The effect of the serotonin transporter polymorphism (5-HTTLPR) on amygdala function: a meta-analysis

Article metrics


The 5-HTTLPR polymorphism has been widely regarded as a potential genetic risk factor for affective disorders. Consistent with this, this polymorphism has been associated with altered amygdala responses at rest and in response to aversive stimuli. However, the strength of this association remains uncertain. We sought to synthesize existing data on the association between the 5-HTTLPR polymorphism and amygdala activation and ascertain the strength of evidence for this association. Meta-analytic techniques were applied to data from relevant published studies and unpublished data sets to obtain an estimate of the likely magnitude of effect of any association. The large number of studies allowed us to apply a formal test of publication bias, as well as explore the impact of various study-level characteristics on the magnitude of the observed effect size. Our meta-analysis indicated that there is a statistically significant but small effect of 5-HTTLPR on left and right amygdala activity. However, there was considerable between-study heterogeneity, which could not be fully accounted for by the study design and sample characteristics that we investigated. In addition, there was evidence of excess statistical significance among published studies. These findings indicate that the association between the 5-HTTLPR and amygdala activation is smaller than originally thought, and that the majority of previous studies have been considerably under powered to reliably demonstrate an effect of this size.


Since its discovery in 1996,1 a functional polymorphism (5-HTTLPR) in the promoter region of the serotonin transporter gene (SLC6A4) has increasingly been regarded as a potential genetic risk factor for affective disorders. One of the best described findings regarding the association between the 5-HTTLPR and vulnerability to affective disorders has come from imaging genetics studies, which have suggested increased amygdala responses to aversive stimuli in carriers of the short (S) allele of the 5-HTTLPR. This association was first reported in a study which suggested that the influence of this polymorphism could account for over 20% of the variance in the amygdala response to aversive face stimuli.2 The association between the S allele and heightened amygdala activity has subsequently been replicated in healthy volunteers3, 4, 5, 6, 7, 8 and further reported in patients with social phobia,9 panic disorder10 and major depression.5, 11 Given that increased amygdala activation has been associated with depression and anxiety,12, 13, 14 this finding is consistent with the idea that the S allele of the 5-HTTLPR is a risk factor for affective disorders.15

Although the frequency of replication appears impressive, there is an ongoing discussion about the effect size of the association between the 5-HTTLPR and amygdala responses. A previous meta-analysis16 suggested that this association is significant and accounts for about 10% of phenotypic variance. Since the publication of this meta-analysis,16 there has been an influx of further studies investigating the link between the 5-HTTLPR and amygdala activation and there has been some evidence of growing inconsistency in the field.

It is possible that some of these inconsistencies may be due to variation in individual study characteristics, such as the imaging method used (for example, functional magnetic resonance imaging (MRI), positron emission tomography and perfusion) or aspects of the task design or analysis approach. For example, while many studies used tasks that involved the presentation of affective stimuli, there has been a broad range of stimuli used (for example, faces, words and pictures) and the task that participants have been required to do (for example, passive viewing, gender discrimination and explicit recognition of facial expression) has varied across studies. Further, the way in which neuroimaging data have been analyzed, and in particular the contrast used within the subtraction method, has varied across studies. Many previous studies have compared amygdala responses to threatening stimuli with non-emotional stimuli (for example, see refs 2,3), but others have used higher level comparisons such as fearful faces compared with neutral faces (for example, see refs 5,10). Such differences may well account for some of the heterogeneity in effect size across studies, especially as some studies have reported that the association of the 5-HTTLPR with amygdala activation is mainly driven by a greater deactivation of amygdala responses to neutral or control stimuli in S allele carriers rather than increased responses to negative stimuli per se.4 Meta analysis allows the indirect investigation of these study design characteristics by comparing pooled effect size estimates grouped by study characteristic. In principle, this may clarify why some studies fail to detect an effect even though the effect is present, and inform the design and conduct of future primary studies.

