Epilepsy & Behavior 32 (2014) 92–99

Contents lists available at ScienceDirect

Epilepsy & Behavior journal homepage: www.elsevier.com/locate/yebeh

Functional MRI of facial emotion processing in left temporal lobe epilepsy Jerzy P. Szaflarski a,b,c,⁎, Jane B. Allendorfer a,b,c, Heidi Heyse c,d, Lucy Mendoza c, Basia A. Szaflarski a, Nancy Cohen a a

Department of Neurology, UAB Epilepsy Center, University of Alabama at Birmingham, Birmingham, AL, USA Center for Imaging Research, University of Cincinnati Academic Health Center, Cincinnati, OH, USA Department of Neurology, University of Cincinnati Academic Health Center, Cincinnati, OH, USA d Department of Psychiatry, University of Cincinnati Academic Health Center, Cincinnati, OH, USA b c

a r t i c l e

i n f o

Article history: Received 2 November 2013 Revised 15 January 2014 Accepted 20 January 2014 Available online 11 February 2014 Keywords: Emotion processing fMRI Temporal lobe epilepsy Face processing task Seizure control

a b s t r a c t Temporal lobe epilepsy (TLE) may negatively affect the ability to recognize emotions. This study aimed to determine the cortical correlates of facial emotion processing (happy, sad, fearful, and neutral) in patients with wellcharacterized left TLE (LTLE) and to examine the effect of seizure control on emotion processing. We enrolled 34 consecutive patients with LTLE and 30 matched healthy control (HC) subjects. Participants underwent functional MRI (fMRI) with an event-related facial emotion recognition task. The seizures of seventeen patients were controlled (no seizure in at least 3 months; LTLE−sz), and 17 continued to experience frequent seizures (LTLE+sz). Mood was assessed with the Beck Depression Inventory (BDI) and the Profile of Mood States (POMS). There were no differences in demographic characteristics and measures of mood between HC subjects and patients with LTLE. In patients with LTLE, fMRI showed decreased blood oxygenation level dependent (BOLD) signal in the hippocampus/parahippocampus and cerebellum in processing of happy faces and increased BOLD signal in occipital regions in response to fearful faces. Comparison of groups with LTLE+sz and LTLE−sz showed worse BDI and POMS scores in LTLE+sz (all p b 0.05) except for POMS tension/anxiety (p = 0.067). Functional MRI revealed increased BOLD signal in patients with LTLE+sz in the left precuneus and left parahippocampus for “fearful” faces and in the left periarcheocortex for “neutral” faces. There was a correlation between the fMRI and Total Mood Disturbance in the left precuneus in LTLE−sz (p = 0.019) and in LTLE+sz (p = 0.018). Overall, LTLE appears to have a relatively minor effect on the cortical underpinnings of facial emotion processing, while the effect of seizure state (controlled vs. not controlled) is more pronounced, indicating a significant relationship between seizure control and emotion processing. © 2014 Elsevier Inc. All rights reserved.

1. Introduction Facial emotion processing is considered part of social intelligence, a concept separate from general (or cognitive) intelligence; the ability to recognize facial emotions provides the greatest amount of clues regarding the emotional state of the observed individual [1,2]. Further, the ability to “mentalize” or understand and manipulate other people's behaviors is a major underlay of social interaction and social integration, while incorrect recognition of facial emotions may lead to increased psychosocial stress [3]. Because the brain regions that are responsible for facial emotion recognition and processing including the temporolimbic areas of the Papez circuit are also involved directly or indirectly in the generation or propagation of the interictal and ictal events in patients with TLE, it is important to understand how TLE affects emotion processing and what factors modify facial emotion recognition [1,4–7]. ⁎ Corresponding author at: UAB Epilepsy Center, Department of Neurology, University of Alabama at Birmingham, 312 Civitan International Research Center, 1719 6th Avenue South, Birmingham, AL 35242-0021, USA. E-mail address: szafl[email protected] (J.P. Szaflarski). 1525-5050/$ – see front matter © 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.yebeh.2014.01.012

In neuroimaging experiments, the ability to activate the regions that are responsible for emotion processing may depend on the nature and content of the presented task that examines the emotion recognition — whether the processing of the stimuli occurs via language modality vs. nonlanguage modality [5,8] and whether the task performed during imaging involves sentence processing, face or eye deviation recognition, or higher level processing, e.g., recognizing faux pas [5,6,8–12]. Further, many fMRI studies use blocked design and black-and-white depictions of emotions in which correctly and incorrectly recognized emotions are combined in the analyses possibly leading to habituation and decreased ability to capture activations in some brain regions, e.g., the amygdala [13,14]. Recent behavioral and imaging studies have documented that patients with TLE before and/or after temporal lobectomy have difficulties with the recognition of facial emotions with the main focus on fear perception [2,15–20], auditory emotions [2], and faux pas [21,22]. Thus, the objective of the present study was to elucidate the cortical correlates of facial emotion processing in patients with left temporal lobe epilepsy (LTLE) using event-related fMRI with jittered baseline (to decrease the possibility of habituation) and pictures of emotional

