Abstract
Background: Acute encephalopathy with biphasic seizures and late reduced diffusion (AESD) often causes various neurological sequelae, necessitating early and objective differentiation of AESD from a febrile seizure (FS). Therefore, we developed a scoring system that predicts AESD onset using only early laboratory data. Methods: We selected patients with AESD or FS admitted to the Tottori University Hospital between November 2005 and September 2020 and collected laboratory data from onset to discharge in patients with FS and from onset to the second neurological events in patients with AESD. Results: We identified 18 patients with AESD and 181 patients with FS. In comparison with patients with FS, patients with AESD showed statistically significant increases in ammonia (NH3), blood sugar (BS), and serum creatinine (Cr) levels, and the white blood cell (WBC) count, and a significant decrease in pH at <3 h from onset. We set the cut-off values and adjusted the weight of each of these parameters based on data obtained <3 h from onset and proposed a scoring system for predicting AESD. This system showed 91% sensitivity and 94% specificity for distinguishing AESD from FS. These accuracies were only slightly improved by the addition of information related to consciousness and seizure duration (sensitivity, 91%; specificity, 96%). Conclusion: NH3, BS, and Cr levels, WBC count, and pH were significantly different between patients with AESD and patients with FS at <3 h from seizure onset. This scoring system using these data may enable the prediction of AESD onset for patients under sedation or without precise clinical information.
Introduction
In pediatric neurology, distinguishing between a "simple" febrile seizure (FS) and the onset of Acute Encephalopathy with Biphasic Seizures and Late Reduced Diffusion (AESD) is a high-stakes race against time. AESD is a devastating syndrome, primarily seen in East Asian children, characterized by an initial seizure followed by a "silent" period of apparent recovery, and then a second round of seizures and consciousness impairment. By the time the signature "bright tree" appearance shows up on an MRI (usually around day 7), permanent neurological damage—such as intellectual disability or paralysis—may have already occurred.
Early intervention (like therapeutic hypothermia or specific vitamin cocktails) can improve outcomes, but clinicians lack objective tools to identify AESD in its earliest hours. Previous scoring systems relied heavily on subjective clinical observations, which can be difficult to assess accurately if a child is sedated or if the clinical report is imprecise.
Methods
The researchers conducted a retrospective study at Tottori University Hospital, analyzing data from 199 pediatric patients (18 AESD, 181 FS) admitted between 2005 and 2020. The team used multivariate regression to identify which laboratory parameters were significantly different between the two groups specifically within the first three hours of seizure onset. They then binarized these parameters using receiver operating characteristic (ROC) curves to find optimal "cut-off" values and assigned integer weights (1 or 2 points) to create a simple, bedside-ready scoring system.
Results
The study identified five key laboratory markers that differed significantly within the 3-hour window: blood sugar (BS), serum creatinine (Cr), ammonia (NH3), pH, and white blood cell (WBC) count.
The proposed Laboratory Prediction Score uses these variables obtained within 3 hours of onset:
- Blood Sugar (BS) ≥ 220 mg/dL: 2 points
- Serum Creatinine (Cr) ≥ 0.35 mg/dL: 2 points
- Ammonia (NH3) ≥ 75 µg/dL: 1 point
- pH ≤ 7.25: 1 point
- WBC count ≥ 25,000/µL: 1 point
A total score of ≥ 3 predicted AESD with 91% sensitivity and 94% specificity. The Area Under the Curve (AUC) for the lab-only model was 0.92, indicating excellent discriminatory power.
Discussion and Insights
The most critical finding is that lab differences are most pronounced in the very early phase (<3h). After 24 hours, many of these markers (like blood sugar and pH) normalize in both groups, making them less useful for differentiation. The researchers hypothesize that high blood sugar (BS) likely reflects a massive surge in cortisol and catecholamines (stress response), while elevated creatinine (Cr) may indicate early energy failure in the central nervous system, as the brain consumes phosphocreatine to buffer ATP during the initial seizure.
While adding clinical data like the Glasgow Coma Scale (GCS) and seizure duration slightly improved the model's specificity (to 96%), the lab-only model performed remarkably well on its own. This is vital for patients who are intubated or sedated, where clinical assessment is limited.
Conclusion
This scoring system provides a robust, objective method for early AESD prediction. The primary limitation is the small sample size of the AESD group (n=18), which necessitates a larger, multi-center prospective study to validate the scoring system. Future work should also examine how early medical interventions, such as intravenous fluids or anticonvulsants, might interact with the predictive power of the score.
Summary updated