There is also increasing concern that the genetic literature may be distorted by various biases, which would lead to a misleading impression of the strength of the association between 5-HTTLPR and amygdala activity. The previous meta-analysis of this literature16 suggested a potential publication bias and concluded that the size of the association may be currently overestimated. In addition, the majority of studies to date have used relatively small sample sizes and there has been a tendency to duplicate data points across studies (for example, see refs 2,6) increasing the likelihood of Type I error. Given that the number of studies in this field has more than doubled since the publication of this previous analysis, it is important to reassess the strength of evidence for the association between 5-HTTLPR and amygdala activity.

We therefore applied established meta-analytic techniques to further synthesize existing data on the association between the 5-HTTLPR polymorphism and amygdala activation and ascertain the strength of evidence for this association. The large number of studies allowed us to apply a formal test of publication bias, as well as explore the impact of various study design characteristics on the magnitude of the observed effect size.

Materials and methods

Selection of studies for inclusion

Genetic association studies of the 5-HTTLPR polymorphism and amygdala activation were identified using a search strategy described previously.16 This included contact with research groups whom we knew from personal communications to hold relevant data for inclusion, to minimize the potential for our results to be distorted by publication bias. Unpublished data from two studies in our own laboratory were also included in the analysis (Z Mannie et al, unpublished data; SE Murphy et al, unpublished data; see Supplementary Information for full details of methods). Studies that reported the effect of the 5-HTTLPR polymorphism on resting state amygdala activity, or on amygdala activity in response to affective stimuli (for example, emotional faces, words and pictures) or the induction of emotional state (for example, public speaking) were included. Studies in children (for example, see ref. 17) or adolescents (for example, see ref. 18) were not included in the analysis.

Data extraction

For each study, the following data were extracted: (1) author (s) and year of publication; (2) data (mean and s.d. of amygdala activity in short and long genotype groups, number of participants, mean age and male/female ratio) and (3) study design. Study design was coded (where applicable) for: sample (healthy volunteer, patient), task requirement (passive viewing of the stimulus, explicit encoding of the stimulus, incidental encoding of the stimulus), stimulus type (faces, words, pictures), stimulus contrast (fixation, shape, neutral image), at rest (yes, no), imaging method (functional MRI, positron emission tomography, perfusion MR), stimulus presentation (unmasked, masked), stimulus emotion (fear, anger, negative), deviation from Hardy–Weinberg equilibrium (no, yes, not stated), ancestry (European, mixed/other) and publication (published, unpublished). Tasks were coded as ‘explicit’ if participants were required to assess the emotion of the affective stimuli presented. For the coding of stimulus presentation, presentation times of 33 ms or less were coded as masked, and longer presentation times as unmasked. Studies were defined as ‘published’ if they appeared as primary studies no later than 2 September 2011. Studies were defined as ‘at rest’ if participants were in a resting state during the scan and not required to engage in any task. Where possible, the data included in the meta-analysis had the following characteristics, to be as close to the original study2 as possible: right amygdala, unmasked, aversive face stimuli (anger and/or fear) and shape (or other low level) contrast. In addition, where data were also available for both hemispheres, the data for the left amygdala were also extracted and included in a subsequent analysis. Some studies reported genotype information that took account of an additional single-nucleotide polymorphism in the L allele of 5-HTTLPR (rs25531 A/G), reported by Hu et al.19 In these studies the Lg allele was treated as functionally equivalent to the S allele and grouped accordingly.

Where data were not reported in an appropriate format in the original study, and attempts to contact study authors unsuccessful, data were extracted manually from figures where possible.

Analysis of data

Effect size estimates (Hedge's g) for individual studies were calculated from mean amygdala activation and corresponding s.d. by 5-HTTLPR genotype where these data were available, or reported effect size estimates, t-statistics or P-values where these data were not available Hedge's g is a measure of standardized mean difference, similar to Cohen's d but including a correction for small sample size.

We used a random effects framework throughout, with gs pooled using DerSimonian and Laird methods.20 A random effects framework assumes that between-study variation is due to both chance or random variation and an individual study effect, and provides an estimate of the range of likely effect sizes across the populations sampled by individual studies. Random-effects models are more conservative than fixed-effects models and generate a wider confidence interval (CI), but give similar results under conditions of low between-study heterogeneity. The significance of the pooled g was determined using a Z-test Between-study heterogeneity was estimated using the I2 statistic. Conventionally, values of 25%, 50% and 75% represent the upper thresholds for low, moderate and high heterogeneity, respectively.

