If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University, Beijing, ChinaStereotactic and Functional Neurosurgery Laboratory, Beijing Neurosurgical Institute, Capital Medical University, Beijing, ChinaBeijing Key Laboratory of Neurostimulation, Beijing, China
Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University, Beijing, ChinaStereotactic and Functional Neurosurgery Laboratory, Beijing Neurosurgical Institute, Capital Medical University, Beijing, ChinaBeijing Key Laboratory of Neurostimulation, Beijing, China
# Address: No.119 South 4th Ring West Road, Fengtai District, Beijing, China.
Affiliations
Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University, Beijing, ChinaStereotactic and Functional Neurosurgery Laboratory, Beijing Neurosurgical Institute, Capital Medical University, Beijing, ChinaBeijing Key Laboratory of Neurostimulation, Beijing, China
Four clinical patterns of automatisms were identified by k-means cluster analysis.
•
Automatisms from temporal and extratemporal localisations showed different semiologic features.
•
Clinical patterns 1, 3, 4 were significantly correlated with mesial temporal lobe epilepsy (MTLE), frontal lobe epilepsy (FLE), and neocortical temporal lobe epilepsy (NTLE), respectively.
•
Manual automatisms occurred earlier in the NTLE group and persisted longer in the MTLE group.
•
MTLE had a significantly higher proportion of automatisms than other groups.
Abstract
Objectives
To identify semiologic features of automatisms correlating to different seizure onset zones (SOZ).
Methods
In total, 204 seizures from 74 patients with either oral or manual automatisms were assessed. Patients were divided into four groups depending on the SOZ into frontal, posterior, neocortical temporal, and mesial temporal cortex groups. A k-means analysis was applied on 11 semiologic features on a multi-criteria scale. Then, the resulting clinical patterns were correlated with the SOZs determined by presurgical anatomy-electroclinical data (25 cases with stereo-EEG).
Results
Four clinical patterns of automatisms with different accompanying symptoms were identified. The clinical features of clusters 1 and 4 were mostly found in temporal epilepsy whereas clusters 2 and 3 were more frequent in extratemporal epilepsy. Cluster 1 was significantly correlated with mesial temporal lobe epilepsy (p = .017) and was characterised by aura, postictal confusion, short automatisms delay. Cluster 3 included 1/3 patients with frontal lobe epilepsy and was characterised by emotionality. Cluster 4 was related to neocortical temporal lobe epilepsy and characterised by dystonia and short automatism delay (p = .011).
Conclusion
The distinct semiologic patterns of automatisms may provide information which may allow clinicians to define the SOZs. These findings could improve diagnostic accuracy and surgical outcome.
Automatisms are defined by more or less coordinated motor activity usually occurring with impaired cognition and often, but not always, followed by amnesia [
]. Automatisms often resemble a voluntary movement and may consist of an inappropriate continuation of preictal motor activity. Oral and manual automatisms are the most common types [
]. Oral automatisms involve the mouth and tongue, including chewing, swallowing, lip-smacking, blowing, and whistling. Manual automatisms affect the distal extremities and include fumbling, picking, and gesticulating movements.
Automatisms were considered one of the most common signs of temporal lobe epilepsy (TLE) [
]. However, recent studies showed that automatisms could occur in seizures with other seizure onset zones (SOZs). These localisations included the frontal lobe [
Occipital lobe epilepsy: electroclinical manifestations, electrocorticography, cortical stimulation and outcome in 42 patients treated between 1930 and 1991. Surgery of occipital lobe epilepsy.
]. As a result, it is difficult to find the SOZs in seizures with automatisms. Some semiologic features are essential for distinguishing different types of epilepsy. For example, the frequency of automatisms has been proposed as a distinctive feature between TLE and frontal lobe epilepsy (FLE) [
Occipital lobe epilepsy: electroclinical manifestations, electrocorticography, cortical stimulation and outcome in 42 patients treated between 1930 and 1991. Surgery of occipital lobe epilepsy.
]. However, it still lacks the approach to localising automatisms seizures’ SOZ by semiologic features.
