Altered regional activity and inter-regional functional connectivity in psychogenic non-epileptic seizures

Although various imaging studies have focused on detecting the cerebral function underlying psychogenic non-epileptic seizures (PNES), the nature of PNES remains poorly understood. In this study, we combined the resting state fMRI with fractional amplitude of low-frequency fluctuations (fALFF) and functional connectivity based on the seed voxel linear correlation approach to examine the alterations of regional and inter-regional network cerebral functions in PNES. A total of 20 healthy controls and 18 patients were enrolled. The PNES patients showed significantly increased fALFF mainly in the dorsolateral prefrontal cortex (DLPFC), parietal cortices, and motor areas, as well as decreased fALFF in the triangular inferior frontal gyrus. Thus, our results add to literature suggesting abnormalities of neural synchrony in PNES. Moreover, PNES exhibited widespread inter-regional neural network deficits, including increased (DLPFC, sensorimotor, and limbic system) and decreased (ventrolateral prefrontal cortex) connectivity, indicating that changes in the regional cerebral function are related to remote inter-regional network deficits. Correlation analysis results revealed that the connectivity between supplementary motor area and anterior cingulate cortex correlated with the PNES frequency, further suggesting the skewed integration of synchronous activity could predispose to the occurrence of PNES. Our findings provided novel evidence to investigate the pathophysiological mechanisms of PNES.

Inter-regional functional connectivity. Relative to HC, a significantly abnormal FC between 6 seeds and a number of cortical regions were observed. Table 3 and Fig. 2 show the results of the between-group FC comparison with the seeds (p < 0.05, AlphaSim corrected). For the seeds with higher fALFF value (left SFG, left precuneus, left paracentral lobule, right postcentral gyrus and left SMA), the PNES group displayed increased FC in the precuneus, anterior and median cingulate cortices, postcentral gyrus and particular regions of the frontal and parietal cortices. For the seed with lower fALFF value (right triangular IFG), we detected decreased FC in the middle frontal gyrus in patients with PNES.
Correlation analyses. The frequency of PNES was not correlated with the fALFF values in any of the seed regions. However, we observed a significant correlation between the mean FC values and the frequency of PNES. As shown in Fig. 3, the FC between SMA and left anterior cingulate cortex (ACC) was positively correlated with the frequency of PNES (r = 0.577, p < 0.05). In addition, there was no significant association between the fALFF values and altered connectivity z scores in any areas with significant connectivity alterations.

Reproducibility.
To test the reproducibility of our findings, we carried out both split-half and leave-one-out validations as suggested by a previous study 25 . For the split-half analysis, 9 PNES patients and 10 healthy controls were randomly selected from the full sample and underwent the same statistical comparison of fALFF. The results are illustrated in Fig. 4A, which replicated most of the findings from the full sample (p < 0.05, AlphaSim corrected). Subsequently, seed-based FC analysis was performed in the other half of the sample. The results showed that the overall previous FC findings (Fig. 2) were partly limited to the spatial extent of the half sample (Fig. 5A, p < 0.05, AlphaSim corrected). Visual inspection indicated that the FC patterns seeded only in left precuneus, left paracentral lobule, left supplementary motor area and right inferior frontal gyrus were highly reproducible. This limitation may be attributed to the relatively small sample size and the split-half analysis, which may have significantly reduced the statistical power of the group comparisons. Therefore, a leave-one-out validation was employed to test the reproducibility and robustness of the previous findings on fALFF and FC without losing statistical power. Specifically, we left one PNES patient out of the sample and performed the same group comparisons based on the permutated sample (i.e., 17 PNES vs. 20 healthy controls ), obtaining a total 18 two sample t-test images (p < 0.05, AlphaSim corrected). For each voxel in the t-test images, the number of tests where this voxel exhibited significant group differences was calculated as the reproducibility of the fALFF and FC differences between the two groups, respectively. The results showed that the difference in fALFF and FC patterns ( Fig. 4B and Fig. 5B) between two groups were similar to that in the full group. Together, these results suggest a high reproducibility of our findings.

Discussion
The present fMRI study investigated the resting state brain function by measuring the regional fALFF and inter-regional FC in patients with PNES. Patients with PNES showed significantly increased synchronous regional brain activity mainly in the dorsolateral prefrontal cortex (DLPFC), parietal cortices and motor areas, as well as decreased regional activity in the triangular IFG. Thus, our results add to literature suggesting that abnormalities of neural synchrony in PNES and extend these findings to the LFO domain. In addition, neural network deficits in PNES patients were revealed by seed analyses, including both increased and decreased FC. Moreover, the correlation analyses demonstrated a significant positive trending correlation between the FC values and the PNES frequency. These findings indicate that changes in the regional cerebral function are related to remote neural activity across a widely distributed brain functional network.
