Thalamocortical circuits in idiopathic generalized epilepsies with different subsyndromes and preserved global functional architecture

To the Editor: As idiopathic generalized epilepsies (IGEs), juvenile myoclonic and absence epilepsies (JME and JAE) typically manifest during adolescence. These disorders share key clinical features like adolescent-onset seizures and high relapse rates after the withdrawal of antiepileptic drug therapy.[1] It has been believed that the thalamocortical system participates in the imbalance of cortical excitation and inhibition of IGEs.[2] However, less is known about whether these abnormalities in thalamocortical circuits are specific to subsyndromes. Moreover, no research has explored whole-brain functional connectivity (FC) by a graph theory approach in IGEs focused on patients with JME and JAE. Magnetic resonance imaging (MRI) images were obtained by use of standard sequences with a Siemens 3T MRI system (Skyra, Siemens, Erlangen, Germany). The system has a head coil with 12 channels. Foam pads were used for reducing head movement, and earplugs were used for reducing the noise of the scanner. Brain functional images were got using gradient-echo T2*-weighted echo planar imaging: 30 5-mm-thick slices, 205 volumes, a repetition time of 2000 ms, an echo time of 30 ms, a field-of-view of 24 cm × 24 cm, a matrix of 64 × 64, an in-plane resolution of 2 × 2 mm, and a flip angle of 90°. Blood oxygen level-dependent functional MRI (fMRI) data were gathered in consecutive runs of 410 s. In the same session, T1-weighted structural brain images were acquired with a three-dimensional spoiled gradient-recalled sequence: 176 axial slices (a thickness of 1 mm, a repetition time of 1900 ms, a echo time of 2.26 ms, a field-of-view of 256 × 256 mm2, a flip angle of 9°, and a matrix of 320 × 320). During MRI scanning, a 32-channel electroencephalogram (EEG) device (EBNeuro Mizar 40, Florence, Tuscany, Italy) with a sampling rate of 4 kHz was applied to record synchronous EEG. BE-MRI Toolbox (La Sapienza, Rome, Italy) was adopted to filter magnetic resonance (MR) artifacts. Two professional epileptologists would identify interictal generalized spike-and-wave discharges (GSWDs) occurring during scanning. The subsequent analysis gave up entire MRI sessions with interictal GSWDs to avoid the impact of interictal epileptic discharges. In the case of a seizure during fMRI scanning, postictal effects were avoided by discarding the EEG-fMRI scanning of the next session. The same patient was required to be scanned again if seizures or spikes were recorded in a session. Only one session without seizures or spikes for participants was conducted. They were matched for every patient and control. DPABI V3.1 (http://rfmri.org/dpabi) and SPM8 software package (http://www.fil.ion.ucl.ac.uk/spm) were leveraged to preprocess structural and functional images. All participants had a consistent volume of data. The first 10 volumes of fMRI data were cancelled. Next, the steps below were taken: slice timing was corrected. All functional volumes were realigned first, and then functional volumes and the T1-volume were coregistered. Each voxel was realigned with the relevant T1-volume, spatially normalized to the space of the Montreal Neurological Institute (MNI), resampled to 3 × 3 × 3 mm3, smoothed by use of a 4-mm Gaussian kernel, and filtered using a band-pass filter (0.01–0.08 Hz). In addition, the confounding of head movement was reduced using Friston’s 24-parameter model. Multiple linear regression analysis was used for regressing out nuisance covariates like cerebrospinal fluid signals as well as white matter. Subjects whose maximum displacement was greater than 2 mm or rotation was above 2° were not included. This study was exclusive of participants with an average framewise displacement (FD) of over 0.2 mm. To explore whether specific thalamocortical circuit alterations exist between JME and JAE, the FC analysis took the 16 bilateral thalamic subregions drawn from human subcortex atlases[3] as seeds. Seed regions of interest (ROIs) are detailed in Supplementary Figure 1, https://links.lww.com/CM9/C751. The seed-based correlation approach was carried out based on DPABI software. Pearson correlation coefficients between all the other voxels in a gray matter mask and the average time series of every seed region were calculated. Later, Fisher’s r-to-z transformation was utilized for converting all correlation maps into z-value ones to raise the normality of correlation distribution. At