We hypothesised seizures with automatisms have some specific clinical patterns that were valuable for localizing SOZ. This study aimed to distinguish these clinical patterns by cluster analysis, then correlate them to the SOZs. We performed cluster analysis on a multiple-criteria scale with 11 clinical features to explore this issue.
2. Methods
2.1 Participants
We included 74 consecutive patients recruited from the Tiantan-Fengtai Epilepsy center between January 2015 and December 2019 who met the following inclusion criteria: (1) had at least one seizure with either manipulative manual or oral automatisms, that lasted longer than three seconds [
] during scalp EEG; (2) had been diagnosed with focal epilepsy; and (3) had undergone surgery with postsurgical follow-up over 12 months and had achieved ILAE (International League Against Epilepsy) class 1–4 surgical outcomes [
]. Detailed noninvasive presurgical evaluations, including medical history, neurological examination, video-EEG monitoring, high-resolution magnetic resonance imaging (MRI), positron emission tomography-computed tomography (PET-CT), and image postprocessing, were performed for all the patients. If the clinical features suggested a possible surgical indication and intracranial EEG was necessary to localise the SOZ, patients were selected to undergo stereoelectroencephalogram (SEEG) exploration. Twenty-five patients underwent SEEG before surgery. All patients gave informed consent before investigation. This research was approved by the Institutional Review Boards of the Beijing Tiantan Hospital. We have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.
2.2 Location of the SOZs
The locations of the SOZs were determined by multidisciplinary presurgical evaluation and SEEG data when necessary. Patients were classified into four groups according to the localisations of SOZs: mesial temporal, neocortical temporal, frontal, and posterior cortex groups. Since the symptoms [
] were different for neocortical TLE (NTLE) and mesial TLE (MTLE), TLE patients were divided into NTLE and MTLE groups. NTLE patients underwent tailored neocortical temporal lobectomy, whereas MTLE patients underwent anterior temporal lobectomy or selective amygdalohippocampectomy. Notably, the SOZs of posterior cortex epilepsy (PCE) patients are located in the parietal and occipital lobes and the occipital border of the temporal lobe [
]. No patient had insular SOZ. Instead, insula involvement was due to early electrophysiological spread after seizure onset (confirmed by SEEG) rather than SOZ. Specifically, there were five patients with early spread areas involving the insula. The SOZs of these patients were located in the orbital frontal cortex (OFC) in three cases (FLE), in the operculum in one case (FLE), and in the superior temporal gyrus in one case (NTLE). In addition to the SOZ location, the semiology and EEG pattern were also considered in the classification.
2.3 Automatisms scale
As in previous studies, a specific criterion was used to group seizures into semiologic clusters by the scores of the clinical features [
]. Based on previous literature, the features of the relevant symptoms, as well as the accompanying symptoms, showed potential value in cluster analysis [
]. In addition, the time feature (when automatisms appeared (delay) and how long it lasted (duration)) showed significant differences amongst the various focal epilepsy types[
]. In contrast, it was difficult to quantify some characteristics of oral and manual automatisms, such as the motor speed of the hypermotor, which reduced its localizing value. Thus, the movement features of automatisms were not included. Instead, nine accompanying symptoms and two quantitative time features (delay and duration) were included in the criteria. The nine accompanying symptoms were defined according to the criteria of ILAE commission reports in 2017 [
] (the referenced literature is listed in Table S1). Each semiologic component was scored for each patient, with values ranging from 0 (=absence of that feature) to a maximum of 2 for major features (=constant sign in each seizure), with the minor feature being scored as 1 (=sign not always present).
2.3.1 Hyperkinesia was characterised by complex movements predominantly involving the proximal limb or axial muscles. It included irregular sequential ballistic movements, an increasing rate of ongoing movements, or the inappropriately rapid performance of a movement [
2.3.2 Emotion was mainly based on facial expression. Emotion included spontaneous joy or euphoria, laughing (gelastic), or crying (dacrystic), amongst other features [
]. Considering the potential value for localisation, we classified the version into a single feature rather than a subcategory of the elementary motor.
2.3.6 Aura was defined as a subjective ictal phenomenon preceding an observable seizure [
]. The presence of aura was determined when patients mentioned feelings after seizures. Since only a few patients in the cohort exhibited auras, which were difficult to categorise in half of the cases, we did not consider the type of aura in the cluster analysis. Detailed aura information is included in Table S2.
