Alterations of Activity in Visual Areas of Left Mesial Temporal Lobe Epilepsy Patients: Evidence from Moving Facial Emotional Tasks

Abstract

Background

Persons with epilepsy suffer from problems with social adaptation and face numerous social issues, even when the seizures are well-controlled. Facial emotion recognition (FER), one of the key components of social cognition, has been shown to be impaired in persons with mesial temporal lobe epilepsy (MTLE).

Objective

To clarify the impairment of neural networks in persons with MTLE, we performed functional magnetic resonance imaging (fMRI) studies of MTLE persons using six dynamic tasks involving FER.

Methods

We evaluated brain responses to realistic dynamic facial emotional expressions involving six basic emotions (fear, happiness, anger, sadness, disgust and surprise) in nine persons with left MTLE and ten healthy controls (HCs) using fMRI.

Results

We noted brain responses to facial emotions in regions related to FER, such as the anterior insular cortex, medial frontal gyrus, superior frontal gyrus and cerebellum by our moving tasks which involved video tasks used to evaluate FER. Persons with left MTLE showed a reduced response in the right calcarine cortex compared to that of HCs.

Conclusions

This is the first study to evaluate task-related fMRI on exposure to six basic emotions involving moving tasks in persons with epilepsy. FER deficit in persons with left MTLE may be partially associated with calcarine activity.

Introduction

Epilepsy is among the most frequent neurological diseases worldwide, with a lifetime prevalence of 6.38 per 1,000 persons. ¹ Almost two-thirds of persons with epilepsy newly treated with an antiseizure medications (ASMs) showed satisfactory seizure control. ² On the other hand, persons with epilepsy suffer from problems with social adaptation and face numerous social issues, including situations involving normal daily activities or employment, even if the seizures are well-controlled.^{3, 4} This has been attributed to impaired social cognitive functioning in epilepsy persons that is defined as information processing that contributes to the correct perception and interpretation of others’ emotions, mental states, dispositions and intentions. ⁵ Persons with epilepsy need to have effective social cognitive functions to have successful social activities.

One of the key components of social cognition is facial emotion recognition (FER), which is frequently reported to be impaired in mesial temporal lobe epilepsy (MTLE). ⁶ Meta-analysis revealed that the most severe deficit in MTLE persons was recognition of fear, and significant deficits were also noted for disgust, sadness and anger compared with healthy controls (HCs). ⁶ Early-onset and a long duration of epilepsy contribute to severity of deficits in FER.^7–10

FER involves an extended network that includes the amygdala and other limbic structures insula, ventral striatum, cingulate cortex and orbitofrontal cortex, ¹¹ and these areas may be impaired by epileptic discharges of MTLE, but details are not clear. Several functional magnetic resonance imaging (fMRI) studies have investigated neural activity in MTLE subjects using paradigms that involve FER. Bernhardt et al reported that in MTLE persons, epileptic discharges contribute to structural and functional alternations in local and distributed networks and these alternations are associated with FER deficits.^{12, 13} A recent fMRI study using static facial emotion expression showed that there is a relatively minor impact of epilepsy on neural activity facilitating FER in left MTLE persons. But the seizure control status is more pronounced. ¹⁴

However, most fMRI studies of FER in MTLE persons evaluated neural networks using photographic tasks or dynamic tasks involving only one expression, such as a fearful expression.^{15, 16} A previous report suggested that dynamic compared with static emotional expressions triggered more widespread activation patterns in emotion-related brain regions. ¹⁷

Therefore, we performed fMRI with more realistic moving tasks involving six basic emotions (fear, happiness, anger, sadness, disgust and surprise) to investigate neural correlates and the effect of epilepsy in left MTLE. Based on previous findings, ¹⁷ we hypothesised that the neural correlates associated with FER impairment will be more widely detected in a moving FER task than in a static FER task, and that fMRI could provide a detailed assessment of neural networks. Since early onset and long-term persistence of epilepsy are associated with lower FER, a secondary objective was to determine the correlation between brain activity related to FER and epilepsy duration and age of onset.

Methods

Participants

Nine left MTLE persons and ten HCs participated in this study. Left MTLE persons were included in the study to exclude the influence of left-right differences in lesions. The diagnosis of left MTLE was made by an epileptologist at Kyoto Prefectural University of Medicine based on clinical syndromes, including lateral signs, left frontotemporal spikes in the interictal period on scalp-recorded EEG. All persons were free of any abnormal MRI finding such as hippocampal sclerosis or other structural disorder.

