Selective serotonin reuptake inhibitors (SSRIs) are a class of antidepressants and have been used to treat mood disorders, such as depression and anxiety disorder. Selective serotonin reuptake inhibitors such as citalopram selectively block serotonin transporter (5-HTT) activity, Reference Meyer, Wilson, Sagrati, Hussey, Carella and Potter1 inhibit the reuptake of serotonin (5-HT) into the presynaptic cell and lead to increasing extracellular levels of 5HT. Reference Bosker, Cremers, Jongsma, Westerink, Wikström and den Boer2 Although clinical trials have shown that SSRIs typically have a delay of several weeks in the onset of their clinical effects, SSRI effects can occur soon after their administration. Reference Taylor, Freemantle, Geddes and Bhagwagar3 The early changes in emotional processing that occur after the acute administration of an SSRI contribute to later mood improvements. Reference Harmer, Goodwin and Cowen4,Reference Pringle, Browning, Cowen and Harmer5 Therefore, understanding the neural mechanisms underlying the acute effects of SSRIs is important for decisions about long-term treatment. However, recent neuroimaging studies have reported inconsistent acute effects of SSRIs on neural responses in emotion-related brain regions such as the amygdala, whose hyperactivity has been suggested as an endophenotype of anxiety disorder and major depression. Reference Drevets6,Reference Ketter, Kimbrell, George, Dunn, Speer and Benson7
Twelve functional magnetic resonance imaging (fMRI) studies have examined the acute effects of SSRIs (single SSRI administration) on neural responses to negative emotions, while eight fMRI studies have examined acute effects of SSRI administration on neural response to positive emotions. Among these studies, four reported increased amygdala response Reference McKie, Del-Ben, Elliott, Williams, del Vai and Anderson8–Reference Grady, Siebner, Hornboll, Macoveanu, Paulson and Knudsen11 whereas eight reported decreased amygdala response to negative emotions Reference Grady, Siebner, Hornboll, Macoveanu, Paulson and Knudsen11–Reference Outhred, Das, Felmingham, Bryant, Nathan and Malhi18 after SSRI v. placebo administration. Two studies reported increased amygdala response to positive emotions, Reference Macoveanu, Rowe, Hornboll, Elliott, Paulson and Knudsen10,Reference Brühl, Jäncke and Herwig19 whereas six studies did not show SSRI effects on amygdala response to positive emotions. Reference Brühl, Kaffenberger and Herwig9,Reference Macoveanu, Rowe, Hornboll, Elliott, Paulson and Knudsen10,Reference Anderson, Juhasz, Thomas, Downey, McKie and Deakin14–Reference Murphy, Norbury, O'Sullivan, Cowen and Harmer16,Reference Brühl, Jäncke and Herwig19 Moreover, our recent meta-analysis Reference Ma20 revealed that the acute effects of antidepressants on positive emotions did not show any convergent activation in healthy adults. Acute antidepressant administration showed discrepant effects on neural responses to negative emotions. Specifically, single antidepressant administration led to convergent increases as well as decreases in a similar neural network underlying negative emotions. Although age, gender or disease conditions may contribute to individual differences in psychometric outcome and discrepant SSRI effects on amygdala response, the genetic structure of individuals also affects drug responses. Reference Weinshilboum21 A sodium-dependent serotonin-transporter-linked polymorphic region (5-HTTLPR), which influences the expression and function of 5-HTT, Reference Heils, Teufel, Petri, Stöber, Riederer and Bengel22,Reference Canli and Lesch23 may be implicated in SSRI efficacy. For example, major depression patients with the long (l) allele of 5-HTTLPR showed better responses to SSRI treatment compared to homozygotes for the short variant (s/s) of 5-HTTLPR. Reference Smeraldi, Zanardi, Benedetti, Di Bella, Perez and Catalano24,Reference Serretti, Kato, De Ronchi and Kinoshita25 However, it remains unknown whether and how 5-HTTLPR polymorphism affects the acute effects of SSRIs on brain activity.