2.3.7 Dystonia refers to sustained contractions of both agonist and antagonist muscles producing twisting movements and abnormal postures [
2.3.8 Autonomic signs refer to alterations in the autonomic nervous system, including heart rhythm disturbances, blood pressure changes, respiratory changes, and hypersalivation, amongst other features [
]. Therefore, we analysed this symptom after seizure termination. In addition to postictal confusion, almost all patients in this cohort exhibited more or less altered consciousness during the seizure; thus, this element would only provide limited information for our cluster analysis. Subsequently, altered consciousness during a seizure was not included, although it was considered a fundamental feature of complex partial seizures.
Based on previous studies, we used two semi quantified features for cluster analysis: delay [
] of automatisms, which corresponds to the first and second parts, respectively, as below. Each seizure was divided into three parts: 1) from the clinical onset to the appearance of automatisms, 2) from the start to the end of automatisms, and 3) from the end of automatisms to seizure termination. The clinical onset corresponded to the first visible change in behaviour or the announced aura; seizure termination corresponded to when symptoms stopped [
]. To minimise bias from slow or rapid seizure spread across the epileptic network, we scaled each of the three parts by dividing the total seizure duration [
]. Notably, the classification was performed at the patient level; therefore, oral and manual automatisms were not distinguished for each patient in this analysis [
]. In contrast, we separately performed quantified analysis for oral and manual automatisms in Section 2.4.3.
2.3.10 The duration measured how long oral or manual automatisms persisted. Patients were ranked two when automatisms lasted longer than two-thirds of the whole duration in most seizures, ranked one if only a part of seizures met this criterion, and ranked zero if none of the seizures met this criterion.
2.3.11 If the interval between the first symptom and the automatisms onset was shorter than one-third, the patient's automatisms delay score was two. When only a subset of seizures met the criteria described above, the patient would be ranked one, and that patient would be scored zero if automatisms always started later than the first one-third.
2.4 Statistical analysis
Two researchers analysed all the seizures independently and assessed the interrater agreement of symptoms by Cohen's kappa [
2.4.1 Cluster analysis and principal component analysis
First, each clinical feature was ranked based on the multiple-criteria scale described in Section 2.3. Then, principal component analysis (PCA) was performed on the ranks. Based on the first three components supplied by PCA, the k-means algorithm [
] was conducted on the automatisms scale consisting of all 11 clinical features. Before clustering analysis, the ‘elbow method’ was used to determine an optimal value for k [
]. The ‘elbow method’ consists of running the k-means for an increasing number of clusters. For each partition, the within-cluster sums of point-to-centroid distances were calculated as a function of the clusters number k (Fig. S1). This criterion was then combined with the clinical features to obtain clusters with sufficient size and clinical homogeneity.
2.4.2 Relationship between clinical features and SOZ
To identify whether some clinical features may occur more frequently in certain seizure types, we computed a correlation matrix between the clinical signs and the SOZ localisations. Two dissimilarity matrices (signs × patients and areas × patients) were calculated, and the Kendall correlation test [
] was then used to assess the significance at p < .05 for each pair of correlations.
2.4.3 Scaled onset time and duration of automatisms
Each seizure was divided into three parts and normalized using the same procedure mentioned in Section 2.3. For each patient, the duration of all habitual seizures was averaged for comparison in the next processing step. The Kruskal–Wallis test was applied to compare the delay of automatisms (the first period) and the duration of automatisms (the second period) across different groups. Post-hoc analyses were performed using Bonferroni correction.
2.4.4 Proportion of automatisms occurrence in habitual seizures
We investigated the habitual seizures recorded on scalp EEG (up to ten seizures) for each patient and calculated the proportion of seizures with automatisms in relation to all habitual seizures. The chi-square test with Bonferroni correction was used to compare the frequency of automatisms occurrence between all pairs of groups.