The effect size estimate of the first published study was compared with the pooled effect size of the remaining studies using a Z-test, and for published studies a meta-regression of individual study effect size against year of publication conducted, as there is evidence for a substantially greater estimate of effect size in the first published study, and decreasing effect size with increasing year of publication.21 Small study bias was also assessed using Egger's test.22 Finally, the number of expected studies with statistically significant results was estimated, based on the average statistical power of individual studies given the likely true effect size, and compared against the number of observed significant studies to test for an excess of statistically significant results in this literature.23, 24 Briefly, the expected number of statistically significant findings is calculated using the achieved power of each individual study assuming that the pooled effect size estimate from the meta-analysis is a reasonable estimate of any true population effect. This is compared against the observed number of statistically significant findings using a binomial test. Our approach was conservative as it took observed significant findings from those individual study effects in our meta-analysis. These did not include adjustment for covariates as may have been the case in the published report of those data.

We also conducted a series of stratified analyses and meta-regression analyses to assess the impact of various study design characteristics. The analyses were conducted using the Comprehensive Meta-Analysis (v2) statistical software package (Biostat, Englewood, NJ, USA). Exact P-values are reported throughout.


Description of studies

A total of 31 studies published between 2002 and 2011,2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45 and 4 unpublished data sets (SE Taylor, personal communication, 17 April 2007; IM Anderson, personal communication 3 August 2009; Z Mannie, unpublished; SE Murphy, unpublished) were identified and met the inclusion criteria. Five studies (ref. 11; IM Anderson, personal communication 3 August 2009,28, 30, 33) reported data separately for patients and controls, and these were treated as independent samples. In all studies except one,28 data were based on independently defined amygdala regions of interest, which were either anatomically or functionally defined. The characteristics of these studies are described in Table 1.

Table 1 Characteristics of studies included in meta-analysis

Data from six published studies were not included, in two cases because these data were reported in other studies included in our meta-analysis,31, 34 in three cases because data were not reported in a format that enabled their inclusion and contact with the study author was not successful in obtaining these data36, 39, 40 and in one case where study characteristics were superficially relevant but on closer inspection turned out to not be appropriate for inclusion.29 This resulted in a total of 34 independent study samples for inclusion and analysis. Data from one unpublished data set included in a previous meta-analysis16 were not included at the request of the researcher (MB Stein, personal communication, 17 August 2009).


When all samples (k=34) were included there was weak evidence of association between 5-HTTLPR and amygdala activation in the predicted direction (g=+0.21, 95% CI 0.00, +0.43, P=0.050), with evidence of substantial between-study heterogeneity (I2=70%). These results are presented in Table 2.

Table 2 Meta-analysis stratified by study-level characteristics

When this analysis was restricted to published studies only, the strength of evidence of association was greater and the effect size estimate increased (g=+0.35, 95% CI +0.15, +0.55, P=0.001), but there remained considerable between-study heterogeneity (I2=60%). Among published studies only (k=29), meta-regression did not indicate an effect of year of publication (P=0.25). The effect size estimate for the first published study (g=+0.97), while larger, did not differ significantly from the pooled effect size estimate for the remaining studies (k=28, g=+0.33, Pdiff=0.11). Egger's test did not indicate any evidence of small study bias (t(27)=1.20, P=0.24), although there was some evidence of asymmetry in a funnel plot of study precision against effect size (Figure 1). However, Ioannidis's test indicated an excess of statistically significant results compared with the number that would be expected given the average statistical power of individual studies (O=11, E=5.5, P=0.027), based on the less conservative estimate of the likely true effect size derived from our meta-analysis (g=0.35). This latter test may be more appropriate, as Egger's test is insensitive when most studies are of similar size.

Figure 1

Study precision (1/SE) is plotted against effect size (g). Asymmetry in the plot in the predicted direction (that is, a relative lack of low accuracy or small studies, which indicate no effect or a direction of effect opposite to that reported in the first published study) suggests possible publication bias. SE, standard error.

