Social adversity, the serotonin transporter (5-HTTLPR) polymorphism and major depressive disorder.
ABSTRACT Recent evidence has suggested that the short allele of the serotonin transporter (5-HTT) gene-linked polymorphic region (5-HTTLPR of the human serotonin gene [SLC6A4]) is associated with increased risk of depressive disorder but only among individuals exposed to social adversity. We report an investigation designed to replicate this finding.
Data were available from a non-clinical sample of 4,175 adult men and women, ages 41-80 years, selected from participants in the European Prospective Investigation into Cancer and Nutrition in Norfolk (EPIC-Norfolk, United Kingdom) study. Evidence of past-year prevalent episodic major depressive disorder (MDD), defined by restricted DSM-IV diagnostic criteria, was assessed through questionnaire. Adverse experiences in childhood and in adulthood (during the five years preceding assessment) were also assessed through self-report. The 5-HTTLPR variant was genotyped according to published protocols.
One-year prevalent MDD criteria were met by 298 study participants. The experience of social adversity (both in childhood and adulthood) was strongly associated with increased rates of past-year prevalent MDD. No gene by environment (GxE) interactions between the 5-HTTLPR genotype, social adversity, and MDD were observed.
This study has not replicated a previous finding of a GxE interaction between the 5-HTTLPR genotype, social adversity, and depression.
- SourceAvailable from: Christoph Anacker[Show abstract] [Hide abstract]
ABSTRACT: We review studies with human and nonhuman species that examine the hypothesis that epigenetic mechanisms, particularly those affecting the expression of genes implicated in stress responses, mediate the association between early childhood adversity and later risk of depression. The resulting studies provide evidence consistent with the idea that social adversity, particularly that involving parent-offspring interactions, alters the epigenetic state and expression of a wide range of genes, the products of which regulate hypothalamic-pituitary-adrenal function. We also address the challenges for future studies, including that of the translation of epigenetic studies towards improvements in treatments.Dialogues in clinical neuroscience 09/2014; 16(3):321-33.
- [Show abstract] [Hide abstract]
ABSTRACT: Considerable evidence suggests a crucial role for the epigenetic regulation of brain-derived neurotrophic factor (BDNF) in the pathophysiology of major depressive disorder (MDD). However, the relationship between BDNF DNA methylation and white matter (WM) integrity in MDD has not yet been investigated. In the current study, we examined the association between the DNA methylation status of the BDNF promoter region and WM integrity in MDD. Sixty patients with MDD and 53 healthy controls underwent T1-weighted structural magnetic resonance imaging (MRI), including diffusion tensor imaging (DTI), to assess their WM integrity. BDNF DNA methylation at 4 CpG sites of the promoter region was also measured.As compared to healthy controls, the MDD group demonstrated reduced fractional anisotropy (FA) in the bilateral anterior and posterior corona radiata (ACR and PCR), genu of the corpus callosum, and the bilateral posterior thalamic radiations. We observed a significant inverse correlation between the DNA methylation of the BDNF promoter region and the FA of the right ACR in MDD patients.Our findings demonstrate a relationship between methylation of the BDNF promoter region and the integrity of the ACR, a key structural component of the emotional and cognitive control network involved in the pathophysiology of MDD. This correlation suggests that BDNF DNA methylation may contribute to structural WM changes in MDD patients.Journal of Affective Disorders 02/2015; 172. · 3.71 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: To ascertain whether genetic variations in the serotonin transporter gene (5-HTTLPR 44-bp insertion/deletion polymorphism) influence an increase in depressive and anxiety symptoms in children and adolescents exposed to high levels of violence. Saliva samples were collected from a group of children who were working on the streets and from their siblings who did not work on the streets. DNA was extracted from the saliva samples and analyzed for 5-HTTLPR polymorphism genotypes. One hundred and seventy-seven children between the ages of 7 and 14 years were analyzed (114 child workers and 63 siblings). Data on socioeconomic conditions, mental symptoms, and presence and severity of maltreatment and urban violence were collected using a sociodemographic inventory and clinical instruments. There was no positive correlation between the 5-HTTLPR polymorphism and presence of mental symptoms in our sample, although the children were exposed to high levels of abuse, neglect, and urban violence. Despite previous studies that associated adult psychiatric disorders with the 5-HTTLPR polymorphism and a history of childhood maltreatment, no such association was found in this sample of children at risk.Revista Brasileira de Psiquiatria 12/2014; 36(4):277-84. · 1.64 Impact Factor
Social Adversity, the Serotonin Transporter
(5-HTTLPR) Polymorphism and Major
Paul G. Surtees, Nicholas W.J. Wainwright, Saffron A.G. Willis-Owen, Robert Luben,
Nicholas E. Day, and Jonathan Flint
Background: Recent evidence has suggested that the short allele of the serotonin transporter (5-HTT) gene-linked polymorphic region
(5-HTTLPR of the human serotonin gene [SLC6A4]) is associated with increased risk of depressive disorder but only among individuals
exposed to social adversity. We report an investigation designed to replicate this finding.