J.P. Szaflarski et al. / Epilepsy & Behavior 32 (2014) 92–99

faces adapted from “NimStim” [23]. The NimStim set of facial expressions is a major upgrade when compared with the previous blackand-white set that uses pictures of facial expressions of predominantly Caucasian subjects of similar ages [14]. The NimStim set includes a choice of over 600 color facial poses collected from 43 professional actors of various ages and ethnic and racial backgrounds and includes validated examples of happy, sad, disgusted, fearful, angry, surprised, neutral, and calm facial expressions. In this study, we utilized the happy, sad, fearful, and neutral faces to extend the probe beyond the most frequently studied emotions of fear or happiness. Based on the previous studies of emotion processing in patients with TLE, we hypothesized that cortical correlates of facial emotion processing would be different in patients with LTLE compared with matched-for-age-andsex HC subjects, indicating the presence of either compensatory activity or cortical plasticity in response to left temporal lobe seizures and that patients with LTLE would have impaired facial emotion recognition. The second objective was to assess the effect of seizure control (controlled vs. not controlled) on cortical correlates of emotion processing. 2. Methods 2.1. Subjects This cross-sectional study was approved by the local ethics review committees at the University of Cincinnati (UC) and at the University of Alabama at Birmingham (UAB). Thirty-four consecutive subjects with LTLE and 30 healthy control (HC) subjects participated in this study. The diagnosis of LTLE was made based on clinical semiology and confirmatory EEG or video/EEG findings. Inclusion criteria also included normal MRI; patients with lesions other than medial temporal lobe sclerosis, e.g., malformations of cortical development or vascular lesions, were excluded. Further, we also excluded patients with suspected temporal neocortical or extratemporal lobe epilepsy. Subjects were considered seizure-free (controlled; LTLE−sz) if they did not experience any seizures in the 3 months preceding the scanning session; all other patients' seizures were considered not controlled (LTLE+sz) [24]. All subjects provided written informed consent prior to study participation and were physically healthy, had no contraindications to receiving an MRI, and had no history of neurological illness, other than epilepsy in subjects with LTLE. Participants with depression were admitted into the study provided that they had no history of suicidal tendencies within 1 year of participation; none of the subjects with epilepsy were diagnosed with other psychiatric conditions other than depression. Urine pregnancy tests were obtained for all female subjects on the day of the study and were negative. Of the 63 subjects with epilepsy who were initially approached for study participation, 10 were excluded after detailed chart review (e.g., due to diagnosis other than LTLE or lesion on MRI), and 15 were excluded after phone screening (e.g., due to claustrophobia, lack of transportation, or comorbid conditions). An additional 4 subjects either did not show for the scanning session (N = 1) or were unable to finish the scanning session (N = 3). Healthy subjects were approached after subjects with LTLE were enrolled to match the groups for age and sex. The data presented here are a part of a larger study focusing on evaluating emotion processing in patients with LTLE using behavioral, physiologic, and neuroimaging measures [25]. 2.2. Assessments The Beck Depression Inventory—II (BDI) and the Profile of Mood States (POMS) were administered prior to the MRI session. The BDI is a self-report measure designed to assess a person's degree or level of depression, with higher scores reflecting greater mood severity [26]. The BDI has been shown to correlate well with the POMS depression/ dejection scale [27]. The POMS is a self-report assessment that measures fluctuations in mood states and provides scores for the following mood

93

scales: tension/anxiety, depression/dejection, anger/hostility, vigor/ activity, fatigue/inertia, and confusion/bewilderment [28]. The Profile of Mood States also assesses Total Mood Disturbance (TMD) that is thought to be a highly reliable and clinically relevant measure of overall mood and mood problems. Total Mood Disturbance for each subject was calculated by subtracting the vigor/activity score from the sum of all other mood scale scores [28,29]. Higher scores on POMS scales and TMD reflect greater mood severity, except for the vigor/activity scale (low score = low vigor/activity). Two-sample t-tests were performed for each assessment score using SAS (Statistical Analysis System version 9.3, Cary, NC) with p b 0.05 considered significant. 2.3. fMRI faces task The fMRI faces task was programmed using E-prime, version 1.1 (Psychology Software Tools, Inc.); this and related fMRI tasks are well known to investigate the brain’s implicit response to images of emotional facial expressions [23,30]. During fMRI, subjects were presented a pseudorandomized series of faces with different emotional expressions (happy, fearful, sad, and neutral) taken from the NimStim set of facial expressions [23]. Using a response box held in their right hand, subjects pressed the left button with their pointer finger when presented with a male face or the right button with their middle finger when presented with a female face. A total of 120 unique images (60 females; 30 per facial expression) were displayed, each for 2 s, followed by a screen with a “+” in the center during the variable interstimulus interval of 1 to 6 s (average interval of 4.02 s) to allow for event-related modeling of each event type (i.e., happy, fearful, sad, and neutral). Immediately after, fMRI subjects were asked to indicate the expression on each of the previously presented faces given the following choices: happy, fearful, sad, neutral, and unknown. Only facial expressions correctly identified on the postscan assessment were used in modeling each emotional expression, and incorrect responses were modeled as separate events in fMRI analysis for each subject. While it is common practice to include all faces in processing each emotional expression, we modeled only correctly identified emotional expressions to strengthen the analysis. Two-sample t-tests were performed for accuracy and response times for the faces task and for the postscan assessment using SAS (Statistical Analysis System version 9.3, Cary, NC) with p b 0.05 considered significant. 2.4. Magnetic resonance imaging Imaging of 30 HC subjects and 23 subjects with LTLE was performed at the Center for Imaging Research (CIR) at the University of Cincinnati using a 4.0-Tesla Varian Whole Body MRI system. An additional 11 subjects with LTLE were scanned in the Civitan Functional Neuroimaging Laboratory (CFNL) at the University of Alabama at Birmingham using a 3.0-Tesla Siemens Magnetom Allegra MR system. All subjects were fitted with an MRI-compatible headset to communicate with research staff and a response device in their right hand to make responses during the faces task. Subjects at the CIR were fitted with video goggles for viewing visual stimuli, while subjects scanned in the CFNL viewed stimuli presented on a screen via a mirror attached to the radio-frequency head coil. After subjects were positioned in the scanner, a three-plane localizer scan was performed for alignment and brain localization, followed by a shim procedure to generate a homogeneous magnetic field. Next, anatomical scans were acquired for localization of brain activation maps. In CIR, a high-resolution T1-weighted three-dimensional (T1-W 3D) anatomical scan was acquired using a modified equilibrium Fourier transform (MDEFT) sequence (TR: 13 ms, TE: 5.3 ms, flip angle: 22°, FOV: 25.6 cm × 25.6 cm × 19.2 cm, matrix: 256 × 192, slice thickness: 1 mm) [31]. In the CFNL, a T1-W 3D anatomical scan was acquired using a magnetization-prepared rapid acquisition with gradient echo (MP-RAGE) sequence (TR: 2300 ms, TE: 2.17 ms, flip angle: 9°, FOV: 25.6 cm × 25.6 cm × 19.2 cm, matrix: 256 × 256, slice thickness:

