Abstracts

Rationale: Accurate delineation of the epileptogenic cortex is indispensable as more than 30% of patients with epilepsy undergo surgical resection of their epileptic focus, and at least 1/3 have seizure recurrence after surgery.Studying the epileptic networks can lead to improved methods for accurate delineation of the epileptogenic zone, ability to predict surgical outcomes and clear understanding of the effects of recurrent seizures on various cortical regions. Hence, we intended to study the local (epileptogenic cortex) and long range network disturbances, in different frequency bands. Methods: Fifty-two patients (M: F=37:15; age=23±8.9years; disease duration 11.4±7.7years) with medically refractory mesial temporal lobe(n=25) and focal neocortical (n=27) epilepsies and 25 age-gender matched healthy controls were longitudinally recruited and studied with MEG. Epileptogenic cortex was surgically resected out, and the clinical outcome was assessed during the follow-up(22.1±10months). Spatial covariance matrix and lead field matrix was computed for a 4min continuous resting state data on each subject specific brain grid. Source time course was extracted from each location using an adaptive spatial filter. The connectivity analysis was performed using imaginary coherence at the level of the epileptogenic cortex and whole brain in delta(1-4Hz), theta(4-8Hz), alpha(8-14Hz), beta(14-30Hz), low-gamma (30-54Hz), high-gamma (55-80Hz) and ripple (80-200Hz) frequency bands. Epileptogenic cortical connectivity (over the surgical resection area) was computed, and voxel-wise comparison was made with the contralateral homologous cortex while global connectivity was computed for each voxel to every other voxel in the brain and contrasted against healthy controls.  Results: With surgery, 44(84.6%) patients had good(Engel-1) and 8(15.4%) patients had poor (Engel-2 to 4) clinical outcome. Patients with increased epileptogenic cortical connectivity within the resection area were more seemingly to achieve postsurgical seizure freedom (91% with Engel I outcome) than those with neutral and (71% seizure free) or decreased (60% seizure free) regional connectivity (P < 0.046, RMANOVA). Compared with the healthy controls, mean global functional connectivity (both mTLE and FNE) was decreased in dorsolateral frontal, temporoparietal, sensorimotor and parietooccipital cortices (P < 0.01, T-test) at all frequency bands, except at 30-54Hz band. The decreased global connectivity was directly related with longer duration of epilepsy and the frequency of seizures (P < 0.01, linear regression). Patients with decreased global connectivity in 14-30Hz, 55-80Hz and 80-200Hz were more likely to achieve seizure freedom. Conclusions: This is the first MEG study which attempted to investigate the role of functional connectivity as a biomarker in patients with epilepsy independent of frequency bands and demonstrates that both regional and global connectivity measures could predict the clinical outcome.  Funding: N.A