Neural markers of symptomatic improvement during antidepressant therapy in severe depression: subgenual cingulate and visual cortical responses to sad, but not happy, facial stimuli are correlated with changes in symptom score
- 73 Views
-
Like (2)
J Psychopharmacol OnlineFirst, published on September 18, 2008 as doi:10.1177/0269881108093589
Original Papers
Neural markers of symptomatic improvement during antidepressant therapy in severe depression: subgenual cingulate and visual cortical responses to sad, but not happy, facial stimuli are correlated with changes in symptom score
Journal of Psychopharmacology 00(00) (2008) 1–14 © 2008 British Association for Psychopharmacology ISSN 0269-8811 SAGE Publications Ltd, Los Angeles, London, New Delhi and Singapore 10.1177/0269881108093589
P Keedwell Cardiff University, Psychological Medicine, Cardiff, UK USA. D Drapier Institute of Psychiatry, Psychological Medicine, London, UK. S Surguladze Institute of Psychiatry, Psychological Medicine, London, UK. V Giampietro Insitute of Psychiatry, Brain Image Analysis Unit, London, UK. M Brammer Insitute of Psychiatry, Brain Image Analysis Unit, London, UK. M Phillips Cardiff University, Psychological Medicine, Cardiff, UK; Department of Psychiatry, University of Pittsburgh, Pittsburgh, PA, USA. Abstract
Resting state activity in the ventral cingulate may be an important neural marker of symptomatic improvement in depression. The number of task related functional magnetic resonance imaging (fMRI) studies correlating blood oxygenation level dependent (BOLD) response with symptomatic improvement is limited and methodologies are still evolving. We measured BOLD responses to sad and happy facial stimuli in 12 severely depressed individuals in the early stages of antidepressant treatment (Time 1) and 12 weeks later (Time 2) using event-related fMRI. We calculated correlations between temporal changes in BOLD response and changes in symptom scores. Most subjects improved markedly by Time 2. At Time 1, depression severity correlated positively with responses to sad stimuli in the right visual cortex, subgenual cingulate, anterior temporal pole and hippocampus and correlated negatively with responses to happy stimuli in left visual cortex and right caudate. Decreases in individual effect sizes of right subgenual cingulate and right visual cortical responses to sad, but not happy, facial stimuli were correlated with decreases in symptom scores. There are contrasting cortical and subcortical responses to sad and happy stimuli in severe depression. Responses to sad stimuli show the strongest correlates of clinical improvement, particularly in the subgenual cingulate.
Key words
antidepressant; depression; fMRI; neuroimaging; treatment
Introduction
There are consolidating findings, mainly from Positron Emission Tomography (PET) studies of resting state activity, which suggest that subgenual cingulate (BA25) activity is increased in depression and decreases following a variety of antidepressant treatments, including vagus nerve stimulation
(Zobel, et al., 2005), the antidepressant medication fluoxetine (Mayberg, et al., 2000), electroconvulsive therapy (Nobler, et al., 2001), Cognitive Behavioural Therapy (CBT) (Goldapple, et al., 2004), transcranial magnetic stimulation (Mottaghy, et al., 2002) and cingulotomy (Dougherty, et al., 2003; Malizia, 1997). Task-related studies using functional magnetic resonance imaging (fMRI) have mainly used facial expressions of affect
Corresponding author: Paul Keedwell, Institute of Psychiatry, Psychological Medicine, Denmark Hill, London SE5 8AZ, UK. Email: p.keedwell@iop.kcl.ac.uk
2
Neural markers of symptomatic improvement during antidepressant therapy
to determine markers of symptomatic improvement during antidepressant therapy, with mixed findings. A cross-sectional study (Gotlib, et al., 2005) found that moderately to severely depressed patients (mean BDI score = 25) had greater responses to the covert recognition of sad faces in the subgenual cingulate (BA25) when compared with healthy volunteers. For happy stimuli, depressed subjects showed greater responses in the pregenual cingulate (BA24/32). However, longitudinal treatment studies using the covert recognition of sad faces to determine indicators of symptomatic recovery have not shown decreases in subgenual cingulate response. For example, Fu, et al. found that responses in the pregenual cingulate gyrus and ventral striatum (Fu, et al., 2004) reduced over time, with changes in blood oxygenation level dependent (BOLD) effect size in the pregenual cingulate being correlated with changes in symptom score. Other studies have reported a decrease in response to the covert recognition of sad stimuli in the visual cortex (Davidson, et al., 2003; Fu, et al., 2004) and the amygdala (Davidson, et al., 2003; Sheline, et al., 2001). A more recent longitudinal study by (Fu, et al., 2007) using happy facial expressions found no effect in the ventral cingulate over time, in contrast to their study of responses to sad faces. Rather, they showed attenuated overall responses in limbic-subcortical and visual cortical regions, which increased after antidepressant treatment compared with healthy controls. Also, trends in response to happy stimuli of different potencies showed increased capacity over time in the hippocampus and visual cortex. The dissociated findings for visual cortical and subcortical response are consistent with the findings of a cross-sectional study, which showed a double dissociation of response in these regions to sad and happy stimuli in depressed and healthy individuals (Surguladze, et al., 2005). Dissociated visual cortical responses are also consistent with behavioural studies showing a perceptual bias in facial emotion recognition toward sadness and away from happiness in depression (Murphy, et al., 1999; Sloan, et al., 2002; Williams, 1996). Hence, in contrast to the resting state studies, the subgenual cingulate has not been identified as a marker of symptom change in studies of response to facial affect. However, PET and fMRI studies have differed in their patient populations, varying from treatment resistant cases in secondary care to previously untreated individuals recruited from the community by advert (Fu, et al., 2004). It is important to clearly specify which subgroup of depressed patients is being studied as this may determine responses to different biological treatments. For example, a review of antidepressant effectiveness in depression (Khan, et al., 2002) found that the proportion of studies favouring antidepressants over placebo increased with the severity of depression; the response to placebo decreased with increasing severity, whereas that to antidepressants increased. Changes in neural response over time in longitudinal studies can occur as a result of factors that are independent of symptomatic recovery, including non-treatment specific neurophysiological effects of antidepressants and regression to the mean. Fu et al (Fu,et al., 2004) included a control group of healthy
volunteers who were scanned on two occasions. Although this controls for regression effects, it cannot control for the fact that individuals were taking medication at follow-up only, making the findings difficult to interpret. Furthermore, it is commonly asserted that changes over time are because of the therapeutic effects of antidepressant treatment, but one cannot state this definitively without a matched placebo-treated group of depressed patients; recovery can occur as a result of placebo response, spontaneous recovery and psychosocial interventions, prescribed or otherwise. However, if one is examining the neural correlates of clinical recovery per se, difficulties of interpretation can be addressed to some extent by (a) examining correlations between changes in individual BOLD responses and changes in individual symptom scores and (b) by examining differential responses within subjects to different stimuli within the same paradigm. In summary, there is some agreement in the literature with regard to the importance of ventral cingulate, visual cortical and subcortical activity as markers of symptom change in response to sad stimuli. However, there is a paucity of literature examining the differential responses of these brain regions to sad and happy stimuli in within-subject comparisons, interactions with changes over the course of treatment and correlations between individual BOLD responses to sad and happy stimuli and changes in symptom score. In this study, we report findings from a naturalistic longitudinal examination of the effects of clinical improvement on neural response to covert presentations of sad and happy facial expressions of increasing intensity in a psychiatric population of depressed individuals. Based on the most frequent findings in the neuroimaging literature summarised above, we hypothesised that, before improvement, there would be an increased response in the ventral cingulate, primary visual/fusiform cortex and subcortical structures involved in emotion processing (parahippopcampus/amygdala and ventral striatum) to sad faces of increasing intensity, when compared with the baseline condition and that this difference would be attenuated upon clinical recovery. Conversely, we predicted an increased response in the pregenual cingulate and attenuated responses in the visual cortex and ventral striatum to happy faces in the early stages of treatment, with a reversal of these changes following improvement. Thus, we predicted a dissociation of response to sad and happy stimuli in areas involved in visual processing, reflecting a well-recognised perceptual bias in depression (Murphy, et al., 1999; Sloan, et al., 2002; Williams, 1996) and associated with neurobiological changes in the ventral cingulate and subcortical structures thought to be involved in emotion processing. We also predicted a reversal of this dissociated pattern with symptomatic improvement in keeping with limited findings of previous longitudinal fMRI studies as summarised above, and consistent with the dissociated pattern of visual and subcortical responses to sad and happy stimuli previously reported in a cross-sectional study of healthy and depressed volunteers (Surguladze, et al., 2005).