94

J.P. Szaflarski et al. / Epilepsy & Behavior 32 (2014) 92–99

1 mm). Prior to the fMRI sessions, a midsagittal localizer scan was performed to locate the anterior commissure (AC) and posterior commissure (PC). In both centers, fMRI images were acquired obliquely at 20° transverse-to-coronal from the AC–PC line in order to reduce signal distortions from the sinuses [32]. Thirty-five slices were positioned for whole-brain coverage; fMRI was performed using a T2*-weighted gradient-echo echo-planar imaging (EPI) pulse sequence (TR: 3 s, TE: 25 ms, FOV: 25.6 cm × 25.6 cm, matrix: 64 × 64, slice thickness: 3 mm with 1 mm interslice gap, flip angle: 85°). For scans acquired at the CIR, a multiecho reference scan to reduce geometric and ghosting artifacts at high field was also performed [33]. 2.5. MRI signal processing and statistical analysis In-house software programmed in Interactive Data Language (www. exelisvis.com) was used to reconstruct scanner images from the CIR, applying the multiecho reference scan correction. During CIR image reconstruction, spatial smoothing of 7.2 mm × 7.2 mm × 7.2 mm (FWHM) was introduced via Hamming filtering. The fMRI data were then processed, analyzed, and visualized using AFNI [34]. After coregistering structural and fMRI scans for each subject, an image coregistration algorithm was used to correct for subject head movements during fMRI [35]. A binary mask was applied to remove extraneous signal outside of the brain, followed by conversion of structural and functional data into Talairach space using the ICB452 brain template (www.loni.ucla.edu) in AFNI and resampling of functional images to a 3 × 3 × 3 mm3 voxel resolution [36]. A similar spatial smoothing of 7.2 × 7.2 × 7.2 mm (FWHM) using a Gaussian filter was performed on fMRI data acquired at the CFNL. Percent signal change calculation was performed, followed by single-subject statistical modeling of the blood oxygenation level-dependent (BOLD) response to each previously defined event type (i.e., happy, fearful, sad, neutral, and incorrect). For each subject, the event times used for fMRI data analysis were extracted from the behavioral data files recorded by E-prime, and each event type was convolved with a canonical hemodynamic response function in AFNI while accounting for subject head motion using motion correction parameters as covariates and for low frequency signal drift using the EPI signal mean and linear, quadratic, and cubic polynomial functions as covariates. Group-level analyses were performed using one-sample t-tests for each emotional facial expression to show overall activation patterns in each group. We determined group differences in neural processing of each emotional facial expression using scanner type as a covariate in regression analysis. Within the group with LTLE, we also compared differences in neural processing of each emotional facial expression between patients with LTLE−sz and those with LTLE+sz, using both scanner type and TMD as covariates in regression analysis. Activation clusters with a voxelwise p b 0.001 and consisting of ≥10 voxels were considered significant [37]. Activation clusters showing significant group differences in processing of each of the emotional facial expressions between patients with LTLE+sz and those with LTLE−sz were defined as regions of interest (ROIs). For each ROI, the mean beta weight was extracted for each subject's BOLD response to the particular emotional facial expression and used in Pearson's correlation analysis with TMD scores. All behavioral data analyses in this study and correlations with the neuroimaging data were determined a priori and were derived from the main hypotheses. Thus, no corrections for multiple comparisons were made, and the significance level was set at α = 0.05. 3. Results Subject demographics, clinical variables, assessment scores, and behavioral performance are summarized in Table 1 for each group (HC and LTLE). There were no significant differences between the LTLE group and the HC group in mean age, sex, and years of education (all p N 0.5); BDI and POMS/TMD scores (all p ≥ 0.1); or in-scanner performance on the faces task (p = 0.84 for accuracy and response times).