3. Results
3.1 Demographic characteristics
Seventy-four patients (35 females, 39 males), accounting for a total of 204 seizures, were included in this study. Fifty-seven patients had a visible lesion on MRI, the median age at epilepsy onset was 14.71 years (range = 1–38), and the median age at surgery was 25.30 years (range = 4–56). Thirty-seven patients had left-sided epilepsy, and thirty-seven patients had right-sided epilepsy. Histopathology was performed in all patients. The most common aetiology was hippocampal sclerosis (23 patients, 31.1%), followed by focal cortical dysplasia (21 patients, 28.4%). All of the 74 patients underwent surgery. The average postsurgical follow-up was 2.72 years (ranging from 1 to 6 years). After surgery, 58 (78.4%) patients were completely seizure-free, with no auras (ILAE 1); eight (10.8%) had only auras and no other seizures (ILAE 2); five (6.8%) patients had one to three seizure days per year ± auras (ILAE 3); and three (4.1%) patients had four seizure days per year to a 50% reduction in baseline seizure days and some auras (ILAE 4). Additional information is listed in Table 1.
3.2 Semiologic features of the four clusters of automatisms
First, the clinical features were compared between the two examiners with a good interrater agreement (Cohen's kappa was 82.3%). We found that the ideal number of clusters was four by the ‘elbow’ method (Fig. S1). By the k-means clustering method, the patients were then divided into four clusters of automatisms. The distribution of eleven clinical features for the four resulting clusters is represented in Fig. 1.
Fig. 1Distribution of the clinical features according to the four clusters of automatisms seizures. Clinical features ranked from 0 to 2 by the automatisms seizures scale, and the axes represent the averaged scores of each sign in each cluster. AT: automatisms.
Cluster 1 (21 patients) was characterised by a strong aura (1.67), postictal confusion (1.62), and short delay of automatisms (1.52), accompanied by moderate dystonia (0.86) and duration of automatisms (0.90). Cluster 2 (18 patients) was characterised by frequent hyperkinetic behaviour (1.00) and moderate version (0.83). Cluster 3 (16 patients) was characterised by strong emotion (1.06), moderate postictal confusion (0.88), elementary motor signs (0.75), autonomic signs (0.69) and vocalization (0.63). Cluster 4 (19 patients) was predominantly characterised by a short delay of automatisms (1.63) and strong dystonia (1.42).
To display the results of clustering analysis intuitively, we projected patients within the new lower-dimensional space supplied by PCA (Fig. S2) (the first three principal components accounting for 49.12% of variance). Roughly, we can see a spatial trend of four grouped points. However, there was no clear boundary between the four clouds of points. Of note, this study only focused on the symptoms and did not rely on intracranial electrophysiology data. Therefore, these results are expected.
3.3 The relationship between semiologic clusters and SOZ localisations
Patients were grouped into four semiologic subgroups, and the SOZs of each group were projected separately onto the mesial and lateral faces of the cerebral cortex (Fig. 2). Except for Cluster 2, patients in other clusters had SOZs distributed in all four anatomical regions. Patients with certain types of SOZs tended to exist in specific semiologic clusters, revealing a correlation between SOZ localisations and the clinical features (Fig. 3A). In addition, using the k-means algorithm, we represented the proportion of patients’ SOZs in relation to the resulting clusters (Fig. 3B, Fig. 3C).
Fig. 2Schematic of SOZ localisations in relation to the four semiologic clusters. Each sphere represents the core of the SOZ for a patient. Cluster 1 contained a significant proportion of patients with mesial-temporal SOZs, whereas Cluster 3 and Cluster 4 had more patients with frontal and neocortical temporal SOZs, respectively. Additionally, the SOZs of Cluster 2 were distributed diffusely. All SOZs are depicted in the left hemisphere. Encircled spheres represent SEEG-defined SOZs.
Fig. 3Correlation between semiologic cluster and SOZ localisations. a) Kendall correlation between SOZ localisations and clinical features as a function of colour (yellow, positive correlation; blue, negative correlation; starred squares, p < .05). Asterisk represents a significant correlation at p < .05; b). Spider plot representing the numbers of groups of SOZ localisations versus the resulting clusters. The axes represent how many patients in each SOZ group were clustered into each semiological group; c) Proportions of anatomical subgroups in relation to semiology clusters. The y-axis represents the percent of each semiological cluster in each SOZ group. FLE: frontal lobe epilepsy, NTLE: neocortical temporal lobe epilepsy, PCE: posterior cortex epilepsy, MTLE: mesial temporal lobe epilepsy.
