Subcortical and ventral prefrontal cortical neural responses to facial expressions distinguish patients with bipolar disorder and major depression

Subcortical and Ventral Prefrontal Cortical Neural Responses to Facial Expressions Distinguish Patients with Bipolar Disorder and Major Depression Natalia S. Lawrence, Andrew M. Williams, Simon Surguladze, Vincent Giampietro, Michael J. Brammer, Christopher Andrew, Sophia Frangou, Christine Ecker, and Mary L. Phillips Background: Bipolar disorder (BD) is characterised by abnormalities in mood and emotional processing, but the neural correlates of these, their relationship to depressive symptoms, and the similarities with deficits in major depressive disorder (MDD) remain unclear. We compared responses within subcortical and prefrontal cortical regions to emotionally salient material in patients with BP and MDD using functional magnetic resonance imaging. Methods: We measured neural responses to mild and intense expressions of fear, happiness, and sadness in euthymic and depressed BD patients, healthy control subjects, and depressed MDD patients. Results: Bipolar disorder patients demonstrated increased subcortical (ventral striatal, thalamic, hippocampal) and ventral prefrontal cortical responses particularly to mild and intense fear, mild happy, and mild sad expressions. Healthy control subjects demonstrated increased subcortical responses to intense happy and mild fear, and increased dorsal prefrontal cortical responses to intense sad expressions. Overall, MDD patients showed diminished neural responses to all emotional expressions except mild sadness. Depression severity correlated positively with hippocampal response to mild sadness in both patient groups. Conclusions: Compared with healthy controls and MDD patients, BD patients demonstrated increased subcortical and ventral prefrontal cortical responses to both positive and negative emotional expressions. Key Words: Bipolar disorder, depression, facial expressions, fMRI

A

bnormalities in emotion processing have previously been demonstrated in patients with affective disorders such as bipolar disorder (BD) and major depressive disorder (MDD) (Phillips et al 2003b), but little is known about the neuropsychological and neuroanatomical correlates of these abnormalities and the extent to which they are state or trait features of the disorders. The neural basis of the reported perceptual bias toward negative, and away from positive, emotional stimuli in MDD patients (David and Cutting 1990; Gur et al 1992; Suslow et al 2001; Surguladze et al, in press) and that of the emotional lability frequently observed in BD patients remain poorly understood, for example. Furthermore, the extent to which neural abnormalities distinguish BD from MDD patients with similar illness severity and medication remains unexplored. Elucidating the nature of abnormalities in neural response to emotive stimuli in BD and MDD will, therefore, help to increase understanding of the functional neuroanatomical basis of vulnerability to, and symptoms of, these psychiatric disorders. Abnormalities in the identification of emotional facial expressions, a process of critical importance for social interaction (Darwin 1872/1965), have previously been demonstrated in patients with BD. Depressed BD patients show impaired recognition of happy and sad facial expressions (Rubinow and Post 1992) and a bias toward identifying neutral facial expressions as sad (George et al 1998; Gur et al 1992; Lior and Nachson 1999). In euthymic patients with BD, findings indicate an enhanced From the Sections of Neuroscience and Emotion (NSL, AMW, SS, CE, MLP) and Neurobiology of Psychosis (SF); Division of Psychological Medicine, Brain Image Analysis Unit (VG, MJB); and Neuroimaging Research Group (CA), Institute of Psychiatry, London, United Kingdom. Address reprint requests to Dr. Natalia S. Lawrence, Section of Neuroscience and Emotion, Box PO69, De Crespigny Park, London SE5 8AF, United Kingdom. Received July 22, 2003; revised November 18, 2003; accepted November 20, 2003.

0006-3223/04/$30.00 doi:10.1016/j.biopsych.2003.11.017

ability to discriminate expressions of disgust (Harmer et al 2002). Studies employing nonfacial emotion processing tasks, including the emotional Stroop task (Williams et al 1996) and the affective go/no-go task (Murphy et al 1999), have demonstrated attentional biases toward negatively valenced stimuli in depressed BD patients (Lyon et al 1999; Murphy et al 1999) and in manic patients, biases toward negative (Lyon et al 1999) and positive (Murphy et al 1999) stimuli. These findings indicate in BD patients impaired recognition of emotional stimuli and an attentional bias toward positive and negative stimuli, with the latter being associated with depressive episodes in particular. In MDD, some studies have reported a generalized emotion identification deficit (Jaeger et al 1987; Rubinow and Post 1992; Persad and Polivy 1993; Mikhailova et al 1996; Asthana et al 1998), while others have demonstrated an attentional bias toward negative emotional material. Studies have demonstrated a specific impairment in the identification of happy expressions (Suslow et al 2001) and a response bias away from identifying expressions as happy (Surguladze et al, in press) and toward identifying expressions as sad (David and Cutting 1990; Gur et al 1992; Hale 1998). Studies employing nonfacial stimuli have further shown negative emotional biases during memory recall (Bradley et al 1996), the emotional Stroop task (Kerr and Phillips 2002; Williams et al 1996), and an affective go/no-go task (Murphy et al 1999). It is unclear whether these abnormalities persist in MDD patients during remission. Studies employing a variety of techniques have highlighted the importance of specific subcortical regions in the response to emotionally salient material (Calder et al 2001; Davis and Whalen 2001), namely, the amygdala, ventral striatum (including caudate nucleus, ventral putamen, and globus pallidus), and dorsomedial nucleus of the thalamus, in addition to the hippocampus and parahippocampal gyrus (Alexander et al 1990; Calder et al 2001). These regions may be involved both in the identification and generation of emotional states (Phillips et al 2003a). There is also increasing evidence for the role of different regions of ventral and dorsal prefrontal cortex in the generation and regulation of emotional states, respectively, in response to emotionally salient BIOL PSYCHIATRY 2004;55:578 –587 © 2004 Society of Biological Psychiatry

N.S. Lawrence et al material (e.g., Beauregard et al 2001; Levesque et al 2003; Mayberg et al 1999; reviewed in Phillips et al 2003a). In the current study, these subcortical, ventral, and dorsal prefrontal regions and the hippocampus, referred to subsequently as “subcortical and prefrontal regions,” were therefore included as regions of interest (ROIs), and between-group differences in blood oxygen level dependent (BOLD) response within activated clusters located within these regions were examined. Previous research has demonstrated increased amygdala activation and decreased right dorsolateral prefrontal cortex responses to fearful facial expressions in a group of stable bipolar patients (Yurgelun-Todd et al 2000, 2001). Additionally, increased subcortical (basal ganglia) activity has been demonstrated in BD patients during performance of a motor task (Caligiuri et al 2003), supporting previous findings of hypermetabolism in striatal (Drevets et al 1997; O’Connell et al 1995; Blumberg et al 2000) and pallidal (Mayberg 2001) regions in manic and depressed BD patients, respectively. These initial findings suggest increased activity in subcortical regions and decreased dorsal prefrontal cortical activity in BD patients in response to emotionally salient stimuli and also during cognitive and motor task performance. The extent to which these findings represent symptom-related activity and the relative magnitude of subcortical and prefrontal cortical responses to positive and negative stimuli in BD remain unexplored. In MDD patients, emerging evidence indicates increased activity within subcortical regions at rest and during task performance (Drevets et al 1992, 1995; Mayberg et al 1999) and a positive correlation between amygdalar metabolism and depression severity (Abercrombie et al 1998; Drevets et al 1992). Increased activation of the left amygdala to masked presentation of facial expressions, which resolves with antidepressant treatment (Sheline et al 2001), and decreased attenuation of the amygdalar response to emotional words (Siegle et al 2002) have also been reported in these patients. Furthermore, studies have consistently demonstrated in patients during a major depressive episode reduced activity within dorsal regions of the prefrontal cortex (e.g., Baxter et al 1989; Buchsbaum et al 1997; Goodwin et al 1993) but increased activity within ventral prefrontal cortex (Drevets et al 1992, 1995), including enhanced response within the rostral anterior cingulate gyrus and orbitofrontal cortex to negative emotional stimuli (Elliott et al 2002; Phillips et al 2003b). Together, these findings indicate increased activity within subcortical and ventral prefrontal cortical regions to negative emotional stimuli and decreased activity within dorsal prefrontal cortical regions in MDD patients. It is unclear, however, whether the magnitude of these abnormal neural responses correlate positively with depressive symptom severity. We wished to examine the nature of functional abnormalities within subcortical and prefrontal cortical regions in response to socially salient emotive stimuli in BD and MDD patients. In particular, we aimed to determine whether in BD abnormalities in subcortical and prefrontal cortical responses to emotional facial expressions were symptom related and whether depressed BD and MDD patients could be distinguished on the basis of these responses. Therefore, we measured neural responses in euthymic and depressed BD patients, depressed MDD patients, and healthy volunteers to facial expressions of positive (happy) and negative (sad and fear) emotions compared with neutral expressions. These emotional expressions represent, respectively, displays of social approval, internal distress, and external threat. We employed expressions of mild and intense emotion to examine whether abnormal neural responses in both patient