Performance on the postscan assessment of emotional facial expression of each face viewed during fMRI was also similar between groups in accuracy (all p N 0.08) but consistently lower for the group with LTLE than the HC group. Average response times on the postscan assessment were also slower overall for subjects with LTLE than HC subjects, with significant differences for happy and sad faces (p = 0.018 and p = 0.011, respectively); however, since this part of the research session was selfpaced (no time constraints on subjects), it is unlikely to be reflective of speed of information processing. Within the group with epilepsy (Table 2), LTLE−sz and LTLE+sz did not differ in average age, sex, age at epilepsy onset, or the number of antiepileptic drugs (AEDs) being used (all p N 0.24); the group with LTLE+sz had significantly less years of education (p = 0.001). The Beck Depression Inventory (BDI) and POMS/TMD scores for the group with LTLE+sz were consistently higher than those for the group with LTLE−sz and significantly different (all p b 0.01) except for the tension/anxiety scale (p = 0.067). Finally, there were no differences between the groups with LTLE−sz and those with LTLE+sz in the accuracy and timing of intrascanner responses (both p = 0.97) and in the postscanning rating of emotions (all p ≥ 0.13). These results summarized in Table 2 are reflective of overall greater mood disturbance in the group with LTLE+sz. Blood oxygenation level dependent (BOLD) activation to each emotional facial expression was consistently lower in patients with LTLE than HC subjects, although the patterns of activation were very similar between groups in showing increased activation in the frontal, temporal, parietal, visual, and cerebellar regions (Fig. 1). Regression analysis revealed three clusters where patients with LTLE showed significantly greater activation than HC subjects (Fig. 2A): in the right culmen/vermis and the right hippocampus/parahippocampus to happy faces and in the right middle/inferior occipital gyrus to fearful faces. We performed additional regression analyses within the group with LTLE (LTLE−sz vs. LTLE+sz). These analyses revealed greater activation in the left precuneus and the left parahippocampus (Figs. 2B 1 and 2, respectively) in the group with LTLE+sz vs. the group with LTLE−sz to fearful faces and greater activation in the subcallosal/ parahippocampus (BA 34) extending to the uncus (Fig. 2B 3) in the group with LTLE+sz vs. the group with LTLE−sz to neutral faces. Further, for each group with LTLE, there was a negative relationship between activation in these ROIs and TMD score (Figs. 2 b1, b2, and b3), although only the BOLD response to fearful faces in the left precuneus showed significant correlations with TMD (Fig. 2 b1) in the group with LTLE+sz (r = − 0.57; p = 0.018) and the group with LTLE− sz (r = −0.56; p = 0.019). 4. Discussion The goal of this study was to examine the neural underpinnings of facial emotion processing and how it is affected by seizure control in a large number of patients with LTLE using event-related fMRI. Two findings emerge from the analyses. The first finding is that there are relatively few differences in how HC subjects and subjects with LTLE process emotions whether based on the results of behavioral or fMRI testing — the observed differences in cortical correlates include fMRI signal changes in the right cerebellum and the right hippocampus/ parahippocampus when processing happy faces and in the right occipital gyrus when processing fearful faces. The second finding is that seizure control (controlled vs. not controlled) in patients with LTLE has a differential effect on cortical representation of emotion processing. First, let us consider the differences in activation patterns between patients with LTLE and HC subjects when processing happy facial emotions — the right cerebellum (culmen/vermis) and the right hippocampus/parahippocampus (Fig. 2A). Both of these regions, especially the hippocampus/parahippocampus, have long been recognized to participate in emotion regulation. In addition to the typically reported components of the Papez circuit (e.g., amygdala, hippocampus, and

J.P. Szaflarski et al. / Epilepsy & Behavior 32 (2014) 92–99

95

Table 1 Demographic, clinical, and performance variables for the 34 subjects with LTLE and 30 HC subjects. LTLE Age Sex, male Education, years Epilepsy history Age at onset Number of AEDs Number of seizures in the past 3 months Mood assessments Beck Depression Inventory (BDI) POMS — TMD POMS — tension/anxiety POMS — depression/dejection POMS — anger/hostility POMS — vigor/activity POMS — fatigue/inertia POMS — confusion/bewilderment Faces task during fMRI Accuracy, % correct Response time, ms Postscan rating of emotions on faces Happy Accuracy, % correct Response time, ms Fearful Accuracy, % correct Response time, ms Sad Accuracy, % correct Response time, ms Neutral Accuracy, % correct Response time, ms

HC

p-Value

41 (12) 7 (21) 15 (2)

39 (11) 8 (27) 15 (3)

0.52 0.77 0.71

27 (14) 1.8 (0.9) 6.0 (12.1)

– – –

– – –

11.8 (11.4) 34.7 (42.3) 10.9 (7.3) 12.5 (13.0) 11.5 (10.7) 15.6 (7.2) 8.8 (6.6) 6.6 (5.7)

0.25 0.60 0.42 0.54 0.10 0.76 0.94 0.33

97.0 (3.5) 858.4 (186.0)

96.8 (4.6) 867.4 (171.1)

0.84 0.84

97.9 (3.5) 1914.1 (474.4)

98.9 (2.2) 1601.3 (541.1)

0.19 0.018

86.9 (13.7) 3516.6 (2144.5)

88.3 (10.8) 2749.1 (1087.7)

0.63 0.072

79.9 (10.8) 3143.0 (1720.3)

84.3 (9.1) 2268.0 (809.4)

0.081 0.011

88.4 (14.3) 2767.3 (994.5)

90.8 (10.8) 2750.6 (1592.4)

0.46 0.43

8.8 (8.9) 29.4 (36.3) 9.5 (6.2) 10.6 (12.9) 7.5 (8.1) 15.0 (7.4) 8.9 (6.0) 8.0 (5.4)

POMS — Profile of Mood States; TMD — Total Mood Disturbance; fMRI — functional MRI; data reported as mean (SD) except for sex, which is reported as frequency (percentages).

cingulate), studies have postulated that the cerebellum partakes in emotional processing via the cerebellar–hypothalamic pathways (“limbic cerebellum”) [38,39]. Indeed, electrical stimulation of the cerebellum via

either a cerebellar pacemaker or excitatory repetitive transcranial stimulation (rTMS) induces improvements in mood, especially in response to happy facial expressions [38,40,41]. Further, cerebellar inhibition with