last, z-FC maps of subjects were obtained for group statistics. A graph theory analysis was conducted through the GRETNA 2.0.0 toolbox (https://www.nitrc.org/projects/gretna/) to investigate whole-brain FC measures. Nodes for whole-brain analysis were defined by parcellating the brain into 160 ROIs in accordance with a functionally defined Dosenbach’s atlas. Pairwise Pearson correlation coefficients among all of these ROIs were calculated, and the results were converted by using Fisher’s Z-transformation. To estimate network metrics, multiple sparsity thresholds from 5% to 40% with an interval of 0.01 were used. Two topological organization parameters, global efficiency and small-worldness, were chosen. They are the most representative of the functional integration of a network. An analysis was made of the area under the curve (AUC) for every network metric with superior sensitivity. One-way analysis of covariance (ANCOVA) was used for comparing age, whereas Kruskal–Wallis analysis of variance (ANOVA) was used for comparing sex among JME, JAE, and healthy control (HC) groups. Disease duration and onset age between JME and JAE were compared using a two-sample t-test. ANCOVA was adopted to examine group effects on FC among epilepsy groups and HCs. Gender, age, and FD values were covariates. Post hoc analysis was performed on significant group effects identified by ANCOVA. The Tukey–Kramer honestly significant difference correction method was used to make multiple comparisons (corrected P <0.05). Gender, age, and FD values were covariates. The resulting statistical map was corrected by the Gaussian random field theory. The threshold was set as P <0.01 and P <0.05 at voxel and cluster levels, respectively, with z scores above 2.33, two-tailed. Once brain regions exhibited significant differences for every seed of the thalamus among groups, two-tailed partial correlation analyses were further conducted by regressing out gender, age, and FD values for detecting the underlying associations between FC with significant differences among groups and clinical measures like disease duration and onset age. A general linear model was used. Every global network parameter and subject group was taken as a dependent and independent variables, respectively. Besides, gender, age, and FD values were used as covariates. This was aimed at identifying significant differences of topological parameters (including small-worldness and global efficiency) in AUC values among JME, JAE, and HC groups. It was a two-sided test performed by SPSS 21.0 (IBM Corp., Armonk, NY, USA). It was considered that a P-value below 0.05 showed statistical significance. To validate the seed-based FC findings, an alternative thalamic parcellation mask from the Human Brainnetome Atlas was utilized as the ROI. Furthermore, Craddock’s functional clustering atlas (comprising 200 ROIs) was used to confirm the reliability of whole-brain FC metrics. This demonstrates that small-worldness and global efficiency showed no significant differences among the three groups (details are provided in Supplementary Material, https://links.lww.com/CM9/C751). Patients were recruited from the Epilepsy Center of the Neurology Department at West China Hospital, Sichuan University, starting in December 2017. This study was approved by the Ethics Committee of Biomedical Research, West China Hospital of Sichuan University (2021 [24]) and registered with the Chinese Clinical Trial Registry under the identifier ChiCTR2000033957. Of those recruited participants, three patients and one HC were not included because of excessive head motion with a rotation of more than 2° and a maximum displacement of over 2 mm. In addition, six patients whose average FD (Jenkinson value) exceeded 0.2 mm were excluded. Finally, 41 patients (20 with JME and 21 with JAE) and 21 HCs were left for further analysis. A record was kept of their clinical details, as presented in Supplementary Table 1, https://links.lww.com/CM9/C751. The three groups were found to be not statistically different in gender, age, and head motion as per average FD (average JME: 0.0875 ± 0.0397, average JAE: 0.1145 ± 0.0573, and average HCs: 0.1161 ± 0.0559; P = 0.15). No difference was detected in disease duration and age of disease onset between the JME and JAE groups. The results of each ANOVA among these three groups are summarized in Supplementary Table 2, https://links.lww.com/CM9/C751. In comparison with HCs (P <0.05, corrected by Gaussian random field), patients with JME demonstrated decreased connectivity between