Methods: Data were available from a non-clinical sample of 4175 adult men and women, ages 41–80 years, selected from
participants in the European Prospective Investigation into Cancer and Nutrition in Norfolk (EPIC-Norfolk, United Kingdom) study.
Evidence of past-year prevalent episodic major depressive disorder (MDD), defined by restricted DSM-IV diagnostic criteria, was
assessed through questionnaire. Adverse experiences in childhood and in adulthood (during the five years preceding assessment) were
also assessed through self-report. The 5-HTTLPR variant was genotyped according to published protocols.
Results: One-year prevalent MDD criteria were met by 298 study participants. The experience of social adversity (both in childhood
and adulthood) was strongly associated with increased rates of past-year prevalent MDD. No gene by environment (GxE) interactions
between the 5-HTTLPR genotype, social adversity, and MDD were observed.
Conclusions: This study has not replicated a previous finding of a GxE interaction between the 5-HTTLPR genotype, social adversity,
Key Words: Depression, stress, serotonin transporter, 5-HTTLPR,
SLC6A4, gene-environment interaction
arthropathies (Scofield et al 1995). Few such interactions, how-
ever, have been successfully characterized in relation to human
psychiatric phenotypes, despite clear indications from nonhu-
man mammal research that they exist. For example, investiga-
tions of the genetic contribution of the serotonin transporter
(5-HTT) to behavioral measures of anxiety in both rodents and
nonhuman primates have indicated a moderating role of social
adversity, most specifically during early stages of development
(Ansorge et al 2004; Bennett et al 2002; Champoux et al 2002).
Recently, Caspi et al (2003) reported that a length polymor-
phism (SLC6A4) in the promoter region of 5-HTT (5-HTTLPR)
mediates the influence of stressful life events on human depres-
sion. They showed that individuals carrying one or more short
(S) allele who were exposed to stressful life events were more
likely to develop depression than those homozygous for the long
(L) allele. They also reported that childhood maltreatment pre-
dicted adult depression only among individuals carrying a copy
of the S allele.
This finding is intriguing for a number of reasons. First, as one
of the few examples of a gene by environment (GxE) interaction,
it provides an opportunity to investigate the phenomenon at a
molecular level. Because of its potential importance, Caspi et al
nteractions between genetic loci and the environment are
known for a wide variety of biological phenotypes, such as
the thalassaemias (Weatherall 2001) and reactive spondylo-
have emphasized the need for its replication. Second, the
interaction effects reported seem to be present in the absence of
a main effect of the SLC6A4 promoter length polymorphism on
depression. Although, historically, studies that have addressed
this question in relation to (unipolar) depression have reached
inconsistent conclusions, a recent large multicenter case-control
study (involving 539 unipolar patients and 821 control subjects;
Mendlewicz et al 2004) and a recent meta-analysis (Lasky-Su et al
2005) have both reported no main effect, indicating that the main
effect, if present, must be very small. Third, the reported effect
was observed on depression occurring in the year before assess-
ment (at age 26) for individuals experiencing stressful life events
in the preceding five years. In contrast, the reported impact of life
events is typically limited to the 1–3 months preceding the onset
of a depressive episode (Brown et al 1973; Kendler et al 1998;
Surtees et al 1986).
We now report an attempt to replicate the GxE interaction
between SLC6A4 (5-HTTLPR) genotype, the experience of social
adversity, and DSM-IV– defined (American Psychiatric Associa-
tion 1994) major depressive disorder (MDD) among a large
non-clinical sample of men and women participating in the
European Prospective Investigation into Cancer and Nutrition in
Norfolk (EPIC-Norfolk), United Kingdom study (Day et al 1999).