Table 2 Demographic, clinical, and assessment variables for the subjects with LTLE who are seizure-free (LTLE − sz; n = 17) and those who are not (LTLE + sz; n = 17). LTLE − sz Age Sex, male Education, years Epilepsy history Age at onset Number of AEDs Mood assessments Beck Depression Inventory Profile of Mood States (POMS) total POMS — tension POMS — depression POMS — anger POMS — vigor POMS — fatigue POMS — confusion Faces task during fMRI Accuracy, % correct Response time, ms Postscan rating of emotions on faces Happy Accuracy, % correct Response time, ms Fearful Accuracy, % correct Response time, ms Sad Accuracy, % correct Response time, ms Neutral Accuracy, % correct Response time, ms

LTLE + sz

p-Value

39 (14) 5 (29) 16 (2)

42 (10) 2 (12) 14 (2)

0.48 0.40 0.001

27 (12) 1.6 (0.9)

26 (17) 2.0 (0.8)

0.80 0.24

4.0 (5.4) 9.3 (23.9) 7.5 (5.1) 4.8 (7.4) 4.2 (6.2) 18.4 (6.7) 5.6 (4.9) 5.6 (4.2)

13.5 (9.3) 49.6 (35.8) 11.4 (6.7) 16.3 (14.8) 10.8 (8.5) 11.6 (6.6) 12.2 (5.3) 10.4 (5.6)

0.001 b0.001 0.067 0.009 0.014 0.005 b0.001 0.008

97.0 (3.0) 2697.2 (1144.9)

97.0 (4.0) 2709.1 (808.2)

0.97 0.97

98.2 (2.4) 1790.7 (376.8)

97.6 (4.4) 2037.6 (538.3)

0.63 0.13

88.0 (9.1) 3671.2 (2820.3)

85.7 (17.3) 3361.9 (1216.8)

0.62 0.68

80.4 (9.5) 3068.2 (1634.3)

79.4 (12.3) 3217.8 (1849.5)

0.80 0.80

90.8 (11.5) 3006.2 (1205.7)

86.1 (16.7) 3033.9 (903.8)

0.35 0.94

Data reported as mean (SD) except for sex, which is reported as frequency (percentages).

96

J.P. Szaflarski et al. / Epilepsy & Behavior 32 (2014) 92–99

A

happy

B

sad

C

fearful

D

neutral

Fig. 1. One-sample t-tests illustrating the overall pattern of BOLD activation for each emotional facial expression in subjects with LTLE (top row) and HC subjects (bottom row). Activation is significant at the voxelwise threshold of p b 0.001 and the cluster threshold of 10 voxels. Images are in radiological convention (the left part of the image shows the right part of the brain).

the use of 1-Hz rTMS increased negative mood [42]. Finally, a recent meta-analysis of “cerebellar” studies documented the involvement of the vermis and cerebellar hemispheres in emotional processing. We believe that the observed differences are related to the disease state with other explanations being less likely. These possibilities include activations related to the decision-making process rather than to emotional processing itself [43] or due to the use of antiepileptic drugs (AEDs), as negative effects of AEDs on brain activation related to cognitive tasks have been previously observed [44]. Another identified difference between the

groups is the decreased BOLD signal change in the right occipital lobe in patients with LTLE when compared with HC subjects (Fig. 2A). The most likely explanation is again the disease state (epilepsy vs. healthy), but other explanations may include the previously observed differences in timing of processing of fearful faces between hemispheres (typically delayed in right hemispheres) with possible contribution by the slight differences in behavioral testing (p = 0.072) and the use of AEDs [44,45]. Of importance is the fact that the observed differences in emotion processing between HC subjects and patients with LTLE are relatively minor. While

J.P. Szaflarski et al. / Epilepsy & Behavior 32 (2014) 92–99

A

B

happy

97

b1

fearful

r=-.57, p=.018

1 1

r=-.56, p=.019

2 b2

r=-.29, p=0.26

L

2

r=-.38, p=0.13

neutral

R fearful

b3

3

r=-.27, p=0.29

3 r=-.33, p=0.20

R

L

Fig. 2. Regression analysis of the BOLD response to emotional facial expressions in healthy control (HC) subjects and subjects with left temporal lobe epilepsy (LTLE). (A) Group activation difference between HC subjects and patients with LTLE, where patients with LTLE exhibited greater activation (in blue) than HC subjects in (1) the right cerebellum (17 voxels; Talairach coordinates: x = 23, y = −29, z = −26) and in (2) the right hippocampus/parahippocampus (10 voxels; Talairach coordinates: x = 34, y = −16, z = −16) to happy faces and in (3) the right middle/inferior occipital gyrus (11 voxels; Talairach coordinates: x = 46, y = −77, z = 2) to fearful faces. (B) Patients with LTLE with uncontrolled seizures (+sz) show greater activation (in red) to fearful faces than those who are seizure-free (−sz) in the (1) left precuneus (20 voxels; Talairach coordinates: x = −14, y = −49, z = 39) and in the (2) left parahippocampus (11 voxels; Talairach coordinates: x = −16, y = −29, z = −7) and to neutral faces in the (3) left subcallosal/parahippocampal gyrus (BA 34) extending into the uncus (15 voxels; Talairach coordinates: x = 20, y = −5, z = −14). For each group with LTLE + sz and for each group with LTLE − sz, there are negative relationships (b1, b2, and b3) between the BOLD response (y-axis) in these ROIs and Total Mood Disturbance (TMD; x-axis) as measured by the POMS, but are only significant for the response to fearful faces in the left precuneus (b1).