Neural markers of symptomatic improvement during antidepressant therapy
3
Methods
Subjects
We recruited 12 subjects, six men and six women, with a mean age of 49 from the inpatient and outpatient departments of the Maudsley Hospital, Decrespigny Park, London. They met International Classification of Diseases Version 10 (ICD-10) criteria for a major depressive episode. Patients with psychotic symptoms, a history of mania or hypomania, comorbid current alcohol or drug misuse, cognitive deficits, organic brain disease, including history of trauma, were excluded. In this naturalistic study, we endeavoured to recruit patients who were severely depressed and who had just commenced antidepressant treatment. Subjects had all been started on new treatments within the last 2 weeks. They were being treated with a variety of different antidepressant treatments, in common with the study by Gotlib et al (Gotlib, et al., 2005). The mean followup time was 12 weeks (range 8–16 weeks).
were acquired at each of 16 near-axial noncontiguous 7-mm thick planes parallel to the intercommissural (AC-PC) line: TE 40 ms, TR 2 s, in-plane resolution 3.44 mm, interslice gap 0.7 mm, matrix size: 64 × 64 pixels.
Neuroimaging data analysis
The data were analyzed using the Institute of Psychiatry software (Brammer, et al., 1997; Bullmore, et al., 1999a). Neural responses to prototypic (100%) emotion were compared with baseline by time series analysis using gamma variate functions (peak responses at 4 and 8 s) to give the best-fit (least-squares) model of the time series of the BOLD response at each intracerebral voxel. A goodness-of-fit statistic, the sum of squares (SSQ) ratio, was then computed at each voxel. This was the ratio of the SSQ of deviations from the mean intensity value due to the model (fitted time series) divided by the SSQ due to the residuals (original time series minus model time series). To the sample, the distribution of SSQ ratio under the null hypothesis that observed values of SSQ ratio were not determined by experimental design (with minimal assumptions), the time series at each voxel was permuted using a waveletbased resampling method. This parametric data-driven, permutation-based approach is set out in detail in Bullmore et al (Bullmore, et al., 1999b). The first step is to fit a BOLD model to the time series data at each voxel. The model is then refitted a large number of times at each voxel after wavelet-based permutation of the time series to destroy the relationship between the stimulus and the experimental response (the validity of this approach has been established in detail in Bullmore et al (Bullmore, et al., 2001)). Any “observed” SSQ ratio (or any other statistic chosen to be tested) could then be compared with its own distribution under the null hypothesis and its probability of chance occurrence assessed. Observed and randomized SSQ ratio maps were transformed into the standard space of Talairach and Tournoux (Talairach and Tournoux, 1998). A generic brain activation map (GBAM) was produced for each experimental condition by testing the median observed SSQ ratio over all subjects (median values were used to minimize outlier effects) at each voxel in standard space against a critical value of the permutation distribution for median SSQ ratio ascertained from the spatially transformed wavelet-permuted data. The 95% confidence limit for voxel occurrence was determined by bootstrapping null data and was such that >1 false-positive voxel would only be observed once in 20 experiments. This approach was extended to cluster level by connecting activated voxels into three-dimensional (3-D) clusters, computing the integral of SSQ ratio over the cluster, doing the same for the permuted data, and using the data to establish the critical threshold for cluster mass at a given level of probability under the null hypothesis of no experiment effect. At the whole brain level, outputs would contain a count of observed clusters and false-positive clusters at different levels of clusterwise significance. Increasingly, conservative calculations could
Functional neuroimaging task
This was identical to the task carried out in a previously published study (Surguladze, et al., 2005). All individuals participated in two 6-min experiments using event-related fMRI. They were presented with a baseline fixation cross, 20 neutral expressions or emotional expressions of either 50% or 100% (prototypical) intensity (n = 20 for each intensity). In one experiment, the emotional expression was of sadness and in the other it was of happiness. Expressions were posed by 10 different volunteers (4 male volunteers) from a standardized computer-morphed series (Ekman and Friesen, 1976). The order of experiment was counter-balanced and within each experiment both the order of expressions and the interstimulus interval (3–8 s) was randomized. During the ISI, patients viewed the fixation cross. Facial expressions were always presented for 2 s.
Image acquisition
Magnetic resonance (MR) images were acquired using a GE Signa 1.5T Neuro-optimised MR system (General Electric, Milwaukee, Wisconsin, USA) for gradient echo echoplanar imaging (EPI) at the Maudsley Hospital, London, United Kingdom. A quadrature birdcage headcoil was used for radio frequency transmission and reception. An inversion recovery EPI data set was acquired at 43 near-axial 3-mm thick planes parallel to the anterior commissure–posterior commissure (ACPC) line: echo time (TE) 73 ms, time to inversion 180 ms, repetition time (TR) 16 s, in-plane resolution 1.72 mm, interslice gap 0.3 mm, matrix size: 128 × 128 pixels. This higher resolution EPI data set was used to register the fMRI data sets acquired from each individual in standard stereotactic space. In all, 180 T2*-weighted images depicting BOLD contrast
4
Neural markers of symptomatic improvement during antidepressant therapy
be carried out (by adjusting the cluster-wise P value) to ensure that there were no false-positive clusters. In this case, this would equate to a GBAM significance level of P < 0.01. The standard space maps were then analyzed by fitting linear trends at each voxel for all individuals at the three intensity levels of stimulation, neutral, 50% and 100% of emotion compared with baseline, with orthogonal polynomial trend analysis. In this study, voxelwise estimates of linear trends in each individual in the two groups for each emotional expression experiment were fitted to a linear model (Surguladze, et al., 2005). The null distribution of the regression slope was obtained by systematic data permutation between the two groups (to produce all possible rearrangements of the data except the original or a random subset of these with large groups sizes) and refitting the model. Following computation of the observed and null distributions of the regression slope, the voxelwise and clusterwise significance of differences in linear trends between groups were obtained by analysis of variance (ANOVA) of these data sets in a manner similar to that for between-group comparisons of GBAMs (Bullmore, et al., 1999b). Comparison of GBAMs of responses to prototypically happy and sad stimuli at Time 1 and Time 2 were obtained by ANOVA. A similar approach was used for the setting of significance levels. The statistical parameters for detection of clusters showing significant differences were set at levels (voxelwise P = 0.05, clusterwise P = 0.005) unless otherwise stated. This level of significance ensured <1 false-positive cluster in each comparison. The statistical parameters for detection of clusters showing significant correlations between Hamilton Rating Scale for Depression (HRSD) score and whole-brain responses (BOLD effect sizes) to prototypically sad or happy faces (100% intensity) were set at levels (voxelwise P = 0.05, clusterwise P = 0.005) unless otherwise stated. This level of significance ensured <1 false-positive cluster in each correlation.
