Charakterystyczne cechy sieci mózgowej w opornej na leki epilepsji: badanie fazy 1-2 dotyczące biomarkerów diagnostycznych opartych na badaniu EEG w spoczynku
A spectral- topological network signature of drug-resistant epilepsy: a phase 1-2 study on resting-state EEG-based diagnostic biomarkers of drug resistance
W skrócie
Badacze badali nagrania elektrycznej aktywności mózgu (EEG) pacjentów z epilepsją, którzy nie reagują na leki, i zdrowych osób. Odkryli, że osoby z opornością na leki mają charakterystyczne zmiany w rodzajach fal mózgowych i sposobie komunikacji między obszarami mózgu. Te cechy można wykorzystać jako nowy test diagnostyczny, który dokładniej rozpoznaje pacjentów opornych na leki i pomoże lekarzom szybciej podjąć właściwe decyzje terapeutyczne.
Oryginalny abstract (angielski)
OBJECTIVE: Drug-resistant epilepsy (DRE) is increasingly recognized as a disorder of large-scale brain networks. Here, we evaluated a candidate resting-state EEG-based biomarker for identifying DRE in a diagnostic context of use. METHODS: We conducted a retrospective observational study (Phase 1-2 biomarker validation) on resting-state EEG recordings on healthy subjects (HS) and people with epilepsy (PwE). In PwE, EEGs were recorded after a second anti-seizure medication trial. The reference standard was longitudinal clinical outcome at 12 months (DRE vs. non-DRE). Index tests included the following quantitative EEG measures: spectral frequency-specific and aperiodic components, along with graph-theory metrics derived from the weighted phase-lag index. Multivariate logistic regression models assessed their discriminative value. RESULTS: We enrolled 120 PwE (60 DRE) and 60 HS. DRE showed a distinct spectral profile, with increased δ (1-4 Hz) power, reduced α (8-12 Hz) power, and a steeper aperiodic slope compared with both HS and non-DRE. Network analysis revealed increased δ-band betweenness centrality and small-world index, alongside reduced global efficiency, indicating a shift toward a more regular and less integrative topology. These findings were independent of epilepsy etiology (p-values = 0.001-0.04). Adding EEG features significantly improved DRE classification compared with clinical variables alone (AUC: 0.83 ± 0.03 vs. 0.71 ± 0.02, p < 0.001). CONCLUSIONS: We revealed convergent spectral and network-level alterations that delineate an intrinsic network signature highly associated with DRE. SIGNIFICANCE: Resting-state EEG metrics show promise as candidate diagnostic biomarkers for DRE, addressing an important unmet clinical need, though external validation is required.