Depression is a common and disabling comorbidity in temporal lobe epilepsy (TLE). Despite its clinical significance, reliable objective biomarkers for depression in TLE remain elusive. In this study, we investigated whether scalp EEG complexity as assessed by sample entropy could serve as a potential biomarker of depression in persons with unilateral TLE.
We analyzed visually normal awake, resting-state scalp EEG from 22 subjects with unilateral TLE who also completed the Beck Depression Inventory-II (BDI-II). Five minutes of visually normal, awake EEG were bandpass filtered (0.5–55 Hz) using a finite impulse response filter, segmented into 1-second epochs, and analyzed for sample entropy (pattern length m = 2, tolerance r = 0.1). For each channel, we computed the mean and standard deviation (SD) of the entropy across epochs.
Nine patients met criteria for depression (BDI-II ≥ 20) and 13 were not depressed (BDI-II < 20). Mean entropy did not differ between the two groups. However, entropy variability (SD) was significantly lower in the depressed group (MANOVA p = 0.002). Regional analysis showed reduced entropy variability in frontal, posterior temporal, and occipital electrodes of the ictal hemisphere and in the posterior temporal region of the non-ictal hemisphere (p < 0.03), with a trend in the non-ictal frontal region (p = 0.07). Across all subjects, reduced variability in entropy in frontal and posterior temporal regions of the ictal hemisphere correlated with higher depression severity (r2 > 0.31, p < 0.007).
Reduced flexibility in EEG complexity within the frontotemporal networks of the epileptogenic hemisphere may serve as a network-level signature of depression in unilateral TLE. Measuring entropy variability provides a non-invasive EEG metric that could enhance clinical evaluation and inform individualized treatment.