Methods and Materials
During 1993–1997, EPIC-Norfolk recruited, through general
practice age-gender registers, a total of 30,414 men and women
(then) ages 40 –74 years and resident in Norfolk, England (Day et
al 1999). The study was approved by the Norwich District Health
Authority Ethics Committee and all participants gave signed
informed consent. During 1996–2000, an assessment of social
and psychological circumstances, on the basis of the Health and
Life Experiences Questionnaire (HLEQ) (Surtees et al 2003b),
was completed by a total of 20,921 participants, representing a
response rate of 73.2% of the total eligible EPIC-Norfolk sample
(28,582). A sample of 5000 participants was selected from the
From Strangeways Research Laboratory and University of Cambridge De-
Causeway, Cambridge; and the Wellcome Trust Centre for Human Ge-
netics (SAGW-O, JF), Headington, Oxford, United Kingdom.
Worts Causeway, Cambridge, CB1 8RN, United Kingdom; E-mail: paul.
Received January 31, 2005; revised June 15, 2005; accepted July 8, 2005.
BIOL PSYCHIATRY 2006;59:224–229
© 2005 Society of Biological Psychiatry
EPIC-Norfolk HLEQ cohort. This sample was originally designed
for a study of neuroticism (Willis-Owen, in press) and included
2500 men and 2500 women with DNA available, selected accord-
ing to extremes of high and low neuroticism scores (assessed
through completion of a 12-item scale; Eysenck et al 1985). The
boundaries of high and low neuroticism varied by gender (high:
scale score ranges 6–12 and 8–12 for men and women, respec-
tively; and low: scale score ranges 0 –1 and 0 –2; see Willis-Owen,
in press for further details). The final study sample consisted of
4175 study participants for whom the SLC6A4 variant was
Mood State Assessment
The HLEQ includes a structured self-assessment approach to
psychiatric symptoms embodying restricted DSM-IV (American
Psychiatric Association 1994) criteria for MDD. The assessment
was designed to identify those EPIC-Norfolk HLEQ participants
thought likely to have met a putative diagnosis of MDD at any
time in their lives. Where any episode was reported, participants
were asked to estimate onset and (if appropriate) offset timings
(see Surtees et al 2000 for full details of this assessment). The
dependent variable was taken as 1-year prevalence of MDD (i.e.,
any episode that was either current at the time of or ended within
1 year of HLEQ completion). In addition, this measure was
refined according to neuroticism score, with cases defined as
those with both 1-year prevalent MDD and high neuroticism, and
with control subjects defined as those free of 1-year prevalent
MDD and with low neuroticism.
Social Adversity Assessment
The HLEQ included an assessment of social adversity, defined
by adverse experience in childhood (0–16 years) and in adult-
hood (represented by stressful life events and enduring personal
Adverse experience in childhood was assessed by the follow-
ing eight circumstances: separation from mother for more than 1
year; hospital stay for two or more weeks; parental divorce;
parental unemployment for several years when they wanted to
be working; an experience that was so frightening as to be
thought about for years following its occurrence; being sent
away from home because of doing something wrong; parental
alcohol or drug use sufficient to cause family problems; and
experience of physical abuse by someone close. An overall
measure of childhood adverse experience was constructed as the
total number of circumstances reported (Surtees et al 2003a;
Wainwright and Surtees 2002a).
