Optimization of 18F-FDG PET/CT semi-quantitative diagnostic model for children with autoimmune encephalitis with epilepsy and negative MRI / 中华核医学与分子影像杂志

ABSTRACT

Objective:To investigate the value of 18F-FDG PET/CT imaging in the diagnosis of suspected autoimmune encephalitis (AE) in children with epilepsy and negative MRI. Methods:From May 2019 to August 2022, 94 suspected AE children (49 males, 45 females; age 1-15 years) with epilepsy and negative MRI who underwent brain 18F-FDG PET/CT imaging at Xinhua Hospital Affiliated to Shanghai Jiao Tong University School of Medicine were retrospectively analyzed. All patients were divided into AE and non-AE groups based on clinical final diagnosis. The effectiveness of visual diagnosis was evaluated. The cortical lesion extent score (S), and SUV max, SUV mean and minimum of SUV (SUV min) of cortical lesions (L), basal ganglia (B) and thalamus (T) were measured and SUV ratios (SUVR) of L/B or L/T were obtained. Independent-sample t test or Mann-Whitney U test was used to analyze data. Binary logistic regression analysis was used to screen the diagnostic factors of AE, and a diagnostic model was established. The diagnostic efficiency was evaluated by ROC curve analysis and Delong test. Results:There were 53 cases in AE group and 41 cases in non-AE group. Based on visual analysis, the sensitivity, specificity and accuracy of 18F-FDG PET/CT for AE were 100%(53/53), 43.9%(18/41) and 75.5%(71/94), respectively. Differences of LSUV max, LSUV mean, LSUV min, L/BSUVR max, L/BSUVR mean, L/BSUVR min, L/TSUVR max, L/TSUVR mean, L/TSUVR min and S between AE and non-AE groups were statistically significant ( z=-6.74, t values: from -8.51 to -3.97, all P<0.001). ROC curve analysis showed that the AUC of L/BSUVR max was the highest (0.914) among visual analysis and semi-quantitative parameters. Logistic regression analysis showed that S (odds ratio ( OR)=11.40, 95% CI: 2.18-59.52, P=0.004), L/BSUVR max( OR=13.19, 95% CI: 2.11-82.51, P=0.006) and L/TSUVR max( OR=9.66, 95% CI: 1.57-59.55, P=0.015) were independent diagnostic factors for AE. Regression model was established: P=1/(1+ e - x), x=2.433×S+ 2.580×L/BSUVR max+ 2.267×L/TSUVR max-3.802. The AUC of this model was 0.948, with the sensitivity, specificity and accuracy of 98.1%(52/53), 90.2%(37/41) and 94.7%(89/94), respectively. The diagnostic efficacy of the optimized scoring system was consistent with the pre-optimization model, and were both superior to L/BSUVR max(both z=2.01, both P=0.040). Conclusion:The diagnostic model and scoring system based on the semi-quantitative analysis of 18F-FDG PET/CT have better diagnostic efficacy for AE and are superior to semi-quantitative parameters alone.