BIOL PSYCHIATRY 2004;55:578 –587 579 groups occurred both to subtle and prototypical expressions of facial emotion, the former approximating more closely expressions observed in everyday life. We therefore reasoned that any between-group differences in neural response to such stimuli would emerge more clearly in response to the subtle expressions. We did not specifically wish to determine the effect of emotional category or intensity per se on patterns of subcortical and prefrontal cortical response to these stimuli, however, nor did we wish to examine between-group differences in neural response to faces per se, since these were beyond the focus of our study. Based on current data we predicted that: 1. Bipolar disorder patients would demonstrate an enhanced response in subcortical regions previously identified as important in processing positive and negative emotional expressions to all expressions, including an enhanced amygdalar response to fearful expressions; and 2. Major depressive disorder patients would demonstrate increased activity within these subcortical regions to negative but not positive facial expressions. Findings to date did not permit us to make specific predictions regarding the nature of prefrontal cortical responses to facial expressions in BD and MDD patients compared with healthy volunteers, although they would suggest decreased dorsal prefrontal cortical responses in both patient populations.

Methods and Materials Participants Twelve participants with a diagnosis of bipolar I affective disorder (DSM-IV criteria, American Psychiatric Association 1994) and 9 participants with a diagnosis of major depressive disorder (DSM-IV criteria, American Psychiatric Association 1994) were recruited from the South London and Maudsley National Health Service Trust. Eleven healthy control subjects (CON) were recruited from the local community. Ethical approval was obtained from the Ethical Committee of the South London and Maudsley Trust and Institute of Psychiatry. All participants signed a statement of informed consent. All participants were right-handed (Edinburgh Handedness Inventory, Oldfield 1971) and matched for age (overall mean 41 ⫾ 11 years) and gender ratio (c. 40% females). The BD and CON groups were matched for years of education (respectively, mean 15.4 ⫾ SD 1.6 and mean 17.1 ⫾ SD 3), but the MDD group had fewer years of education than the CON group (13.44 ⫾ 2.45, p ⬍ .05). Exclusion criteria included a history of head injury, illicit substance abuse, and comorbid diagnoses. Mean duration of illness in BD patients was 15.4 ⫾ 13.4 years, and in MDD patients, mean duration of illness was 8 ⫾ 5 years. Bipolar disorder patients without any episodes of mania or major depression in the previous 6 months were recruited for the study. There were no residual manic symptoms in the BD patients, as assessed using the Mania Rating Scale (Young et al 1978); mania scores ranged from 2 to 7. Depression severity was measured in all groups on the day of testing using the Beck Depression Inventory (BDI) (Beck et al 1961). Mean BDI scores were greater in MDD than BD patients (respectively, 31.8 ⫾ 11.8 and 15.3 ⫾ 9.2), while both patient groups had higher BDI scores than CON group, whose mean BDI score was 2.27 ⫾ 2.28. Three BD patients were euthymic (BDI ⬍ 9) at the time of testing, seven patients reported mild symptoms of depression (BDI 10 –19), and two patients had moderate to severe depression (BDI ⬎ 20). Two MDD patients were mildly depressed, two were moderately depressed (BDI ⬎ 20), and five www.elsevier.com/locate/biopsych

580 BIOL PSYCHIATRY 2004;55:578 –587 were severely depressed (BDI ⬎ 30). All BD patients were taking medication; five (45%) were on selective serotonin reuptake inhibitors (SSRIs), five (45%) were on atypical antipsychotics, and nine (82%) were on mood stabilizers. Three (27%) were taking lithium, four (36%) were taking sodium valproate, 2 (18%) were taking carbamazepine, and one (9%) was taking lamotrigine. Six BD patients were taking combination treatments. All MDD patients were taking antidepressant medication; three (33%) were taking SSRIs, four (44%) were taking selective noradrenaline reuptake inhibitors, one (11%) was taking a monoamine oxidase inhibitor, and one (11%) was taking a tricyclic antidepressant. Procedure Subjects participated in three, 6-minute experiments employing event-related functional magnetic resonance imaging (fMRI), described previously (Surguladze et al 2003). In each experiment, subjects were presented with 10 different facial identities, manipulated by computer software to depict 50% and 100% intensities of one emotion (either sadness, happiness, or fear) (Young et al 2002), in addition to a 100% neutral expression. In one experiment, subjects viewed 20 prototypically happy (10 identities with expressions of 100% happiness, each stimulus presented twice), 20 mildly happy (expressions of 50% happiness), and 20 neutral (100% neutral) expressions. In the other experiments, subjects viewed similar numbers of neutral, mildly and prototypically sad or fearful expressions. Facial stimuli were presented for 2 seconds each in a pseudorandom order. During the interstimulus interval, the duration of which was varied from 3 to 8 seconds according to a Poisson distribution with average interval of 4.9 seconds, subjects viewed a fixation cross. We adopted an event-related design with a variable interstimulus interval to reduce habituation of the BOLD response in regions such as the amygdala, since habituation is a problem in block designs with highly repetitive and predictable stimulus presentation (e.g., Breiter et al 1996). Previous studies have demonstrated that neural responses to emotional stimuli depend on the nature of the task performed. Since performance on an implicit (gender decision) task is more reliably associated with responses in subcortical and extrastriate cortical regions (Morris et al 1996; Phillips et al 1997), we asked participants to decide the gender rather than the emotional expression of each face and press one of two buttons accordingly with the right thumb. All subjects identified the gender of the faces correctly. After the scanning session, participants completed a computerized test of emotion identification and the Recognition Memory Test (Warrington 1984) or the Short Recognition Memory Test for Faces (Warrington 1996) as an indirect measure of nonemotional face perception. The emotion identification task required subjects to identify neutral faces and expressions of mild (50%) and intense (100%) sadness, happiness, fear, and disgust presented on a laptop computer screen in a pseudorandom order. Major depressive disorder and BD patients did not differ from control subjects on the performance of either task. Image Acquisition Magnetic resonance images were acquired using a GE Signa 1.5 Tesla system (General Electric, Milwaukee, Wisconsin) with an operating console and software (Advanced Nuclear Magnetic Resonance, Woburn, Massachusetts) for gradient echo echoplanar imaging (EPI) at the Maudsley Hospital. A quadrature birdcage head coil was used for radiofrequency (RF) transmission and reception. One hundred and eighty T2*-weighted www.elsevier.com/locate/biopsych