between differences in effect size at Time 1 and Time 2 versus differences in HRSD score, for both happy and sad conditions (values at Time 2- values at Time 1 in both cases). We also examined correlations between changes in %BOLD response and changes in HRSD score using the anatomically defined clusters (above), which had showed differences in response to sad stimuli between Time 1 and Time 2.
Results
Behavioural results
The mean drop in depression (HRSD) scores was 56% or 14 points, (t = 5, df = 11, P = 0.000; confidence intervals: 7.8 to 19.6), from the severe range (range = 15 to 35; mean = 25; sd = 0.7) to the mild range (range = 2 to24; mean = 11, sd = 10.4). Of the 12 subjects recruited, 11 improved, seven remitted ((HRSD) score < 10), one had a greater than 50% drop in Hamilton scores and three showed a mild improvement (See Table 1). In addition, there was a significant mean drop of 14.1 points on the Beck Depression Inventory (BDI) from the severe range (mean = 31.8) to the mild range (mean = 19.3); (t = 2.7, df = 11, P = 0.002; confidence intervals: 6.6 to 21.6). There was no positive correlation between (HRSD) scores at Time 1 and Time 2 (r = 0.19, P = 0.556, 2 tailed), reflecting the fact that not all patients improved by the same degree. A positive correlation between (HRSD) score at Time 1 and size of change in depression score by Time 2 would suggest that those with higher initial depression severity did better in terms of symptom improvement. However, although there was a trend toward such a correlation, this did not quite reach significance (r = 0.57, P = 0.051, 2 tailed); rather, there was a signifTable 1 Treatment responses (HRSD scores) HRSD scores Time 1 26 20 30 21 15 22 35 30 27 26 19 27 25 0.7 HRSD scores Time 2 7 3 9 18 2 24 8 7 16 4 14 22 11 10.4 Change in HRSD scores (Time 2 –Time 1) 21 17 21 3 13 −2 27 23 11 22 5 5 14 11.3 Symptomatic response remitted remitted remitted improved remitted not improved remitted remitted >50% response remitted improved improved
Further region of interest analyses
A time series analysis was conducted for each individual’s responses to sad and happy facial expressions of increasing intensity, versus fixation cross, at anatomically defined (Talairach and Tournoux, 1998) coordinates for the regions of bilateral amygdalae, (±23, −4, −16), caudate (±14, 15, 9), putamen (±18, −4, 9) and subgenual cingulate gyri (±2, 6, −6; Mayberg, et al., 1999), at Time 1 and Time 2. Mean effect sizes (%BOLD change) across all individuals at 6 s post stimulus were then compared across Time 1 and Time 2 using paired student ttests. Using clusters identified on maps of correlations (at P = 0.005) between depression severity and neural responses at Time 1 as templates, we extracted mean values of %BOLD change from the GBAMs representing prototypical emotion, happy and sad, versus fixation cross for all 12 depressed individuals. We then carried out multiple regression analyses
Subject
1 2 3 4 5 6 7 8 9 10 11 12 Means SD
Neural markers of symptomatic improvement during antidepressant therapy
5
icant negative correlation between size of change and final depression score at Time 2 (r = −0.66, P = 0.02, 2 tailed). This reflects the lack of variance in depression scores at Time 1 (SD = 0.7) compared with Time 2 (SD = 10.4). Most of the variance at Time 2, therefore, is accounted for by variance in the size of change. There were no significant differences between Time 1 and Time 2 in accuracy in the gender decision tasks t = −0.38, P = 0.710. In both groups, however, the accuracy of gender decision for these expressions was low. This was because of the masked nature of the facial expressions, that is, hair and any other non-facial features were cropped, which renders gender decision difficult. We then performed a 2 × 2 repeated measures ANOVA (time × emotion).There was no main effect of time (F = 0.634, P = 0.427) or emotion (happy, sad; F = 1.226, P = 0.269) but there was an interaction of time × emotion: F = 4.143, P = 0.043. Post-hoc t-tests showed a trend toward shorter reaction times for happy faces (t = 1.828, P = 0.069, CI = −0.008 to 0.216) at Time 2 compared with Time 1 but no significant difference for sad faces (t = −0.643, P = 0.521, CI = −0.115–0.058).
b. Correlational analyses between positive and negative neural responses to prototypic (100% happy and 100% sad) stimuli, versus fixation cross, and depression severity (HRSD score) at time 1 (see Table 2) All correlations were highly significant (P = 0.005) even after correction for multiple comparisons (positive and negative correlations; Bonferroni-corrected threshold for significance P = 0.005/2 = 0.025). Happy faces Positive correlations (increasing BOLD effect with increasing depression severity) were seen in the ventromedial prefrontal cortex (L medial frontal pole, BA10, and R anterior cingulate, BA24) and right posterior cingulate (BA31). Negative correlations (decreasing BOLD effect with increasing depression severity) were seen in the left dorsolateral prefrontal cortex (BA44/45), left posterior cingulate (BA31), left primary visual cortex (BA18) and right caudate. Sad faces (see Figure 1) Positive correlations were seen in the right primary visual cortex (BA17-19), right subgenual cingulate gyrus (BA25), right hippocampus and the right post central
Neuroimaging results
Time-1 analyses a. GBAMs for prototypic (100% happy and 100% sad) stimuli versus fixation cross For 100%, sad stimuli positive responses were seen in the left fusiform gyrus (cluster size 229; Talairach coordinates −40, −78, −13; P = 0.0008, BA18) and the right medial occipital gyrus (cluster size 398; Talairach coordinates 32, −78, −7; P = 0.0002, BA19) of the primary visual cortex. Negative responses were seen in the left occipitotemporal gyrus (cluster size 10; Talairach coordinates −30, −23, −13; P = 0.0077, BA36) and the right temporal pole (cluster size 3; Talairach coordinates 32, 19, 42; P = 0.006; BA38). For 100%, happy stimuli positive responses were seen in the left inferior frontal gyrus (cluster size 161, Talairach coordinates −43, 0.00, 31.35, P = 0.001993, BA44), right cerebellum (cluster size 1109, Talairach coordinates 29, −59, −24, P = 0.000285, BA71), left inferior frontal gyrus (cluster size 243, Talairach coordinates 47, 11, 26, P = 0.001, BA44), postcentral gyrus (cluster size 296, Talairach coordinates −40, −22, 53, P = 0.0002, BA1-3) and dorsal cingulate gyrus (cluster size 327, Talairach coordinates 7, 0, 48, P = 0.0003, BA24). Negative responses were seen in the right occipitotemporal gyrus (cluster size 6, Talairach coordinates 18, −30, −24, P = 0.004, BA36), right uncus (cluster size 5, Talairach coordinates 18, −11, −29, P = 0.004, BA35), left middle temporal gyrus/medial occipital gyrus (cluster size 4, Talairach coordinates −36, −70, 20, P = 0.004, BA39/19) and frontal pole (cluster size 6, Talairach coordinates 14, 59, 9, P = 0.004, BA10).
