The current study uniquely demonstrates that seizures alter the functional connectivity of eloquent cortical areas and that these alterations are predictive of clinical neurological deficit. We also provide direct evidence that seizure-induced alterations of connectivity in functional networks, which may be distant from iEEG-defined SOZ or presumptive epileptogenic MRI lesions, are associated with neurological impairments. We also provide the first evidence that invasion of function-specific areas of eloquent cortex by ictal connectivity dynamics are selectively related to impairment of the relevant functional domain (e.g. ictal desynchronization of motor cortex is selectively relevant to motor impairment). Hand motor function was selected to test this hypothesis as it represents a relatively simple, robust network, for which the implicated cortical regions are reliably identified through cortical stimulation 
. Based on our findings, we speculate that ictal phase desynchronization and disruption of functional connectivity within a variety of distributed brain networks may underlie the broad spectrum of impairments affecting children with epilepsy. This view is consistent with EEG evidence linking gamma-band synchronization to the formation of distributed neuronal coalitions supporting a variety of cognitive and perceptual processes 
as well findings of transient ictal desynchrony during aberrant emotional behaviour 
. The observed association between ictal reduction gamma-band synchrony within Rolandic cortex and motor deficit may therefore represent a mechanism through which epileptic seizures exert long-lasting effects on cortical network dynamics and consequently neuropsychological function.
It remains unclear why in the present study seizure-induced changes in functional connectivity were found to be independently associated with motor weakness, although this finding is supported by clinical associations between longer duration of epilepsy and increased baseline functional impairment. One possible explanation is that prolonged desynchronization and disconnection of functional networks may facilitate network plasticity within the Rolandic cortex, resulting in clinical motor deficit. It has been previously shown that the capacity of individual neurons to exhibit adaptive changes or plasticity is influenced by gamma synchronization 
. Furthermore, coherence of oscillatory gamma-band EEG activity has been previously studied as a basis for cognitive processes necessitating neuronal plasticity, such as learning and memory 
. A critical implication of our findings is that aberrant gamma synchrony may act to pathologically decorrelate neurons comprising a functional circuit, resulting in long-lasting disruptions in connectivity and thus motor impairments.
Another explanation for our findings involves the role of interictal discharges in disrupting networks beyond the ictal period, further contributing to network destabilization. Using EEG-correlated functional MRI, it has been reported that interictal epileptic discharges disrupt resting-state networks in a manner analogous to task performance 
. However, it has also been shown that patients with epilepsy exhibit resting-state network impairments during interictal periods without interictal epileptic discharges and that functional connectivity is negatively correlated with disease duration 
. Based on the findings of the current study, we hypothesize that dysfunctional network integration may be related to repeated ictal desychronization and functional disconnection within brain networks supporting motor function. This viewpoint is further buttressed by recent demonstration that phase correlations among gamma oscillations in distributed neuronal coalitions contribute to the formation of task specific network interactions involving the motor system 
A wealth of literature has also recently emerged suggesting that pathological high frequency oscillations (pHFOs; above 80 Hz) are a signature of cortical epileptogenicity 
. One study showed that the presence of ictal motor symptoms was related more to pHFO amplitude in Rolandic cortex than in the SOZ, and that augmentation of ripple-band pHFOs (80–200 Hz) occurred approximately 400 ms prior to EMG onset of ictal motor phenomenon 
. In contrast to studying oscillation amplitude, we examined phase-locking synchrony and graph theoretical properties of gamma-band networks and showed that ictal disturbances in connectivity are associated with baseline motor deficit. Furthermore, we demonstrated that the magnitude of motor impairment was correlated with the extent of desynchronization and functional disconnection within Rolandic cortex. Using multivariate analysis, we have also shown that these findings are more predictive of deficit than epileptogenicity (SOZ presence) in the Rolandic cortex. It is also interesting that in the current study, the most significant frequency was the high-gamma band (81–150 Hz) (p<0.01), suggesting that ictal desychronization was strongest within the gamma-band, which has been reliably implicated in the formation of networks supporting cognition, perception and motor control. Finally, while previous studies have shown that the SOZ is itself functionally disconnected 
, we show that ictal disconnection of eloquent cortical areas independent of the SOZ location is associated with neurological deficits. We speculate that this may be a leading mechanism for neurological and cognitive impairment in children with epilepsy.
The present study possesses numerous advantages over previous works. Importantly, large subdural grids were used for electrocorticography, allowing us to capture a greater number of nodes and to more accurately apply graph theoretical analyses to map local and inter-regional connectivity involving Rolandic cortex. Secondly, electrophysiological synchrony and graph theoretical properties were associated with clinical and neuropsychological impairments. Finally, the epileptogenic pathology was the same (focal cortical dysplasia) for all children. Our main limitation is the absence of a control group, which has been previously reported for intracranial EEG 
The current report is the first to bridge the gap in knowledge between disturbances in functional connectivity caused by epilepsy and clinical impairments observed in affected children. We show that seizure-induced desychronization of Rolandic cortex is associated with contralateral motor hand deficits independently of the location of the SOZ. Furthermore, ictal disruptions in functional connectivity, shown by local declustering of the Rolandic cortex, are also associated with motor deficit following adjustment for SOZ location. Finally, our study has the advantage of correlating observed seizure-induced changes with clinical and neuropsychological outcomes, rather than statistical differences in BOLD signal on fMRI. These findings were highly significant in the gamma frequency range, and disturbance of network dynamics involving motor cortex were selectively related to motor function. We present evidence for a plausible mechanism for network impairment due to epileptic seizures.