Autoimmune encephalitis (AE) is a major noninfectious cause of encephalitis, accounting for 20% to 30% of cases [1]. Its incidence in the United States from 1995 to 2015 was estimated at 0.8 per 100,000 person-years, comparable to that of infectious encephalitis, which was approximately 1.0 per 100,000 person-years, and has increased over time because of improved recognition [2]. AE is an immune-mediated disorder that targets neuronal surface and synaptic proteins and remains an important but often underdiagnosed cause of pediatric encephalitis [3,4].

Diagnosis relies on clinical features supported by cerebrospinal fluid (CSF), electroencephalography (EEG), and magnetic resonance imaging (MRI) findings [5,6]. Detection of neuronal autoantibodies, particularly anti–N-methyl-D-aspartate receptor (NMDAR) antibodies, is central to diagnosis and is typically performed in serum or CSF. CSF testing has higher sensitivity than serum testing, with reported sensitivity of up to 99%–100% in CSF compared with 68%–86% in serum, whereas specificity remains high in both sample types at >99% [7,8]. However, antibody-negative AE can occur and contributes to diagnostic uncertainty, particularly because EEG and MRI findings may be normal or nonspecific, which can lead to misdiagnosis and delayed treatment [9,10].

In low-resource settings such as Indonesia, the diagnosis and treatment of AE are difficult because access to MRI, EEG, CSF analysis, and neuronal autoantibody testing is limited. Although NMDAR antibody testing is available in some tertiary centers, other antibody panels are largely unavailable [11-14]. Consequently, physicians often rely on clinical judgment and corticosteroid responsiveness when diagnostic tests are unavailable or delayed [15]. These challenges are compounded by incomplete insurance coverage for immunotherapy, limited availability of first- and second-line treatments, and systemic referral barriers that can reduce follow-up effectiveness and worsen clinical outcomes [16].

This study aimed to describe the clinical characteristics, diagnostic challenges, therapeutic approaches, and outcomes of pediatric patients with AE in Indonesia. The findings are expected to provide insight into real-world clinical practice and inform the development of more effective management strategies in similar settings.

A total of 14 children with AE were included in this study. Based on established diagnostic criteria, six of 14 patients (42.9%) were classified as having definite AE, and eight of 14 patients (57.1%) were classified as having probable AE. The distribution of clinical features, investigations, management, and outcomes is summarized in Table 1.

In this study, seizures were the predominant presenting symptom and occurred in nearly all cases. Emotional disturbances and other psychiatric abnormalities were also common. These findings are consistent with previous studies emphasizing the broad clinical spectrum of pediatric AE [5,17,18]. Seizures in AE are associated with disease activity and may be resistant to standard antiseizure treatment before immunotherapy is initiated [18,19]. In addition, Gunawan et al. [20] identified a moderately significant association between seizure frequency and NMDAR antibody positivity in pediatric epilepsy. Despite this clinical heterogeneity, seizures and psychiatric manifestations may serve as important early clinical indicators of AE, especially in pediatric populations. However, their overlap with infectious, metabolic, and other neurological disorders remains a major diagnostic challenge [16]. Prominent psychiatric symptoms may further delay diagnosis by leading to initial misclassification and referral to psychiatric services [21]. In our study, many patients were initially diagnosed with alternative conditions, including post-dengue encephalopathy, schizophrenia, or infectious meningoencephalitis, before AE was considered.

Clinical manifestations in children differ from those in adults, likely because of developmental differences in neuronal networks and receptor expression. Children with AE often present with multiple neuropsychiatric symptoms, including movement disorders, agitation, insomnia, and seizures. In contrast, adults more commonly present with memory impairment, psychiatric manifestations, and central hypoventilation [9,22]. Tumor associations are also less common in children than in adults [23]. These age-related differences underscore the difficulty of diagnosing AE in children based solely on clinical presentation.

CSF analysis, which is commonly used as a supplementary diagnostic tool, showed limited sensitivity and specificity. In our cohort, only 50% of patients had elevated white blood cell counts, and protein elevation was rare, consistent with previous studies reporting variable and generally moderate CSF abnormalities [5,17,24]. Olive-Cirera et al. [25] also reported CSF pleocytosis in patients later diagnosed with noninflammatory diseases, such as epilepsy and primary psychiatric disorders, highlighting the limited specificity of this finding. Therefore, these disorders should be considered in the differential diagnosis of AE. In Indonesia, CSF analysis may be underused; a minority of neurologists perform lumbar puncture because of inadequate equipment, patient instability, or concern about complications [26].

The EEG findings in this study predominantly showed background slowing, consistent with previous pediatric AE studies [5,23,27]. EEG abnormalities are not specific to AE; however, EEG remains a valuable clinical tool for early diagnosis, particularly when antibody testing is unavailable. EEG can be performed rapidly, does not require anesthesia, and permits repeated bedside monitoring, making it especially useful in children. Extreme delta brush, observed in one patient, may serve as a distinctive indicator of anti-NMDAR encephalitis and has been associated with disease severity in previous studies [28,29]. Nevertheless, EEG availability in Indonesia remains limited, with only 26.9% of healthcare facilities reporting access, which may hinder electrophysiological confirmation [30].