Table 2 Size
Correlations between neural responses and HRSD score Tal(x) Tal(y) Tal(z) Side BA Cerebral region
Negative correlation with neural responses to sad stimuli 67 11 −59 −40 R 71 Cerebellum 19 −18 −37 −29 L 71 Cerebellum −2 R 38 Ant temporal pole 6 40 18 38 −18 −81 −2 L 18 Lingual gyrus 10 −50 −11 20 L 20 Postcentral gyrus 23 50 −48 20 R 22 Superior temporal gyrus Positive correlation with neural responses to sad stimuli −78 −7 R 17 Lingual gyrus 10 7 4 25 −48 −7 R 19 Fusiform gyrus 3 4 15 −7 R 25 Subgenual cingulate gyrus 3 25 −52 −2 R 19 Lingual gyrus 3 4 −70 −2 R 18 Lingual gyrus 3 14 −33 4 R — Hippocampus 3 47 −22 37 R 1 Postcentral gyrus (S1) Negative correlation with neural responses to happy stimuli 35 −39 30 15 L 45/46 Dorsolateral prefrontal cortex 26 −22 −30 31 L 31 Posterior cingulate 12 −36 −78 −13 L 18 Primary visual cortex (fusiform gyrus) 10 14 15 9 R — Caudate 9 −7 −63 4 L 18 Primary visual cortex (lingual gyrus) Positive correlation with neural responses to happy stimuli 42 7 −52 26 R 31 Posterior cingulate gyrus 16 −14 48 26 R 9/10 Medial frontal lobe 6 11 15 26 R 24 Anterior cingulate P < 0.005.
6
Neural markers of symptomatic improvement during antidepressant therapy
Figure 1 Correlations between absolute scores on the HRSD and percentage change in BOLD signal in response to 100% sad faces versus baseline cross at Time 1. Positive correlations (hot colours) were seen in the right primary visual cortex (crosshairs at cluster 7, −78, −7; additional cluster at 4, 25, −48; BA17-19), right subgenual cingulate gyrus (SG; 4, 15, −7; BA25), right hippocampus (14, −33, 4) and the right post central gyrus (47, −22, 37; BA1). Negative correlations (cool colours) were demonstrated in the bilateral cerebellum (11, −59, −40; −18, −37, −29; BA71), left primary visual cortex (−18, −81, −2; BA18), right superior temporal gyrus (50, −48, 20; BA22), left postcentral gyrus (−50, −11, 20; BA20) and right anterior temporal lobe (40, 18, 2; BA38). Radiological images.
gyrus (BA1). Negative correlations were seen in the bilateral cerebellum (BA71), left primary visual cortex (BA18), right superior temporal gyrus (BA22), left postcentral gyrus (BA20) and right anterior temporal lobe (BA38). Correlations with depression severity at Time 2 were not calculated because most patients had remitted thereby making any interpretation of the results problematic. c. Trend analysis: the effect of increasing emotion intensity For sad stimuli, there was a linear increase in BOLD response to increasing emotion intensity in the right fusiform gyrus (BA37) and the right cerebellum (BA71). There were significant decreases in neural response to increasing emotion intensity in the left supramarginal gyrus (BA40). For happy stimuli, there was a linear decrease in BOLD response to increasing intensity
in the L cerebellum (BA71) and L supramarginal gyrus (BA40). There were no significant increases in neural response to increasing emotion intensity for happy stimuli. d. Time series extraction at regions of interest: the effect of emotion intensity in the right visual cortex to sad stimuli Using the GBAM for sad stimuli as a template, we extracted the percentage of BOLD change in the right visual cortex (BA18, Talairach Coordinates 32, −78, −7) at each TR following a sad stimulus (total number of TRs = 10; TR duration = 2 s), at Time 1, averaged for all 12 depressed individuals and for each intensity of emotion (neutral, 50% and 100%). This showed a linear relationship between sadness intensity and %BOLD response in the left visual cortex at Time 1 (see Figure 2).
Neural markers of symptomatic improvement during antidepressant therapy
7
Figure 2 Time series of percent change in blood oxygenation level dependent (BOLD) signal to neutral and both intensities of sad expression in the left visual cortex (BA18, Talairach Coordinates −40, −78, −12: cluster derived from GBAM of responses to 100% sad stimuli versus fixation cross), averaged for 12 depressed individuals over 10 x TRs of 2 seconds duration (Time 1).
Longitudinal analyses a. The interaction between emotions of both intensities versus time Happy faces There were no significant effects of happy emotion (50% + 100% happiness vs fixation cross) versus time. Sad faces (see Figure 3) Sad stimuli (50% + 100% versus fixation cross) were associated with greater responses in the primary visual cortex bilaterally at Time 1, BA17-BA19, compared with Time 2. In contrast, there were greater responses in the left ventrolateral prefrontal cortex, BA47, (−51, 18, −6), at Time 2 than at Time 1. (Talairach coordinates are given in parentheses). b. The interaction between trends in emotion intensity by time (see Table 3, Figure 4) Happy faces A factorial analysis of linear trends in neural response to happy faces of increasing intensity (neutral, 50% happy and 100% happy (all versus baseline fixation cross)) by time (Time 1 = before improvement; Time 2 = after improvement) showed a stronger linear relationship at Time 2 versus Time 1 in the primary visual cortex, especially on the right (BA17; see Figure 1D), right frontal pole (BA10) and right dorsolateral prefrontal cortex (BA9/46). This positive interaction with time was also seen in a smaller area of the left visual cortex (BA18) and in the bilateral cerebellum (BA71). Negative relationships with time (greater linear responses at Time 1 than Time 2) were seen in the right cerebellum (BA71).
(P < 0.005 in each case; corresponding to <1 false-positive cluster per comparison). Sad faces A similar analysis of linear trends in neural response to sad faces of increasing intensity by time revealed a stronger linear relationship at Time 2 versus Time 1 in the left ventrolateral prefrontal cortex (BA47) and left anterior temporal pole (BA38). Negative relationships with time (greater linear responses at Time 1 than Time 2 were seen in the bilateral primary visual cortex (BA17-19; see Figure 1 D), right greater than left, right supramarginal gyrus (BA40) and bilateral cerebellum (BA71). (P < 0.005 in each case; corresponding to < 1 false-positive cluster per comparison).
