). Compared with non-infected patients, those with influenza virus infection had a higher mortality rate (11/22 vs. 7/45, 50.0% vs. 15.6%). Patients with prodromal infection to encephalopathy interval ≤24 h had higher mortality than those with >24 h (11/24 vs. 7/43, 45.8% vs. 16.3%). Mortality was markedly higher in patients with GCS <5 (14/23 vs. 4/44, 60.9% vs. 9.1%). Conversely, corticosteroid treatment within 24 h of onset reduced mortality (2/20 vs. 15/43, 10.0% vs. 34.9%).

To visually represent these findings, a Sankey diagram was generated showing viral infection, prodromal-to-encephalopathy interval, GCS score, corticosteroid administration timing, and survival status (Figure 4A). These results suggest that children with ANE who have influenza virus infection, prodromal-to-encephalopathy interval ≤24 h, or GCS <5 should be closely monitored for mortality risk, and corticosteroid therapy should be initiated within 24 h whenever possible.

Analysis of clinical features and long-term outcomes revealed that GCS <5, the presence of hemorrhage, cystic degeneration, or atrophy on follow-up brain MRI, higher Wong 2006 MRI scores, and elevated cerebrospinal fluid protein levels were associated with worse EGOS scores at long-term follow-up. Conversely, corticosteroid administration within 24 h of onset and ≥6 months of standardized hospital-based rehabilitation were associated with better long-term outcomes and lower EGOS scores (Table 5 and Figures 2D–F).

To identify independent predictors of long-term prognosis, univariate and multivariate logistic regression analyses were performed. The results demonstrated that hemorrhage and brain atrophy on follow-up MRI were independent risk factors for poor long-term outcomes, whereas corticosteroid therapy within 24 h of onset was an independent protective factor (Table 6). A Sankey diagram illustrating the distribution of long-term prognostic factors among 49 children showed that patients with milder EGOS scores during follow-up rarely had hemorrhage (7/14, 50%) or atrophy (12/14, 85.7%), and most received early corticosteroid treatment within 24 h (11/14, 78.6%) (Figure 4B).