A meta-analysis of the association between DRD4 polymorphism and novelty seeking.
ABSTRACT A meta-analytical review of 20 studies (n = 3907) of the association between DRD4 polymorphism and novelty seeking suggests the following conclusions: (a) on average, there is no association between DRD4 polymorphism and novelty seeking (average d = 0.06 with 95% CI of +/- 0.09), where 13 reports suggest that the presence of longer alleles is associated with higher novelty seeking scores and seven reports suggest the opposite; (b) there is a true heterogeneity among the studies (ie, unknown moderators do exist) but the strength of the association between DRD4 polymorphism and novelty seeking in the presence of any (unknown) moderator is likely to be weak; (c) search for moderators has not yielded any reliable explanation for the variability among studies. We propose that to find such moderators, theory-driven research for potential interaction, coupled with larger sample sizes should be employed. The growing availability of powerful statistical techniques, high-throughput genotyping and large numbers of polymorphic markers such as single nucleotide polymorphisms makes such proposed studies increasingly feasible.
- SourceAvailable from: János Török[Show abstract] [Hide abstract]
ABSTRACT: There is increasing evidence that the genetic architecture of exploration behavior includes the dopamine receptor D4 gene (DRD4). Such a link implies that the within-individual consistency in the same behavior has a genetic basis. Behavioral consistency is also prevalent in the form of between-individual correlation of functionally different behaviors; thus, the relationship between DRD4 polymorphism and exploration may also be manifested for other behaviors. Here, in a Hungarian population of the collared flycatcher, Ficedula albicollis, we investigate how males with distinct DRD4 genotypes differ in the consistent elements of their behavioral displays during the courtship period. In completely natural conditions, we assayed novelty avoidance, aggression and risk-taking, traits that were previously shown repeatable over time and correlate with each other, suggesting that they could have a common mechanistic basis. We identified two single-nucleotide polymorphisms (SNP554 and SNP764) in the exon 3 of the DRD4 gene by sequencing a subsample, then we screened 202 individuals of both sexes for these SNPs. Focusing on the genotypic variation in courting males, we found that "AC" heterozygote individuals at the SNP764 take lower risk than the most common "AA" homozygotes (the "CC" homozygotes were not represented in our subsample of males). We also found a considerable effect size for the relationship between SNP554 polymorphism and novelty avoidance. Therefore, in addition to exploration, DRD4 polymorphisms may also be associated with the regulation of behaviors that may incur fear or stress. Moreover, polymorphisms at the two SNPs were not independent indicating a potential role for genetic constraints or another functional link, which may partially explain behavioral correlations.Ecology and Evolution 04/2014; 4(8):1466-79. · 1.66 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: The aim of the study was to examine functional brain activity in response to unpleasant images in individuals with the 7-repeat (7R) allele compared to individuals with the 4-repeat (4R) allele of the dopamine receptor D4 (DRD4) gene (VNTR in exon 3). Based on the response ready hypothesis, individuals with the DRD4-4R/7R genotype were expected to show greater functional brain activity in response to unpleasant compared to neutral stimuli in specific regions of the frontal, temporal, parietal and limbic lobes, which form the networks involved in attentional, emotional, and preparatory responses. Functional Magnetic Resonance Imaging activity was studied in 26 young adults (13 with the DRD4-4R/7R genotype and 13 with the DRD4-4R/4R genotype). Participants were asked to look at and subjectively rate unpleasant and neutral images. Results showed increased brain activity in response to unpleasant images compared to neutral images in the right temporal lobe in participants with the DRD4-4R/7R genotype versus participants with the DRD4-4R/4R genotype. The increase in right temporal lobe activity in individuals with DRD4-4R/7R suggests greater involvement in processing negative emotional stimuli. Intriguingly, no differences were found between the two genotypes in the subjective ratings of the images. The findings corroborate the response ready hypothesis, which suggests that individuals with the 7R allele are more responsive to negative emotional stimuli compared to individuals with the 4R allele of the DRD4 gene.Psychiatry Research Neuroimaging 11/2014; · 2.83 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: A vast body of empirical evidence highlights the contribution of maternal sensitivity to multiple features of children's development. Long appreciated, too, is that parenting effects frequently vary across children, often moderated by characteristics of the child such as early temperament. Most such work has been guided by conceptions of vulnerability stemming from negative or difficult temperament that interacts with insensitive, harsh or otherwise unsupportive parenting to undermine child well-being. Belsky's (1997b; 2005) differential-susceptibility framework challenges such diathesis-stress thinking, highlighting the fact that the very individuals who seem most susceptible to environmental adversity––including negatively emotional infants, toddlers, and preschoolers––may also benefit the most from developmentally supportive rearing. In other words, some children may be more affected than others by both highly sensitive and insensitive parenting. Evidence consistent with this view is reviewed, including research on temperament-X-parenting and gene-X-environment interaction. Finally, potential implications of the differential susceptibility perspective regarding the understanding of parenting effects are discussed.