One cortical area where we observed significantly increased regional synchronous activity in PNES was the left PFC. Previous cerebral blood flow studies have demonstrated that PNES-related disorders exhibit altered cerebral activity in the prefrontal cortex 26 , which are believed to have implications for deficits in emotion regulation and executive function. Given that PNES is often accompanied by many diffuse psychological, psychiatric and somatoform symptoms 23,27 , the enhanced activation of PFC may be very likely to contribute to the impairment of executive control function and the manifestation of emotional triggers as somatosensory symptoms. Additionally, we noted that impairment of consciousness  Table 2. and reduced self-control are key features of most PNES (94% of cases in the reported sample). Both PFC and precuneus are components of the default mode network (DMN) associated with awareness 28 . A recent resting state networks study suggested that abnormalities in DMN may be important for the specific alterations of consciousness in PNES 29 .The increased spontaneous activity in PFC and precuneus in patients may therefore cause its awareness impairment or even loss. The precuneus, part of the DMN, is thought to be involved in self-consciousness and self-related mental representations during rest 30 . Increased regional fALFF in PNES was observed in the precuneus, and the functional integration of this region was also found to be abnormal. Our observations are also coherent with precuneus dysfunction that has been identified in patients with other conversion disorders 31 . These imaging findings provide converging evidence that suggest the function of PFC and precuneus in disrupting the normal integrative functions of consciousness in PNES.
We also observed hyperactivity in particular regions belonging to SMN, which is involved in self-initiated and involuntary muscle movements. It is important to note that PNES were characterized by stereotyped motor phenomenon. In our patients with PNES, several of the cases showed hypermotor or trembling movements of the extremities. The increased fALFF in SMN may thereby reflect altered motor control functions in PNES. These results are consistent with the findings from other work that showed abnormalities of the motor areas in PNES 17 . In a structural MRI study 32 , PNES patients exhibited abnormal cortical atrophy of the motor and premotor regions. In another study that employed the fMRI technique, van der Kruijs et al. 5 found that increased FC was involved in the sensorimotor cortex. Therefore, our findings of increased spontaneous activity in the motor regions together with previous studies strengthen the view that the SMN plays an important role in the pathogenesis of PNES.
It is potentially interesting that PNES displayed significantly decreased fALFF in the triangular IFG. Pars triangularis is part of the ventrolateral prefrontal cortex (VLPFC) and has been shown to play a role in cognitive control of memory 33 . A previous study demonstrated that cognitive integrative functions for dealing with social stress and memory are impaired in patients with PNES 34 . A recent study employing a data-driven technique indicated lower long-range functional connectivity density (FCD) of the triangular IFG in PNES 35 . For the regional cerebral function perspective, our findings provided new evidence for the impairments of the working memory in patients with PNES. More task-related studies are needed to identify the neuromechanism that contribute to altered spontaneous activity as observed in the current study.
Alterations of FC in PNES have attracted attention as a potential network level substrate of this disorder. Increased connectivity within the frontal-parietal and SMN has been reported 17,29 . A study involving the use of EEG synchronization found that PNES appeared to be associated with decreased prefrontal and parietal synchronization 4 .These inconsistent findings may be due to the different methodologies and parameters used, as well as the different characteristics of the participants. In this study, we demonstrated that the widespread altered inter-regional connectivity beyond the regional activity alterations in the PFC and motor areas was also related to PNES.
The altered connectivity pattern of PFC may provide new insights into how prefrontal areas affect motor system in patients with PNES. Although the PNES showed decreased coactivation in VLPFC, the DLPFC displayed enhanced connectivity with motor regions. A task-related study has found that conversion paralysis patients exhibited altered connection between DLPFC and various sensorimotor nodes during the formation of action plans, whereas ventromedial PFC was not functionally connected to the motor system 36 , thereby indicating a distinct connectivity patterns for different parts of the PFC. DLPFC plays an important function in the circuitry of emotional control, and it monitors attention resources and guides emotional responses 37 . The involvement of increased DLPFC-SMN connection is consistent with previous fMRI studies that have identified dysfunctional connectivity 5 and increased long range FCD 35 between emotional regions and SMN in PNES. As a result of this alteration, there is a failure for the PNES in the coordination and balance of different mental and somatic functions. Another main observation is the increased connectivity between ACC and SMA, and the enhanced connectivity is positively correlated with the frequency of PNES. The conversion disorder has been hypothesized to affect the abnormal inhibition of motor systems by limbic regions 38 .The proposed interaction model between limbic and motor control regions is consistent with previously increased amygdala connectivity with SMA in patients with conversion disorder 26 . Our results may add to these findings and suggest an altered functional organization that involves the limbic regions and the sensorimotor cortex in PNES. ACC is an important region functionally connected with SMA 39 , and the connectivity between them may be involved in conveying information about negative emotion to the motor areas associated with executing goal-directed behavior 40 . Accordingly, it is reasonable to speculate that the ACC is a meeting place between cognitive and motivational processes; hence, altered connectivity between ACC and SMA in PNES may generate uncontrolled motor output. Furthermore, the altered connectivity between the SMA and ACC is correlated with the frequency of PNES, thereby suggesting a skewed integration of synchronous activity predisposed to the occurrence of PNES episodes. The results may be important therapeutically, such as selectively reducing or modulating FC, which may improve the disrupt balance and so prevent PNES events.
