The concept of "limbic encephalitis", introduced by Corsellis JA et al. in 1968, is the first historical record of autoimmune encephalitis research[14]. However, it was not until 2016 that a consensus diagnostic criterion was proposed. Graus stated that their guidelines should be applied with caution to children, particularly those younger than 5 years[8]. Children differ from adults in their clinical presentations, paraclinical findings, autoantibody profiles, treatment response, and long-term outcomes. The clinical approach to the diagnosis of autoimmune encephalitis in pediatric patients proposed in 2020 suggested a PET test if available[12]. Seizure is a prevalent symptom of autoimmune encephalitis in children. AED therapy alone is not effective for AE. Failure to diagnose AE in time will delay treatment, resulting in a poor prognosis. FDG PET is a useful tool for the diagnosis of AE, but up to now, there was no report about the application of FDG PET for the differential diagnosis of AE from the patients with seizures. We established a diagnostic model based on semi-quantitative ratios to help in the identification of AE.
Recent semi-quantitative studies indicate that AE is characterized by lobar hypometabolism, most commonly involving the parietal and occipital cortices and hypermetabolism, most commonly involving the basal ganglia [11, 15]. Ratios comparing uptake at the caudate nuclei to the parieto-occipital cortex have been confirmed to increase during the symptomatic phase of anti-VGKC antibody encephalitis[16]. A study proved the good early diagnostic efficacy of the lower cortex/striatum metabolic ratio without limitation to any antibody types[17]. The results of the above studies all suggest that involvement of the cortical and basal ganglia is a prominent manifestation of AE. Our result is consistent with the existing findings. The ratio of the lesion to basal ganglia was significantly lower in the AE group, which can be used to differentiate between AE and non-AE. We found that the ratio of the lesion to thalamus also had a good discriminatory ability.
There were significant differences in the number and distribution of lesions between the AE and non-AE groups. In our study, 60% of non-AE patients had multiple lobes involved. An analysis of 119 pediatric patients with drug-resistant epilepsy showed that the proportion of multiple lobes involved was 47%[18]. A portion of children with epilepsy show a normal glucose metabolic pattern[19]. The affected regions often showed focal cortical dysplasia and hippocampal sclerosis on histopathology[20]. AE seems to be characterized by multiple lesions and lobe involvement. The lesions of AE can even involve the whole cerebral cortex. The following are possible reasons for the wide range of AE lesions: On the one hand, antigens in the CNS, including some receptors, ion channels, and neurotransmitter-related enzymes, are responsible for essential physiological functions, and their distribution is extensive. For example, NMDAR density is high in the occipital lobe and parietal association cortices[21]. On the other hand, a study on anti-NMDAR encephalitis suggested serum antibodies diffuse into the cortical gray matter in the early stage [22]. We believe that the extensive range of lesions could help diagnose AE in children with epilepsy.
The present study might provide a novel method for the diagnosis of AE. We used a combination of three PET factors to identify AE in children with seizures. The diagnostic model based on three PET factors yielded high diagnostic values. The diagnostic model could achieve better diagnostic performance than visual analysis or semi-quantitative ratios alone. In addition, these factors are easy to obtain. So, the diagnostic model is simple and convenient, allowing widespread use among clinicians.
The present study has several limitations. First, this is a single-center retrospective study performed with a risk of selection bias. Second, this study's sample size may be relatively small. Third, most AE patients included were in the non-acute phase, so it was limited to clarifying the metabolic changes during different disease stages. Finally, prospective studies are needed to confirm the diagnostic performance of the diagnostic model in AE patients.