Method
Experimental design
We combined genetic neuroimaging, pharmacological technique and a psychological paradigm of emotion recognition to provide a better understanding of the neurogenetic mechanism of the acute effects of SSRIs on emotions. Specifically, we examined whether 5-HTTLPR modulates the acute effects of SSRIs on neural response to emotional facial expressions. Using a double-blind, within-subject, counterbalanced design (Fig. 1), we scanned s/s and l/l homozygotes during perception of fearful, happy and neutral facial expressions using fMRI after administration of a single oral dose of a SSRI (30 mg citalopram) and a matched placebo tablet with at least a 7-day washout period (at least 7 days apart, ranging from 7 to 29 days, mean = 14.3 days, s.d. = 6.7). Given that l/l compared with s/s homozygotes have been shown to exhibit increased 5-HTT expression and 5-HT uptake, Reference Lesch, Bengel, Heils, Sabol, Greenberg and Petri26 and a greater increase in 5HT neurotransmission after acute blockade of 5-HT reuptake, Reference Whale, Quested, Laver, Harrison and Cowen27 blocking 5-HTT activity with citalopram may produce stronger effects on brain activity in l/l than s/s homozygotes. Thus, we predicted greater SSRI effects on the neural responses to emotional facial expressions in l/l than s/s homozygotes.
Fig. 1 Experimental procedure. Each individual participated in two fMRI sessions separated by a 7-day washout. In each session, functional scanning was conducted 2 h after citalopram or placebo administration. A block design was used during scanning. There were two functional scans lasting 334 s in each (citalopram or placebo) session. Each functional scan consisted of eight blocks (two blocks of each stimulus category) of 14 stimuli (12 faces and two repeats) from the same stimulus category. Each face stimulus was presented for 1500 ms; participants were then asked to keep their gaze on a fixation cross for 500 ms. Two successive blocks were separated by a 10 s fixation cross. fMRI: functional magnetic resonance imaging.
Participants
Fifty Chinese men participated in this study as paid volunteers. We recruited only men to avoid any potential gender difference in citalopram effects. We used polymerase chain reaction to determine the genotypes of 5-HTTLPR. (See online supplementary material for DNA isolation and analysis.) Two l/l and two s/s homozygotes were excluded from the data analysis due to excessive head movement during scanning (exclusion criterion: 6 mm head motion or 3° for head rotation). Thus 23 s/s (18–23 years, mean = 19.4, s.d. = 1.8) and 23 l/l homozygotes (18–23 years, mean = 18.6, s.d. = 1.4), who did not differ significantly in age (t (44) = 1.296, P = 0.202), were included in the data analysis. All participants were right-handed and had normal or corrected-to-normal vision. All participants had no history of cardiac, hepatic, renal, pulmonary, neurological, psychiatric or gastrointestinal disorders, medication/drug use, and no family history of major depression or bipolar affective disorder. The study protocol was approved by the local ethics committee of the Department of Psychology at Peking University. Written informed consent was obtained prior to the study from all participants.
General procedure
We used a 2 × 2 × 2 factorial design with emotion (fearful v. neutral faces, or happy v. neutral faces) and treatment (citalopram v. placebo) as the between-subjects variables and genotype (s/s v. l/l genotype) as the between-subjects variable. The order of citalopram and placebo treatment was counterbalanced within each genotype group and across the s/s and l/l genotype groups. We used a single oral administration of 30 mg citalopram, which is within the dose range (20–60 mg) and has been used in previous studies. Reference Crockett, Clark, Hauser and Robbins28,Reference Nandam, Hester, Wagner, Cummins, Garner and Dean29 We administered a single dose to examine the acute effects of SSRIs. Pharmacokinetic and plasma hormone concentration studies have shown that, in men, citalopram plasma concentration reaches a peak around 2 h. Reference Penttilä, Syvälahti, Hinkka, Kuusela and Scheinin30,Reference Jiang, Rong, Xu, Chen, Xie and Chen31 Thus, functional scanning was conducted about 2 h after citalopram or placebo administration when citalopram reached its peak effect. During the waiting period, participants sat on a comfortable couch, resting or reading. A researcher explained the task and procedure to participants and participants practised a few trials to become familiar with the task 30 min before scanning. Before taking citalopram or the placebo, we evaluated each participant’s harm-avoidance tendency using the harm avoidance subscale from the Tridimensional Personality Questionnaire. Reference Cloninger, Svrakic and Przybeck32 At the beginning and end of the citalopram or placebo session, we asked participants to complete the Positive and Negative Affect Scale Reference Watson, Clark and Tellegen33 to monitor their mood changes.