PowerPoint slide

Stratified analyses of all data indicated that there was a significant moderating effect of Hardy–Weinberg equilibrium status, with larger effect size among studies where genotype frequencies deviated from equilibrium, or where this was not reported. There was no significant moderating effects of any other study-level characteristics (see Table 2), including imaging method (functional MRI, positron emission tomography and perfusion MR), task requirement (passive viewing of the stimulus, explicit encoding of the stimulus, incidental encoding of the stimulus), stimulus contrast (fixation, shape and neutral image), stimulus type (faces, pictures), sample (healthy volunteer, patient), at rest (yes, no) and ancestry (European, mixed/other). The effect of stimulus presentation (unmasked, masked) was not investigated due to the small number of studies using masked presentation of images. These findings were not altered substantially when restricted to published studies only. Notably, in nearly all cases the high levels of between-study heterogeneity persisted when analyses were stratified by study-level characteristics.

Meta-regression, where data were available, did not indicate a significant relationship between individual study effect size estimate and the proportion of male participants (k=29, P=0.25), but did indicate a significant positive relationship with the average age of participants (k=27, slope=+0.02, P=0.010). When unpublished data were included, the relationship between age and effect size still held, and weak evidence for a positive relationship with sex ratio emerged (k=33, slope=+0.01, P=0.067).

Excluding data from one study2 that were partially (68% of sample) reported elsewhere6 did not alter these results substantially. Similarly, removing one study,28 comprising two samples, which did not use an independently defined ROI did not alter these results substantially.

The effect size estimate from our meta-analysis (g=0.35) indicated that a sample size of N=236 would be required to achieve 80% power at an α level of 0.05, assuming a LL genotype frequency of 40%. Power calculations were performed using G*Power.46

In the left amygdala, there was evidence of a similar size of association between 5-HTTLPR and amygdala activation. When all available samples (k=21) were included there was a significant effect of genotype in the predicted direction (g=+0.22, 95% CI 0.00, +0.43, P=0.046). When this analysis was restricted to published studies only (k=19), the effect size estimate remained the same but the strength of evidence of association was weaker (g=+0.22, 95% CI −0.02, +0.45, P=0.067). These results are presented in Table 3.

Table 3 Meta-analysis of left amygdala


Our meta-analysis indicated that there is a marginally statistically significant but small effect of 5-HTTLPR on left and right amygdala activity. Under a random-effects model, the effect size estimates within the 95% CIs ranged from g=0.00 to 0.43. The size of this effect for the right amygdala increased when only published studies were included in the analysis (ranging from g=0.15 to 0.55) but there was evidence of excess statistical significance among these. In all cases, there was evidence of between-study heterogeneity, which could not be fully accounted for by the study design and sample characteristics that we investigated.

The association between the S allele and increased amygdala responses to aversive stimuli is well documented2, 3, 5, 6, 7, 9, 10, 11, 31 and has contributed to a widely accepted view that this polymorphism represents a robust and important risk factor for affective disorders. A previous meta-analysis estimated that the 5-HTTLPR may account for about 10% of the variance in amygdala activation; however, it was noted that this may have been an overestimate in light of evidence, suggesting a potential publication bias.16 The current meta-analysis, which includes over double the number of published studies, indicated a high degree of heterogeneity across samples. It is noteworthy that the random-effects model addresses a different research question to the fixed-effects model; specifically, it provides a range of effect sizes, which may exist across the different populations sampled, rather than the best estimate of effect size in a single population. This is appropriate given the substantial between-study heterogeneity we observed, even in our stratified analyses. This analysis suggests that the effect size of the association between the 5-HTTLPR and amygdala activation is smaller than previously thought, accounting for 1% of the variance in amygdala activation. Importantly, even at the upper bound of the range of estimates, this analysis suggests that the 5-HTTLPR accounts for 4.6% of variance amygdala activation.