Retrospective assessment of adverse event experience in
adulthood was restricted to 16 specific events and a further
undefined event of personal significance. Specific events in-
volved serious illnesses (injuries or assaults) experienced by the
participant (or a first degree relative), relationship events (con-
cerning separation, divorce, termination of pregnancy), work
events (retirement, redundancy, or being fired) and loss experi-
ences (through death of first degree relatives). The undefined
event provided an opportunity for the participant to describe
(and rate) the lifetime experience of any other particularly
stressful experience. Participants were asked to provide details of
event timing to within a year of occurrence and a rating of their
degree of upset (rated on a four-point scale) associated with each
event experienced. These events were selected on the basis of
the questionnaire version of the List of Threatening Experiences
(LTE-Q) (Brugha et al 1985). Concurrent validity of the LTE-Q
assessed against the Life Events and Difficulties Schedule (LEDS;
Brown and Harris 1978) has shown both high specificity and
sensitivity (Brugha and Cragg 1990) (see Surtees and Wainwright
1998, 2000 for further details). A calendar-based Personal Life
Chart (PLC) was designed to allow each participant to record the
date, description, and onset/offset times of (up to six) prolonged
difficulties in their lives. The design format provided a brief way
of representing periods of personal difficulty and stems from the
methodological framework of the LEDS (Brown and Harris 1978)
approach to the assessment of long-term difficulties adapted for
use within a questionnaire format and to record experiences
across a lifetime. Only those long-term difficulties reported to
have been experienced by the study participant or close family
(restricted to spouse/partner, parent, sibling, or child) were
included. Adverse experience in adulthood was represented by
the total number of moderately or extremely upsetting life events
and/or a period of long-term difficulty reported to have been
experienced during the past 5 years. In addition, these measures
were refined to include only those adverse events and difficulties
reported to have been experienced in the year preceding MDD
episode onset (or time of HLEQ assessment for those with no
1-year prevalent MDD).
Deoxyribonucleic acid (DNA) was extracted from fresh 9-mL
EDTA whole blood samples (Whatman International, Ely, United
Kingdom) and arrayed onto 96 well plates. Two oligonucleotide
primers (5=-GGCGTTGCCGCTCTGAATGC-3=, 5=-GAGGGACTGA-
GCTGGACAACCCAC-3=) were used to generate SLC6A4 allele-
specific fragments (484-base pair [bp], 528-bp) (Lerman et al 1998)
by polymerase chain reaction (PCR). Genotype was then ascer-
tained by agarose gel electrophoresis of PCR products, and visual-
ized by transillumination. The PCR was initially performed in a
increased to a final volume of 15 ?L (16.8 ng DNA) where samples
failed to produce amplification. Each 10-?L reaction included 1 ?L
NH4buffer, .25 ?L dNTP (40 mmol/L), 2.35 ?L 2xPolymate (Bio-
Line, London, United Kingdom), 0.3 ?L MgCl2(50 mmol/L), 1 ?L
primers, and .1 ?L BioTaq DNA Polymerase (BioLine). The PCR
consisted of a 5-min denature step at 95°C (1 cycle), 95°C for 30
seconds, 62°C for 45 seconds, and 72°C for 1 min (35 cycles), and
72°C for 4 min (1 cycle).
Logistic regression was used to investigate the presence of
GxE interaction for past-year prevalent MDD. Models were fit for
the main effect of SLC6A4 (5-HTTLPR) genotype (included as
L/L; L/S and S/S, assuming a multiplicative effect for the S allele
and, correspondingly, a 1 degree of freedom test of trend [allele
test]) and main effect of environment (with childhood adverse
experience classified as 0, 1, and 2 or more and adult life events
or long-term difficulty experience as 0, 1, 2, 3, and 4 or more
included as continuous variables, again with a 1 degree of
freedom test of trend), and with subsequent addition of GxE
interaction (through inclusion of the cross-product term for a 1
degree of freedom test of multiplicative interaction). All models
included gender as a covariate. Secondary analyses were per-
formed to confirm the absence of significant GxE interaction in
these data. Analyses were repeated 1) for the refined measure of
adverse experience in adulthood (in the year before MDD onset),
2) for lifetime MDD, 3) for the combined measure of past-year
prevalent MDD and neuroticism, and 4) for past-year prevalent
MDD by gender.
P.G. Surtees et al
BIOL PSYCHIATRY 2006;59:224–229 225
A total of 4175 genotype calls were made from a total of 4416
individuals genotyped (a 94.5% genotype call rate). Of the 4175
study participants, 2225 were men and 1950 were women. The
age range of the sample was 41–80 years (mean 60.3 years, SD
9.1). Genotype frequencies were: L/L homozygotes, n ? 1391
(33.3%); L/S heterozygotes, n ? 2029 (48.6%); and S/S homozy-
gotes, n ? 755 (18.1%). Allele frequencies were in Hardy-
Weinberg equilibrium [?2(1) ? .10, p ? .75]. The 1-year preva-
lence of MDD (in this neuroticism-enriched sample) was 7.1% (n ?
298) and lifetime MDD prevalence was 18.3% (n ? 767).