Functional imaging task
Four categories of facial expressions (fearful, happy, neutral and scrambled; 24 images in each category, half male models and half female models, from the Chinese Facial Affective Picture System) Reference Gong, Huang, Wang and Luo34 were used during scanning. Scrambled faces were made by segmenting each face image (300 × 336 pixels) into 50 × 56 square grids (6 × 6 pixels) that were then randomly arranged. This rearranged picture was then masked by a stencil of the original face shape, resulting in a pixellated face on a grey background. Stimuli used during the scanning were delivered through a LCD projector onto a rear projection screen (SINORAD SA-9900). Each stimulus subtended a visual angle of 3.85° × 4.31° (width × height) at a viewing distance of 83 cm. Participants completed a typical repetition-detection task by responding to repeated facial expressions via a button press (Fig. 1). After scanning, participants were asked to rate the emotion intensity of 12 fearful and 12 happy faces randomly selected from the two categories on an 11-point Likert scale (0 = not at all fearful/happy; 10 = extremely fearful/happy).
Imaging parameters
We acquired images of the brain using a Siemens 3-Tesla Trio MRI scanner at the Beijing MRI Center for Brain Research. We obtained blood oxygenation level-dependent gradient-echo echo-planar images using a 12-channel head coil (64 × 64 × 32 matrix with 3.44 × 3.44 × 5.0 mm spatial resolution, repetition time (TR) = 2000 ms, echo time (TE) = 30 ms, flip angle = 90°, field of view = 24 × 24 cm) while the participants performed the repetition-detection task. We subsequently acquired a high-resolution, T1-weighted structural image (256 × 256 × 144 matrix with a spatial resolution of 1 × 1 × 1.33 mm, TR = 2530 ms, TE = 3.37 ms, inversion time (TI) = 1100 ms, flip angle = 7°).
Imaging analysis
Preprocessing
We used the Statistical Parametric Mapping software, version 8 (Wellcome Trust Centre for Neuroimaging; http://www.fil.ion.ucl.ac.uk/spm) to analyse the functional image data; we analysed data using the general linear model for a block design. We realigned the functional images within and across runs to correct for head movement. We included six movement parameters (translation: x, y, z; rotation: pitch, roll, yaw) in the statistical model. We normalised the functional images to a standard Montreal Neurological Institute EPI template and we then spatially smoothed them using a 4-mm, full-width, half-maximum isotropic Gaussian kernel.
Regions of interest analysis
We obtained parameter estimates of signal intensity from structurally defined amygdala to assess genotype treatment interaction on emotional processes. We identified the amygdala using the Automated Anatomical Labeling (AAL) atlas (AAL templates) and we selected it using the Wake Forest University School of Medicine PickAtlas software toolbox. We subjected the parameter estimates of signal intensity in the left and right amygdala to repeated-measures analyses of variance (ANOVA) with genotype (s/s v. l/l) as a between-subjects variable, and treatment (citalopram v. placebo) and emotion (fearful v. neutral facial expressions or happy v. neutral facial expressions) as between-subjects independent variables. To further clarify whether differential SSRI effects on amygdala activity were driven by neural response to emotional or neutral facial expressions, we calculated the contrast value related to fearful, happy and neutral facial expressions v. fixation cross, as used in previous studies. Reference Heinz, Smolka, Braus, Wrase, Beck and Flor35 We then subjected these contrast values to genotype treatment ANOVAs for fearful, happy and neutral facial expressions, respectively. We included the treatment (citalopram v. placebo) sequence as a covariate in region of interest (ROI) and whole-brain analyses to control for any potential effect of treatment order. We also conducted analyses that included treatment order as a between-subject variable in the ANOVA and this did not show significant treatment order effects.