Perhaps the most striking result to emerge from this analysis is the power analysis that suggests, if the least conservative effect size (from only published studies) is accepted, a sample of 236 participants would be needed to reliably observe an association of this size. Importantly, this analysis also indicated that, to date, no single study has been sufficiently powered to reliably demonstrate an effect of this size. Indeed, the largest published study is a perfusion MR study, which showed no evidence of association (g=0.00) with a sample size of N=183.43 Although it is important to take into account that this large scale study assessed baseline perfusion in the amygdala, rather than the amygdala response to affective stimuli, our stratified analysis suggested that method (for example, functional MRI vs positron emission tomography vs perfusion) did not significantly moderate the effect size estimate (see Table 2).

Given the high degree of heterogeneity across samples, it is possible that the true effect size may genuinely differ across populations or study designs. The current meta-analysis provided the opportunity to examine whether there are any potential moderating factors that may be responsible for the variability in the published data. A variety of experimental paradigms and analysis methods have been used across the studies to date, and one possibility is that these differences account for the failure of some of the studies to replicate the original effect. However, our analysis found no significant moderating influence of study design characteristics, with the exception of Hardy–Weinberg equilibrium status (deviation from Hardy–Weinberg equilibrium is sometimes taken to indicate potential genotyping error). Other factors, such as experimental paradigm, imaging modality and experimental contrast, did not appear to substantially influence the pooled effect size estimate. Overall, our stratified analyses suggest that the variability seen across studies is mainly driven by inadequate sample sizes, and possibly genotyping quality, rather than differences in study design. It should be noted, however, that in many cases there were relatively few studies within an individual stratum of a stratified analysis, so that the power to detect statistically significant effects of study-level characteristics was often limited. Our meta-regression analysis did highlight one population characteristic that may influence effect size, namely age, with studies using older samples associated with a larger effect size.

The use of intermediate phenotypes to dissect the effects of genetic polymorphisms is an attractive strategy and has been applied to a number of polymorphisms. However, there have been a number of recent meta-analyses that have failed to confirm apparently reliable associations,47, 48 suggesting that the power to identify the consequences of single-genetic polymorphisms using this approach may be more modest than previously assumed. A theoretical basis for this unreliability has been identified by Flint and Munafò,49 who suggest that the contribution of specific genes to endophenotypes may not necessarily be any larger than that it is to illness phenotypes. A particular danger lies in the assumption that intermediate phenotypes necessarily confer greater statistical power; if this is not the case, but studies use this assumption to justify the use of small sample sizes, the rate of false-positive findings among those studies that achieve nominal statistical significance will increase, which will be a consequence of the resulting low statistical power achieved.50, 51 Future studies of the way in which genetic variation influences the processing of emotional information may have to use very large data sets with simultaneous measurements of several genes and appropriate mathematical modeling. Such a task is challenging but may equate more closely to the complex neurobiology of the brain mechanisms involved in emotional processing and their accompanying clinical disorders.


  1. 1

    Lesch KP, Bengel D, Heils A, Sabol SZ, Greenberg BD, Petri S et al. Association of anxiety-related traits with a polymorphism in the serotonin transporter gene regulatory region. Science 1996; 274: 1527–1531.

  2. 2

    Hariri AR, Mattay VS, Tessitore A, Kolachana B, Fera F, Goldman D et al. Serotonin transporter genetic variation and the response of the human amygdala. Science 2002; 297: 400–403.

  3. 3

    Bertolino A, Arciero G, Rubino V, Latorre V, De Candia M, Mazzola V et al. Variation of human amygdala response during threatening stimuli as a function of 5′HTTLPR genotype and personality style. Biol Psychiatry 2005; 57: 1517–1525.

  4. 4

    Canli T, Omura K, Haas BW, Fallgatter A, Constable RT, Lesch KP . Beyond affect: a role for genetic variation of the serotonin transporter in neural activation during a cognitive attention task. Proc Natl Acad Sci USA 2005; 102: 12224–12229.

  5. 5

    Dannlowski U, Ohrmann P, Bauer J, Kugel H, Baune BT, Hohoff C et al. Serotonergic genes modulate amygdala activity in major depression. Genes Brain Behav 2007; 6: 672–676.