some may consider this unusual and unexpected, this is not the case in view of the previous studies of emotion processing in TLE. Recent behavioral and imaging studies have documented that patients with TLE frequently experience difficulties in recognition of facial emotions including fear perception, auditory emotions, and faux pas, but these deficits are usually limited to patients with RTLE. For example, Adolphs et al. showed that patients with LTLE who had temporal lobe resection performed similarly to HC subjects on recognizing fear and prosody [15], and Anderson et al. showed that of patients with LTLE who underwent temporal lobectomy, less than 10% suffered any problems with fear, disgust, sadness, anger, happiness, or surprise recognition [16]. Similar results were obtained by Meletti et al. who showed that for happiness, sadness, fear, disgust, and anger recognition, patients with LTLE performed similarly to HC subjects (~10% difference) [18]. Finally, a small imaging study that enrolled patients with LTLE (N = 5) and compared them with healthy controls (N = 14) also found similarities rather than differences between the groups [30]. Thus, the lack of major differences between patients with LTLE and HC subjects is not surprising. The substantial N in our study supports the claim from the previous behavioral and small imaging studies that the differences between HC subjects and patients with LTLE are minimal. The effect of seizure control on facial emotion processing has not been examined in the imaging studies to date. In fact, most studies have enrolled patients who either have poorly controlled TLE or have already had temporal lobe resection; other studies did not report on the seizure control in the enrolled patients. What we have found when comparing patients with LTLE+sz and those with LTLE−sz is that patients with LTLE+sz require more significant involvement of the left precuneus and the left parahippocampus for the processing of fearful faces and that these patients depend more on the left subcallosal/parahippocampal gyrus for processing of neutral faces (Fig. 2B). These relationships were either significantly correlated with TMD (left precuneus) or trended in the same direction for the other two regions indicating a relationship between these regions and

emotion processing. The subcallosal gyrus and the parahippocampal gyrus (periarcheocortex) are part of the limbic system [46] and, thus, are involved in emotion processing as tested in this study. Further, the amygdalae (directly adjacent to and connected to the periarcheocortex) are known to participate in facial emotion recognition, which may be affected more in right TLE (RTLE) than in LTLE [30]. However, the presence of activation in the precuneus and the observed differences between groups in the context of this study are somewhat less clear. The precuneus has been implicated in deciphering contextual information [47]. Some reports indicate it to be involved in memory-related mental imagery [48], while its contribution to and participation in reflective self-awareness and default mode are unquestionable [49,50]. Thus, the presence of the effect of seizure status on the involvement of the left precuneus and the left hippocampus in fear processing and the differences in recognizing neutral emotions in periarcheocortex need to be interpreted in this context and in the context of presence/absence of mood problems (TMD). Clearly, the effects of seizures on mood observed here are typical and comparable to the effects of seizures on mood in other studies that have examined these issues in patients with seizures that are or are not controlled [51]. Limitations of the current study need to be considered. First, the patients were not assessed for the presence of medial temporal lobe sclerosis (MTS), and this cannot be inferred from the T1 anatomical images collected as part of the study. Since the presence of MTS may be associated with decreased ability of emotion recognition, the presence/ absence of MTS and its severity need to be included as covariates in future studies [2,52]. Also, we did not perform specific prescan neuropsychological evaluation of emotion recognition. While subtle differences between the groups are possible, the fact that the postscan emotion rating was similar between groups (all p N 0.08) makes this unlikely. This possibility will need to be considered in future studies. Another limitation of this study is the lack of comparison to a group with RTLE. It is now recognized that patients with RTLE (nondominant TLE) may have more difficulties with emotion recognition than patients with LTLE