Cluster 1 included almost half of the MTLE patients (13/23) and approximately one-third of the NTLE patients (5/14). In contrast, PCE patients (2/16) and frontal patients (1/21) were rarely included in Cluster 1. Cluster 1 was significantly correlated with MTLE (p = .017). Cluster 2 included approximately half of the FLE patients (11/21) and several PCE patients (7/16). Cluster 3 included almost one-third of the FLE patients (7/18), several NTLE patients (3/14) and PCE patients (4/16), and only 2 MTLE patients (2/23). Cluster 3 was only significantly correlated with FLE patients (p = .023). Cluster 4 consisted of several NTLE patients (6/14), with which it was significantly correlated (p = .011), and a few FLE patients (2/21) and PCE patients (3/16). Eight of the 23 patients of Cluster 4 were MTLE patients.
In Fig. S2, the clinical features of Clusters 1 & 4 processed by PCA were closest to each other, as well as Clusters 2 & 3. In comparing this observation with the semiologic clusters in Fig. 1, the short delay to automatisms was one of the typical features between Clusters 1 and 4. In terms of the SOZ locations, Clusters 1 and 4 were essentially temporal, and Clusters 2 and 3 were often extratemporal, as illustrated in Fig. 2, which is consistent with Fig. 3B.
3.4 Scaled onset and duration of automatisms
We performed a quantitative analysis to compare the scaled onset time and the duration of automatisms between groups. The time of oral and manual automatisms was calculated separately. Regarding oral automatisms, neither the onset time nor the duration showed significant differences across these four groups (Fig. 4). Manual automatisms occurred significantly earlier in patients in the NTLE group than in those in the FLE group (p=.002). In addition, manual automatisms persisted longer in the MTLE group than in the PCE group (p=.034). Other comparisons of features were not significant. The range and median of automatisms delay and duration are listed in Tables S3 & S4, respectively.
Fig. 4Graphical illustration of the scaled timestamp of three ictal parts. a) For each patient, the ictal seizure was divided into three parts along the x-axis: 1) from the clinical onset to the appearance of automatisms (either oral or manual), 2) from the start to the end of automatisms, and 3) from the end of automatisms to the termination of the last clinical signs. The patient order is sorted according to the duration of the normalised first part. The upper and lower rows represent oral and manual automatisms, respectively. b) The comparison of timestamp characteristics at the group level. The blue and red bars represent the delay and duration of automatisms, respectively, which correspond to the first and second parts of the x-axis in plot a). The y-axis represents the normalised timestamps at the group level. Asterisks denote significant differences at the p < .05 level (after multiple comparison correction). For both a) and b), the upper and lower rows refer to oral and manual automatisms, respectively. AT: automatisms, FLE: frontal lobe epilepsy, NTLE: neocortical temporal lobe epilepsy, PCE: posterior cortex epilepsy, MTLE: mesial temporal lobe epilepsy.
3.5 Proportion of automatisms occurrence in habitual seizures
We calculated the proportion of automatisms in all seizures for each patient and compared it across the four groups. In Fig. 5, each bar represents the percent of seizures with automatisms relative to each patient's total number of seizures. In MTLE patients, 76 out of 109 seizures appeared with automatisms. The other three groups showed a lower proportion of automatisms occurrence. In particular, the proportions were 62/118 for the FLE group (p=.048), 35/84 for the NTLE group (p<.001), and 31/83 for the PCE group (p<.001) (p values refer to the comparisons between MTLE and other groups).
Fig. 5Proportion of seizures with and without automatisms. a) Proportion of automatisms occurrence in relation to all habitual seizures for each patient. All of the automatisms ratios are re-ordered ascendingly. The x-axis represents patients’ ID, and the y-axis represents the automatisms seizure ratios at the patient level. b) The total proportion of automatisms occurrence at the group level. The x-axis represents four SOZ groups, and the y-axis represents the total number of seizures with automatisms. Asterisks denote significant differences at the p < .05 level (after multiple comparison correction). FLE: frontal lobe epilepsy, NTLE: neocortical temporal lobe epilepsy, PCE: posterior cortex epilepsy, MTLE: mesial temporal lobe epilepsy.