Whole-brain analysis
We used a two-level, hierarchical, random-effects model for whole-brain analysis. At the individual level, we modelled the onset and duration of each block using a general linear model for five conditions: fearful, happy, neutral, scrambled facial expression and fixation cross. We used a box-car function to convolve with a canonical haemodynamic response in each condition. The contrasts of ‘fearful v. neutral’ and ‘happy v. neutral’ identified brain regions responsive to positive and negative emotional facial expressions, respectively. We then performed a whole-brain factorial design to reveal the genotype treatment interaction on the neural response to fearful or happy facial expressions. Significant activations were identified using a threshold of P<0.05 (topological false discovery rate (FDR) corrected for multiple comparisons correction).
Results
Behavioural results
Response accuracy was high in the repetition-detection tasks during scanning (mean accuracy>95%). We subjected response accuracy and reaction times to 2 (emotion: fearful v. neutral facial expressions, or happy v. neutral facial expressions) × 2 (treatment: citalopram v. placebo) × 2 (genotype: s/s v. l/l) ANOVAs, but did not show any significant effect (P>0.05).
We subjected self-reported mood changes from pre- to post-scan to a 2 (treatment: citalopram v. placebo) × 2 (genotype: s/s v. l/l) ANOVA. This did not show any significant effect (F<1, Table DS1), suggesting that citalopram treatment did not have an effect on the participants’ mood. Harm avoidance scores did not differ between s/s and l/l homozygotes (18.2 v. 16.0, t(44) = 1.07, P = 0.291). Participants rated fearful faces as less fearful after citalopram than placebo administration (F(1,43) = 5.44, P = 0.024), but this effect did not differ significantly between s/s and l/l homozygotes (F(1,43) = 2.44, P = 0.126, Fig. DS1). Relative to l/l homozygotes, s/s homozygotes reported greater emotional intensity of happy facial expressions (F(1,43) = 7.30, P = 0.010), but neither the main effect of treatment nor its interaction with genotype was significant (P>0.35).
Functional magnetic resonance imaging
Region of interest analyses of amygdala response to fearful v. neutral facial expressions in the placebo session revealed significantly stronger left amygdala response in s/s than l/l genotype groups (F(1, 43) = 4.15, P = 0.048, Fig. 2a), replicating previous findings. Reference Hariri, Mattay, Tessitore, Kolachana, Fera and Goldman36,Reference Hariri, Drabant, Munoz, Kolachana, Mattay and Egan37 We assessed the differential citalopram effects on amygdala response to fearful v. neutral facial expressions in the two genotypes by conducting a 2 (emotion: fearful v. neutral) × 2 (treatment: citalopram v. placebo) × 2 (genotype: s/s v. l/l) ANOVA of parameter estimates of signal intensity. This revealed a significant main effect of emotion in both left and right amygdala (F(1,43) = 5.40 and 4.12, P= 0.025 and 0.048), indicating greater amygdala responses to fearful than neutral facial expressions. Similar to previous findings, Reference Grady, Siebner, Hornboll, Macoveanu, Paulson and Knudsen11,Reference Bigos, Pollock, Aizenstein, Fisher, Bies and Hariri12,Reference Ma20 we found a significant main effect of treatment on left amygdala response (F(1,43) = 5.14, P = 0.028), indicating that, relative to placebo, acute administration of citalopram in healthy individuals increased activation in the left amygdala to fearful v. neutral facial expressions.
Fig. 2 Brain imaging results. (a) Contrast values of fearful v. neutral faces of the left and right amygdala. (b) Contrast values of the left and right amygdala response to fearful, neutral and happy faces compared to fixation. (c) Whole-brain treatment × genotype interaction on the contrast of fearful v. neutral faces. This revealed significant activations in the left and right amygdala and insula (*P<0.05, **P<0.01).
