Personalized neurometabolic signature predicts seizure outcomes of laser ablation in mesial temporal lobe epilepsy.

BACKGROUND

Mesial temporal lobe epilepsy (mTLE) is a common form of drug-resistant epilepsy and seizure outcomes after minimally invasive laser ablation remain suboptimal. Current imaging-guided strategies often fail to capture individual variability in seizure foci. This study aimed to develop a personalized neuroimaging biomarker to improve surgical planning and predict outcomes.

METHODS

Thirty patients with mTLE (16 women, 53.3%; age range 17-59 years) who underwent magnetic resonance-guided laser interstitial thermal therapy were retrospectively analyzed. The asymmetry index (AI) from [18F]fluorodeoxyglucose positron emission tomography ([18F]FDG PET) defined the personalized NeuroMetabolic Signature (pNMS). Prognostic thresholds and optimal pNMS ablative rate were explored using restricted cubic spline (RCS) analysis and Youden's index as statistical methods for identifying cutoffs. A generalized additive model (GAM) was applied to examine imaging-derived features associated with pNMS.

RESULTS

Here we show that the AI of PET metabolic values significantly predicted seizure outcomes (odds ratio = 1.43, P = 0.02), with -0.06 as the threshold for defining pNMS (P for non-linearity = 0.04). A hippocampal pNMS ablative rate of 39.79% is significantly associated with seizure freedom (Pearson χ2 = 10.16, P = 0.001; balanced accuracy = 0.83). Hippocampal atrophy contributes most to pNMS expression (Shapley value = -0.026), and correlates with metabolic asymmetry (Pearson's r = 0.47, P < 0.01).

CONCLUSIONS

The pNMS provides an individualized imaging marker for guiding laser ablation and predicting postoperative seizure outcomes. This approach supports more precise surgical planning and may improve long-term prognosis in patients with mesial temporal lobe epilepsy.