HCs were recruited via in-house advertising to staff of Kyoto Prefectural University of Medicine (KPUM) and their families and subjects’ families. Exclusion criteria were a history of other neurological or psychiatric disorders, and abnormal MRI findings. All participants gave written informed consent. The study was approved by the ethics committee of KPUM (approval number: ERB-C-767-5).

Overview of the Experiment

After a participant arrived for MRI at the Kajiicho Medical Imaging Center, the experimenter demonstrated the moving FER task. Then, the participant was escorted to the MRI scanner room. In the MRI scanner, the participant performed three trials of the task to become acclimatised to the task and response buttons. They also wore MRI-compatible glasses when necessary. After confirming that the participant understood the task, they performed one run while being subjected to fMRI scanning.

Moving FER Task

We used a standardised moving FER task and we achieved more than 80% agreement in 76 healthy young students.^{18, 19} The task consisted of six basic emotions (anger, surprise, sadness, fear, disgust and happiness) without sounds. The moving FER task during fMRI was programmed by the NBS Presentation system (Neurobehavioral Systems, Inc.). The facial emotion block consisted of six trials. In the facial emotion trial, a participant was instructed to gaze at a fixed cross for 2.5 s. Then, one of the facial emotions (anger, surprise, sadness, fear, disgust and happiness) was presented for 2.5 s as a moving face image changing from neutral to emotional with emotional expression shown for 2 s, and subsequently, they had to select which emotion was presented from two choices (the correct answer and an incorrect one) by pressing the button presented for 2.5 s. Within one run, a facial emotion block was presented 12 times in total and each facial emotion block was randomly presented two times. The control block consisted of two trials. Within a trial, a participant was instructed to gaze at a fixed cross for 2.5 s, and subsequently, showed instructions to press the indicated button for 2.5 s. For fMRI, all participants performed one run of the moving FER task (Figure 1). Participants were presented with emotional faces on a screen via a mirror attached to a head coil and selected an answer using a response button. We recorded accuracy and response times while scanning.

Figure 1. Experimental Design.

Notes: In the control condition, participants were instructed to press the indicated button, and in the task condition, after the moving task was presented, they choose one from two options. In one block, the control condition was performed twice, and the task condition was performed six times. A block consisting of a control and a task condition was repeated 12 times.

Imaging Acquisition Protocol

Brain images were acquired using a 3T Philips Achieva 3.0 Quasar Dual (Royal Philips, Japan) with a 32-channnel head coil. Anatomical images were acquired using T1-weighted three-dimensional magnetisation-prepared rapid gradient-echo (3D MPRAGE) with 3.0 T scan of the entire brain in 170 sagittal slices (magnetisation prepared rapid gradient echo sequence, with repetition time (TR) and echo time (TE) = shortest, flip angle = 9°, field-of-view (FOV) = 256 × 256 mm ² , voxel size = 1.0 × 1.0 × 1.2 mm and slice thickness = 1.2 mm). fMRI was performed using T2*-weighted echo planar imaging with TR = 2,500 ms, TE = 30 ms, FOV = 212 × 212 mm, matrix = 64 × 64 mm, voxel size = 3.31 × 3.31 × 3.20 mm ² , matrix size = 64, no interslice slip, slice thickness = 3.2 mm with 0.8 mm interslice gap and interleaved collection. The scan time for one run was about 660 s (11 min) with 6 dummy volumes and 264 volumes.

Statistical Analysis

All analyses were conducted using Statistical Parametric Mapping (SPM)12 (The Welcome Trust Centre for Neuroimaging, Institute of Neurology, University College, London, UK) on MATLAB Release 2019b (The MathWorks, Massachusetts, USA). After discarding the first six volumes, images were realigned, coregistered, normalised to the MNI template and re-sampled at a voxel size of 2 mm ³ , detrended using detrending software to correct signal drift, and smoothed with an 8 mm full-width at half maximum Gaussian kernel.

The fMRI data were then corrected for motion effects using Artifact Detection Tools, which include software for comprehensive analysis of sources of artifacts in time-series data, including spiking and motion for post-processing fMRI data.

On first-level analysis, preprocessed MRI data were analysed using the general linear model (GLM) and SPM default canonical haemodynamic response function defined by the onset and duration of the facial emotion block. The following contrasts were created a) anger, b) surprise, c) sadness, d) fear, e) disgust and f) happiness.