Molecular Psychiatry (2002) 7, 712–717
2002 Nature Publishing Group
All rights reserved 1359-4184/02 $25.00
ORIGINAL RESEARCH ARTICLE
A meta-analysis of the association between DRD4
polymorphism and novelty seeking
AN Kluger1, Z Siegfried2and RP Ebstein3
1School of Business Administration, The Hebrew University of Jerusalem, Jerusalem, Israel;2Department of Human
Nutrition and Metabolism, Faculty of Medicine, The Hebrew University of Jerusalem, Jerusalem, Israel;3Research
Laboratory, S Herzog Memorial Hospital, Jerusalem, Israel
A meta-analytical review of 20 studies (n = 3907) of the association between DRD4 polymor-
phism and novelty seeking suggests the following conclusions: (a) on average, there is no
association between DRD4 polymorphism and novelty seeking (average d = 0.06 with 95% CI
of ± 0.09), where 13 reports suggest that the presence of longer alleles is associated with
higher novelty seeking scores and seven reports suggest the opposite; (b) there is a true
heterogeneity among the studies (ie, unknown moderators do exist) but the strength of the
association between DRD4 polymorphism and novelty seeking in the presence of any
(unknown) moderator is likely to be weak; (c) search for moderators has not yielded any
reliable explanation for the variability among studies. We propose that to find such moder-
ators, theory-driven research for potential interaction, coupled with larger sample sizes should
be employed. The growing availability of powerful statistical techniques, high-throughput gen-
otyping and large numbers of polymorphic markers such as single nucleotide polymorphisms
makes such proposed studies increasingly feasible.
Molecular Psychiatry (2002) 7, 712–717. doi:10.1038/sj.mp.4001082
Keywords: meta-analysis; personality; polymorphism (genetics); dopamine D4 receptor; novelty
Evidence based on twin and adoption studies suggests
that human personality traits are partially heritable.1
However, the specific genes responsible for trait varia-
bility are not known. Human personality is usually
assessed using self-report questionnaires and one such
questionnaire is the tridimensional personality ques-
tionnaire or TPQ designed by Cloninger2which has
been widely used in genetic studies. The TPQ is based
on a biosocial theory of personality that draws on
human and animal work to suggest that behavior is
mediated by certain neurotransmitters which underlie
three basic and largely heritable dimensions, called
‘Novelty Seeking (NS),’ ‘Harm Avoidance (HA)’ and
‘Reward Dependence (RD).’ Cloninger suggested that
the trait of NS is mediated by genetic variability in
dopamine transmission whereas HA is mediated by
serotonergic transmission and RD by noradrenergic
transmission. Research undertaken to test this hypoth-
esis identified the dopamine D4 receptor gene (DRD4)
exon III polymorphism3,4as a candidate gene that may
be associated with the NS trait.5,6This polymorphism
is an unusually variable repeat region in the third cyto-
plasmic loop of the receptor coding for 16 amino acids.
Correspondence: AN Kluger, School of Business Administration,
The Hebrew University, Mt Scopus, Jerusalem 91905, Israel.
Received 12 November 2001; revised 11 January 2002; accepted
16 January 2002
The two most common repeats in most populations are
the 4 and 7. However, reports concerning the associ-
ation between the DRD4 exon III polymorphism and
the personality trait of NS have yielded an inconsistent
pattern. Whereas initial reports suggested that long
alleles (mostly representing the 7 repeat since the 6 and
8 repeats are relatively rare) are associated with higher
scores on the personality scale of NS, others have sug-
gested that either the two are not associated or even
the short alleles (usually 4 repeats) are associated with
higher NS scores. This inconsistent pattern has been
noted in several review articles which offered a narra-
tive review of the pertinent reports.7,8Yet, these narra-
tive reviews were inconclusive as well.