N.S. Lawrence et al images depicting BOLD contrast were acquired at each of 16 near-axial noncontiguous 7-mm thick planes parallel to the intercommissural (anterior commissure-posterior commissure [AC-PC]) line: echo time (TE) ⫽ 40 milliseconds, repetition time (TR) ⫽ 2 seconds, in-plane resolution ⫽ 3.44 mm, interslice gap ⫽ .7 mm, matrix size ⫽ 64 ⫻ 64 pixels. High-resolution inversion recovery EPI images were acquired for subsequent localization of functional activation (3-mm thick, near-axial slices: TE ⫽ 73 milliseconds, time to inversion [TI] ⫽ 180 milliseconds, TR ⫽ 16 seconds, in-plane resolution ⫽ 1.72 mm, interslice gap ⫽ .3 mm, matrix size ⫽ 128 ⫻ 128 pixels). fMRI Data Analysis Individual Analysis. Before time series analysis, data were processed to remove low-frequency signal changes and motionrelated artifacts (Bullmore et al 1999a) and smoothed using a Gaussian filter (full-width half-maximum [FWHM] 7.2 mm). The responses at each voxel were then analyzed using Gamma variate functions (peak responses weighted from 4 to 8 seconds) convolved with each contrast vector (i.e., experimental condition) to model the BOLD response. Following least-squares fitting of this model, a goodness of fit statistic (sum of square ratio [SSQ]) composed of the ratio of model to residual sum of squares was calculated for each contrast. The distribution of the same statistics under the null hypothesis of no experimental effect was then calculated by wavelet-based resampling of the time series at each voxel and refitting the models to the resampled data (Bullmore et al 2001; Breakspear et al, in press). An experimentally derived null distribution of the goodness of fit statistic was then derived by following this procedure 10 times at each intracerebral voxel and combining the resulting data. Activations for any contrast at any required p value can then be determined by obtaining the appropriate critical values from the null distribution. Individual brain activation maps were produced for each subject within each emotion experiment for neutral expressions compared with the fixation cross baseline, and for each emotion intensity compared with neutral expressions. Group Mapping To extend inference to the group level, the observed and randomized SSQ ratio maps were transformed into standard space (Talairach and Tournoux 1988) by a two-stage process (Brammer et al 1997) using spatial transformations computed for each subject’s high-resolution structural scan. A group activation map was produced for each experimental condition by testing the median observed SSQ ratio over all subjects at each voxel in standard space (median values were used to minimize outlier effects) against a critical value of the permutation distribution for median SSQ ratio ascertained from the spatially transformed wavelet-permuted data (Brammer et al 1997). For greater sensitivity and to reduce the multiple comparison problem encountered in fMRI, hypothesis testing was carried out at the cluster level using methods developed by Bullmore et al (1999b). Essentially, this involved first thresholding at a voxelwise probability of false activation of .025, combining all three-dimensional (3D) contiguous activated voxels into clusters and then assessing the probability of occurrence of clusters under the null hypothesis by reference to the distributions produced when the null data, produced by time-series permutation, were similarly analyzed. At a clusterwise probability of false activation of .001 in this series of experiments, the expected number of false-positive activations over the whole of standard space was less than 1 (.5).

N.S. Lawrence et al Group Differences Several functional ROIs were selected for further analysis from the 50% intensity versus neutral and the 100% intensity versus neutral expression maps for each emotion from each group. These functional ROIs included clusters of significant activation located in subcortical regions (amygdala, ventral striatum, thalamus), the hippocampus/parahippocampal gyrus, ventral prefrontal regions involved in the perception of emotional stimuli, and dorsal prefrontal cortical regions believed to be involved in the regulation of emotion. As outlined in the introduction, our analytical approach was to examine activation in areas suggested (from the literature) to be important in emotional processing and which were significantly activated by our tasks. Care was taken to select ROIs from all three groups, and approximately 20 ROIs were selected for each of the six emotional versus neutral contrasts (three categories of emotion, each at two intensities). The SSQ ratio, as a measure of the mean power of neural response, was then extracted from each functional ROI in each participant. The SSQ is analogous to a Z score but, unlike a Z score, does not rely on the assumptions of normal distribution of data. Mean SSQ values within ROIs for each emotional versus neutral contrast were compared across all three groups using nonparametric between-subject comparisons (Kruskal-Wallis test for k independent groups) with SPSS software (SPSS, Inc., Chicago, Illinois). Outliers were removed before these between-group comparisons and a p-value of .01 was considered significant to minimize type I errors. Regions of interest showing a significant overall group effect at p ⫽ .01 were further examined for differences between separate groups using specific nonparametric pairwise contrasts. Mean SSQ values in these ROIs were also extracted from the control condition (neutral expressions versus fixation cross in each of the three different emotion experiments) to control for any between-group differences in baseline neural response to faces per se within each experiment. Finally, simple nonparametric (Spearman) correlation analyses were carried out using SPSS to determine whether SSQ values in each ROI showed a significant correlation with depressive symptoms (BDI score), both within each group and across all subjects. In addition, since the MDD group had completed fewer years of education than the CON group, correlations between SSQ and years of education were computed within each group to determine any influence of this factor on brain activation. Medication effects on neural response during task performance were assessed within each patient group by dividing patients into subgroups determined by whether they were taking or not taking a particular type of medication. In ROIs in which significant betweengroup differences in SSQ had been demonstrated in response to emotion-neutral expression contrasts, mean SSQs were compared between patient subgroups with Mann–Whitney U tests in SPSS. Bipolar disorder patients were divided into subgroups taking or not taking lithium, antipsychotic, and antidepressant medications. Since all MDD patients were taking antidepressant medication, they were divided into four subgroups (levels 1– 4) according to the criteria of Sackeim (2001), with those within levels 1 to 2 coded as the low-dose subgroup and those within levels 3 to 4 coded as the high-dose subgroup.

Results Neural Responses to Facial Expressions Group activation maps for 50% and 100% emotional versus neutral expressions for happy, sad, and fearful faces were