At the second level, all contrast images from the first level analysis were used to perform a 6 (emotion) × 2 (group) factorial analysis of variance (ANOVA) to investigate the effect of emotion and group on the brain response to each facial emotion. We defined a cluster-forming threshold as P < .001 uncorrected, and the cluster level threshold as P < .05 corrected for multiple comparisons using the family-wise error (FWE) rate.

As the post-hoc analysis, a region of interest (ROI) approach was used to clarify the influence of epilepsy on facial emotion perception. Based on the results of ANOVA, brain regions showing a significant effect of group were used as ROIs. From ROIs, contrast estimates of six contrasts (anger, surprise, sadness, fear, disgust and happiness) were extracted and differences were tested between groups. For this analysis, Bonferroni adjustment was applied for multiple testing correction and P < .008 (0.05/6) was set as significant.

The associations between the brain response to emotion and duration of illness or onset age were also tested using extracted contrast estimates.

Results

Participants

Group comparison of left MTLE persons (N = 9) and HCs (N = 10) did not show significant differences between age, sex or years of education (P > .14). In the left MTLE group, the mean age at epilepsy onset was 56.2 years old, mean duration with epilepsy was 10.6 years and mean number of antiepileptic drugs was 1.78 (Table 1). There were no obvious imaging abnormalities in the subjects of either group.

Table 1. Demographics and Clinical Data.

	LMTLE Persons (n = 9)	HCs (n = 10)	P Value
Agea	66.9 ± 11.6	61.3 ± 19.8	.47
Sex (male/female)b	4/5	5/5	.65
Years of educationa	13.7 ± 1.97	15.7 ± 2.34	.14
Clinical data
Age at epilepsy onset	56.2 ± 21.0	–	–
Years with epilepsy	10.6 ± 12.6	–	–
Seizure-free period (years)	3.0 ± 2.67	–	–
Number of ASMs	1.78 ± 0.916

Notes: ^aWilcoxon chi-squared test.

^bFisher’s exact test.

Mean ± Standard deviation.

ASM: Antiseizure medication.

Behavioural Results

There were no significant differences between the two groups regarding the accuracy and reaction time of the FER task (P > .21). Total accuracy was 75.4 % in left MTLE subjects and 73.9 % in HCs (P = .75). The total reaction time was 1,952 ms in left MTLE persons and 1,865 ms in HCs (P = .21) (Table 2).

Table 2. Performance Data.

	Left MTLE (n = 9)	HCs (n = 10)	P Value
Total Facial Emotion Taska
Accuracy (%)	75.4	73.9	.75
Reaction time (ms)	1952	1865	.21
Anger
Accuracy (%)	81.3	80.5	.92
Reaction time (ms)	1895	1871	.87
Surprise
Accuracy (%)	86.6	86.1	.95
Reaction time (ms)	1883	1836	.67
Sadness
Accuracy (%)	81.3	80.5	.94
Reaction time (ms)	1793	1729	.63
Fear
Accuracy (%)	57.6	58.3	.94
Reaction time (ms)	2268	2034	.33
Disgust
Accuracy (%)	65.7	63	.86
Reaction time (ms)	2052	1964	.63
Happiness
Accuracy (%)	79.9	75.0	.64
Reaction time (ms)	1821	1758	.67

Note: ^aMean, two sample t-test.

Imaging Results

The Main Effect of Emotion

The main effect of emotion on the brain response to facial emotion for all participants was observed in a facial emotion processing network by fMRI study (Figure 2). The main effect of emotion was significant at the superior frontal gyrus (SFG) ([x, y, z] = [−6, 28, 44], P_{FWE-corrected} < .001 z = 6.49, cluster size = 1,844), medial frontal gyrus (MFG) ([x, y, z] = [46, 30, 30], P_{FWE-corrected} < .001 z = 6.34, cluster size = 1,546), ([x, y, z] = [−46, 24, 26], P_{FWE-corrected} < .001 z = 6.2, cluster size = 3,470), the anterior insular cortex ([x, y, z] = [40, 24, −2], P_{FWE-corrected} < .001 z = 5.76, cluster size = 378) and cerebellum ([x, y, z] = [16, −76, −30], P_{FWE-corrected} < .001 z = 5.67, cluster size = 711) (Table 3, Figure 2). These data suggested that the regions activated in the present task are closely involved in FER.

