A generalized epilepsy network derived from brain abnormalities and deep brain stimulation

Abstract

Idiopathic generalized epilepsy (IGE) is a brain network disease, but the location of this network and its relevance for treatment remain unclear. We combine the locations of brain abnormalities in IGE (131 coordinates from 21 studies) with the human connectome to identify an IGE network. We validate this network by showing alignment with structural brain abnormalities previously identified in IGE and brain areas activated by generalized epileptiform discharges in simultaneous electroencephalogram-functional magnetic resonance imaging. The topography of the IGE network aligns with brain networks involved in motor control and loss of consciousness consistent with generalized seizure semiology. To investigate therapeutic relevance, we analyze data from 21 patients with IGE treated with deep brain stimulation (DBS) for generalized seizures. Seizure frequency reduced a median 90% after DBS and stimulation sites intersect an IGE network peak in the centromedian nucleus of the thalamus. Together, this study helps unify prior findings in IGE and identify a brain network target that can be tested in clinical trials of brain stimulation to control generalized seizures.

Fig. 1. Coordinate locations and ALE.

Coordinates (n = 131), shown as red spheres, were heterogeneously distributed across the brain (A). An ALE analysis identified consistent brain abnormalities in the bilateral thalamus (B). Notably, only 17% of all coordinates were in the thalamus (C). *Note that spheres of coordinates may be assigned to multiple lobes or regions when located at the borders between them. Source data are provided as a Source Data file. ALE activation likelihood estimation.

Fig. 2. Coordinate network mapping.

Study-level coordinates (A, left) were used as seeds in an atlas of human brain connectivity (i.e., a human connectome) to identify the functional brain network connected to these coordinates. (A, right). Coordinate networks for each study (n = 21) were then overlapped to identify a common brain network deriving an IGE network (B). These same functional connections were found to be specific to IGE compared to coordinates from neuroimaging abnormalities in neurodegenerative diseases (C) or randomly distributed coordinates (D). FDR false discovery rate, IGE idiopathic generalized epilepsy.

Fig. 3. Multimodal validation.

Brain regions atrophied in IGE (white outlines) as identified in the worldwide ENIGMA study were part of the IGE network and overlapped more with this network compared to a null distribution of randomly selected brain regions (A). Brain regions activated by epileptiform discharges during simultaneous EEG-fMRI (white dots) in generalized epilepsies (n = 6) show higher overlap with the IGE network compared to focal epilepsies (n = 6) in a two-sided t-test (t = 2.955, df = 9.7124, P = 0.0149, 95% CI [0.038, 0.389]). This is shown in the boxplot where the center is the median overlap with the IGE network bound by the 25th and 75th percentile of the data. The whiskers extend from the lower and upper quartile to the minimum and maximum, respectively. The points on the plot each mark a unique EEG-fMRI study on either generalized or focal epilepsy (B). Lesion locations (white outlines) associated with epilepsy (n = 347) overlap with negative functional connections in the IGE network more than lesions not associated with epilepsy (n = 1126) in a two-sided t-test (t = 12.69, df = 581.9, P < 2e-16, 95% CI [0.2,+∞]). The distribution of the data is shown in a raincloud plot which displays individual epilepsy and control cases in the jitter plot. The boxplot shows the median network overlap of these two groups by the center line and is bound by the 25th and 75th percentiles of the data. The whiskers of the plot mark data that fall within 1.5 times the interquartile range (C). Scalp EEG electrodes (white circles) in frontocentral regions (Cz) overlap with positive functional connections of the IGE network, while electrodes in anterior frontal regions (Fp1/Fp2) overlap with negative functional connections (D). Source data are provided as a Source Data file. EEG-fMRI electroencephalogram-functional magnetic resonance imaging, IGE idiopathic generalized epilepsy.

Fig. 4. Alignment with the somato-cognitive action network.

The motor cortex homunculus (A, panel adapted under the Creative Commons Attribution (CC-BY) license from Gordon et al. 2023) includes effector and inter-effector regions (B, left). Inter-effector regions (n = 6) align with the IGE network (B, right) and show higher overlap compared to effector regions (n = 5) in a two-sided t-test (t = 3.96, df = 7.17, P = 0.005, 95% CI[0.081, 0.38]). The boxplot shows the center line as the median network overlaps for each effector and inter-effector region (dots) enclosed by the 25th and 75th percentiles of the data, while the whiskers extend to the maximum and minimum (C). The peak nodes in the SCAN (D) were used as a seed in the human connectome to generate a whole-brain map of the SCAN (E, left), which was spatially similar to the identified IGE network (spatial r = 0.81; E, right). Coordinates of brain abnormalities in IGE were most positively connected to the SCAN and most negatively connected to the default mode network (F). Source data are provided as a Source Data file. IGE idiopathic generalized epilepsy, SCAN somato-cognitive action network.

Fig. 5. Relevance for deep brain stimulation.

DBS electrode locations implanted to treat drug-resistant generalized seizures in patients with IGE (n = 21) were localized with Lead-DBS software and plotted in relation to the CM (A, red). Notably, the IGE network peaked in the CM of the thalamus (B), which was the most functionally connected thalamic nucleus (C). Seizures reduced a median 90% after CM-DBS in 21 patients with IGE (D). The IGE network was projected onto a publicly available ultra-high resolution ex vivo brain aligned to MNI space (E, warm colors), and DBS electrodes intersected with the peak of the IGE network in the CM (mesh). Source data are provided as a Source Data file. AV anterior ventral nucleus, CM centromedian nucleus, DBS deep brain stimulation, Hb habenular nucleus, IGE idiopathic generalized epilepsy, LGN lateral geniculate nucleus, MD mediodorsal nucleus, MGN medial geniculate nucleus, Pul pulvinar nucleus, SCAN somato-cognitive action network, VA ventral anterior nucleus, VLa ventral lateral anterior nucleus, VLpd ventral lateral posterior nucleus (dorsal part), VLpv ventral lateral posterior nucleus (ventral part), VPL ventral posterior lateral nucleus.

Fig. 6. Illustration of the potential clinical translation to image-guided DBS.

The DBS electrode locations of an independent patient with IGE treated with CM DBS was plotted in relation to (1) the peak voxel of the IGE network (MNI coordinate: x = −9.05, y = −21.07, z = −0.07) in the thalamus (A, red), (2) previously reported DBS sweetspots in IGE (B, green) and LGS (B, purple), and (3) discriminative fiber tracts associated with improved generalized seizure control after CM DBS (C, pink). This IGE network peak converged on a similar location to these optimal DBS sites, yet 4 mm closer to the sweetspot derived from IGE versus LGS patients (D). CM the centromedian nucleus, DBS deep brain stimulation, IGE idiopathic generalized epilepsy, LGS Lennox–Gastaut Syndrome.

Fig. 7. A convergent generalized epilepsy network.

The IGE network (A) was spatially similar to the CM DBS network (B), and a convergent IGE network was identified (C). CM the centromedian nucleus, DBS deep brain stimulation, IGE idiopathic generalized epilepsy.