BIOL PSYCHIATRY 2004;55:578 –587 581 generated as described above. Numerous brain regions were consistently activated more to emotional faces than to neutral faces. These included the cerebellum, bilateral fusiform gyri, middle and superior temporal gyri, several subcortical regions (amygdala, parahippocampal cortex, thalamus, caudate, putamen), ventral and dorsal prefrontal cortex, precuneus, and inferior parietal cortex. Full details (coordinates, cluster size) of the group activation maps are available on request. To test our specific predictions, we focus here on between-group comparisons of neural responses within our specific theoretical and functional ROIs in subcortical, ventral, and dorsal prefrontal regions to these emotion-neutral contrasts. Tables 1, 2, and 3 describe the ROIs selected from the group maps for detailed comparison for each emotion-neutral condition. These regions showed significant overall between-group differences in neural response (SSQ) using the Kruskal-Wallis test for three groups (p ⱕ .01). Results of post hoc pairwise betweengroup contrasts (Mann–Whitney U scores) are given in the final column. Tables 1, 2, and 3 also indicate which ROIs were significantly correlated with depression scores and years of education within groups and in which ROIs there were significant effects of medication. Differences in Neural Response to Fearful Expressions Bipolar disorder patients demonstrated a significantly greater neural response to mild fear than either CON subjects or MDD patients in the right globus pallidus/anterior thalamus (Table 1, Figure 1). In response to intense fear, BD patients showed a significantly larger neural response than either CON subjects or MDD patients in a cluster encompassing the left amygdala and ventrolateral prefrontal cortex (Brodmann’s area [BA] 47) (Table 1, Figure 2). Closer examination of this cluster revealed that the between-group difference in activation in the ventrolateral prefrontal cortex (22 voxels) was greater than that within the amygdala (10 voxels). Control subjects demonstrated a significantly greater neural response to mild fear than both patient groups in a cluster encompassing the right amygdala and hippocampus and in the right dorsolatersal prefrontal cortex (Table 1). Both CON subjects and BD patients demonstrated a significantly greater response to mild fear than MDD patients in the left medial prefrontal cortex. In response to intense fear, both CON subjects and BD patients showed a greater neural response than MDD patients in the right globus pallidus/anterior thalamus. Differences in Neural Response to Happy Expressions Bipolar disorder patients had significantly greater neural responses to expressions of mild happiness than both CON subjects and MDD patients in three ROIs: within a large cluster extending from the left uncus, ventrally, to the amygdala and the caudate nucleus/putamen; in the ventromedial prefrontal cortex; and in the right ventrolateral prefrontal cortex (Table 2, Figure 3). Control subjects also showed greater neural responses than MDD patients in the left amygdala/caudate-putamen cluster. Closer inspection of between-group differences in activation in the different subregions in this cluster indicated that between-group differences in activation were greatest within the uncus (58 voxels), caudate nucleus (9 voxels), and putamen (18 voxels). In response to expressions of intense happiness, CON subjects had significantly greater neural responses than BD and MDD patients in the right parahippocampal gyrus and in a large cluster encompassing bilateral regions of thalamus, midbrain, caudate nucleus, and left amygdala (Table 2). All subregions contributed to the between-group differences in neural response. Control subwww.elsevier.com/locate/biopsych

582 BIOL PSYCHIATRY 2004;55:578 –587

N.S. Lawrence et al

Table 1. Functional ROIs Showing Between-Group Differences in Activation in Fear Condition Brain Region (including Brodmann Area) and Group Map of Origin 50% Fear vs. Neutral R globus pallidus/anterior thalamus from BDc R amygdala/hippocampus from CONe L medial PFC (BA 10, 11, 47) from CONd,e R dorsolateral PFC (BA 44) from CON 100% Fear vs. Neutral L amygdala/ventrolateral PFC (BA 47) from BD R parahippocampal gyrus/globus pallidus/ anterior thalamus from CONe

Center-of-Mass (R/L, A/P, S/I)

Cluster Size (Number Voxels)

11

⫺4

4

43

20

⫺6

⫺11

36

⫺30 48

44 9

⫺8 27

22 25

⫺25

10

⫺18

47

9

⫺6

1

58

CON ⬎ Other Groups

⬎BDa ⬎MDDb ⬎MDDb ⬎MDDb ⬎BDa

⬎ MDDb

BD ⬎ Other Groups

MDD ⬎ Other Groups

⬎CONa ⬎MDDb ⬎MDDa ⬎MDDb

⬎CONa ⬎MDDb ⬎MDDa

All ROIs showed a significant main effect of group (p ⫽ .01) in Kruskal-Wallis test comparing all three groups. ROI, region of interest; R, right; L, left; A, anterior; P, posterior; S, superior; I, inferior; CON, control group; BD, bipolar disorder; MDD, major depressive disorder; PFC, prefrontal cortex; BA, Brodmann Area. a p ⬍ .05; significant between-group differences in pairwise contrasts following a main group effect. b p ⬍ .01; significant between-group differences in pairwise contrasts following a main group effect. c Lithium medication reduces activation in BD. d MDD patients show greater activation than control subjects to baseline contrast (neutral faces vs. fixation cross). e Inverse correlation with Beck Depression Inventory score across all three groups.

jects and BD patients also showed greater neural responses to intense happiness than MDD patients in the right ventral and right dorsolateral prefrontal cortex.

sadness than CON subjects and MDD patients in the right ventral prefrontal cortex and in a large cluster including ventral and dorsal right anterior cingulate gyrus. To mild sadness, CON subjects showed significantly greater neural responses than both patient groups in the orbitofrontal cortex and greater response than BD patients in the right putamen and right dorsolateral prefrontal cortex. Major depressive disorder patients also showed significantly greater neural responses to mild sadness than BD patients in the right putamen. In the intense sadness condition, CON subjects

Differences in Neural Response to Sad Expressions Bipolar disorder patients showed a significantly greater neural response to expressions of mild sadness than either CON subjects or MDD patients in the left hippocampus and left ventral prefrontal cortex (Table 3, Figure 4). Bipolar disorder patients also showed significantly greater neural responses to intense

Table 2. Functional ROIs Showing Between-Group Differences in Happy Condition Brain Region (including Brodmann Area) and Group Map of Origin 50% Happy vs. Neutral L uncus/amygdala/caudate/putamen from BD

Center-of-Mass (R/L, A/P, S/I)

Cluster Size (Number Voxels)

CON ⬎ Other Groups

BD ⬎ Other Groups

⬎ MDDa

⬎ CONa ⬎ MDDb ⬎ CONb ⬎ MDDb ⬎ CONa ⬎ MDDa

⫺25

6

⫺16

131

ventromedial PFC (BA 10, 11) from BDc

11

49

⫺6

244

R ventrolateral PFC (BA 47) from BD

19

24

⫺6

41

⫺4

⫺5

⫺1

87

17

⫺28

⫺13

15

37 47

44 15

4 21

147 88

100% Happy vs. Neutral Thalamus/midbrain/amygdala/globus pallidus/caudate from CONd,f,g R parahippocampal gyrus from CONg R ventral PFC (BA 10, 47, 45, 46) from BD R dorsolateral PFC (BA 44, 45, 9) from CONe,g

⬎ BDa ⬎ MDDb ⬎ BDb ⬎ MDDb ⬎ MDDa ⬎ MDDb

MDD ⬎ Other Groups

⬎ MDDb ⬎ MDDa

All ROIs showed a significant main effect of group (p ⫽ .01) in Kruskal-Wallis test comparing all three groups. ROI, region of interest; R, right; L, left; A, anterior; S, superior; I, inferior; CON, control group; BD, bipolar disorder; MDD, major depressive disorder; PFC, prefrontal cortex; BA, Brodmann Area. a p ⬍ .05; significant between-group differences in pairwise contrasts following a main group effect. b p ⬍ .01, significant between-group differences in pairwise contrasts following a main group effect. c High doses of antidepressant medication increases activation in MDD. d Antidepressant medication increases activation in BD. e Antipsychotic medication reduces activation in BD. f Inverse correlation with education in MDD group. g Inverse correlation with Beck Depression Inventory score across all three groups.

www.elsevier.com/locate/biopsych

BIOL PSYCHIATRY 2004;55:578 –587 583

N.S. Lawrence et al Table 3. Functional ROIs Showing Between-Group Differences in Sad Condition Brain Region (including Brodmann Area) and Group Map of Origin 50% Sad vs. Neutral L hippocampus from BDe,f

Center-of-Mass (R/L, A/P, S/I)

Cluster Size (Number Voxels)

⫺26

⫺20

⫺3

24

⫺21

27

2

193

17 7

13 49

8 ⫺29

28 11

43

11

39

25

5

32

28

181

Ventral PFC (BA 10, 11) from BD

25

57

⫺6

28

R dorsal cingulate gyrus (BA 24) from CONd,g R dorsolateral PFC (BA 44, 45, 9) from CONg