Further region of interest analyses
Effect sizes for responses to sad and happy facial expressions of increasing intensity were extracted from anatomically defined (Talairach and Tournoux, 1998) coordinates for the regions of bilateral amygdalae, (±23, −4, −16), caudate (±14, 15, 9), putamen (±18, −4, 9) and subgenual cingulate gyri (±2, 6, −6) at Time points 1 and 2. Changes where effect sizes became more or less negative over Time (off-off comparisons) were disregarded because their relationship to the task was unclear. For happy stimuli, the only statistically significant change in effect size occurred in the region of the right putamen (mean at Time 1 = 0.064, mean at Time 2 = −0.015, difference in means = 0.079, 95% CI = 0.071 to 0.087, t = 2.95, df = 22, P < 0.01) but effect sizes were small. For sad stimuli, there were significant decreases in effect size at Time 2 in the bilateral amygdala, left caudate, right
8
Neural markers of symptomatic improvement during antidepressant therapy
Figure 3 The main effect of Time on neural responses to sad facial expressions (50% and 100% intensity versus fixation cross, cross hairs at −14, −88, 5 and −51, 18, −6). There were no significant main effects of happy stimuli (50% and 100%) versus time, (see text).
Table 3 Factorial analysis of time by emotion intensity (neutral, 50%, 100%, vs fixation cross), calculated separately for happy and sad stimuli, respectively (P < 0.005) Size Happy faces. Time 1 > Time 2 15 2 Happy faces. Time 2 > Time 1 45 6 4 15 18 15 8 4 5 Sad faces. Time 1 > Time 2 46 26 26 35 18 18 15 13 Sad faces. Time 2 > Time 1 38 14 Tal(x) Tal(y) −70 −63 −93 −93 −81 −85 −74 52 22 44 19 −89 −81 −67 −70 −37 −30 −63 −59 19 19 Tal(z) −24 −29 −2 −13 −7 −18 −35 20 26 26 31 4 −7 −2 26 53 48 −13 −46 −7 −29 Side BA Cerebral region
25 29 7 −14 −32 −22 0 36 47 40 51 14 −32 47 29 43 47 32 −14 −51 −32
R R R L L L R R R R R R L R R R R R L L L
71 71 17 17 18 71 71 10 9 46 44 17 18 19 19 40 40 71 71 47 38
Cerebellum Cerebellum Primary visual cortex (V1) Primary visual cortex (V1) Primary visual cortex (V2,V3) Cerebellum Cerebellum Frontal pole Dorsolateral prefrontal cortex Dorsolateral prefrontal cortex Inferior frontal gyrus (Broca’s) Primary visual cortex (V1) Primary visual cortex (V2,V3) Primary visual cortex Primary visual cortex Supramarginal gyrus Supramarginal gyrus Cerebellum Cerebellum Ventrolateral prefrontal cortex Ant temporal pole
Neural markers of symptomatic improvement during antidepressant therapy
9
Figure 4 Factorial analysis: The interaction of time (Time 1 vs Time 2) versus trends in emotion intensity (neutral, 50% and 100% emotion versus fixation cross). Cold colours represent the effect of sad stimuli where the gradient of linear trends in neural responses were more positive at Time 1 than Time 2. Hot colours represents the effect of happy stimuli where the gradient of linear trends in neural responses were more positive at Time 2 than Time 1. There were no other significant interactions. Crosshairs are at 4, −95, −6 in both figures. Radiological images.
subgenual cingulate gyrus and right putamen. In contrast, there were significant increases in effect size at Time 2 in the left subgenual cingulate (with a negative effect size becoming more positive) and right caudate (see Figure 5).
Correlations between percent BOLD change and change in depression score
Using clusters identified on the correlation maps for depression severity as templates, (above) we calculated correlations between percent BOLD change at these clusters and change in depression severity (HRSD score) for both happy and sad conditions (values at Time 2 subtracted by values at Time 1 in both cases). There was a strong positive correlation (r = 0.863, P < 0.001) between the size of reduction in % BOLD response at the right subgenual cingulate (BA25: 4, 15, −4) and reduction in HRSD score (Figure 6). In other words, those who showed the greatest reduction in subgenual response to sad stimuli had the best response to antidepressant treatment. In addition, there was a weaker correlation between change in BOLD response to sad stimuli in the right primary visual cortex (r = 0.584, P = 0.046) and reduction in HRSD score, but this did not remain significant after correcting for multiple comparisons. In addition, BOLD changes in the following brain areas were correlated with each other: BA17 and BA18 of the right primary visual cortex (r = 0.752, P = 0.005), the right hippocampus and the primary visual cortex (BA19, r = 0.625,
P = 0.030), and the right subgenual cingulate (BA25) and BA19 of the right primary visual cortex (r = 0.662, P = 0.019). For happy stimuli, there were no significant correlations between BOLD change and change in depression score when using this technique, although a negative correlation between change in right hippocampal responses and change in depression severity nearly reached significance (r = −0.558, P = 0.058). Similarly, when we used as a template clusters that had showed a negative correlation between depression severity and responses to happy stimuli at Time 1, no significant correlations were found between change in BOLD response in these locations and change in HRSD score. There were, however, significant positive correlations in BOLD response between the following: the left posterior cingulate gyrus (precuneus; BA31; −14, −41, 37) and the left hippocampus (−25, −30, −2; r = 0.871, P = 0.000), the left primary visual cortex (BA19; −25, −67, −7) and the left hippocampus (−25, −33, −7) (r = 0.755, P = 0.004), the left cuneus (BA31; −7, −63, 9) and the hippocampus (−32, −30, 4; r = 0.736, P = 0.006) and between the left insula (−40, −7, 4) and the hippocampus (−25, −33, −7). We also examined correlations between changes in % BOLD response and changes in HRSD score using anatomically defined clusters, which had showed differences in response to sad stimuli between Time 1 and Time 2, namely the bilateral amygdalae, (±23, −4, −16), bilateral caudate (±14, 15, 9), the bilateral putamen (±18, −4, 9) and bilateral subgenual cingulate gyrus (±2, 6, −6). For sad stimuli, there
10
Neural markers of symptomatic improvement during antidepressant therapy
Figure 5 Region of interest analysis (ROIs) of responses to sad stimuli: changes in %BOLD response 6 seconds post stimulus at Time 1 and Time 2, using the GBAMs of responses to 100% sad stimuli versus fixation cross. Significance levels are shown above charts of mean value changes for each ROI: amyg(R) = right amygdala (mean difference = 0.28%, 95% CI = 0.179 to 0.237%, df = 22, t = 4.05, P < 0.001), amyg(L) = left amygdala (mean difference = +0.115%, 95% CI = 0.085 to 0.145%, df = 22, t = 2.61, P < 0.05), BA25(R) = right subgenual cingulate (mean difference = −0.076%, 95% CI = 0.061 to 0.091%, df = 22, t = 2.81, P < 0.01), BA25(L) = left subgenual cingulate (mean difference = −0.066%, 95% CI = 0.055 to 0.072%, df = 22, t = 3.3, P < 0.01), caud(R) = right caudate (mean difference = +0.092%, 95% CI = 0.081 to 0.110%, df = 22, t = 3.83, P < 0.001), put(R) = right putamen (mean difference = −0.035%, 95% CI = 0.027 to 0.043%, df = 22, t = 2.3, P < 0.05), caud(L) = left caudate (mean difference = −0.069%, 95% CI = 0.060 to 0.078%, df = 22, t = 4.32, P < 0.001), put(L) = left putamen (mean difference = −0.056%, 95% CI = −0.163 to 0.276%, df = 22, t = 0.145, P > 0.05). NS = not significant.