Past-year prevalence of MDD was 5.6%, 8.5%, and 10.1%,
respectively, for those who reported 0, 1, and ? 2 adverse
experiences in childhood, and was 2.9%, 5.5%, 11.2%, 14.5%, and
26.8%, respectively, for those who reported 0, 1, 2, 3, and ? 4 life
events or difficulties. No differences were observed by genotype,
either in the mean number of reported adverse experiences in
childhood (.78 for L/L, .77 for L/S, and .80 for S/S, p ? .75) or
adverse experiences during adulthood (1.14 for L/L, 1.07 for L/S,
and 1.05 for S/S, p ? .19).
Table 1 shows the number (and percentage) of study partic-
ipants with past-year prevalent MDD by SLC6A4 (5-HTTLPR)
genotype and according to the number of adverse experiences in
childhood. Table 2 provides the same data according to the
number of life events or long-term difficulties experienced
during the five years before assessment. Table 3 shows the odds
ratios (OR) for the main effect of SLC6A4 genotype and social
adversity and their interaction on past-year prevalent MDD. The
table confirms the absence of any main effect for genotype and
the presence of strong main effects for environment (as repre-
sented by either adverse experience in childhood or adult life
events or difficulties) and reveals that no GxE interactions were
observed in these data.
Adverse adult events or difficulties were reported in the year
before MDD onset/HLEQ completion by 930 participants (732
[17.5%] reported one and 198 [4.7%] reported more than one
event and/or difficulty in this 1-year period). This refined mea-
sure of adult adverse events showed a stronger association with
past-year MDD (OR ? 2.11, 95% confidence interval [CI]1.78–
2.51 per event) but, again, no GxE interaction with SLC6A4
(5-HTTLPR) genotype was observed (OR ? .91, 95% CI .71–
1.17). Nor was there a significant interaction between genotype
and adverse events experienced in childhood or between geno-
type and adult adverse experience (in the past 5 years) on the
basis of a lifetime MDD phenotype. Finally, we found no
evidence for a significant interaction between genotype and
adverse experience in childhood or adult adverse experience for
the combined measure of past-year prevalent MDD and neurot-
icism (n ? 276 cases and n ? 2018 control subjects).
A significant interaction was observed for past-year prevalent
MDD and adverse experience in childhood among men (p ?
.040) such that adverse experiences in childhood were associ-
ated with increased rates of past-year prevalent MDD for L/L
homozygotes (OR ? 1.69, 95% CI 1.17–2.44) but not for L/S
heterozygotes (OR ? 1.26, 95% CI .91–1.75) or S/S homozygotes
(OR ? .82, 95% CI .46–1.48). This finding is in the opposite
direction of that previously reported for a GxE interaction (Caspi
Table 1. Past-year Prevalent Major Depressive Disorder (MDD) According
to SLC6A4 (5-HTTLPR) Genotype and the Number of Adverse Experiences
Adverse Experience in Childhood
0 (n ? 2,103) 1 (n ? 1,265)
?2 (n ? 807)
MDD, major depressive disorder.
a115 missing values.
Table 2. Past-year Prevalent Major Depressive Disorder (MDD) According to SLC6A4 (5-HTTLPR) Genotype and the
Number of Adult Events and Difficulties Experienced in the Past Five Years
Adverse Life Events or Long-Term Difficulties
0 (n ? 1,766) 1 (n ? 1,127) 2 (n ? 680) 3 (n ? 373)
?4 (n ? 229)
MDD, major depressive disorder.
a115 missing values
Table 3. Odds Ratios (OR) and 95% Confidence Intervals (CI) for the Main
Effects of SLC6A4 (5-HTTLPR) Genotype (G) and Social Adversity (E) and
Their Interaction (G?E) (Adjusted for Gender) on Past-year Prevalent
Major Depressive Disorder, Where Social Adversity Represented by A)
Childhood Adverse Experience and B) Adult Events or Difficulties in the
Past Five Years
OR95% CIOR 95% CI
G (per S allele)
E (per adverse exposure)
.90 (.73–1.11).96 (.84–1.09)
226 BIOL PSYCHIATRY 2006;59:224–229
P.G. Surtees et al
et al 2003; Figure 2). No GxE interactions were observed for
adverse experience in childhood and past-year prevalent MDD
among women or for adult adverse experience and past-year
prevalent MDD in either men or women. With Bonferroni
correction, according to the number of subgroup analyses per-
formed, the interaction for adverse experience in childhood and
past-year prevalent MDD in men was not significant (required p
value for nine tests, p ? .006).