16 47

18 26

29 13

15 85

⫺44

12

28

52

L ventral PFC (BA 11, 47, 45, 24, 32) from BD R putamen from CON Orbitofrontal cortex (BA 11) from CONg R dorsolateral PFC (BA 9) from CONc 100% Sad vs. Neutral R anterior cingulate (BA 9, 32) from BD

L dorsolateral PFC (BA 44, 9) from CONg

CON ⬎ Other Groups

⬎ BDb ⬎ BDb ⬎ MDDa ⬎ BDb

⬎ MDDa ⬎ BDb ⬎ MDDb ⬎ BDb ⬎ MDDb

BD ⬎ Other Groups ⬎ CONb ⬎ MDDa ⬎ CONa ⬎ MDDa

MDD ⬎ Other Groups

⬎ BDa

⬎ CONb ⬎ MDDb ⬎ CONb ⬎ MDDa ⬎ MDDb

All ROIs showed a significant main effect of group (p ⫽ .01) in Kruskal-Wallis test comparing all three groups. ROI, region of interest; R, right; L, left; A, anterior; P, posterior; S, superior; I, inferior; CON, control group; BD, bipolar disorder; MDD, major depressive disorder; PFC, prefrontal cortex; BA, Brodmann Area. a p ⬍ .05; significant between-group differences in pairwise contrasts following a main group effect. b p ⬍ .01; significant between-group differences in pairwise contrasts following a main group effect. c Lithium medication increases activation in BD. d Antidepressant medication increases activation in BD. e Positive correlation with Beck Depression Inventory score in BD group. f Positive correlation with Beck Depression Inventory score across all three groups. g Inverse correlation with Beck Depression Inventory score across all three groups.

showed increased neural responses compared with BD and MDD patients in the right and left dorsolateral prefrontal cortex (Table 3). MDD Patients At the chosen statistical threshold of p ⫽ .01 for betweengroup differences in neural response, there were no regions in which MDD patients alone demonstrated increased activation to emotional versus neutral facial expressions compared with both CON subjects and BD patients; however, MDD patients did show a trend (significant at a lower statistical threshold of p ⫽ .05) toward increased neural responses to mild sadness compared with CON subjects in the left parahippocampal gyrus (including hippocampus) and in the pulvinar nucleus of the left thalamus. Major depressive disorder patients also showed a trend (p ⬍ .05) toward increased neural responses to intense fear in the left dorsomedial prefrontal gyrus (compared with CON subjects and BD patients) and the right ventrolateral prefrontal cortex (compared with BD patients). Differences in Neural Response to Neutral Expressions Versus Fixation Cross The ROIs in which significant between-group differences were demonstrated in the emotional versus neutral expression contrasts were examined for differences in baseline activation to faces per se. Significant between-group differences were found in one cluster: MDD patients showed greater neural responses to neutral faces versus fixation cross than CON subjects in a region within the left medial prefrontal cortex that was activated to mild fear in CON subjects and BD patients. Since this baseline

difference in neural response was in the opposite direction to that found in response to mild fear, it is unlikely that this was confounding the pattern of results for between-group differences in neural response to emotion-neutral contrasts. Correlations Between Magnitude of Neural Response, Depression Severity, and Years of Education Neural responses to emotional faces showed a negative correlation with depression scores across all subjects in 10 of the 23 ROIs listed in Tables 1, 2, and 3; however, since our three experimental groups differed in the magnitude of depression severity (MDD ⬎ BD ⬎ CON), these correlations merely reflected the significant differences in neural response between the three groups in regions where this response decreased from CON to BD to MDD patients. Indeed, controlling for the effect of group using a partial correlation removed these negative correlations between neural response and depression. We therefore focused on correlations with depression symptoms that were also present within groups. Only one ROI (the left hippocampus activated to 50% sad in BD patients) showed a positive correlation with depression across all groups. Similarly, in BD patients alone, a near-significant positive correlation was demonstrated between depression severity and neural response to expressions of mild sadness in this region (correlation coefficient .56, p ⫽ .057). In MDD patients, there was a significant positive correlation between depression severity and response to mild sadness in the left parahippocampal gyrus, the region in which there was a nonsignificant trend toward greater activation in MDD patients than CON subjects (correlation coefficient .75, p ⬍ .05). In MDD, there was a significant negative correlation between www.elsevier.com/locate/biopsych

584 BIOL PSYCHIATRY 2004;55:578 –587

Figure 1. Globus pallidus/thalamic responses to expressions of mild fear versus neutral expressions in CON subjects and BD and MDD patients. Regions demonstrating significant neural responses to this contrast are demonstrated on axial slices (4 mm above the anterior commissure). The graphs show the mean (⫾ SEM) of the mean power of response (SSQ value; goodness of fit statistic) to expressions of mild fear versus neutral expressions. Positive values indicate greater responses to fearful expressions, and negative values indicate greater responses to neutral faces. CON, control; BD, bipolar disorder; MDD, major depressive disorder; SEM, standard error of the mean; SSQ, sum of square ratio; L, Left; R, Right.

years of education and magnitude of neural response to expressions of intense happiness within the large thalamus/caudate/ putamen cluster activated in control subjects (correlation coefficient ⫺.79, p ⬍ .05). Major depressive disorder patients showed a reduced neural response in this region relative to CON subjects and were less educated. The inverse nature of this correlation suggested, however, that the less educated MDD patients had a greater rather than smaller neural response within this region. Education differences did not, therefore, contribute to the overall between-group effect. There were no other correlations between years of education and magnitude of neural response within any other ROIs in any group. The Effect of Psychotropic Medication on Neural Responses to Emotional Versus Neutral Expressions All medication effects were significant at p ⬍ .05. In BD patients, lithium medication normalized neural responses toward those observed in CON subjects in two regions: in the right globus pallidus/thalamus in response to expressions of mild fear and in the right dorsal prefrontal cortex in response to mild sadness. Similarly, BD patients taking antidepressant medication showed a normalization (increase) in neural responses toward those observed in CON subjects in two regions: in the thalamus/ midbrain to intense happy expressions and in the right dorsal cingulate gyrus to intense sad expressions. Since the above medication effects diminished rather than increased differences in neural responses between BD patients and CON subjects, it is unlikely that medication effects contributed to the observed pattern of between-group differences in activation. Bipolar disorder patients taking antipsychotic medication demonstrated reduced neural responses within the right dorsolateral prefrontal cortex to expressions of intense happiness compared with those not taking antipsychotic medication. Finally, MDD patients taking higher doses of antidepressants showed increased neural responses to mild happiness in the ventromedial prefrontal www.elsevier.com/locate/biopsych

N.S. Lawrence et al

Figure 2. Left amygdala/ventrolateral PFC responses to expressions of intense fear versus neutral expressions in CON subjects and BD and MDD patients. Regions demonstrating significant neural responses to this contrast are demonstrated on axial slices (17 mm below the anterior commissure). The graphs show the mean (⫾ SEM) power of response (SSQ value; goodness of fit statistic) to expressions of intense fear versus neutral expressions. CON, control; BD, bipolar disorder; MDD, major depressive disorder; SEM, standard error of the mean; SSQ, sum of square ratio; PFC, prefrontal cortex.

cortex compared with MDD patients taking lower doses. Again, these latter medication effects diminished between-group differences in neural response and were not, therefore, contributing to the overall pattern of results.