Figure 2. Main Effect of Emotion.

Notes: Activations for main effect of emotion are presented at P < .001.

This figure represents the significant clusters shown in Table 3.

Table 3. Main Effect of Emotion.

Region	Hemisphere	Cluster Size	MNI Coordinates			Z Value	P FWE-corrected
			X	Y	Z
Superior frontal gyrus	L	1,844	–6	28	44	6.49	< .001
Medial frontal gyrus	R	1,546	46	30	30	6.34	< .001
Medial frontal gyrus	L	3,470	–46	24	26	6.2	< .001
Anterior insular	R	378	40	24	–2	5.76	< .001
Cerebellum exterior	R	711	16	–76	–30	5.67	< .001
Precuneus	L	1,538	14	–60	24	4.8	< .001
Superior frontal gyrus	R	130	10	52	2	4.49	.032

The Main Effect of Group

The main effect of group between left MTLE and HC on the brain response to facial emotion was shown in the Figure 3. The main effect of group was significant at the right calcarine cortex ([x, y, z] = [10, −76, 8], P_{FWE-corrected} = .035, z = 4.13, cluster size = 166) (Table 4). In the next step, the right calcarine response to each facial emotion was compared between groups by post-hoc analysis. However, no significant difference was observed in the right calcarine response between groups (P > .12) (Figure 4).

Figure 3. Main Effect of Group.

Notes: Activations for main effect of group are presented at P < .001.

Significant difference between left MTLE and HCs was observed at Rt calcarine (circled areas in the image).

Table 4. Main Effect of Group.

Region	Hemisphere	Cluster Size	MNI Coordinates			Z Value	P FWE-corrected
			X	Y	Z
Calcarine	R	166	10	–76	8	4.13	.035

Note: FWE: Family-wise error.

Figure 4. Post-hoc t-Test (Main Effect of Group).

Note: Brain activation in right-calcarine responses to each facial emotion did not show a significant difference between left MTLE and HCs (P > .12).

The Interaction Effect of Group and Emotion

No significant clusters were found in the interaction effects of groups and emotions.

Association Between the Brain Response to Emotion and Duration of Illness or Onset Age

We investigated the association between brain activation in the right calcarine and clinical data (duration of epilepsy and onset age) for each emotion to clarify the relationship between calcarine activity and epilepsy. We noted negative correlations between the calcarine response to fear or happiness and duration of illness at a trend level (P = .069, r = −0.63 and P = .15, r = −0.52 respectively). We also observed an association between the calcarine response to fear and onset age at a trend level (P =.092, r = 0.59) (Figure 5).

Figure 5. Correlation Between Contrast Estimate in Calcarine and Clinical Data.

Note: A trend-level correlation between clinical information (A: duration of illness, B: onset age) and contrast estimate of the right calcarine were observed in the left MTLE.

Correlation of Accuracy and Reaction Time Data with the Calcarine Response

The accuracy and reaction time data were analysed for correlation with the contrast estimate in calcarine cortex for each expression. No significant correlations were found for accuracy and reaction time with the contrast estimate in calcarine for any of the facial expressions with Bonferroni adjustment (Ps > .05/6).

Discussion

We investigated neural network differences in FER between left MTLE persons and HCs to clarify the brain networks associated with FER deficit in persons using fMRI with six dynamic facial expressions. It was reported that persons with MTLE showed different results in FER between those with left MTLE and right MTLE, ²⁰ so only persons with left MTLE were evaluated to keep the cases uniform.

We identified brain activation in facial emotion processing network areas by the present moving FER task, including the anterior insula cortex, SFG, MFG and cerebellum. The anterior insula cortex is integral in emotional awareness and empathy, being involved in the processing and expression of emotions.^{21, 22} MFG is considered to be involved in general emotional processing, and by virtue of its connections with specific areas of the temporal lobe, this region plays a critical role in complicated elements of emotional processing such as social interactions. ²³ Previous studies also suggested that the cerebellum plays an important role in emotion and behavior. ²⁴ Several studies provided neuroimaging evidence suggesting the involvement of the cerebellum in support of sensory, emotional, attentional and cognitive process independent of motor involvement.^{25, 26} Based on these reports, it is likely that these regions activated in this task are closely involved in FER and that our dynamic task could appropriately evaluate FER.