the left posterior division of inferior ventroanterior thalamus (THA-VAip-lh, ROI1) and right ( T = −3.57, P = 0.0007) and left orbital frontal cortex (OFC) (T = −3.89, P = 0.0003). They also exhibited decreased connectivity between left medial ventroposterior thalamus (THA-VPm-lh, ROI5) and right dorsolateral prefrontal cortex (DLPFC) (T = −4.28, P <0.0001), left lateral dorsoanterior thalamus and left occipital lobe (T = −3.66, P = 0.0005), as well as right dorsoposterior thalamus and cerebellum (T = −4.01, P = 0.0002). Patients with JAE showed decreased connectivity between THA-VAip-lh (ROI1) and right cerebellum (T = −4.26, P <0.0001) and right putamen (T = −4.02, P = 0.0002). They also exhibited decreased connectivity between the right posterior division of inferior ventroanterior thalamus and cerebellum (T = −3.98, P = 0.0002) and right middle occipital gyrus (T = −4.30, P <0.0001). Moreover, they demonstrated decreased connectivity between THA-VPm-lh (ROI5) and left (T = −3.73, P = 0.0004) and right OFC (T-value = −3.12). In patients with JAE, decreased connectivity was detected between the left lateral ventroposterior thalamus and cerebellum (T = −3.86, P = 0.0003). Decreased connectivity was noted between right lateral ventroposterior thalamus and right DLPFC (T = −4.35, P <0.0001) and cerebellum (T = −3.73, P = 0.0004) in patients with JAE. Furthermore, decreased connectivity was noticed between right medial dorsoanterior thalamus and left DLPFC (T = −4.43, P <0.0001). Compared with HCs, JME and JAE groups exhibited decreased FC in thalamic-cerebellum circuits and the prefronto-thalamic network. Prefrontal-thalamic hypoconnectivity manifests in the two syndromes and involves the dorsal lateral prefrontal cortex and OFC [Supplementary Figure 2 and Supplementary Table 3, https://links.lww.com/CM9/C751]. Compared with patients with JAE, those with JME demonstrated increased connectivity between THA-VAip-lh (ROI1) and right (T = 4.04, P = 0.0002) and left putamens (T = 3.67, P = 0.0005), as well as between right anterior division of inferior ventroanterior thalamus and left (T = 4.15, P = 0.0001) and right putamens (T = 4.83, P <0.0001). This pattern of altered striatal-thalamic connectivity suggests its potential as a neural biomarker for distinguishing subsyndromes within IGEs [Supplementary Figure 2 and Supplementary Table 4, https://links.lww.com/CM9/C751]. The three groups showed no differences in global efficiency (F = 1.849, P = 0.167) and small-worldness (F = 2.343, P = 0.105) (detailed in Supplementary Material, https://links.lww.com/CM9/C751). These results aligned with prior research that global efficiency was intact in IGEs.[4] They indicate that IGEs involve specific thalamocortical network abnormalities with a relatively complete functional architecture of the whole brain. Patient with JME or JAE exhibited no significant associations in the greatly different FC of thalamus seeds and clinical variables (all P values >0.05, with FDR correction). A few limitations can be found in this study. Validation in larger cohorts with longitudinal follow-up is warranted due to the modest sample size. It is necessary to consider residual confounding from antiepileptic drugs despite the inclusion of 13 drug-naive patients. While preprocessing mitigated motion effects, some residual influence remains. The focus on thalamocortical circuits and whole-brain topology excluded other potentially relevant networks. Altered thalamic connectivity was not significantly correlated with clinical variables across groups. Notably, the findings from seed-based connectivity analyses showed limited convergence with graph-theoretical approaches. Longitudinal studies are needed to establish clinical correlations. This study attempts to evaluate subtype specific thalamic-cortical circuitry and whole-brain topologic organization in IGE at the seed-based level and the whole-brain scale. Our results support the notion that both JME and JAE are associated with thalamo-cortical network abnormalities but with preserved global functional architecture. Specifically, both the JME and JAE group exhibited decreased FC in thalamic-cerebellar circuits and prefronto-thalamic network. The difference of thalamocortical circuit between JME and JAE was in striatal-thalamic circuit. We highlighted the convergence and divergence of thalamocortical circuit dysfunctions in patients with JME and JAE, which may provide crucial insights into pathophysiological mechanisms of the two IGE subtypes. Conflicts of interest None.