Discussion The roles of subcortical, ventral, and dorsal prefrontal cortical regions in the response to emotionally salient stimuli have been previously highlighted, with specific patterns of functional abnormalities within these regions reported in BD and MDD patients (Phillips et al 2003a, 2003b). We aimed to examine the extent to which patterns of activation within these regions to emotionally salient stimuli distinguished BD and MDD patients. In particular, we wished to determine whether a pattern of increased subcortical response to facial expressions of positive and negative emotion characterized BD patients. As predicted, BD patients showed increased responses compared with CON subjects in subcortical regions to positive and negative facial expressions: within the left amygdala/caudate nucleus/putamen to expressions of mild happiness; within the right globus pallidus/thalamus to expressions of mild fear; and within the left amygdala/ventrolateral prefrontal cortex to expressions of intense fear. Bipolar disorder patients also demonstrated increased activation within the ventral prefrontal cortex to expressions of mild happiness and mild and intense sadness compared with CON subjects and increased activation to all facial expressions compared with MDD patients in similar subcortical and ventral prefrontal cortical regions. There were no correlations between the magnitude of response within the above regions and depression severity in the BD patients, suggesting that the pattern of increased subcortical and ventral prefrontal response to happy and fearful facial expressions in BD patients was not symptom-related. We employed one self-rating scale for depression severity, however; it is possible that other measures of depression severity may have

N.S. Lawrence et al

Figure 3. Left amygdala and ventromedial PFC responses to expressions of mild happiness versus neutral expressions in CON subjects and BD and MDD patients. Regions demonstrating significant neural responses to this contrast are demonstrated on axial slices (17 mm below the anterior commissure). The graphs show the mean (⫾ SEM) power of response (SSQ value; goodness of fit statistic) to expressions of mild happiness versus neutral expressions. CON, control; BD, bipolar disorder; MDD, major depressive disorder; SEM, standard error of the mean; SSQ, sum of square ratio; L, left; PFC, prefrontal cortex.

provided further information regarding the nature and magnitude of depressive symptoms in both patient groups, and these may have correlated with neural responses to the emotional stimuli. This could be addressed in future studies. Our finding of increased activation in thalamic, pallidal, and caudate/putamen regions in response to mild fear and mild happiness in BD patients supports recent work showing a similar pattern of response during performance of a motor task (Caligiuri et al 2003) and suggests a role for these regions in the pathophysiology of bipolar disorder. Bipolar disorder patients also showed increased activation in the left amygdala/ventrolateral prefrontal cortex to intense fear and in the left uncus/amygdala to mild happiness. Yurgelun-Todd et al (2000) demonstrated increased activation in the left amygdala to fearful but not happy faces in BD patients, whereas our data suggest that clusters encompassing the left amygdala were activated to expressions of fear and happiness in BD. It is possible that the observed increase in amygdalar and pallidal responses in BD patients may have been secondary to enlarged bilateral amygdalar and pallidal volumes, previously reported in patients with this disorder (Altshuler et al 1998, 2000; Strakowski et al 1999); however, BD patients did not demonstrate increased responses in these regions to all emotional expressions or to faces per se, so it is difficult to explain our findings entirely in terms of regional increases in volume in this population. Overall, the observed pattern of enhanced subcortical and ventral prefrontal cortical responses to a wide range of emotional stimuli and, in particular, to expressions of mild intensity in BD patients was largely in support of our predictions and may indicate in these patients an enhanced perceived salience of positive and negative emotional stimuli. This could contribute to the increased lability of mood observed in this group. Compared with the two other groups, BD patients demonstrated increased activation within the left hippocampus to expressions of mild sadness. In all subjects and in BD patients to

BIOL PSYCHIATRY 2004;55:578 –587 585

Figure 4. Left hippocampal responses to expressions of mild sadness versus neutral expressions in CON subjects and BD and MDD patients. Regions demonstrating significant neural responses to this contrast are demonstrated on sagittal slices (25 mm left of the anterior commissure-posterior commissure line). The graphs show the mean (⫾ SEM) power of response (SSQ value; goodness of fit statistic) to expressions of mild sadness versus neutral expressions and the correlation between depression symptoms (BDI) and neural response in the hippocampus in BD patients (correlation coefficient ⫽ .56, p ⫽ .057). CON, control; BD, bipolar disorder; MDD, major depressive disorder; SEM, standard error of the mean; SSQ, sum of square ratio; L, left; BDI, Beck Depression Inventory.

a near-significant extent, the magnitude of this activation correlated positively with depression severity. Increased activation within the left hippocampus may, therefore, represent a state feature of depression. The role of the hippocampus in memory retrieval is well described (Suzuki 2003). It is therefore possible that during depression mildly sad expressions in others evoke negative memories and associated contexts, particularly in more severely depressed individuals. Interestingly, in a previous report, a response to antidepressant treatment after 6 weeks in patients with MDD was associated with decreased hippocampal metabolism (Mayberg et al 2000), and hippocampal dysfunction is often implicated in long-term depression (Davidson et al 2002). Contrary to our prediction and previous studies demonstrating a negative attentional bias in MDD patients (David et al 1990; Gur et al 1992; Hale 1998), MDD patients in the current study did not demonstrate increased activation compared with CON subjects within any subcortical regions to negative facial expressions, although they demonstrated an enhanced response within the putamen to expressions of mild sadness compared with BD patients. At a lower statistical threshold (p ⫽ .05), MDD patients did demonstrate increased activation compared with CON subjects within the pulvinar nucleus of the thalamus and the left hippocampus and parahippocampal gyrus to expressions of mild sadness, with parahippocampal activation to these expressions again correlating positively with depression severity. The main finding with respect to MDD patients, therefore, was of a general “emotional blunting” in terms of reduced neural responses to all emotional stimuli (which has been previously reported; e.g., Elliott et al 2002), but with some suggestion of increased neural responses to negative, specifically mildly sad, facial expressions. Our data do not, therefore, support findings of increased amygdalar activity to negative facial expressions per se in MDD patients but instead suggest in these patients increased activity www.elsevier.com/locate/biopsych

586 BIOL PSYCHIATRY 2004;55:578 –587 within other regions important in emotion processing, subcortical regions, and hippocampus/parahippocampal gyrus, to mildly sad facial expressions. Compared with both patient groups, CON subjects demonstrated increased activation within the right parahippocampal gyrus and thalamic, uncus, and caudate nucleus to expressions of intense happiness. There were no regions in either patient group showing increased activation to these expressions compared with CON subjects, suggesting that expressions of intense happiness may have been perceived as particularly salient by CON subjects but not by BD and MDD patients. This is in support of previous findings suggesting attentional biases toward happy information in controls (Williams et al 1997) and away from happy expressions in MDD patients (e.g., Surguladze et al, in press), although it remains unclear whether there is a similar attentional bias in depressed BD patients. Increased activation in CON subjects compared with all patients was also observed in the right amygdala to expressions of mild but not intense fear. These findings may relate to the perception by CON subjects more than patients of the mild fear stimuli as particularly ambiguous, since previous studies have implicated the amygdala in the normal processing of ambiguous or potentially threatening information (Davis and Whalen 2001). This could be addressed in more detail in future studies examining neural responses to different intensities of fearful expression in BD patients and CON subjects. Control subjects also demonstrated increased activation compared with all patients within the right dorsolateral prefrontal cortex to expressions of mild fear and within bilateral dorsolateral prefrontal cortex to expressions of intense sadness. Dorsal prefrontal cortical regions have been associated with the suppression of emotional responses (Levesque et al 2003) and may be involved in the regulation of emotion per se (Phillips et al 2003a). It is therefore possible that CON subjects were able to suppress emotional responses to a greater extent than both patient groups to expressions of intense sadness in particular, although this requires further exploration. All subjects showed a high level of accuracy in discriminating the gender of faces during the scan, indicating that all subjects were attending to the stimuli equally well. Furthermore, neither patient group demonstrated abnormal performance on the (postscan) facial expression identification task, despite showing altered patterns of subcortical and prefrontal neural response to facial expressions of emotion. It is possible that the employment of more sensitive measures of subjective response to facial expressions than emotion identification accuracy (which is examined here), such as response bias (Surguladze et al 2003, in press) or the identification of covert rather than overt presentations of these expressions (for example, in a masking paradigm, e.g., Esteves and Ohman 1993), may uncover any alterations in subjective response to these stimuli in BD and MDD patients. Since there was only one significant group difference in baseline neural response to neutral faces versus fixation cross and this was in the opposite direction to the between-group difference in emotionneutral contrasts, it is unlikely that the observed differences to emotional facial expressions were the result of group differences in baseline neural activity. Rather, our findings indicate specific between-group differences in response to emotional expressions. It is also possible that medication may have contributed to some of the between-group differences in subcortical and prefrontal cortical response; however, the observed effects of medication diminished between-group differences in neural response, “normalizing” these responses in the patient groups. It is therefore unlikely that psychwww.elsevier.com/locate/biopsych