Figure 6 Sad stimuli: correlation between change in absolute HRSD scores and change in BOLD response (%) at the right subgenual cingulate gyrus (BA25; 7, 15, -7). BOLD = Blood Oxygenation Level Dependent. HDRS = Hamilton Rating Scale for Depression.".
Neural markers of symptomatic improvement during antidepressant therapy
11
was a positive correlation between putamen BOLD change and HRSD score (r = 0.600, P = 0.039) and a negative correlation between left subgenual cingulate BOLD change and HRSD score (r = −0.618, P = 0.032), but these associations did not remain significant after correction for multiple comparisons. In addition, BOLD changes in the left subgenual cingulate were correlated with BOLD changes in the left putamen (r = 0.713, P = 0.009) and right amygdala (r = 0.633, P = 0.019). For happy stimuli, there was a positive correlation between change in right subgenual cingulate response and change in HRSD score (r = 0.606, P = 0.037), although, again, this did not remain significant after correcting for multiple comparisons. In addition, BOLD changes in the right subgenual cingulate were positively correlated with BOLD changes in the right putamen (r = 0.620, P = 0.032), and BOLD changes in the left subgenual cingulate positively correlated with BOLD changes in the right caudate (r = −0.671, P = 0.017). Only the final correlation remained significant after correcting for multiple comparisons.
Medication effects
We were interested in examining the potential effects of clinical recovery on neural responses. Antidepressant dose was kept constant for Time 1 and Time 2 thereby controlling for any independent physiological effects of antidepressant prescribing per se on the BOLD response. Equivalent medication dose varied between subjects as did the type of medication. In theory, relative dose could have had an independent effect on the relative size of the difference in BOLD responses at Time 1 and 2 between subjects. Using an approach devised by Sackeim (Sackheim, 2001), dose equivalents were derived for each participant and coded from 1 (low dose) to 4 (high dose).There was no correlation between dose score and severity of depression at Time 1 or with degree of clinical improvement (calculated as HRSD score at Time 2 – HRSD score at Time 1).
Discussion
Whole brain analysis of differences in response to increasing intensity of sad and happy faces over time showed a dissociation of visual cortex responses in depressed individuals to sad and happy faces pretreatment and posttreatment: before treatment, depressed individuals showed increased responses in visual areas to sad stimuli of increasing intensity but no such relationship to happy stimuli. Following clinical improvement, however, the pattern of responses was reversed: there was a positive relationship between visual cortex responses and happy stimuli of increasing intensity but no such relationship with sad stimuli. In further support of these findings, factorial trend analyses within subjects at Time 2, confirmed greater responses in primary visual cortex to happy faces (of increasing intensity) than to sad faces. Similarly, at Time 1, there was a greater response to sad rather than happy faces in the primary visual cortex.
These findings are consistent with psychological and neurophysiological studies suggesting an attentional bias away from happy stimuli and toward sad stimuli in depressed individuals and with the findings of previous fMRI studies by our group and other researchers. However, this is the first time that responses to both happy and sad stimuli of increasing intensity have been studied before and after clinical improvement. A previous cross-sectional study, using the same paradigm, showed that the visual responses of healthy subjects to sad and happy stimuli are the reverse of the depressed pattern (Surguladze, et al., 2005). Taken together, these findings suggest a trend toward ‘normalisation’ of visual cortical responses to sad facial expressions following clinical improvement. We also noted greater responses to happy faces in the pregenual cingulate (BA10/24) at Time 1. This is consistent with the findings of (Gotlib, et al. (2005) and a previous study, which showed increased responses to happy mood induction in BA10/24/32 in depressed individuals but not in healthy volunteers (Keedwell, et al., 2005b). It was suggested that this medial prefrontal response was because of the discordant nature of the external stimuli compared with the prevailing low mood, and this assessment being important for determining the rewarding potential of stimuli (Elliott, et al., 2004, 2000; Knutson, et al., 2001). Fu, et al. did not find a similar increase in the pregenual area, but as already discussed, participants were taking antidepressants at follow-up only and they may have been less severely depressed. Correlations with depression severity as measured by the HRSD, at Time 1, are consistent with comparisons over time (associated with recovery). These analyses showed that the greater the depression severity the greater the response of the primary visual cortex to sad stimuli. Conversely, decreasing depression severity was associated with increasing visual cortex responses to happy stimuli. They also replicate previous findings by our research group (Surguladze, et al., 2005). In addition, we have showed that decreasing depression scores are associated with increasing responses to happy stimuli in the right caudate, left dorsolateral prefrontal cortex and left posterior cingulate. The dorsolateral prefrontal cortex is thought to have reduced resting state activity in depressed individuals (Mayberg, et al., 1999, 2000), and there is increased activity in this area following successful antidepressant treatment (Mayberg, et al., 1999, 2000).The finding of a negative correlation with responses in the ventral striatum is consistent with our hypothesis and a study linking attenuated responses to happy mood induction with depressive anhedonia (Keedwell, et al., 2005a). It is also consistent with a large body of research linking increased responses in this area with rewarding stimuli (Blood and Zatorre, 2001; Breiter, et al., 1997; Keedwell, et al., 2005a; Surguladze, et al., 2003). In addition, Fu, et al. have recently shown an attenuation of subcortical response to the covert recognition of happy faces (Fu, et al., 2007). Additionally, higher depression scores were associated with greater responses to sad stimuli in the right hippocampus and right subgenual cingulate. The increased response in the subgenual cingulate is consistent with the work of Gotlib, et al.