98

J.P. Szaflarski et al. / Epilepsy & Behavior 32 (2014) 92–99

(dominant TLE) [15,17]. Thus, future studies will need to include a group with RTLE in order to better characterize the differences between patients with LTLE and those with RTLE in emotion recognition and include measures of language lateralization to determine the specific influences of hemispheric dominance on recognition of facial emotions. We also cannot exclude the possibility that some of the results of the LTLE+sz vs. LTLE−sz analyses are related to differences in education between the groups. Thus, future studies with larger numbers of subjects will need to control for the level of education and/or IQ during enrollment or analysis stages of the studies. Finally, a potential limitation to the study is the possibility of Type I errors (i.e., falsely rejecting the null hypothesis) since correction for multiple comparisons was not performed in the regression analyses. However, correction for multiple comparisons increases the likelihood of Type II errors (i.e., failing to reject the null hypothesis when it is not true), particularly in fMRI data where the effects may be small or subtle. In order to minimize both types of errors, we incorporated the strategy of Lieberman and Cunningham of setting a stringent voxelwise threshold (p b 0.001) and a smaller cluster threshold (10 voxels) [37]. Similar concerns may arise regarding correlation analyses, although these analyses and statistical significance were determined a priori. 5. Conclusion In this study, we evaluated the effects of LTLE on the ability to process and recognize facial emotions and the effects of seizure control (controlled vs. not controlled) on the cortical underpinnings of processing facial emotion while controlling for mood state. Our finding is that the differences in how healthy controls and patients with LTLE process emotions whether based on the results of behavioral or fMRI testing are minimal. This means that LTLE does not have a major effect on emotional processing as evaluated in this study. On the other hand, seizure control (controlled vs. not controlled) has a significantly more pronounced effect on the cortical underpinnings of emotional processing, particularly of fearful faces. Acknowledgments Support of this study was provided by the Charles Shor Foundation for Epilepsy Research. Results of this study were presented, in part, at the 10th European Congress on Epileptology (London, UK) and the 65th Annual Meeting of the American Epilepsy Society (Baltimore, MD). The authors acknowledge Drs. James Eliassen and Erik Nelson for their help in designing the fMRI probe utilized in this study. References [1] Bar-On R, Tranel D, Denburg NL, Bechara A. Exploring the neurological substrate of emotional and social intelligence. Brain 2003;126(Pt 8):1790–800. [2] Bonora A, Benuzzi F, Monti G, Mirandola L, Pugnaghi M, Nichelli P, et al. Recognition of emotions from faces and voices in medial temporal lobe epilepsy. Epilepsy Behav 2011;20(4):648–54. [3] Frith CD, Frith U. Interacting minds — a biological basis. Science 1999;286(5445): 1692–5. [4] Adolphs R, Tranel D, Damasio H, Damasio A. Impaired recognition of emotion in facial expressions following bilateral damage to the human amygdala. Nature 1994;372(6507):669–72. [5] Calarge C, Andreasen NC, O'Leary DS. Visualizing how one brain understands another: a PET study of theory of mind. Am J Psychiatry 2003;160(11):1954–64. [6] Fletcher PC, Happe F, Frith U, Baker SC, Dolan RJ, Frackowiak RS, et al. Other minds in the brain: a functional imaging study of “theory of mind” in story comprehension. Cognition 1995;57(2):109–28. [7] Stone VE, Baron-Cohen S, Knight RT. Frontal lobe contributions to theory of mind. J Cogn Neurosci 1998;10(5):640–56. [8] Hofer A, Siedentopf CM, Ischebeck A, Rettenbacher MA, Verius M, Felber S, et al. Gender differences in regional cerebral activity during the perception of emotion: a functional MRI study. Neuroimage 2006;32(2):854–62. [9] Baron-Cohen S, Ring HA, Wheelwright S, Bullmore ET, Brammer MJ, Simmons A, et al. Social intelligence in the normal and autistic brain: an fMRI study. Eur J Neurosci 1999;11(6):1891–8.

