Abstracts

Rationale: Even though surgery is by far the most effective therapeutic option for patients with drug-resistant focal epilepsy, up to 40% of operated patients do not remain seizure-free in the long term. One possible explanation for these failures is that seizures are generated by distributed neuronal networks involving spatially distinct, but functionally interconnected brain areas. Delineating these networks in order to inform clinical practice, however, remains difficult. One important hypothesis states that seizure networks might exhibit a high degree of internal connectivity while being relatively disconnected from the remainder of the brain. Intrinsic functional connectivity MRI (iFC-MRI) can assess the functional connectivity of the entire human brain non-invasively. Cortico-cortical evoked potentials (CCEPs), an invasive procedure utilizing intracranial EEG (iEEG) electrodes, allow assessment of the direction of connectivity between brain regions (Keller et al., Proc Natl Acad Sci U S A. 2011;108:10308-13). Here, we used iFC-MRI and CCEPs to test the "internally hyperconnected, externally hypoconnected" hypothesis. Methods: We recorded iFC-MRI and CCEPs in 10 patients with drug-resistant epilepsy undergoing iEEG monitoring. For iFC-MRI acquisition, patients rested in the scanner with their eyes closed for 7 minutes. We computed the whole-brain connectivity to each electrode. For CCEPs, we stimulated pairs of neighboring electrodes while recording the iEEG from the remaining electrodes. We averaged responses using bipolar channels to match the bipolar stimulation sites. We considered significant CCEP responses during the early response period (10-50 ms) as evidence of connectivity (1-sample t-tests, FDR-corrected across timeframes and electrodes for each stimulation site to an alpha level of 0.01). We reviewed the ictal iEEG to map out electrodes involved in seizure onset and spread. We compared these seizure networks with iFC-MRI and CCEP maps. Results: The correspondence between the seizure and CCEP networks is shown for 2 patients (Figures 1-2, A: seizure network; B: CCEP responses to stimulation of the seizure onset zone; C-D: CCEP responses to stimulation of the early spread zone). In both cases, stimulating the seizure onset or early spread zone activated a distributed network of brain areas that overlapped with the brain areas involved in seizure activity, supporting the "internally hyperconnected, externally hypoconnected" hypothesis. In Figure 1, the correspondence between CCEP and iFC-MRI connectivity is also apparent (Keller et al., 2011). Conclusions: CCEPs can reveal the effective connections underlying seizure networks. The correspondence between CCEPs and iFC-MRI opens the possibility of prospectively delineating seizure networks for better targeting of electrode implantation and guidance of resective surgery.