This investigation was designed to attempt replication of a
recent report of interaction between the experience of social
adversity, the SLC6A4 (5-HTTLPR) genotype, and past-year prev-
alent MDD (Caspi et al 2003). Our results, on the basis of a
sample of 4175 study participants, did not support such a GxE
To our knowledge, five other human studies, to date, have
focused on the relationship between the SLC6A4 genotype,
social adversity, and mood-related phenotypes (Eley et al 2004;
Gillespie et al 2005; Grabe et al 2005; Kaufman et al 2004;
Kendler et al 2005). Of these studies, one (Gillespie et al 2005)
reported no evidence of replication, and four studies provided
evidence for replication, though on the basis of different catego-
rizations of genotype (full [Kaufman et al 2004], additive [Eley et
al 2004], recessive [Kendler et al 2005], and dominant [Grabe et al
2005]). In addition, two of these studies found a result only when
on the basis of a data subset (in both cases in women and not in
men [Eley et al 2004; Grabe et al 2005]). Care is required in
interpreting studies that only partially replicate an original study
finding, because multiple testing can distort the type I error rate.
Marginal evidence followed by partial replication does not
provide strong evidence of association (Colhoun et al 2003). In
addition, all of these replications were on the basis of small
samples, of more limited statistical power than the population
presented here. Our study carries approximately twice the
number of MDD cases as in the original report by Caspi et al
(2003); however, sample size requirements for interactions are
approximately four times as great as those required for the
detection of main effects of the same magnitude (Cooper 2003).
Therefore, the current study, as in the other studies cited, remains
underpowered to detect interactions other than those of large
magnitude (Hwang et al 1994; Smith and Day 1984).
Associations between genotype(s) and complex phenotypes
are likely to be of modest effect size and large sample sizes with,
perhaps, thousands of cases required to detect and confirm
associations at appropriate levels of statistical significance
(Keavney et al 2004; Pharoah et al 2004; Zondervan and Cardon
2004). Where large sample sizes have been unavailable, many
genetic association studies have employed liberal type I error
rates (e.g., 5%) on the basis that, as few variants are studied, the
need to correct for multiple comparisons is reduced (Freimer and
Sabatti 2004). These studies should be followed by definitive
replication (Thomas and Clayton 2004).
This study differs markedly both in design and methodology
from the report by Caspi et al (2003). These differences could
explain our non-replication of study findings. In particular,
participants in the original Dunedin Multidisciplinary Health and
Development Study were a representative birth cohort with
repeated follow-up assessments with the GxE analyses on the
basis of the cohort at age 26 years. In marked contrast, this study
was on the basis of a single retrospective questionnaire assess-
ment of a population cohort with age range 41–80. Therefore,
consideration needs to be given to the possibility that either the
GxE interaction is restricted to younger age groups (for example,
one recent partial replication was on the basis of a younger
cohort [Kendler et al 2005]) or that the strength of the interaction
varies with age and that the large and older age range in this
study might have obscured any potentially positive result.
In addition, the assessment of phenotype differed between
the two studies. Participants in the birth cohort study were
assessed with the Diagnostic Interview Schedule (Robins and
Regier 1991) at ages 18, 21, and 26 years. An important conse-
quence of these assessments is that most of the depressive
episodes identified in the study were likely to be the participants’
first such experience. In contrast, in the present study, a single
comprehensive structured self-assessment was employed to pro-
vide evidence of each participant’s history of MDD episodes
(Surtees et al 2000). These data suggest that, of those 298
participants who met one-year prevalent DSM-IV MDD criteria,
for only 9.4% was this their first episode, giving further emphasis
to the difference between this and the original birth cohort study.
In addition, clinical observations and epidemiological evidence
have grown to support a “kindling” hypothesis (Post 1992; Post
and Weiss 1998) that the strength of the association between
social adversity and the onset of MDD decays with progressively
increasing number of previous episodes (Kendler et al 2000).