N.S. Lawrence et al otropic medication contributed to the between-group differences in neural response. In summary, our findings indicate that BD patients compared with CON subjects and MDD patients demonstrated increased subcortical and ventral prefrontal cortical activation to all categories of emotional expression at mild intensity and to expressions of intense fear and sadness. This pattern of activation was not correlated with depressive symptom severity, except for hippocampal responses to mild sadness. These findings support the conceptualization of BD in terms of an increased response within neural systems important for emotion processing and provide a functional neuroanatomical basis for the emotional lability observed in these patients. This study examined euthymic BD patients, as well as BD and MDD patients with a range of depressive symptoms. Future studies should extend these investigations to BD patients suffering from severe depression or during a manic episode, in addition to remitted MDD patients, to further understand the neurobiological basis of the vulnerability to and symptoms of BD and MDD. This study was supported by the James McDonnell-Pew Foundation and the Wellcome Trust. We would like to thank Krish Singh for the development of the structural template in Talairach space on which the functional data are presented in Figure 1 (for display purposes only). Abercrombie HC, Schaefer SM, Larson CL, Oakes TR, Lingren KA, Holden JE, et al (1998): Metabolic rate in the right amygdala predicts negative affect in depressed patients. Neuroreport 9:3301–3307. Alexander GE, Crutcher MDD, DeLong MR (1990): Basal ganglia-thalamocortical circuits: Parallel substrates for motor, oculomotor, “prefrontal” and “limbic” function. Prog Brain Res 85:119 –146. Altshuler LL, Bartzokis G, Grieder T, Curran J, Jimenez T, Leight K, et al (2000): An MRI study of temporal lobe structures in men with bipolar disorder or schizophrenia. Biol Psychiatry 48:147–162. Altshuler LL, Bartzokis G, Grieder T, Curran J, Mintz J (1998): Amygdala enlargement in bipolar disorder and hippocampal reduction in schizophrenia: An MRI study demonstrating neuroanatomic specificity. Arch Gen Psychiatry 55:663–664. American Psychiatric Association (1994): Diagnostic and Statistical Manual of Mental Disorders, 4th ed. Washington, DC: American Psychiatric Press. Asthana HS, Mandal MK, Khurana H, Haque-Nizamie S (1998): Visuospatial and affect recognition deficit in depression. J Affect Disord 48:57–62. Baxter LR, Schwartz JM, Phelps ME, Mazziotta JC, Guze BH, Selin CE, et al (1989): Reduction of prefrontal cortex glucose metabolism common to three types of depression. Arch Gen Psychiatry 46:243–250. Beauregard M, Levesque J, Bourgouin P (2001): Neural correlates of the conscious self-regulation of emotion. J Neurosci 21:1–6. Beck AT, Ward C, Mendelson M, Mock J, Erbaugh J (1961): An inventory for measuring depression. Arch Gen Psychiatry 4:561–571. Blumberg HP, Stern E, Martinez D, Ricketts S, de Asis J, White T, et al (2000): Increased anterior cingulate and caudate activity in bipolar mania. Biol Psychiatry 48:1945–1952. Bradley BP, Mogg K, Millar N (1996): Implicit memory bias in clinical and non-clinical depression. Behav Res Ther 34:865–879. Brammer MJ, Bullmore ET, Simmons A, Williams SCR, Grasby PM, Howard RJ, et al (1997): Generic brain activation mapping in fMRI: A nonparametric approach. Magn Reson Imaging 15:763–770. Breakspear M, Brammer M, Robinson P (in press): Construction of multivariate surrogate sets from nonlinear data using the wavelet transform. Physica D. Breiter HC, Etcoff NL, Whalen PJ, Kennedy WA, Rauch SL, Buckner RL, et al (1996): Response and habituation of the human amygdala during visual processing of facial expression. Neuron 17:875–887. Buchsbaum MS, Wu J, Siegel V, Hackett E, Trenary M, Abel L, et al (1997): Effect of sertraline on regional metabolic rate in patients with affective disorder. Biol Psychiatry 41:15–22. Bullmore ET, Brammer MJ, Rabe-Hesketh S, Curtis V, Morris RG, Williams SCR, et al (1999a): Methods for diagnosis and treatment of stimulus-correlated motion in generic brain activation studies using fMRI. Hum Brain Mapp 7:38 –48.