Notably, there was a significant three-way interaction of emotion × treatment × genotype on response in both left and right amygdala (F(1, 43) = 10.47 and 8.42, P = 0.002 and 0.006, Fig. 2a), suggesting that acute administration of citalopram increased response in the left and right amygdala to fearful v. neutral faces in the l/l genotype (left: (F(1,21) = 10.87, P = 0.003; right: F(1,21) = 4.57, P = 0.045) but did not significantly affect amygdala response in the s/s genotype (left: F(1,21) = 1.53, P = 0.229; right: F(1,21) = 1.42, P = 0.248). To further assess whether the three-way interaction was driven by citalopram effect on amygdala response to fearful or neutral facial expressions, we examined the treatment × genotype interaction on amygdala response in response to fearful and neutral facial expressions (v. fixation), respectively. We found significant treatment × genotype interactions on amygdala response to fearful facial expressions (left: F(1,43) = 5.48, P = 0.024; right: F(1,43) = 4.14, P = 0.048, Fig. 2b), but not to neutral facial expressions (P>0.2). Thus, the differential citalopram effects on amygdala response in the l/l and s/s genotypes arose from amygdala response to fearful rather than neutral facial expressions. Similar analyses of amygdala response to happy facial expressions failed to show any significant effect (P>0.1, Fig. 2b).
To further confirm the distinct citalopram effects on amygdala response between the two genotypes of 5-HTTLPR, we conducted a whole-brain analysis to test the interaction between genotype (s/s v. l/l) and treatment (citalopram v. placebo) on the contrast of fearful v. neutral facial expressions. This revealed significant interactions of genotype × treatment in left and right amygdala response (left: –18/–2/–20, T = 2.98; right: 30/–4/–28, T = 3.13, P<0.05, topological FDR corrected), further confirming the differential citalopram effects on amygdala response. The whole-brain analysis also revealed significant interactions of genotype × treatment in left and right insula response (left: –34/–12/16, T = 3.91; right: 40/–4/12, T = 3.35, P<0.05, topological FDR corrected; Fig. 2c), suggesting variations of citalopram effects on insula response as a function of 5-HTTLPR polymorphism (Fig. 3). A similar whole-brain analysis of the contrast of happy v. neutral facial expressions did not show any significant brain activation.
Fig. 3 Contrast values calculated from the left and right insula. (a) Contrast values of fearful v. neutral faces in the left and right insula revealed by the genotype treatment interaction in the whole-brain analysis; (b) contrast values of faces v. fixation in the left and right insula. The effect of citalopram was significant only for fearful faces (*P<0.05, **P<0.01; ***P<0.001). SSRI: selective serotonin reuptake inhibitor.
Discussion
Current research has focused on elucidating the neural substrates underlying the heterogeneity of the acute effects of SSRIs in healthy volunteers. We have provided pharmacogenetic and neuroimaging evidence that 5-HTTLPR polymorphism modulates the acute effects of citalopram on neural activity related to emotion processing. Specifically, we found that the acute administration of citalopram (v. placebo) increased amygdala and insula activity in the l/l but not the s/s genotype of 5-HTTLPR. The genotype differences in the acute effects of an SSRI were evident in the neural responses to fearful but not happy facial expressions, indicating that such genotype differences do not reflect the non-specific effects of citalopram, such as changes in arousal or drowsiness. Our results indicate that an individual’s genetic structure may influence the acute effects of citalopram on neural response to negative emotions and that 5-HTT is a key molecular transporter that contributes to the differential neural responses to negative emotions in healthy adults. Our findings provide a possible neurogenetic mechanism for understanding the previous inconsistent results regarding how the acute administration of SSRIs modulates neural responses to negative emotions.
Self-reported mood was not influenced by 5-HTTLPR genotype or citalopram administration. Therefore, 5-HTTLPR modulation of the effect of citalopram on amygdala response cannot be simply attributed to general mood changes. This is consistent with a cognitive neuropsychological model of SSRI action that emotional processes are sensitive to early SSRI-induced changes in the absence of mood variation. Reference Harmer, Goodwin and Cowen4,Reference Pringle, Browning, Cowen and Harmer5 The post-scan rating scores suggested decreased emotion-intensity rating of fearful facial expressions after citalopram v. placebo treatment. However, this effect was the same in the s/s and l/l genotypes. Similarly, 7 days of SSRI treatment decreased amygdala response to fearful facial expressions but failed to induce significant changes in clinical depression ratings. Reference Godlewska, Norbury, Selvaraj, Cowen and Harmer38 Thus amygdala response may be more sensitive to how 5-HTTLPR modulates the acute administration of citalopram and thus provides a neurobiological index of genetic influences on the effect of SSRIs.