Therefore, these observations combine to compound the differ-
ences between this study and the original report by Caspi et al
(2003). Our sample was enriched by participants meeting MDD
criteria through implementation of an extreme-phenotype design
to permit evaluation of an associated hypothesis (Willis-Owen, in
press). As expected, the 1-year prevalence of MDD was much
greater in the extreme high neuroticism group (13.2%) than in
the low neuroticism group (1.1%). It remains possible that the
absence of individuals with intermediate neuroticism scores
might have acted to obscure any GxE interaction.
Our measures of adverse exposure history, both in childhood
and adulthood, were reported at rates that closely matched those
reported in the original study; however, in the context of a
research program designed to investigate the association be-
tween diet and chronic disease incidence, the assessment of
sexual abuse was considered inappropriate. Availability of the
relative timings of measures of MDD onset, offset, and adverse
event exposure in adulthood also permitted inclusion of only
those events that occurred in the year before episode onset, and
importantly, permitted exclusion of event exposures that fol-
lowed episode onset. This refined measure of adverse exposure,
again, provided no evidence for GxE interaction in these data.
While population stratification could, theoretically, have im-
pacted on our findings, this is unlikely to be an important
consideration (Cardon and Palmer 2003), given the design
characteristics of the EPIC-Norfolk HLEQ study.
We acknowledge that study limitations might follow from the
collection of episodic MDD state and social adversity exposure
history data through self-report, through the retrospective recall
of early adverse experience and through the potential confound-
ing of current emotional state with recall of adverse experience.
In the EPIC-Norfolk HLEQ cohort (n ? 20,921; age range 41–80;
SD 9.3), the lifetime prevalence of MDD is 15.4% and annual
MDD prevalence is 5.2%. These MDD prevalence estimates (and
age-gender distributions) are comparable to those from inter-
view-based assessments from United Kingdom and international
psychiatric epidemiology studies (Surtees et al 2000). In addition,
analysis of EPIC-Norfolk HLEQ episodic mood state data has
revealed only a small amount of compression of reported MDD
P.G. Surtees et al
BIOL PSYCHIATRY 2006;59:224–229 227
episodes (clustering of episodes in the immediate pre-assess-
ment period) (Surtees et al 2000) and has aided elucidation of the
association between adverse event exposure history (including
those experienced in childhood) and the onset (and recurrence)
of depression (Wainwright and Surtees 2002a, 2002b).
The GxE evidence reported by Caspi et al (2003) represents
an exciting development in psychiatric genetic research. There-
fore, the finding has provoked considerable interest. The small
sample sizes of the original study and of the replication studies so
far reported could account for differences between findings. In
addition, the failure of the current study to replicate the finding
of Caspi et al could be due to differences in research design and
measures employed; however, the original finding was of mod-
est statistical significance, and the authors emphasized the re-
quirement for replication. Subsequent evidence has been,
largely, on the basis of studies that only partially replicate the
original finding. In addition, large multicenter case-control stud-
ies and meta-analyses suggest that if there is any direct associa-
tion between SLC6A4 genotype and (unipolar) depression, it
must be very small. Collectively, we believe that these consider-
ations suggest that evidence, to date, is not supportive of a GxE
interaction between the SLC6A4 genotype, social adversity, and
EPIC-Norfolk is supported by program grants from the Medical
Research Council UK (G9502233, G0300128) and Cancer Re-
search UK (C865/A2883), with additional support from the
European Union, Stroke Association, British Heart Foundation,
Department of Health, Food Standards Agency, and the Well-
We thank the study participants, the general practitioners
who took part in this study, and staff associated with the research
American Psychiatric Association (1994): Diagnostic and Statistical Man-
the 5-HT transporter alters emotional behavior in adult mice. Science
experience and serotonin transporter gene variation interact to influ-
ence primate CNS function. Mol Psychiatry 7:118–122.
Brown GW, Harris TO (1978): Social Origins of Depression: A Study of Psy-
chiatric Disorder in Women. London: Tavistock Publications.
Brown GW, Sklair F, Harris TO, Birley JLT (1973): Life-events and psychiatric
disorders. Part 1: Some methodological issues. Psychol Med 3:74–87.
Brugha T, Bebbington P, Tennant C, Hurry J (1985): The list of threatening
experiences - a subset of 12 life event categories with considerable
long-term contextual threat. Psychol Med 15:189–194.
and validity of a brief life events questionnaire. Acta Psychiatr Scand
Cardon LR, Palmer LJ (2003): Population stratification and spurious allelic
association. Lancet 361:598–604.