12
Neural markers of symptomatic improvement during antidepressant therapy
who showed a greater response to covert sad facial expressions in BA25 in depressed patients versus healthy controls (Gotlib, et al., 2005). In addition, previous work has shown increased activity in this area at rest in depressed subjects and in healthy volunteers during sad mood induction (Mayberg, et al., 1999). However, another study found no increased activity in this area following sad mood induction in acutely depressed subjects (Liotti, et al., 2002). This latter finding was thought to be due to a ceiling affect, with the authors suggesting that the activation of BA25 was already too high for mood provocation to have an effect. Our study has used covert sad emotion recognition rather than sad mood provocation, which may account for the discrepancy. The hippocampus is often activated by the retrieval of memories with emotional content (Kensinger and Schacter, 2005) and is dependent on emotional context at the time of retrieval (Fenker, et al., 2005). Sad memories are accessed more easily during depression (Bradley, et al., 1996). Therefore, those individuals who were more depressed were probably more likely to trigger sad memories in response to the mood-congruent facial stimuli, and hence activate the hippocampus, than those with lower symptom scores. We carried out a region of interest analysis for happy and sad stimuli based on previous findings in the literature. Sad stimuli were associated with the greatest changes in effect size over time. We showed significant decreases in effect size at Time 2 in the bilateral amygdala, left caudate, right subgenual cingulate gyrus and right putamen in line with our predictions and previous longitudinal treatment studies using the covert recognition of sad facial expressions with the bilateral amygdalae as regions of interest (Davidson, et al., 2003). The decrease in right subgenual cingulate activity at Time 2 is in keeping with a previous resting-state PET study on the effect of SSRI treatment (Mayberg, et al., 1999, 2000) and with our finding of a positive correlation between depression severity and right subgenual cingulate responses at Time 1. There was an apparent increase in left subgenual cingulate activity at Time 2, but this change in effect size was small and largely represented recovery from a negative value at Time 1; hence, is unlikely to represent a task-related effect. Responses to sad stimuli in the caudate and putamen were broadly attenuated over time, consistent with previous research showing increases in ventral striatal responses to sad stimuli in depressed individuals compared with healthy individuals (Fu, et al., 2004; Surguladze, et al., 2005), and a decrease in ventral striatal response in depressed individuals following treatment (Fu, et al., 2004). Conversely, despite showing a negative correlation between depression severity and responses in the right caudate to happy faces at Time1, we were unable to show the hypothesised increase in ventral striatal activity to happy faces as depression scores diminished. The findings suggest that subcortical responses may be less sensitive markers of clinical improvement for happy stimuli than for sad stimuli. We have shown a strong and highly statistically significant positive correlation between decreases in % BOLD response to sad faces in the subgenual cingulate (BA25) and decreases in
symptom severity during antidepressant treatment. This finding is consistent with a number of previous PET studies examining markers of clinical response to a variety of antidepressant treatments (Zobel, et al., 2005, Mayberg, et al., 2000, Nobler, et al., 2001, Mottaghy, et al., 2002, Dougherty, et al., 2003; Malizia, 1997). It supports a recent approach to the management of treatment resistant depression, which uses deep brain stimulation of the white matter tracts adjacent to BA25; this has shown a 60% response rate (Mayberg, et al., 2005). Our finding differs from that of Fu, et al., who found that pregenual rather than subgenual cingulate activity decreased with symptom improvement, despite the fact that they used very similar stimuli (Fu, et al., 2004). However, there are significant differences in overall study design: in our study, the participants were taking antidepressants during baseline and at follow-up, not just at follow-up. Furthermore, our subjects had different clinical profiles reflecting the naturalistic nature of this study. The findings do not merely represent a regression to the mean in BOLD responses to emotional stimuli per se because (a) there was a differential response to sad and happy facial stimuli within subjects at Time 1, and these distinct responses changed in different ways over time (b) the change in BOLD response to sad facial stimuli was highly correlated with change in symptom score (in other words the effect was not seen in those who did not get better while under treatment). There were responses to sad stimuli that appeared to be greater at Time 2 than at Time 1 despite symptom reduction. Depressed subjects’ responses to both intensities of sad faces showed a greater response in BA47 at Time 2 compared with Time 1. These findings are consistent with previous work, which showed that BA47 was not activated in the severely depressed in response to sad mood induction, whereas mild and remitted depressives did activate this area. It was concluded that partial or complete recovery appears to ‘unlock’ ruminatory processes (Liotti, et al., 2002). Previous studies have shown a positive correlation between degree of negative rumination and activity in this area (Dunn, et al., 2002; Keedwell, et al., 2005a). A ruminative coping strategy in response to negative life events is a risk factor for depression (NolenHoeksema and Larson, 1999; Nolen-Hoeksema, 2000). Hence, a persistently exaggerated response to sad stimuli in BA47 may be a vulnerability marker for depression, increasing the chances of relapse. This may represent a trait or scarring effect of depressive illness. The increase in BA47 responsiveness at Time 2 is unlikely to represent a late effect of antidepressants per se because no such increase was seen for happy stimuli. We have shown that responses in R BA25 and R visual cortex to sad stimuli represent both markers of symptom severity pretreatment and markers of symptomatic improvement. However, the variance in depression score at Time 1 was small compared with the variance at Time 2 (see Table 1). In other words, the correlation between change in BOLD effect size and change in symptom score is more statistically and clinically significant than any correlations between BOLD responses and initial severity.
Neural markers of symptomatic improvement during antidepressant therapy
13
There were some limitations in our study method that reflect its naturalistic design. First, the lack of a placebotreated group limited our analysis to a consideration of correlates of symptomatic improvement. Second, the sample size was modest at n = 12. However, the prospective repeated measures design and the use of regression analyses increase the statistical power. Hence, the positive correlation between reduced BA25 effect sizes and symptom reduction was robust and highly significant. Individuals were taking a variety of antidepressants in common with many previous studies (e.g. (Gotlib, et al., 2005)), but there was no significant relationship between medication dose and severity of depression at Time 1. In any case, it is unlikely that different antidepressants would have had differential effects on responses to happy and sad stimuli – one would expect a constant neurophysiological effect across both types of stimuli. For the purposes of longitudinal comparisons, antidepressant dose was consistent at Time 1 and Time 2. Although it is possible that antidepressants could exert different physiological effects over time that are independent of dose and therapeutic action, this would not account for the dissociated responses to happy and sad stimuli and the correlations between change in BOLD effect size and symptom reduction in regions of interest. In summary, there are contrasting cortical and subcortical responses to sad and happy stimuli in severe depression, which appear to remediate following clinical recovery. This article is unique in showing a double dissociation of primary visual cortex responses over time: severely depressed individuals in the early stages of treatment with antidepressants show increased and decreased visual cortex responses to sad and happy faces respectively, but following clinical improvement this pattern is reversed and comes to resemble the pattern seen in healthy individuals in other studies. The change in visual cortex sensitivity to sad stimuli over time is accompanied by reduced responses in the ventral cingulate, bilateral amygdala/parahippocampus and ventral striatum, in common with previous findings. However, we have found that subgenual cingulate sensitivity, rather than pregenual cingulate sensitivity, appears to be the important marker of therapeutic response in severely depressed patients found in secondary care. Changes in effect size of individual BOLD responses in right BA25 and right visual cortex to sad facial stimuli show correlations with reductions in symptom scores over time but with BA25 responses having by far the strongest correlation. In conclusion, the right subgenual cingulate response to sad stimuli could represent a useful marker of therapeutic outcome in severe depression.