[10] Binder JR, Gross WL, Allendorfer JB, Bonilha L, Chapin J, Edwards JC, et al. Mapping anterior temporal lobe language areas with fMRI: a multicenter normative study. Neuroimage 2011;54(2):1465–75. [11] Blair RJ, Morris JS, Frith CD, Perrett DI, Dolan RJ. Dissociable neural responses to facial expressions of sadness and anger. Brain 1999;122(Pt 5):883–93. [12] Vollm BA, Taylor AN, Richardson P, Corcoran R, Stirling J, McKie S, et al. Neuronal correlates of theory of mind and empathy: a functional magnetic resonance imaging study in a nonverbal task. Neuroimage 2006;29(1):90–8. [13] Breiter HC, Etcoff NL, Whalen PJ, Kennedy WA, Rauch SL, Buckner RL, et al. Response and habituation of the human amygdala during visual processing of facial expression. Neuron 1996;17(5):875–87. [14] Ekman P, Friesen WV. Pictures of facial affect; 1976 [Palo Alto, CA: Consulting Psychologists Press;]. [15] Adolphs R, Tranel D, Damasio H. Emotion recognition from faces and prosody following temporal lobectomy. Neuropsychology 2001;15(3):396–404. [16] Anderson AK, Spencer DD, Fulbright RK, Phelps EA. Contribution of the anteromedial temporal lobes to the evaluation of facial emotion. Neuropsychology 2000;14(4): 526–36. [17] McClelland III S, Garcia RE, Peraza DM, Shih TT, Hirsch LJ, Hirsch J, et al. Facial emotion recognition after curative nondominant temporal lobectomy in patients with mesial temporal sclerosis. Epilepsia 2006;47(8):1337–42. [18] Meletti S, Benuzzi F, Cantalupo G, Rubboli G, Tassinari CA, Nichelli P. Facial emotion recognition impairment in chronic temporal lobe epilepsy. Epilepsia 2009;50(6): 1547–59. [19] Meletti S, Benuzzi F, Rubboli G, Cantalupo G, Stanzani Maserati M, Nichelli P, et al. Impaired facial emotion recognition in early-onset right mesial temporal lobe epilepsy. Neurology 2003;60(3):426–31. [20] Yamada M, Murai T, Sato W, Namiki C, Miyamoto T, Ohigashi Y. Emotion recognition from facial expressions in a temporal lobe epileptic patient with ictal fear. Neuropsychologia 2005;43(3):434–41. [21] Giovagnoli AR, Franceschetti S, Reati F, Parente A, Maccagnano C, Villani F, et al. Theory of mind in frontal and temporal lobe epilepsy: cognitive and neural aspects. Epilepsia 2011;52(11):1995–2002. [22] Schacher M, Winkler R, Grunwald T, Kraemer G, Kurthen M, Reed V, et al. Mesial temporal lobe epilepsy impairs advanced social cognition. Epilepsia 2006;47(12):2141–6. [23] Tottenham N, Tanaka JW, Leon AC, McCarry T, Nurse M, Hare TA, et al. The NimStim set of facial expressions: judgments from untrained research participants. Psychiatry research 2009;168(3):242–9. [24] Szaflarski JP, Rackley AY, Lindsell CJ, Szaflarski M, Yates SL. Seizure control in patients with epilepsy: the physician vs. medication factors. BMC Health Serv Res 2008;8:264. [25] Allendorfer JB, Heyse H, Mendoza L, Nelson EB, Eliassen JC, Storrs JM, et al. Physiologic and cortical response to acute psychosocial stress in left temporal lobe epilepsy. Epilepsy Research 2014 [under review]. [26] Beck AT, Steer RA, Brown GK. Manual for Beck Depression Inventory—II. San Antonio, TX: Psychological Corporation; 1996. [27] Griffith NM, Szaflarski JP, Szaflarski M, Kent GP, Schefft BK, Howe SR, et al. Measuring depressive symptoms among treatment-resistant seizure disorder patients: POMS depression scale as an alternative to the BDI-II. Epilepsy Behav 2005;7(2):266–72. [28] McNair D, Lorr M, Droppleman L. POMS: Profile of Mood States. San Diego, CA: EITS/ Educational and Industrial Testing Service; 1992. [29] Szaflarski JP, Szaflarski M, Hughes C, Ficker DM, Cahill WT, Privitera MD. Psychopathology and quality of life: psychogenic non-epileptic seizures versus epilepsy. Med Sci Monit 2003;9(4):CR113–8. [30] Benuzzi F, Meletti S, Zamboni G, Calandra-Buonaura G, Serafini M, Lui F, et al. Impaired fear processing in right mesial temporal sclerosis: a fMRI study. Brain research bulletin 2004;63(4):269–81. [31] Lee JH, Garwood M, Menon R, Adriany G, Andersen P, Truwit CL, et al. High contrast and fast three-dimensional magnetic resonance imaging at high fields. Magn Reson Med 1995;34(3):308–12. [32] Deichmann R, Gottfried JA, Hutton C, Turner R. Optimized EPI for fMRI studies of the orbitofrontal cortex. Neuroimage 2003;19(2 Pt 1):430–41. [33] Schmithorst VJ, Dardzinski BJ, Holland SK. Simultaneous correction of ghost and geometric distortion artifacts in EPI using a multiecho reference scan. IEEE Trans Med Imaging 2001;20(6):535–9. [34] Cox RW. AFNI: software for analysis and visualization of functional magnetic resonance neuroimages. Comput Biomed Res 1996;29(3):162–73. [35] Cox RW, Jesmanowicz A. Real-time 3D image registration for functional MRI. Magn Reson Med 1999;42(6):1014–8. [36] Talairach J, Tournoux P. Co-planar stereotaxic atlas of the human brain. New York: Thieme Medical Publishers, Inc.; 1988. [37] Lieberman MD, Cunningham WA. Type I and Type II error concerns in fMRI research: re-balancing the scale. Soc Cogn Affect Neurosci 2009;4(4):423–8. [38] Schutter DJ, van Honk J. The cerebellum on the rise in human emotion. Cerebellum 2005;4(4):290–4. [39] Zhu JN, Yung WH, Kwok-Chong Chow B, Chan YS, Wang JJ. The cerebellar–hypothalamic circuits: potential pathways underlying cerebellar involvement in somatic– visceral integration. Brain research reviews 2006;52(1):93–106. [40] Heath RG. Modulation of emotion with a brain pacemaker. Treatment for intractable psychiatric illness. J Nerv Ment Dis 1977;165(5):300–17. [41] Schutter DJ, Enter D, Hoppenbrouwers SS. High-frequency repetitive transcranial magnetic stimulation to the cerebellum and implicit processing of happy facial expressions. J Psychiatry Neurosci 2009;34(1):60–5. [42] Schutter DJ, van Honk J. The cerebellum in emotion regulation: a repetitive transcranial magnetic stimulation study. Cerebellum 2009;8(1):28–34.

J.P. Szaflarski et al. / Epilepsy & Behavior 32 (2014) 92–99 [43] Stoodley CJ, Schmahmann JD. Functional topography in the human cerebellum: a meta-analysis of neuroimaging studies. Neuroimage 2009;44(2):489–501. [44] Szaflarski JP, Allendorfer JB. Topiramate and its effect on fMRI of language in patients with right or left temporal lobe epilepsy. Epilepsy Behav 2012;24(1):74–80. [45] Stefanics G, Csukly G, Komlosi S, Czobor P, Czigler I. Processing of unattended facial emotions: a visual mismatch negativity study. Neuroimage 2012;59(3):3042–9. [46] Schneider JF, Vergesslich K. Maturation of the limbic system revealed by MR FLAIR imaging. Pediatric radiology 2007;37(4):351–5. [47] Lundstrom BN, Ingvar M, Petersson KM. The role of precuneus and left inferior frontal cortex during source memory episodic retrieval. Neuroimage 2005;27(4):824–34. [48] Fletcher PC, Frith CD, Baker SC, Shallice T, Frackowiak RS, Dolan RJ. The mind's eye — precuneus activation in memory-related imagery. Neuroimage 1995;2(3):195–200.

99

[49] Fransson P, Marrelec G. The precuneus/posterior cingulate cortex plays a pivotal role in the default mode network: evidence from a partial correlation network analysis. Neuroimage 2008;42(3):1178–84. [50] Kay BP, Meng X, Difrancesco MW, Holland SK, Szaflarski JP. Moderating effects of music on resting state networks. Brain Res 2012;1447:53–64. [51] Taylor RS, Sander JW, Taylor RJ, Baker GA. Predictors of health-related quality of life and costs in adults with epilepsy: a systematic review. Epilepsia 2011;52(12):2168–80. [52] Ver Hoef LW, Williams FB, Kennedy RE, Szaflarski JP, Knowlton RC. Predictive value of hippocampal internal architecture asymmetry in temporal lobe epilepsy. Epilepsy Res 2013;106(1–2):155–63.