In this study, MRI abnormalities predominantly involved gray matter structures, including the hippocampus and cortical areas of the temporal, parietal, and occipital lobes. MRI findings in AE are often variable and nonspecific, with similar abnormalities reported in noninflammatory disorders [25]. Previous studies have shown that MRI abnormalities are present in approximately 50% to 70% of pediatric AE cases [5,27], most often as increased T2/fluid-attenuated inversion recovery signal intensity. However, a substantial proportion of patients have normal imaging findings, indicating that MRI alone is insufficient for diagnosis and must be interpreted within the appropriate clinical context.

These diagnostic barriers are more pronounced in resource-limited healthcare systems such as Indonesia. MRI availability is often constrained by limited facility capacity and scheduling challenges, resulting in long waiting periods. National data indicate that MRI capacity remains insufficient for the population’s needs, with approximately 1.11 MRI units per million people and uneven regional distribution [31]. Similar challenges have been reported in India, where limited access to advanced diagnostics and long waiting times for evaluation have impeded the diagnosis and treatment of anti-NMDAR encephalitis [21]. Such delays can postpone diagnosis and immunotherapy, potentially affecting clinical outcomes. In addition, young or acutely ill children may be unable to cooperate during MRI and may require sedation to obtain adequate images, further complicating and prolonging the diagnostic process.

In our cohort, six of 14 patients had anti-NMDAR encephalitis. All positive results were identified in CSF, supporting previous studies showing the superior sensitivity of CSF testing compared with serum testing [5]. Antibody testing remains the most reliable diagnostic method for AE [6]. However, delayed testing and possible false-negative results create diagnostic challenges. In Indonesia, testing is largely limited to anti-NMDAR antibodies at selected tertiary centers, whereas comprehensive neuronal antibody panels are generally inaccessible [11-14]. Similar limitations have been documented for other neurological disorders, further restricting access and contributing to incomplete diagnostic assessment [32].

All patients in this study received IV MP as primary immunotherapy, in accordance with recent guidelines [33,34]. Of the six patients who received IV MP alone, four showed clinical improvement, including three with substantial recovery and one with slight improvement, whereas two died despite treatment initiation at 19 and 28 days, respectively. Three patients received IVIG in addition to IV MP, and all showed substantial improvement, supporting the role of IVIG as a first-line or adjunctive therapy in pediatric AE. These findings are consistent with previous studies reporting favorable outcomes associated with IVIG administration [5]. However, access to IVIG remains limited because it is expensive and not covered by insurance. Similar barriers have been reported in other resource-limited settings, including the Philippines and India, where immunomodulatory therapies are difficult to access and costly, limiting optimal treatment and access to second-line agents [21,35]. AZA was used as second-line or maintenance therapy in selected cases [17]. In our cohort, six patients received AZA: three showed substantial improvement, whereas three showed only slight improvement. AZA may be a practical option in settings where rituximab is unavailable or unaffordable [36].

Our study showed a median hospital stay of 14 days (range, 7 to 37), which is shorter than the 54.9 days (range, 4 to 150) reported by Pruetarat et al. [37]. Their study also showed that patients with poor outcomes had longer hospital stays, consistent with our finding that the two patients who died had hospital stays of 27 and 37 days. The mRS findings indicated that most patients in our cohort improved over time. Seven of the 14 patients had scores of 0–2, indicating good outcomes. Four of these patients received combination therapy. Lee et al. [38] reported that 62.2% of patients had mRS scores of 0–1 after 1 year, with treatment response identified as the strongest predictor of outcome. In addition, a multicenter study by Pruetarat et al. [37] found that eight of 14 patients had favorable outcomes at the 12-month follow-up.

In Indonesia, the diagnosis and treatment of AE are challenging because of health-system barriers and limited resources. These barriers include shortages of specialists, uneven healthcare distribution, inefficient referral pathways, and limited intensive care unit capacity, all of which may delay referral for critically ill patients [39]. These challenges are compounded by low clinical suspicion and restricted access to essential diagnostic modalities, including lumbar puncture, laboratory testing, and neuronal antibody assays, often resulting in incomplete evaluations [26]. In our cohort, diagnostic and treatment delays were substantial, with median times of 48 days to MRI, 25 days to CSF antibody testing, and 19 days to initiation of IV MP. These delays likely contributed to the frequent use of empirical first-line immunotherapy before test results were available. Dubey et al. [40] reported shorter median times to MRI (18 days) and CSF evaluation (22 days) and showed that shorter times to diagnosis and treatment were associated with better clinical outcomes.

In conclusion, seizures and psychiatric symptoms are important early signs of pediatric AE. CSF findings, EEG, and MRI provide supportive but often nonspecific information, whereas antibody detection in CSF, especially anti-NMDAR antibody detection, remains the most reliable diagnostic indicator. In Indonesia, restricted access to diagnostic testing and immunotherapy, together with systemic referral barriers, contributes to delayed diagnosis and treatment and may adversely affect outcomes. Early clinical recognition and prompt empirical immunotherapy are essential, particularly when confirmatory testing is unavailable.

Delayed diagnosis and limited access to antibody testing remain substantial barriers to the management of AE. Strengthening centralized laboratories, including antibody testing in national health insurance coverage, improving referral systems, and using teleneurology may improve access to care and support earlier identification and treatment. National guidelines tailored to resource-limited settings are needed and should emphasize rapid clinical identification, exclusion of common mimickers, and timely empirical treatment. These measures would strengthen the translational significance of this research.