This study also has certain limitations. First, the sample size was modest and a relatively weak correction (AlphaSim program) was used for multiple comparisons in our study. A larger sample is needed to replicate and support these results. Second, this study employed a cross-sectional design and is thus Figure 2. Group differences of inter-regional FC maps between seeds and the rest voxels of the brain. Six abnormal fALFF regions (superior frontal gyrus, precuneus, paracentral lobule, postcentral gyrus, supplementary motor area, and inferior frontal gyrus) were defined as seeds for FC analyses. Compared to healthy controls, patients with PNES showed increased (red) and decreased (blue) connectivity based on these seeds. The results were corrected by AlphaSim(all the clusters survived p < 0.05, a combined threshold of p < 0.01 with a minimum cluster sizes of 116,115,162,150,79 and 157 voxels, respectively). Color bar indicates the T score. More details of altered FC regions are described in Table 3. can not provide a test-retest reliability [41][42][43] . A follow-up design could provide longitudinal data to clarify abnormal spontaneous neural activity and functional network deficits in PNES. In addition, although ALFF is believed to reflect spontaneous neural activity, there is a lack of understanding about its exact physiological nature. Finally, no significant correlation was observed when we examined the relationship between fALFF and FC in the PNES group. We speculate that the sample size was not enough to provide sufficient evidence to describe the relationship between them. A further study is required to better elucidate that whether the changes in the regional neuronal activity could cause an alteration in the inter-regional functions of patients with PNES.
Taken together, the findings of abnormal fALFF and FC in PNES were actually derived from the examination at two different analytical levels. The higher fALFF in the PFC and sensorimotor system suggests the regional abnormalities of neural synchrony in PNES, whereas the altered FC in the  fronto-parietal cortex and motor regions indicates that the alterations in the regional cerebral function is related to remote inter-regional functional network deficits. Accordingly, we claim that the dysfunction of the fronto-parietal cortex and SMN may be important factor that could affect the emotion regulation, executive control and motor function of PNES. Overall, our findings from both regional neural function and inter-regional network perspective provided novel evidence to investigate the pathophysiological mechanisms of PNES.

Methods
Participants. The subjects were recruited from our previous studies 17 . A total of 20 patients with PNES (7 males, mean age: 19.65 ± 7.56 years) and 20 controls (8 males, mean age: 21.85 ± 1.70 years) were recruited from West China Hospital. No significant difference in age (t = 1.27, p = 0.12) or sex (x = 0.11, p = 0.74) was found between the groups. PNES was diagnosed by the experienced neurologists using clinical descriptions of symptoms and long-term video-EEG monitoring based on the recent recommendations 44,45 . The detailed inclusion and exclusion criteria are described in the Supplementary materials. Four of the 20 patients were taking antiepileptic drugs before the diagnosis of PNES. All drugs were discontinued at least two weeks prior to MRI examination. The demographic and clinical characteristics for all patients are presented in Table 1. PNES frequency was assessed according to the reports of the patient or their family members. The median frequency was 2.5 per month (with a range of one per week to eight per week). Two patients have a long duration of disease with 18 years, because of which the two patients were excluded from the analysis to ensure the illness conditions of the patients are similar as possible. The HC neither had neurologic/psychiatric disorders evaluated using the SCID-Non-Patient Version, nor did they take any psychotropic medication within the past six months. This study was approved by the local Ethics Committee of West China Hospital and was carried out in accordance with the approved guidelines. All participants gave informed consent to participate in the investigation. Data preprocessing. Data preprocessing was performed using the SPM8 package (http://www.fil.