N.S. Lawrence et al Bullmore ET, Long C, Suckling J, Fadili J, Calvert GA, Zelaya F, et al (2001): Coloured noise and computational inference in neurophysiological (fMRI) time series analysis: Resampling methods in time and wavelet domains. Hum Brain Mapp 12:61–78. Bullmore ET, Suckling J, Overmeyer S, Rabe-Hesketh S, Taylor E, Brammer MJ (1999b): Global, voxel and cluster tests, by theory and permutation, for a difference between two groups of structural MR images of the brain. IEEE Trans Med Imaging 18:32–42. Calder AJ, Lawrence AD, Young AW (2001): Neuropsychology of fear and loathing. Nat Rev Neurosci 2:352–363. Caligiuri MP, Brown GG, Meloy MJ, Eberson SC, Kindermann SS, Frank LR, et al (2003): An fMRI study of affective state and medication on cortical and subcortical brain regions during motor performance in bipolar disorder. Psychiatry Res 123:171–182. Darwin C (1872/1965): The Expression of the Emotions in Man and Animals. Chicago: University of Chicago Press. David AS, Cutting J (1990): Affect, affective disorder and schizophrenia: A neuropsychological investigation of right hemisphere function. Br J Psychiatry 156:491–495. Davidson RJ, Pizzagalli D, Nitschke JB, Putnam K (2002): Depression: Perspectives from affective neuroscience. Annu Rev Psychol 53:545–574. Davis M, Whalen PJ (2001): The amygdala: Vigilance and emotion. Mol Psychiatry 6:3–34. Drevets WC, Price JL, Simpson JR, Todd RD, Reich T, Vannier M, et al (1997): Subgenual prefrontal cortex abnormalities in mood disorders. Nature 386:824 –827. Drevets WC, Price JL, Videen TO, Todd RD, Raichle ME (1995): Metabolic abnormalities in the subgenual prefrontal cortex and ventral striatum in mood disorders. Neurosci Abstr 21:260. Drevets WC, Videen TO, Price JL, Preskorn SH, Carmichael ST, Raichle ME (1992): A functional anatomical study of unipolar depression. J Neurosci 12:3628 –3641. Elliott R, Rubinsztein JS, Sahakian BJ, Dolan RJ (2002): The neural basis of mood-congruent emotional processing biases in depression. Arch Gen Psychiatry 59:597–604. Esteves F, Ohman A (1993): Masking the face: Recognition of emotional facial expressions as a function of the parameters of backward masking. Scand J Psychol 34:1–18. George MS, Huggins T, McDermut W, Priti P, Rubinow D, Post RM (1998): Abnormal facial emotion recognition in depression: Serial testing in an ultra-rapid-cycling patient. Behav Modif 22:192–204. Goodwin GM, Austin MP, Dougall N, Ross M, Murray C, O’Carroll RE, et al (1993): State changes in brain activity shown by the uptake of 99 mTcexametazime with single photon emission tomography in major depression before and after treatment. J Affect Disord 29:243–253. Gur RC, Erwin RJ, Gur RE, Zwil AS, Heimberg C, Kraemer HC (1992): Facial emotion discrimination: II. Behavioral findings in depression. Psychiatry Res 42:241–251. Hale WW (1998): Judgment of facial expressions and depression persistence. Psychiatry Res 80:265–274. Harmer CJ, Grayson L, Goodwin GM (2002): Enhanced recognition of disgust in bipolar illness. Biol Psychiatry 51:298 –304. Jaeger J, Borod J, Peselow E (1987): Depressed patients have atypical hemispace biases in the perception of emotional chimeric faces. J Abnorm Psychol 96:321–324. Kerr N, Phillips M (2002): Selective attention and emotional processing in bipolar disorder. Biol Psychiatry 51:207S. Levesque J, Eugene F, Joanette Y, Paquette V, Mensour B, Beaudoin G, et al (2003): Neural circuitry underlying voluntary suppression of sadness. Biol Psychiatry 53:502–510. Lior R, Nachson I (1999): Impairments in judgment of chimeric faces by schizophrenic and affective patients. Int J Neurosci 97:185–209. Lyon HM, Startup M, Bentall RP (1999): Social cognition and the manic defense: Attributions, selective attention, and self-schema in bipolar affective disorder. J Abnorm Psychol 108:273–282. Mayberg H (2001): Is there a future for PET studies of metabolism and blood flow in affective disorders research. Biol Psychiatry 49:6S. Mayberg HS, Brannan SK, Tekell JL, Silva A, Mahurin RK, McGinnis S, et al (2000): Regional metabolic effects of fluoxetine in major depression: Serial changes and relationship to clinical response. Biol Psychiatry 48:830 –843. Mayberg HS, Liotti M, Brannan SK, McGinnis S, Mahurin RK, Jarabek PA, et al (1999): Reciprocal limbic-cortical function and negative mood: Converging PET findings in depression and normal sadness. Am J Psychiatry 156:675–682. Mikhailova ES, Vladimirova TV, Iznak AF, Tsusulkovskaya EJ, Sushko NV (1996): Abnormal recognition of facial expression of emotions in de-

BIOL PSYCHIATRY 2004;55:578 –587 587 pressed patients with major depression disorder and schizotypal personality disorder. Biol Psychiatry 40:697–705. Morris JS, Frith CD, Perrett DI, Rowland D, Young AW, Calder AJ, et al (1996): A differential neural response in the human amygdala to fearful and happy facial expressions. Nature 383:812–815. Murphy FC, Sahakian BJ, Rubinsztein JS, Michale A, Rogers RD, Robbins TW, et al (1999): Emotional bias and inhibitory control processes in mania and depression. Psychol Med 29:1307–1321. O’Connell RA, van Heertum RL, Luck D, Yudd AP, Cueva JE, Billick SB, et al (1995): Single-photon emission computed tomography of the brain in acute mania and schizophrenia. J Neuroimaging 5:101–104. Oldfield RC (1971): The assessment and analysis of handedness: The Edinburgh Inventory. Neuropsychologia 102:97–113. Persad S, Polivy J (1993): Differences between depressed and nondepressed individuals in the recognition of and response to facial cues. J Abnorm Psychol 102:358 –368. Phillips ML, Drevets WC, Rauch SL, Lane RD (2003a): The neurobiology of emotion perception I: The neural basis of normal emotion perception. Biol Psychiatry 54:504 –514. Phillips ML, Drevets WC, Rauch SL, Lane RD (2003b): The neurobiology of emotion perception II: Implications for major psychiatric disorders. Biol Psychiatry 54:515–528. Phillips ML, Young AW, Senior C, Calder AJ, Perrett D, Brammer M, et al (1997): A specific neural substrate for perception of facial expressions of disgust. Nature 389:495–498. Rubinow DR, Post RM (1992): Impaired recognition of affect in facial expression in depressed patients. Biol Psychiatry 31:947–953. Sackeim HA (2001): The definition and meaning of treatment-resistant depression. J Clin Psychiatry 62(suppl 16):1–17. Sheline YI, Barch DM, Donnelly JM, Ollinger JM, Snyder AZ, Mintun MA (2001): Increased amygdalar response to masked emotional faces in depressed subjects resolves with antidepressant treatment: An fMRI study. Biol Psychiatry 50:651–659. Siegle G, Steinhauer SR, Thase ME, Stenger A, Carter CS (2002): Can’t shake that feeling: Event-related fMRI assessment of sustained amygdala activity in response to emotional information in depressed individuals. Biol Psychiatry 51:693–707. Strakowski SM, DelBello MP, Sax KW, Zimmerman ME, Shear PK, Hawkins JM, et al (1999): Brain magnetic resonance imaging of structural abnormalities in bipolar disorder. Arch Gen Psychiatry 56:254 –260. Surguladze SA, Brammer MJ, Young AW, Andrew C, Travis MJ, Williams SCR, et al (2003): A preferential increase in the extrastriate response to signals of danger. Neuroimage 19:1317–1328. Surguladze S, Young AW, Senior C, Brebion G, Travis MJ, Phillips ML (in press): Recognition accuracy and response bias to happy and sad facial expressions in patients with major depression. Neuropsychology. Suslow T, Junghanns K, Arolt V (2001): Detection of facial expressions of emotions in depression. Percept Mot Skills 92:857–868. Suzuki WA (2003): Declarative versus episodic: Two theories put to the test. Neuron 38:135–144. Talairach J, Tournoux P (1988): Co-Planar Stereotaxic Atlas of the Human Brain. New York: Thieme, Stuttgart. Warrington EK (1984): Recognition Memory Test. Windsor, UK: NFER-NELSON Publishing Company. Warrington EK (1996): Short Recognition Memory Test for Faces. Hove, UK: Psychology Press. Williams JMG, Mathews A, McLeod C (1996): The emotional stroop task and psychopathology. Psychol Bull 120:3–24. Williams JMG, Watts FM, Macleod C, Mathews A (1997): Cognitive Psychology and Emotional Disorders, 2nd ed. New York: John Wiley & Sons Ltd. Young AW, Perrett DI, Calder AJ, Sprengelmeyer R, Ekman P (2002): Facial Expressions of Emotion: Stimuli and Tests (FEEST). Bury St. Edmunds, UK: Thames Valley Test Company. Young RC, Biggs JT, Ziegler VE, Meyer DA (1978): A rating scale for mania: Reliability, validity and sensitivity. Br J Psychiatry 133:429 –435. Yurgelun-Todd DA, Gruber SA, Kanayama G, Killgore WD, Baird AA, Young AD (2000): FMRI during affect discrimination in bipolar affective disorder. Bipolar Disord 2:237–248. Yurgelun-Todd DA, Gruber SA, Kilgore WDS, Young AD (2001): Differential fMRI activation in bipolar patients during affect discrimination. Biol Psychiatry 49:279S.

www.elsevier.com/locate/biopsych