Accurate localisations of SOZs can improve surgical outcomes. However, it is challenging to find the SOZs according to semiologic features. Oral alimentary and manual automatisms are most common in patients with TLE. However, several reports also noted that other patients exhibited automatisms [
]. In this study, we applied a cluster analysis on the semiological features. We found four clinical patterns of automatisms that correlate with specific SOZs. Moreover, the separation of temporal and extratemporal epilepsy was demonstrated by the automated cluster analysis. The symptoms of the four patterns identified in this study could provide essential information in SOZ localisations for focal seizures with automatisms.
4.2 The clinical patterns and their relationship with SOZ localisations
The symptomatic patterns of Clusters 1 and 4 were similar, as were the patterns of Clusters 2 and 3 (Fig. 1 and Fig. S2). Clusters 1 and 4 were essentially temporal, and Clusters 2 and 3 were often extratemporal (Fig. 2 and Fig. 3B). Previous literature has reported distinct automatisms features in different subtypes of TLE [
]. However, the comparison between TLE and extratemporal lobe epilepsy has not been studied previously. Our results suggested that temporal and extratemporal localisations can be separated by automated cluster analysis.
Patients included in Cluster 1 showed more frequent aura, postictal confusion, and earlier onset of automatisms, accompanied by moderate dystonia. MTLE is characterised by auras [
]. Since Cluster 1 consisted of almost half of MTLE patients, it is reasonable that the clinical features of this cluster were similar to those exhibited by patients with MTLE. In contrast, only a few NTLE patients (5/21) were classified into Cluster 1. Accordingly, Maillard [
] suggested that postictal confusion was more common in medial and medial-lateral temporal lobe seizures than in lateral TLE. To some extent, these findings have practical value to help distinguish the automatisms of MTLE from that present in other subtypes.
Cluster 2 (18 patients) was characterised by a prominent hyperkinetic character and head/eye version. This cluster only contained FLE and PCE patients. In the presurgical evaluation, when the clinical features of Cluster 2 appeared, it is suggested that the SOZ may be located in the frontal or posterior cortex. The hyperkinetic features seen in Cluster 2 were not linked to specific lobar localisations, as illustrated in Fig. 2. The complex patterns of SOZs were consistent with growing literature indicating that hyperkinesia can be seen in frontal, temporal, insular, or parietal seizures [
]. The majority of these patients had SOZs in the frontal lobe, especially in the OFC. In addition, a study reported frequent oral or manual automatisms in a subgroup of patients in whom epileptic activity arose from the OFC and quickly spread to the temporal lobe [
]. In line with these previous findings, ten patients in Cluster 2 had SOZs covering the orbital-frontal cortex, which may explain the high hyperkinesia score. Patients in Cluster 2 were also characterised by contralateral versions, which predominantly occurred in PCE patients (7/18). In agreement with our finding, previous studies reported the localisation value of the versive symptom in PCE [
In Cluster 3, emotion was the most prominent sign and was significantly correlated with the FLE group. Given the diffuse SOZ distribution in this cluster, it is reasonable to expect that it is characterised by only one strong feature and several moderate features, such as autonomic symptoms. The epileptic focus might involve the ventrolateral prefrontal cortex/cingulate gyrus for gestural motor seizures [
] are significant symptomatic zones of autonomic manifestations. In our work, three patients in Cluster 3 had SOZs involving the insula cortex (patients 1, 10, 18), which may explain why the autonomic sign was a feature of Cluster 3. Interestingly, some “temporal-like” symptoms have been reported in insular epilepsy [
Cluster 4 mainly included NTLE and MTLE patients. The main characteristics of this cluster were the early occurrence of automatisms and frequent dystonia. These findings are expected since Cluster 4 had half of the TLE patients and showed strong dystonia. Ipsilateral automatisms with dystonia might be helpful for localising SOZ in TLE patients [
] examined automatisms in TLE subtypes. However, there have been no quantitative cluster analyses regarding automatisms. Cluster 4 had a noticeable short delay, suggesting an early onset of automatisms in TLE patients compared to extratemporal epilepsy.