The 5-HTTLPR also modulated the acute effects of citalopram on insula response to fearful facial expressions. The insula plays a key role in how emotions are processed, such as detection of emotional salience Reference Craig39 and anticipation of emotional stimuli. Reference Simmons, Matthews, Stein and Paulus40 Chronic SSRI treatment attenuates insula activation during perception of emotional facial expressions Reference Arce, Simmons, Lovero, Stein and Paulus41 and during anticipation of emotional stimuli Reference Simmons, Arce, Lovero, Stein and Paulus42 in healthy volunteers. The acute administration of an SSRI, however, increased insula response. Reference Anderson, Del-Ben, Mckie, Richardson, Williams and Elliott13 Our findings suggest that 5-HTT plays a pivotal role in producing the differential insula response in the s/s and l/l genotypes. Taken together, our results suggest that the 5-HTTLPR genotype may influence the effect of citalopram on neural response to negative emotions in different nodes (e.g. amygdala and insula) of the emotional neural network in a similar vein.
Implications
Our pharmacogenetic and neuroimaging findings have important implications. First, whereas amygdala response is modulated by the intensity of both fearful and happy facial expressions, Reference Morris, Frith, Perrett, Rowland, Young and Calder43 the acute administration of an antidepressant in healthy volunteers did not show convergent response to positive emotions. Reference Ma20 Some studies have reported that the acute administration of an SSRI facilitates recognition of happy facial expressions and increased amygdala response to happy facial expressions. Reference Macoveanu, Rowe, Hornboll, Elliott, Paulson and Knudsen10,Reference Outhred, Das, Felmingham, Bryant, Nathan and Malhi18 Other studies have failed to show significant SSRI effects on neural or behavioural responses to happy facial expressions. Reference Brühl, Kaffenberger and Herwig9,Reference Del-Ben, Deakin, McKie, Delvai, Williams and Elliott15,Reference Murphy, Norbury, O'Sullivan, Cowen and Harmer16,Reference Brühl, Jäncke and Herwig19 Similarly, the acute administration of citalopram did not change subjective feelings or amygdala response related to happy facial expressions in both genotypes in the current study. Thus, 5-HTT may play a less important role in affective and neural responses to positive (relative to negative) emotion. Anhedonia (i.e. the inability to obtain pleasure from natural rewards and decreased interest in most activities) and negative emotional bias are two core features of depression. Depressed patients tend to interpret happy facial expressions as neutral, fail to experience pleasure from activities they previously experienced as rewarding and lack reward-motivated behaviours. By contrast, they pay more attention to negative emotions, remember negative affective materials better and are more likely to classify ambiguous/neutral facial expressions as negative. Reference Ma20 Thus our findings suggest that citalopram may have distinct therapeutic implications for the treatment of different symptoms in depressed patients. Alternatively, the lack of 5-HTTLPR modulation of the acute effects of an SSRI on amygdala response to happy facial expressions may be due to the absence of amygdala activation to happy facial expressions in our experiment. This should be clarified in future research.