  6. 6

    Hariri AR, Drabant EM, Munoz KE, Kolachana BS, Mattay VS, Egan MF et al. A susceptibility gene for affective disorders and the response of the human amygdala. Arch Gen Psychiatry 2005; 62: 146–152.

  7. 7

    Heinz A, Braus DF, Smolka MN, Wrase J, Puls I, Hermann D et al. Amygdala-prefrontal coupling depends on a genetic variation of the serotonin transporter. Nat Neurosci 2005; 8: 20–21.

  8. 8

    Rao H, Gillihan SJ, Wang J, Korczykowski M, Sankoorikal GM, Kaercher KA et al. Genetic variation in serotonin transporter alters resting brain function in healthy individuals. Biol Psychiatry 2007; 62: 600–606.

  9. 9

    Furmark T, Tillfors M, Garpenstrand H, Marteinsdottir I, Langstrom B, Oreland L et al. Serotonin transporter polymorphism related to amygdala excitability and symptom severity in patients with social phobia. Neurosci Lett 2004; 362: 189–192.

  10. 10

    Domschke K, Braun M, Ohrmann P, Suslow T, Kugel H, Bauer J et al. Association of the functional -1019C/G 5-HT1A polymorphism with prefrontal cortex and amygdala activation measured with 3 T fMRI in panic disorder. Int J Neuropsychopharmacol 2006; 9: 349–355.

  11. 11

    Dannlowski U, Ohrmann P, Bauer J, Deckert J, Hohoff C, Kugel H et al. 5-HTTLPR biases amygdala activity in response to masked facial expressions in major depression. Neuropsychopharmacology 2008; 33: 418–424.

  12. 12

    Sheline YI, Barch DM, Donnelly JM, Ollinger JM, Snyder AZ, Mintun MA . Increased amygdala response to masked emotional faces in depressed subjects resolves with antidepressant treatment: an fMRI study. Biol Psychiatry 2001; 50: 651–658.

  13. 13

    Shin LM, Orr SP, Carson MA, Rauch SL, Macklin ML, Lasko NB et al. Regional cerebral blood flow in the amygdala and medial prefrontal cortex during traumatic imagery in male and female Vietnam veterans with PTSD. Arch Gen Psychiatry 2004; 61: 168–176.

  14. 14

    Stein MB, Goldin PR, Sareen J, Zorrilla LT, Brown GG . Increased amygdala activation to angry and contemptuous faces in generalized social phobia. Arch Gen Psychiatry 2002; 59: 1027–1034.

  15. 15

    Clarke H, Flint J, Attwood AS, Munafo MR . Association of the 5- HTTLPR genotype and unipolar depression: a meta-analysis. Psychol Med 2010; 40: 1767–1778.

  16. 16

    Munafò MR, Brown SM, Hariri AR . Serotonin transporter (5-HTTLPR) genotype and amygdala activation: a meta-analysis. Biol Psychiatry 2008; 63: 852–857.

  17. 17

    Thomason ME, Henry ML, Paul Hamilton J, Joormann J, Pine DS, Ernst M et al. Neural and behavioral responses to threatening emotion faces in children as a function of the short allele of the serotonin transporter gene. Biol Psychol 2010; 85: 38–44.

  18. 18

    Lau JY, Goldman D, Buzas B, Fromm SJ, Guyer AE, Hodgkinson C et al. Amygdala function and 5-HTT gene variants in adolescent anxiety and major depressive disorder. Biol Psychiatry 2009; 65: 349–355.

  19. 19

    Hu XZ, Lipsky RH, Zhu G, Akhtar LA, Taubman J, Greenberg BD et al. Serotonin transporter promoter gain-of-function genotypes are linked to obsessive-compulsive disorder. Am J Hum Genet 2006; 78: 815–826.

  20. 20

    DerSimonian R, Laird N . Meta-analysis in clinical trials. Control Clin Trials 1986; 7: 177–188.

  21. 21

    Ioannidis JP, Ntzani EE, Trikalinos TA, Contopoulos-Ioannidis DG . Replication validity of genetic association studies. Nat Genet 2001; 29: 306–309.

