A new path has opened to diagnose epilepsy through blood-based immune cell analysis, eliminating the need for time-consuming electroencephalogram (EEG) tests or expensive magnetic resonance imaging (MRI) scans.
Seoul National University Hospital announced Wednesday that a joint research team — comprising professors Joo Eun-jin and Lee Sang-kun of Neurology, professor Shin Yong-won of Critical Care Medicine, and professor Hong Sang-bin of the Hospital Medicine Center — has demonstrated for the first time in the world that changes in systemic immune patterns revealed through a novel blood-based approach are deeply linked to epilepsy diagnosis and brain atrophy.
Epilepsy is diagnosed when seizures — accompanied by symptoms such as convulsions, loss of consciousness, or behavioral changes caused by abnormal electrical activity in the brain's neural networks — recur at least twice without specific triggers, separated by intervals of more than 24 hours. Although epilepsy has long been considered solely a problem of the brain itself, recent findings have increasingly shown that abnormalities in the systemic immune system are deeply involved in its onset. However, EEG, essential for diagnosing epilepsy, records the electrical activity of brain cells in real time through electrodes attached to the scalp, and has the limitation that abnormal brain waves may not be detected during seizure-free periods. While MRI is useful for confirming whether structural brain abnormalities such as tumors, vascular diseases, or malformations are causing seizures, MRI alone is insufficient to confirm an epilepsy diagnosis. Positron emission tomography (PET) and single-photon emission computed tomography (SPECT) also require considerable time for testing and result interpretation, creating an urgent need for simple blood-test-based diagnostic indicators.
The research team focused on T cells, the body's core immune cells that fight external pathogens. T cells in the blood carry a unique genetic sequence structure called a T-cell receptor (TCR) on their surface, which can identify invaders inside the body. They are referred to as "immune barcodes," likened to barcodes that contain different information for each product. Healthy people normally carry a wide variety of immune barcodes evenly distributed in their blood to prepare for diverse threats. When a specific pathogen invades the body and customized T cells that recognize it proliferate intensively, the overall number and diversity of immune barcodes within a limited blood volume actually decrease. This is known immunologically as clonal expansion.
The team hypothesized that if epilepsy is a disease accompanied by activation of the systemic immune system, only certain immune barcodes would proliferate abnormally in patients' blood, reducing overall diversity, and they began verifying this. They recruited a total of 100 people, including 45 epilepsy patients and 55 healthy controls, and analyzed the genetic sequences of T-cell receptors in peripheral blood samples. The epilepsy patient group was subdivided into 14 patients with well-controlled drug response, 22 with refractory epilepsy, and 9 with neuroinflammation-related epilepsy, in order to examine differences according to disease severity.
The results showed that the epilepsy patient group had significantly lower overall immune barcode diversity than the healthy control group. In particular, the phenomenon of intensive proliferation of only specific immune barcodes was even more pronounced in patients with refractory epilepsy unresponsive to medication or those with neuroinflammation-accompanied epilepsy. This indicates, the researchers explained, that epilepsy is not merely a problem confined to the brain but rather a state of imbalance in the systemic immune system.
Based on the T-cell receptor analysis results, the team developed an artificial intelligence (AI) model that can assist in epilepsy diagnosis and severity assessment. After applying nine machine learning algorithms to 18 combinations of immune data to build a total of 162 diagnostic prediction models, performance verification showed that the model trained intensively only on T-cell receptor genetic combination pattern data, without additional information such as age or sex (using the random forest algorithm), demonstrated the best discriminative ability. This model distinguished epilepsy patients from healthy individuals with an average accuracy of 80% based solely on blood analysis data. The area under the curve (AUC), which represents the overall discriminative power and reliability of AI prediction, also recorded 0.80, proving its excellent performance as a non-invasive diagnostic method.
It is also notable that these changes in blood immune indicators are directly connected to actual physical structural changes in the brain. Additional analysis of data from 21 patients who underwent brain imaging confirmed that as immune barcode diversity decreased, brain atrophy occurred, with reduced volume in the thalamus and basal ganglia regions. These regions play crucial roles in the generation and propagation of seizures in the brain. This is interpreted as an important link showing that systemic immune activation leads to epilepsy-related neurodegeneration.
"This study will serve as important evidence for monitoring the progress of epilepsy patients and exploring new treatment strategies through immune modulation in the future," Professor Shin said. Professor Joo emphasized, "This study suggests that epilepsy, which has been recognized only as a problem of the brain itself, should be viewed from the perspective of the systemic immune system," adding, "We have taken one step closer to realizing precision medicine based on the immune status of individual patients."
The research findings were published in a recent issue of the international journal Annals of Clinical and Translational Neurology.