We investigated the main effect of group between left MTLE and HCs, but ANOVA revealed decreased activation only at the right calcarine cortex in left MTLE persons.

A recent study showed that visual scanning patterns in MTLE persons differed from those in a control group. MTLE persons with impaired FER tended to perform more diffuse eye-tracking over faces and exhibit cognitive dysfunction. ¹⁵ These different scanning patterns may be related to the reduced activation in the calcarine, the primary visual cortex. Another possibility is an attention deficit in MTLE persons. While the moving tasks contain more information than pictures, participants need to devote more attention to the tasks because the expressions are transient. ***In a resting state fMRI study, an altered fMRI amplitude with low-frequency fluctuation in the occipital lobe in MTLE persons was demonstrated to be correlated with altered scores on visuospatial attention examinations. ²⁷ Low attention to tasks may lead to hypoactivation in the calcarine cortex. Based on these reports, we consider that the reduced calcarine activity in the FER task is due to impaired face recognition or impaired attention to the task.

There were no significant group differences between MTLE persons and HCs regarding the accuracy of FER tasks during fMRI in this study. This may be because the responses were selected from a choice of only two, the correct answer and an incorrect one. The moving task employed in this study was previously reported as a reproducible evaluation of FER in healthy subjects,^{18, 19} and we consider that there is no problem with the content of the task. Since the purpose of this study was to identify group differences in neural activity during the FER task, the lack of significant differences in the accuracy of the task did not lead to a problem regarding its design.

On the other hand, fMRI provided a detailed assessment of the differences in neural networks between the two groups that could not be assessed by the percentage of correct responses to the FER task. Right calcarine responses to fearful facial expressions were correlated with age of onset or duration of epilepsy at the trend level but was not statistically significant. A previous study showed that an earlier onset and a longer duration of epilepsy were associated with poorer performance in FER tasks.^{7, 28} Brain connectivity studies in persons with focal epilepsy demonstrated widespread network alterations extending beyond the epileptogenic zone and connectivity patterns potentially related to the duration and severity of disease. ²⁹ These reports suggest that although the calcarine is far from the epileptogenic focus of MTLE, extensive network changes beyond the epileptogenic region occurs in persons with epilepsy, and that brain activity in the calcarine is decreased in relation to disease duration and severity.

Study Limitation

There were several limitations of this study. First, the statistical power was limited due to the small sample size, which increased the probability of false negatives. Further research with a larger sample size is needed to confirm the present results. In addition, although only persons with left temporal lobe epilepsy were evaluated in this study, we would like to examine facial expression cognitive function in persons with right temporal lobe epilepsy as well. And although this study is biased toward the elderly, future studies should also include younger patients, because MTLE can occur in a wide range of age groups.

Another limitation was that the ceiling effect on task performance was observed because the task was a forced two-choice task, and there was no significant difference in behavioural results. Also, there was no significant difference in brain activity for each expression in the calcarine between the groups. Furthermore, we could not exclude the influences of ASMs. However, a previous study reported that the only ASM showing a possible negative effect on emotion recognition abilities was phenobarbital, ³⁰ and the participants of this study did not take that medication. Thus, we consider the effect of ASMs on FER to be minimal.

Conclusion

To our knowledge, this is the first study to evaluate task-related fMRI on exposure to six basic emotions involving moving tasks in epilepsy persons. Although there was no significant difference in the rate of correct responses to the FER task performed during fMRI compared to controls, a detailed evaluation using fMRI revealed that left MTLE persons show lower regional brain activity in the calcarine area, on recognising facial emotion expression, indicating that FER deficit in MTLE persons may be partially associated with calcarine activity. Our results indicate that not only the limbic system but also the decreased activity of the calcarine may be influencing the FER process of left MTLE.

Authors’ Contribution

Kanako Oya: Data collection and analysis, paper writing and formatting; Akihiro Tanaka: Conceptualization, review, proof-reading and correspondence; Yuko Nakamura: fMRI data analysis; Daisuke Ueno: Moving task creation and data collection; Naoki Akamatsu; Conceptualization; Toshiki Mizuno: Interpretation the results and supervision the project.

Data Availability Statement

The deidentified data that support the findings of this study are available from the corresponding author upon request.

Statement of Ethics

Necessary ethical clearances and informed consent were received and obtained respectively before initiating the study from all participants. This study was approved by the ethics committee of KPUM (Approval Number: ERB-C-767-5).