References
Blood, AJ, Zatorre, RJ (2001) Intensely pleasurable responses to music correlate with activity in brain regions implicated in reward and emotion. Proc Natl Acad Sci U S A 98: 11818–11823. Bradley, BP, Mogg, K, Millar, N (1996) Implicit memory bias in clinical and non-clinical depression. Behav Res Ther 34: 865–879. Brammer, M, Bullmore, ET, Simmons, A, Williams, SCR, Grasby, PM, Howard, RJ, et al. (1997) Generic brain activation mapping in functional magnetic resonance imaging: a nonparametric approach. Magn Reson Imaging 15: 763–770. Breiter, HC, Gollub, RL, Weisskoff, RM (1997) Acute effects of cocaine on human brain activity and emotion. Neuron 19: 591–611. Bullmore, ET, Brammer, M, Rabe-Hesketh, S, Curtis, V, Morris, R, Williams, SCR, et al. (1999a) Methods for the diagnosis and treatment of stimulus correlated motion in generic brain activation studies using fMRI. Hum Brain Mapp 7: 38–48. 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, M (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: 32–34. Davidson, RJ, Irwin, W, Anderlie, MJ, Kalin, NH (2003) The neural substrates of affective processing in depressed patients treated with venlafaxine. Am J Psychiatry 160: 64–75. Dougherty, DD, Weiss, AP, Cosgrove, R, Alpert, NM, Cassem, EH, Nierenberg, AA, et al. (2003) Cerebral metabolic correlates as potential predictors of response to cingulotomy for major depression. J Neurosurg 99: 1010–1017. Dunn, RT, Kimbrell, TA, Ketter, TA, Frye, MA, Willis, MW, Luckenbaugh, DA, et al. (2002) Principal components of the Beck Depression Inventory and regional cerebral metabolism in unipolar and bipolar depression. Biol Psychiatry 51: 387–399. Ekman, P, Friesen, WV (1976) Pictures of Facial Affect. Palo Alto: Consulting Psychologists. Elliott, R, Friston, KJ, Dolan, RJ (2000) Dissociable neural responses in human reward systems. J Neurosci 20: 6159–6165. Elliott, R, Newman, JL, Longe, OA, William Deakin, JF (2004) Instrumental responding for rewards is associated with enhanced neuronal response in subcortical reward systems. Neuroimage 21: 984–990. Fenker, DB, Schott, BH, Richardson-Klavehn, A, Heinze, HJ, Duzel, E (2005) Recapitulating emotional context: activity of amygdala, hippocampus and fusiform cortex during recollection and familiarity. Eur J Neurosci 21: 1993–1999. Fu, CH, Williams, SC, Brammer, MJ, Suckling, J, Kim, J, Cleare, AJ, et al. (2007) Neural responses to happy facial expressions in major depression following antidepressant treatment. Am J Psychiatry 164: 540–542. Fu, CHJ, Williams, CR, Cleare, AJ, Brammer, M, Walsh, N, Kim, J, et al. (2004) Attenuation of the neural response ot sad faces in major depression by antidepressant treatment. Arch Gen Psychiatry 61: 877–889. Goldapple, K, Segal, Z, Garson, C, Lau, M, Bieling, P, Kennedy, S, et al. (2004) Modulation of cortical-limbic pathways in major depression: treatment-specific effects of cognitive behavior therapy. Arch Gen Psychiatry 61: 34–41. Gotlib, IH, Sivers, H, Gabrieli, JD, Whitfield-Gabrieli, S, Goldin, P, Minor, KL, et al. (2005) Subgenual anterior cingulate activation to
Acknowledgments
We are grateful for the technical support provided by Chris Andrew and David Gasston of the Centre for Neuroimaging Sciences. We also acknowledge the generous contribution from the Goldsmith Charitable Foundation.
14
Neural markers of symptomatic improvement during antidepressant therapy
valenced emotional stimuli in major depression. Neuroreport 16: 1731–1734. Keedwell, PA, Andrew, C, Williams, SCR, Brammer, MJ, Phillips, ML (2005a) The neural correlates of anhedonia in major depressive disorder. Biol Psychiatry 58: 843–853. Keedwell, PA, Andrew, C, Williams, SCR, Brammer, MJ, Phillips, ML (2005b) A double dissociation of ventromedial prefrontal cortical responses to sad and happy stimuli in depressed and healthy individuals. Biol Psychiatry 58: 495–503. Kensinger, EA, Schacter, DL (2005) Retrieving accurate and distorted memories: neuroimaging evidence for effects of emotion. Neuroimage 27: 167–177. Khan, A, Leventhal, RM, Khan, SR, Brown, WA (2002) Severity of depression and response to antidepressants and placebo: an analysis of the Food and Drug Administration database. J Clin Psychopharmacol 22: 40–45. Knutson, B, Fong, GW, Adams, CM, Varner, JL, Hommer, D (2001) Dissociation of reward anticipation and outcome with event-related fMRI. Neuroreport 12: 3683–3687. Liotti, M, Mayberg, HS, McGinnis, S, Brannan, SL, Jerabek, P (2002) Unmasking disease-specific cerebral blood flow abnormalities: mood challenge in patients with remitted unipolar depression. Am J Psychiatry 159: 1830–1840. Malizia, A (1997) Frontal lobes and neurosurgery for psychiatric disorders. J Psychopharmacol 11: 179–187. Mayberg, HS, Brannan, SK, Tekell, JL, Silva, JA, 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, Jerabek, 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. Mayberg, HS, Lozano, AM, Voon, V, McNeely, HE, Seminowicz, D, Hamani, C, et al. (2005) Deep brain stimulation for treatmentresistant depression. Neuron 45: 651–660. Mottaghy, FM, Keller, CE, Gangitano, M, Ly, J, Thall, M, Parker, JA, et al. (2002) Correlation of cerebral blood flow and treatment
effects ofrepetitive transcranial magnetic stimulation in depressed patients. Psychiatry Res 115: 1–14. Murphy, FC, Sahakian, BJ, Rubinsztein, JS, Michael, A, Rogers, RD, Robbins, TW, et al. (1999) Emotional bias and inhibitory control processes in mania and depression. Psychol Med 29: 1307–1321. Nobler, MS, Oquendo, MA, Kegeles, LS, Malone, KM, Campbell, CC, Sackeim, HA, et al. (2001) Decreased regional brain metabolism after ECT. Am J Psychiatry 158: 305–308. Nolen-Hoeksema, S (2000) The role of rumination in depressive disorders and mixed anxiety/depressive symptoms. J Abnorm Psychol 109: 504–511. Nolen-Hoeksema, S, Larson, J (1999) Coping With Loss. Mahwah (NJ): Erlbaum. Sackheim, HA (2001) The definition and meaning of treatment resistant depression. J Clin Psychiatry 62 (Suppl 16): 1–17. Sheline, YI, Barch, D, Donnelly, JM, Ollinger, JM, Snyder, AZ, Mintun, MA (2001) Increased amygdala response to masked emotional faces in depressed subjects resolves with antidepressant treatment: an fMRI study. Biol Psychiatry 50: 651–658. Sloan, DM, Bradley, MM, Dimoulas, E, Lang, PJ (2002) Looking at facial expressions: dysphoria and facial EMG. Biol Psychol 60: 79–90. Surguladze, S, Brammer, MJ, Keedwell, P, Giampietro, V, Young, AW, Travis, MJ, et al. (2005) A differential pattern of neural response toward sad versus happy facial expressions in major depressive disorder. Biol Psychiatry 57: 201–209. Surguladze, S, Keedwell, PA, Phillips, M (2003) Neural systems underlying affective disorders. Adv Psychiatr Treat 9: 446–455. Talairach, J, Tournoux, P (1998) Co-planar Stereotactic Atlas of the Human Brain. Stuttgart, Germany: Georg Thieme Verlag. Williams, JM, Mathews, A, MacLeod, C (1996) The emotional Stroop task and psychopathology. Psychol Bull 120: 3–24. Zobel, A, Joe, A, Freymann, N, Clusmann, H, Schramm, J, Reinhardt, M, et al. (2005) Changes in regional cerebral blood flow by therapeutic vagus nerve stimulation in depression: an exploratory approach. Psychiatry Res 139: 165–179.
Readers
Add Comment