Caspi A, Sugden K, Moffitt TE, Taylor A, Craig IW, Harrington H, et al (2003):
the 5-HTT gene. Science 301:386–389.
Champoux M, Bennett A, Shannon C, Higley JD, Lesch KP, Suomi SJ
(2002): Serotonin transporter gene polymorphism, differential early
rearing, and behavior in rhesus monkey neonates. Mol Psychiatry
associations with complex outcomes. Lancet 361:865–872.
mon complex disease. Ann Intern Med 139:437–440.
Day N, Oakes S, Luben R, Khaw K-T, Bingham S, Welch A, et al (1999):
EPIC-Norfolk: Study design and characteristics of the cohort. Br J Cancer
Eley TC, Sugden K, Corsico A, Gregory AM, Sham P, McGuffin P, et al (2004):
Gene-environment interaction analysis of serotonin system markers
with adolescent depression. Mol Psychiatry 9:908–915.
Freimer N, Sabatti C (2004): The use of pedigree, sib-pair and association
studies of common diseases for genetic mapping and epidemiology.
Gillespie NA, Whitfield JB, Williams B, Heath AC, Martin NG (2005): The
relationship between stressful life events, the serotonin transporter
Grabe HJ, Lange M, Wolff B, Völzke H, Lucht M, Freyberger HJ, et al (2005):
Mental and physical distress is modulated by a polymorphism in the
5-HT transporter gene interacting with social stressors and chronic dis-
ease burden. Mol Psychiatry 10:220–224.
size estimation to detect gene-environment interaction in case-control
Kaufman J, Yang B-Z, Douglas-Palumberi H, Houshyar S, Lipschitz D, Krystal
JH, Gelernter J (2004): Social supports and serotonin transporter gene
moderate depression in maltreated children. Proc Natl Acad Sci U S A
Keavney B, Palmer A, Parish S, Clark S, Youngman L, Danesh J, et al (2004):
Lipid-related genes and myocardial infarction in 4685 cases and 3460
controls: Discrepancies between genotype, blood lipid concentrations,
and coronary disease risk. Int J Epidemiol 33:1002–1013.
Kendler KS, Karkowski L, Prescott CA (1998): Stressful life events and major
depression: Risk period, long-term contextual threat and diagnostic
stressful life events and a serotonin transporter polymorphism in the
prediction of episodes of major depression: A replication. Arch Gen Psy-
ous episodes in the etiology of major depression in women: An evalua-
tion of the “kindling” hypothesis. Am J Psychiatry 157:1243–1251.
Lasky-Su JA, Faraone SV, Glatt SJ, Tsuang MT (2005): Meta-analysis of the
association between two polymorphisms in the serotonin transporter
Lerman C, Shields PG, Audrain J, Main D, Cobb B, Boyd NR, et al (1998): The
role of the serotonin transporter gene in cigarette smoking. Cancer
Mendlewicz J, Massat I, Souery D, Del-Favero J, Oruc L, Nothen MM, et al
orders: No evidence of association in a large European multicenter
recurrent affective disorder. Am J Psychiatry 149:999–1010.
anxiety, and obsessive-compulsive disorders: The role of serotonergic
mechanisms in illness progression. Biol Psychiatry 44:193–206.
Robins LN, Regier DA, editors (1991): Psychiatric Disorders in America: The
of peptide from its own sequence and similar peptides from bacteria -
implications for spondyloarthropathies. Lancet 345:1542–1544.
confounding and interaction effects. Int J Epidemiol 13:356–365.
course and the onset of depression. In: Harris T, editor. Where Inner and
Outer Worlds Meet. Psychosocial Research in the Tradition of George W.
Brown. London: Routledge 171–194.
Surtees PG, Miller PM, Ingham JG, Kreitman NB, Rennie D, Sashidharan SP
(1986): Life events and the onset of affective disorder. A longitudinal
general population study. J Affect Disord 10:37–50.
Surtees PG, Wainwright NWJ (1998): Adversity over the life course: Assess-
ment and quantification issues. Stress Medicine 14:205–211.
228 BIOL PSYCHIATRY 2006;59:224–229
P.G. Surtees et al