ion.ucl.ac.uk/spm) and included the removal of the first five volumes, corrected for time delay between slices and head motion, spatial normalization to the Montreal Neurological Institute (MNI) space and smoothed by convolution with an 8-mm full-width half-maximum Gaussian kernel. Finally, the six head motion parameters as nuisance covariates were regressed to reduce the effects of head motion, because a recent study demonstrated that head motion had a confounding effect on ALFF analysis 46 . fALFF Calculation. The fALFF analysis was carried out using the REST software (http://resting-fmri. sourceforge.net). After preprocessing in SPM, a linear trend was removed. To calculate ALFF measure at each voxel, the time series were transformed to the frequency domain by using fast Fourier transform. The power spectrum was then computed and square root-transformed at each voxel. Considering that the power of a given frequency is proportional to the square of the amplitude of this frequency component, the square root was computed at each frequency of the power spectrum and the averaged square root was obtained across 0.01-0.08 Hz at each voxel. This averaged square root was taken as the ALFF 13 . fALFF is the fraction of ALFF in a given frequency band to the ALFF over the entire frequency range detectable in the given signal 14 . fALFF of each voxel was computed for each subject, and it was further divided by the global mean value to reduce the global effects of variability across participants 13 . Functional connectivity analysis. FC was examined using a seed voxel linear correlation approach 47,48 . As discussed later, significant changes in fALFF measurements of PNES were detected in 4 clusters after AlphaSim correction (Table 2). We employed a seed-based approach whereby the mean time series for each participant were extracted from these clusters by masking the Automated Anatomical Labeling (AAL) template 49 . This way, the resulting six masks were defined as seeds for further FC analyses. To perform the analyses, the time series of fMRI data for each voxel were filtered by applying a bandpass filter (0.01-0.08 Hz). Each time series was further corrected for the effect of six head motion parameters by linear regression to reduce the influence of head motion. Recent studies have showed that functional connectivity analysis is sensitive to gross head motion effects 50,51 ; hence, the mean frame-wise displacement (FD) was calculated to further determine the comparability of head movement across groups 50 . The largest FD of all subjects was less than 0.2 mm. Two-sample t-test showed that there was no significant difference in FD between the two groups (0.116 ± 0.210 for HC and 0.117 ± 0.041 for PNES; p = 0.930). In addition, the averaged signals from the specific CSF and white matter masks were regressed based Scientific RepoRts | 5:11635 | DOi: 10.1038/srep11635 on the previous resting state fMRI studies 52,53 . Since removing the global signal may introduce a shift in the distribution of correlation coefficients (mainly the presence of negative correlations) and make biological interpretation ambiguous 54,55 , we did not include the global signal as a covariate variable. Then,  a reference time series was extracted by averaging the filtered signal of all voxels within the six regions showing a significant change in fALFF values. For each seed, a linear correlation analysis was conducted by calculating the correlation coefficient between the time series of the seed region and the time series of each voxel across the whole brain. The resultant r values were converted to z maps using Fisher's r-to-z transformation 56 to improve the Gaussianity of their distribution. Finally, we obtained 6 z-score maps of the seed to all other brain voxels for each subject.

Statistical analysis. The statistical analyses involved in a comparison between PNES patients and
healthy controls in terms of (1) regional cerebral fALFF values, (2) inter-regional FC between seeds and the rest voxels of the brain. We first performed a random effect one-sample t-test on all individual fALFF maps for the two groups respectively. This procedure produced significant fALFF statistical maps for both groups. The significance level for each group was set at p < 0.01 with FDR correction. We then made a mask that combines the two fALFF statistical maps to analyze the differences in fALFF between patients and controls. According to a previous study 14 , such a mask can remove the influence of large vessels and constrain our analyses within regions generating potentially meaningful LFOs. These same analyses were performed on FC maps, and the threshold of one-sample t-test was set at p < 0.01 corrected with FDR. Two-sample t-tests with age and gender as covariates were performed to identify the regions that showed significant differences in fALFF and FC between the two groups. The resulting statistical fALFF maps were corrected for multiple comparisons to a significant level of p < 0.05 by combining the height threshold of p < 0.01 with a minimum cluster size of 75 voxels, using the AlphaSim program in the REST software (http:/resting-fmri.sourceforge.net).The same statistical approaches as those described for the analyses of fALFF were used for thresholding the FC maps. All the statistical maps were corrected by AlphaSim at a significant level of p < 0.05 by combing height threshold of p < 0.01 with a minimum  cluster sizes of 116, 115, 162, 150, 79 and 157 voxels, respectively. Coordinates were reported in the Montreal Neurological Institute space provided in SPM8. Three Pearson correlation analyses were performed: 1) fALFF values in regions with alterations were correlated with PNES frequency (the average number of episodes that patients experienced for a month); 2) FC in regions with significant group differences was correlated with PNES frequency; 3) fALFF values in those regions with abnormalities were correlated with FC. Pearson correlation was calculated with age, gender and duration of disease as covariates to be regressed. The threshold of p < 0.05 was considered to be significant for these correlation analyses.