  22. 22

    Egger M, Davey Smith G, Schneider M, Minder C . Bias in meta-analysis detected by a simple, graphical test. BMJ 1997; 315: 629–634.

  23. 23

    Ioannidis JP . Excess significance bias in the literature on brain volume abnormalities. Arch Gen Psychiatry 2011; 68: 773–780.

  24. 24

    Ioannidis JP, Trikalinos TA . An exploratory test for an excess of significant findings. Clin Trials 2007; 4: 245–253.

  25. 25

    Brown SM, Hariri AR . Neuroimaging studies of serotonin gene polymorphisms: exploring the interplay of genes, brain, and behavior. Cogn Affect Behav Neurosci 2006; 6: 44–52.

  26. 26

    Canli T, Congdon E, Todd Constable R, Lesch KP . Additive effects of serotonin transporter and tryptophan hydroxylase-2 gene variation on neural correlates of affective processing. Biol Psychol 2008; 79: 118–125.

  27. 27

    Canli T, Qiu M, Omura K, Congdon E, Haas BW, Amin Z et al. Neural correlates of epigenesis. Proc Natl Acad Sci USA 2006; 103: 16033–16038.

  28. 28

    Friedel E, Schlagenhauf F, Sterzer P, Park SQ, Bermpohl F, Strohle A et al. 5-HTT genotype effect on prefrontal-amygdala coupling differs between major depression and controls. Psychopharmacology 2009; 205: 261–271.

  29. 29

    Furmark T, Appel L, Henningsson S, Ahs F, Faria V, Linnman C et al. A link between serotonin-related gene polymorphisms, amygdala activity, and placebo-induced relief from social anxiety. J Neurosci 2008; 28: 13066–13074.

  30. 30

    Furmark T, Henningsson S, Appel L, Ahs F, Linnman C, Pissiota A et al. Genotype over-diagnosis in amygdala responsiveness: affective processing in social anxiety disorder. J Psychiatry Neurosci 2009; 34: 30–40.

  31. 31

    Heinz A, Smolka MN, Braus DF, Wrase J, Beck A, Flor H et al. Serotonin transporter genotype (5-HTTLPR): effects of neutral and undefined conditions on amygdala activation. Biol Psychiatry 2007; 61: 1011–1014.

  32. 32

    Lee BT, Ham BJ . Serotonergic genes and amygdala activity in response to negative affective facial stimuli in Korean women. Genes Brain Behav 2008; 7: 899–905.

  33. 33

    Shah MP, Wang F, Kalmar JH, Chepenik LG, Tie K, Pittman B et al. Role of variation in the serotonin transporter protein gene (SLC6A4) in trait disturbances in the ventral anterior cingulate in bipolar disorder. Neuropsychopharmacology 2009; 34: 1301–1310.

  34. 34

    Smolka MN, Buhler M, Schumann G, Klein S, Hu XZ, Moayer M et al. Gene-gene effects on central processing of aversive stimuli. Mol Psychiatry 2007; 12: 307–317.

  35. 35

    Surguladze SA, Elkin A, Ecker C, Kalidindi S, Corsico A, Giampietro V et al. Genetic variation in the serotonin transporter modulates neural system-wide response to fearful faces. Genes Brain Behav 2008; 7: 543–551.

  36. 36

    Williams LM, Gatt JM, Schofield PR, Olivieri G, Peduto A, Gordon E . Negativity bias’ in risk for depression and anxiety: brain-body fear circuitry correlates, 5-HTT-LPR and early life stress. Neuroimage 2009; 47: 804–814.

  37. 37

    Brockmann H, Zobel A, Schuhmacher A, Daamen M, Joe A, Biermann K et al. Influence of 5-HTTLPR polymorphism on resting state perfusion in patients with major depression. J Psychiatr Res 2011; 45: 442–451.

  38. 38

    Dannlowski U, Konrad C, Kugel H, Zwitserlood P, Domschke K, Schoning S et al. Emotion specific modulation of automatic amygdala responses by 5-HTTLPR genotype. Neuroimage 2010; 53: 893–898.