Second, the early changes in emotional processing after a single SSRI dose contribute to later mood improvements. Reference Taylor, Freemantle, Geddes and Bhagwagar3–Reference Pringle, Browning, Cowen and Harmer5 Thus, understanding the neural mechanisms underlying the acute effects of SSRIs has important implications for long-term treatment. Our findings of distinct acute effects of an SSRI on emotion-related neural activity in s/s and l/l homozygotes suggest that one’s genetic structure should be considered when treating mood disorders with SSRIs. However, it remains unknown whether and how 5-HTTLPR modulates the chronic SSRI effects that lead to decreased response to negative emotions in both healthy controls and depressed patients. Reference Ma20 Pharmacogenetic studies of antidepressant response reported better responses of the ‘l’ variant (compared with the ‘s’ variant) of 5-HTTLPR to SSRIs in mood disorder patients. Reference Serretti, Kato, De Ronchi and Kinoshita25,Reference Smits, Smits, Schouten, Stelma, Nelemans and Prins44,Reference Malhotra, Murphy and Kennedy45 Our fMRI results of differential acute effects of citalopram on amygdala and insula response in l/l and s/s homozygotes are consistent with the distinct SSRI effects on clinical assessment of improvement in patients with major depression. Reference Smeraldi, Zanardi, Benedetti, Di Bella, Perez and Catalano24 These results together suggest more pronounced SSRI effects in l/l than in s/s homozygotes and are consistent with the findings of increased 5-HTT expression Reference Lesch, Bengel, Heils, Sabol, Greenberg and Petri26 and increased 5-HT neurotransmission after acute blockade of 5-HT reuptake Reference Whale, Quested, Laver, Harrison and Cowen27 in l/l homozygotes than ‘s’ carriers. Future research on genetic differences in chronic SSRI treatment may open an avenue for prescribing personalised antidepressants.
Third, our finding that citalopram increased amygdala response to fearful facial expressions in l/l homozygotes suggests a more anxiogenic action of acute SSRI treatment in this group. This seems counterintuitive given that mood disorder patients with the ‘l’ variant (compared with the ‘s’ variant) of 5-HTTLPR showed a better SSRI response. Reference Serretti, Kato, De Ronchi and Kinoshita25,Reference Smits, Smits, Schouten, Stelma, Nelemans and Prins44,Reference Malhotra, Murphy and Kennedy45 However, opposite early effects of SSRIs in depression have been reported in clinical treatment. The acute administration of SSRIs may increase anxiety in some mood disorder patients. Reference Mir and Taylor46 Our findings thus provide a potential neurogenetic explanation for the opposite early effect of SSRIs in clinical trials. Future research should address whether and how early SSRI-induced changes in emotional processes can predict long-term treatment efficacy and whether one’s genetic structure modulates the early and chronic effects of SSRIs in a similar way. Future research should also address whether such increased amygdala response to fearful facial expressions is dose-dependent; i.e. whether a lower or higher dose of citalopram could modulate this effect.
Finally, although the ‘s’ allele frequency is different between Asian and White populations, Reference Kunugi, Hattori, Kato, Tatsumi, Sakai and Sasaki47 to date, we know little about whether the ‘s’ allele frequency in a population is associated with a specific pattern of 5-HTTLPR modulation of brain activity. Previous research showed an association between the ‘s’ allele and hyperactivity in the amygdala mainly in White populations. Reference Hariri, Mattay, Tessitore, Kolachana, Fera and Goldman36,Reference Hariri, Drabant, Munoz, Kolachana, Mattay and Egan37,Reference Murphy, Norbury, Godlewska, Cowen, Mannie and Harmer48 Our study involving Chinese healthy volunteers in the placebo group also showed stronger amygdala response to fearful (v. neutral) facial expressions in the s/s than l/l genotype group. This is similar to the observation of 5-HTTLPR effects on amygdala response in White populations. Reference Hariri, Mattay, Tessitore, Kolachana, Fera and Goldman36,Reference Hariri, Drabant, Munoz, Kolachana, Mattay and Egan37,Reference Murphy, Norbury, Godlewska, Cowen, Mannie and Harmer48 In addition, we found that l/l compared to s/s homozygotes tended to show greater amygdala response in the contrast of neutral facial expressions v. fixation in the placebo condition, replicating the previous findings in White populations. Reference Heinz, Smolka, Braus, Wrase, Beck and Flor35 These results suggest similar 5-HTTLPR effects on amygdala activity regardless of the difference in the ‘s’ allele frequency in populations. Pharmacogenetic studies have reported inconsistent 5-HTTLPR effects on antidepressant drug response in White and Asian groups experiencing major depression. Reference Malhotra, Murphy and Kennedy45 However, the improved response to SSRIs in the ‘l’ variant of 5-HTTLPR seemed to be independent of ethnic differences, Reference Serretti, Kato, De Ronchi and Kinoshita25 which may be confirmed by future research.
