Diagnostic model based on semi-quantitative metabolic ratios of 18F‑FDG PET in suspected autoimmune encephalitis children with seizures

Abstract

Background Pediatric autoimmune encephalitis (AE) with seizures is prone to be misdiagnosed as epilepsy. However, epilepsy and AE are treated differently. We aimed to find a method that can identify AE in children with seizures through F-18-fluorodeoxyglucose(18F-FDG) positron emission tomography(PET) with good diagnostic specificity. Results We retrospectively collected data from suspected autoimmune encephalitis children with seizures from May 14, 2019, to April 30, 2021, who had brain 18F-FDG PET/CT. 59 pediatric patients were included, of which 35 were clinically diagnosed with AE. FDG uptake alterations were investigated by visual analysis and Statistical Parametric Mapping(SPM) 12 software. The proportion of cortical hypometabolism and hypermetabolism in the basal ganglia and thalamus was significantly higher in the AE group. The score of the cortical lesion was higher in the AE group. 100% of AE patients had multiple lesions on PET, while 60% were in the non-AE group. The lesion quantity feature, the score of cortical lesion range, and four metabolic ratios were selected for multivariate logistic regression analysis. The results showed that the score of cortical lesion range, the ratio of the maximum standard uptake value of the lesion to basal ganglia (L/BSUVRmax), the ratio of the maximum standard uptake value of the lesion to thalamus(L/TSUVRmax), were independent diagnostic factors of AE. A diagnostic model that combined the three PET diagnostic factors had good diagnostic performance, with the area under the curve(AUC), sensitivity, and specificity at 0.910, 91.2%, and 88%, respectively. Conclusions The established diagnostic model that combines three PET diagnostic factors could achieve better diagnostic performance than visual analysis or semi-quantitative ratios alone and shows potential for clinical applications in differentiating AE from epileptic children.