  39. 39

    Demaree HA, Pu J, Jesberger J, Feeny N, Jeng L, Everhart DE et al. 5HTTLPR predicts left fusiform gyrus activation to positive emotional stimuli. Magn Reson Imaging 2009; 27: 441–448.

  40. 40

    Fukudo S, Kanazawa M, Mizuno T, Hamaguchi T, Kano M, Watanabe S et al. Impact of serotonin transporter gene polymorphism on brain activation by colorectal distention. Neuroimage 2009; 47: 946–951.

  41. 41

    Gillihan SJ, Rao H, Wang J, Detre JA, Breland J, Sankoorikal GM et al. Serotonin transporter genotype modulates amygdala activity during mood regulation. Soc Cogn Affect Neurosci 2010; 5: 1–10.

  42. 42

    Schardt DM, Erk S, Nusser C, Nothen MM, Cichon S, Rietschel M et al. Volition diminishes genetically mediated amygdala hyperreactivity. Neuroimage 2010; 53: 943–951.

  43. 43

    Viviani R, Sim EJ, Lo H, Beschoner P, Osterfeld N, Maier C et al. Baseline brain perfusion and the serotonin transporter promoter polymorphism. Biol Psychiatry 2010; 67: 317–322.

  44. 44

    von dem Hagen EA, Passamonti L, Nutland S, Sambrook J, Calder AJ . The serotonin transporter gene polymorphism and the effect of baseline on amygdala response to emotional faces. Neuropsychologia 2011; 49: 674–680.

  45. 45

    O’Nions EJ, Dolan RJ, Roiser JP . Serotonin transporter genotype modulates subgenual response to fearful faces using an incidental task. J Cogn Neurosci 2011; 23: 3681–3693.

  46. 46

    Buchner A, Erdfelder E, Faul F . How to use G*Power., 1997.

  47. 47

    Barnett JH, Scoriels L, Munafo MR . Meta-analysis of the cognitive effects of the catechol-O-methyltransferase gene Val158/108Met polymorphism. Biol Psychiatry 2008; 64: 137–144.

  48. 48

    Munafò MR, Freimer NB, Ng W, Ophoff R, Veijola J, Miettunen J et al. 5-HTTLPR genotype and anxiety-related personality traits: a meta-analysis and new data. Am J Med Genet B Neuropsychiatr Genet 2009; 150B: 271–281.

  49. 49

    Flint J, Munafo MR . The endophenotype concept in psychiatric genetics. Psychol Med 2007; 37: 163–180.

  50. 50

    Green AE, Munafo MR, DeYoung CG, Fossella JA, Fan J, Gray JR . Using genetic data in cognitive neuroscience: from growing pains to genuine insights. Nat Rev Neurosci 2008; 9: 710–720.

  51. 51

    Sterne JA, Davey Smith G . Sifting the evidence-what's wrong with significance tests? BMJ 2001; 322: 226–231.

Download references


This work was supported by funding from the Medical Research Council and the Wellcome Trust.

Author information

Correspondence to S E Murphy.

Ethics declarations

Competing interests

PJC has been a paid member of advisory boards of Eli Lilly, Lundbeck, Servier and Wyeth. He has received remuneration for scientific advice given to legal representatives of GlaxoSmithKline. CJH is on the advisory board for p1vital and also owns shares in the company. She is a company director of Oxford Psychologists and has received consultancy payments from p1vital, Servier and GSK. She has also participated in paid speaking engagements for Eli Lily and Lundbeck. SEM has received consultancy payments from p1Vital and has participated in paid speaking engagements for Lilly UK. MRM, RN, ZMM and BRG declare no conflict of interest.

Additional information

Supplementary Information accompanies the paper on the Molecular Psychiatry website

Supplementary information

PowerPoint slides

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Murphy, S., Norbury, R., Godlewska, B. et al. The effect of the serotonin transporter polymorphism (5-HTTLPR) on amygdala function: a meta-analysis. Mol Psychiatry 18, 512–520 (2013) doi:10.1038/mp.2012.19

Download citation


  • amygdala
  • 5-HTTLPR
  • meta analysis
  • serotonin transporter

Further reading