Limitations
A potential limitation of the present study was the sample size. To examine the power of our sample size, we calculated the effect size and conducted a power analysis. These analyses suggested that our study yielded a large effect size and appropriate sample size (see online supplementary material for details). We also reviewed published studies that examined 5-HTTLPR genotype differences on brain activity and studies that examined antidepressant effects on emotional neural responses. According to a recent meta-analysis, 26 fMRI studies examined 5-HTTLPR genotype differences in emotional reactivity. Reference Murphy, Norbury, Godlewska, Cowen, Mannie and Harmer48 The mean sample size of these 26 studies was 38. Mode of sample size was 28. Our recent meta-analysis summarised antidepressant effects on emotional responses. Reference Ma20 Twenty-one fMRI studies used between-subjects treatment (the same design as the current study) to examine antidepressant effects on brain responses to emotions in healthy participants. The average sample size of these 21 studies was 15. Considering the effect size of the current work and comparing the sample size of published studies, the sample size of our study allowed us to reach a conclusion (albeit with caution) on 5-HTTLPR genotype differences in SSRI effects on emotional neural activity.
Another possible limitation was that only men were recruited for the current study. This did not allow us to test whether 5-HTTLPR modulates the SSRI effect on amygdala responses in men and women in a similar fashion. Previous research has revealed similar 5-HTTLPR effects on the reactivity of the amygdala to affective stimuli in men and women. Reference Hariri, Drabant, Munoz, Kolachana, Mattay and Egan37 A recent study showed that acute treatment with escitalopram induced changes in response to negative stimuli in the left amygdala and this effect decreased with an increasing number of ‘l’ alleles of 5-HTTLPR in healthy White women. Reference Outhred, Das, Dobson-Stone, Felmingham, Bryant and Nathan49 However, it is difficult to compare our results with previous findings because of differences in participants’ ethnicity, subtypes of SSRIs and stimuli/paradigms. Thus, whether there is a 5-HTTLPR × gender interaction of SSRI treatment is still an open question.
The current study did not find any significant effect of treatment or genotype on participants’ behavioural performances in the repetition-detection tasks during scanning. The lack of significant effect of treatment or genotype on behavioural performance may be driven by: (a) a ceiling effect of behavioural responses, since the repetition-detection task was easy; (b) a limited number of response-required trials, i.e. eight response-required trials for each stimulus category; or (c) the implicit emotional task used in our study, as antidepressant treatment produced distinct effects on implicit and explicit emotional processes. Reference Ma20 A more sensitive and direct behavioural measure may reveal the treatment and genotype effect on behavioural performance.
A neurogenetic mechanism in SSRI response
While there has been increasing interests in pharmacogenetic studies to identify predictors of drug efficacy, Reference Rabl, Scharinger, Müller and Pezawas50,Reference Malhotra, Zhang and Lencz51 our findings of 5-HTTLPR modulation of the effects of SSRIs shed new light on genetic constraints on the acute effects of SSRIs on neural activity underlying emotional processing. Our results suggest possible functional mechanisms by which acute administration of SSRIs may exert effects on emotional responses. Our findings suggest that the ‘small’ effect size of psychometric outcome measures of drug response Reference Rabl, Scharinger, Müller and Pezawas50,Reference Malhotra, Zhang and Lencz51 may be improved by using a pharmacogenetic imaging approach that can help to identify intermediate endophenotypes of drug response.
Funding
This work was supported by the National Natural Science Foundation of China (Project 91332125, 31421003, 31470986), the National Basic Research Program of China (973 Program 2010CB833901, 2010CB833903), Beijing Municipal Natural Science Foundation (Z111107067311058), and the Ministry of Education of China (Project 20130001110049).
Acknowledgements
We thank Siyang Luo, Yi Liu, Yifan Zhang, Zhenhao Shi, Xiangyu Zuo, Dian Yu, Rui Sun, Xiaoyang Li for their help with recruitment of participants. We thank Sook-Lei Liew for proofreading of this paper.
eLetters
No eLetters have been published for this article.