Quantitative epileptiform burden and electroencephalography background features predict post-traumatic epilepsy

Abstract

Background: Post-traumatic epilepsy (PTE) is a severe complication of traumatic brain injury (TBI). Electroencephalography aids early post-traumatic seizure diagnosis, but its optimal utility for PTE prediction remains unknown. We aim to evaluate the contribution of quantitative electroencephalograms to predict first-year PTE (PTE₁).

Methods: We performed a multicentre, retrospective case-control study of patients with TBI. 63 PTE₁ patients were matched with 63 non-PTE₁ patients by admission Glasgow Coma Scale score, age and sex. We evaluated the association of quantitative electroencephalography features with PTE₁ using logistic regressions and examined their predictive value relative to TBI mechanism and CT abnormalities.

Results: In the matched cohort (n=126), greater epileptiform burden, suppression burden and beta variability were associated with 4.6 times higher PTE₁ risk based on multivariable logistic regression analysis (area under the receiver operating characteristic curve, AUC (95% CI) 0.69 (0.60 to 0.78)). Among 116 (92%) patients with available CT reports, adding quantitative electroencephalography features to a combined mechanism and CT model improved performance (AUC (95% CI), 0.71 (0.61 to 0.80) vs 0.61 (0.51 to 0.72)).

Conclusions: Epileptiform and spectral characteristics enhance covariates identified on TBI admission and CT abnormalities in PTE₁ prediction. Future trials should incorporate quantitative electroencephalography features to validate this enhancement of PTE risk stratification models.

Keywords: EEG; EPILEPSY; TRAUMATIC BRAIN INJURY.

Figure 1:. Quantitative Electroencephalography (QEEG) Prediction Models for First-year Post-traumatic Epilepsy (PTE₁).

(A) % Epileptiform Abnormalities (EAs) for all one-hour windows for all patients (0% corresponds to grey, higher % corresponds to darker red/blue); each block represents an one-hour window; y-axis represents individual patients sorted by total recording duration (top: longest duration). (B) Same as panel A but for % suppression distribution. (C) Area Under the receiver-operating-characteristic Curve (AUC) comparison; AUC for forward-selected QEEG logistic regression (orange): 0.69 (95% CI, 0.60–0.78); test AUC for cross-validated ridge logistic regression based on TBI mechanism and Computed Tomography (Mechanism+CT, grey): 0.61 (0.51–0.72), and test AUC for cross-validated ridge logistic regression based on TBI mechanism, Computed Tomography, and QEEG (Mechanism+CT+QEEG, green): 0.71 (0.61–0.80); shaded areas represent the bootstrapped (n=1000) 95% confidence intervals. (D) Same as panel C but showing calibration errors for QEEG logistic regression: 0.06 (0.02–0.12), Mechanism+CT ridge regression: 0.06 (0.02–0.12), Mechanism+CT+QEEG ridge regression: 0.08 (0.04–0.15). (E) Feature importance for Mechanism+CT+QEEG ridge regression; features were sorted by the importance measure; each boxplot visualizes the distribution of penalized coefficients across 8 folds.