Our study found correlations between symptomatic and anatomical groups, although this relationship was not a one-to-one correspondence. Specifically, patients with the same SOZs may have various semiologic features, and patients with different SOZs may have similar symptoms. Ictal discharge in SOZ may not necessarily cause the related symptoms, fast propagation to the distant eloquent cortex could elicit some clinical signs [
]. Therefore, many misleading symptoms may lead to an inaccurate localisation diagnosis, which prompted us to develop the cluster analysis presented in this study. The relationship between semiology and localisations reflected the intricate epileptic network underlying automatisms. This epileptic network was likely to rely on both cortical and subcortical components [
]. Oral automatisms, another frequent symptom of TLE, may be associated with connectivity between temporal lobe structures and insular-opercular regions [
]. However, these findings remain speculative, and a consensus has yet to be reached. Because it is difficult to define the exact anatomical location and specific spreading network, clarifying the precise location and extent was not our research's main aim. Instead, we aimed to use accompanied symptoms and semiquantitative features to differentiate automatisms from various SOZs at the lobar level.
4.3 The onset delay and duration of automatisms and the frequency of automatisms occurrence
Both the spatial and temporal aspects of seizure discharge played a role in clinical features [
]. Accordingly, we hypothesised that the onset time and duration of automatisms might also vary across different types of seizures. In this sense, we analysed oral and manual automatisms separately. Manual automatisms of NTLE began earlier than FLE. Accordingly, a study by Kotagal [
] suggested an early onset of manual automatisms in TLE. Furthermore, we found that the manual automatisms seen in the NTLE group occurred earlier than in the MTLE group, although this difference did not reach significance. In line with these findings, Maillard reported the delayed automatisms in MTLE rather than other subtypes of TLE [
]. In terms of duration, we found that MTLE patients experienced a longer duration of manual automatisms than patients with posterior cortex SOZs. Previous research also recorded a more extended duration of automatisms in temporal seizures than in other seizure types [
], which was controlled by normalisation in our study.
We also calculated the proportion of automatisms occurrence in all recorded seizures during video-EEG monitoring. Other studies suggested that patients with TLE [
], experienced more frequent automatisms, corroborating an observation in our cohort. To some extent, the higher frequency in the MTLE group may be related to the mechanism of automatisms. In particular, manual automatisms sometimes coincided with contralateral dystonic posturing [
Although our work may have significant clinical value for increasing localisation accuracy, there are still several weaknesses: 1) some patients did not undergo SEEG; as a result, we were unable to localise the SOZs at more precise level; 2) we did not analyse the first ictal behaviour change, given that a high number of patients started with behavioural arrest or consciousness change, which reduced the value of the initial symptom; 3) the lateralisation of manual automatisms was not considered because manual automatisms did not constantly occur in the ipsilateral or contralateral side of the SOZ; and 4) in the MTLE group, we only included patients diagnosed with unambiguous hippocampal sclerosis; all cases with the pathology of FCD ⅢA were excluded. Hence, this group only contained a modest number of MTLE patients.
5. Conclusion
The semiologic patterns for automatisms seizures had significant value to define the SOZs. Notably, considering the complex propagation network, the relationship partially overlaps across semiologic patterns and SOZs.
Declaration of Competing Interest
None.
Acknowledgement
Funding: This research was funded by Capital's Funds for Health Improvement and Research (2020–2–1076), National Key R&D Program of China (2018YFC0115401), and National Natural Science Foundation of China (No. 82071457, 81771399).
Occipital lobe epilepsy: electroclinical manifestations, electrocorticography, cortical stimulation and outcome in 42 patients treated between 1930 and 1991. Surgery of occipital lobe epilepsy.
Some Methods for Classification and Analysis of MultiVariate Observations.
in: Proc. of the fifth Berkeley Symposium on Mathematical Statistics and Probability. University of California Press,
Cam1967: 281-297 (Ed J. NeymanL. M. Le)
00369-1&id=gr1.jpg){kind=link}
00369-1&id=gr2.jpg){kind=link}
00369-1&id=gr3.jpg){kind=link}
00369-1&id=gr4.jpg){kind=link}
00369-1&id=gr5.jpg){kind=link}