An alternative to the narrative review is a meta-
analysis that integrates all the available data numeri-
cally. A meta-analysis has three advantages over the
narrative review. First, it can assess the effect (the
strength of the association) in the population more
efficiently than any of the original reports by utilizing
the increase in sample size. Second, it can assess the
variability of the effect in the population to determine
whether inconsistent reports reflect mere statistical
fluctuation (ie, no two reports are likely to yield ident-
ical results even if in the population the effect is con-
stant, merely due to random sampling errors). Third, if
it is determined that some of the between-studies
inconsistency is not random, then one can investigate
the potential moderators (or modulating factors) of the
DRD4-novelty seeking meta-analysis
AN Kluger et al
association under investigation. However, meta-ana-
lytic techniques have their own methodological limi-
tations—including a potential publication bias against
negative or non-significant results,9mixing different
types of variables as if they were measuring the same
construct,10and statistical biases in the meta-analysis
estimates.11Thus, our goal in this paper is to review
the literature regarding the link between the DRD4
exon III polymorphism and NS utilizing the advantages
of the meta-analysis, while safeguarding as much as
possible against the potential limitations of the meta-
To reap the maximum benefit from a meta-analysis,
as many reports with similar methodology as possible
should be reviewed. In the pertinent literature regard-
ing the DRD4 polymorphism and NS, several types of
research designs were used both with regards to the
grouping of the genetic polymorphism and with
regards to the specific instrument used for personality
assessment. Comparison between genetic polymor-
phisms is sometimes assessed by comparing most com-
mon specific genotypes (eg, 4,4 vs 4,7) and at times by
using a ‘short’ vs ‘long’ classification (2–5 vs 6–8
repeats). Since most studies used the short vs long
scheme we used this classification scheme as well.
Most of the literature has used Cloninger’s TPQ instru-
ment and only a few used the NEO-PI12–14and the Kar-
olinska Scale of Personality.15Therefore, to ensure
homogeneity (to avoid the problem of ‘mixing apple
and oranges’10) we restricted our analysis only to those
investigations employing the TPQ. The TPQ also meas-
ures other traits: HA and RD. If the DRD4 polymor-
phism has a unique association with NS, one would
expect no association with the other traits. Hence,
wherever reported we also analyzed the association
between the presence of long vs short alleles with HA
We conducted a computerized search (Medline) with
the following search criteria: ‘(Personality or Novelty
Seeking or Extroversion)
morphism)’. In addition, we used back references from
identified papers to find as many pertinent papers as
possible. We found a total of 19 relevant pap-
ers5,6,13,14,16–30(Table 1) yielding 20 different studies.
For each report we calculated the strength of the
association. Given that the association is calculated
between a dichotomous variable (long vs short allele)
and a continuous variable (NS), the appropriate index
of effect size is the statistics d that measures the stan-
dardized difference in mean NS scores between people
who have and people who do not have the long allele.
That is, in a given study a d of 1 means that the mean
NS of those who have the long allele was one standard
deviation higher on NS than the mean of those who do
not have the long allele. A d of zero means that in a
given study there was absolutely no mean difference
in NS scores between those who carry the long allele
and those who do not. Finally, a d of −0.5, for example,
between the presence of the DRD4 exon III long allele and
Studies included in a meta-analysis of association
Benjamin et al (1996)b
Ebstein et al (1996)
Malhotra et al (1996)
Malhotra et al (1996)
Ebstein et al (1997)
Gelernter et al (1997)
Ono et al (1997)
Sanders et al (1997)
Vadenbergh et al
Noble et al (1998)
Pogue-Geile et al (1998)
Bau et al (1999)
Ekelund et al (1999)
Kuhn et al (1999)
Strobel et al (1999)
Tomitaka et al (1999)d
Benjamin et al (2000)c
Gebhardt et al (2000)
Herbst et al (2000)
Swift et al (2000)
2–5 vs 6–8
p/a = present/absent.
aNS is estimated from a larger sample.
bCorrected personality for age, sex, & ethnicity.
cCorrected personality for age & sex.
dThere is a discrepancy in the d calculated on the basis of
the summary data (means or standard deviations) and the t
value in the paper; the d in this table is based on means and
standard deviation but the d calculated on the basis t-score
suggests d = 0.62.
would mean that the mean NS of the carrier of the long
allele was lower by half a standard deviation than the
mean NS score of those who do not carry a long allele.
Note that d can be converted to a correlation coefficient
(Pearson’s r) but the choice of meta-analysts is to work
with the unit of measurement that is characteristic of
There are two slightly different meta-analytical
approaches to combine research data. Whereas typi-
cally the conclusions produced by the various methods
are similar, in cases where the effect sizes are small
(the association exists but is weak), the methods can
diverge in their conclusions.11Of course, one’s trust in
the meta-analytic results will be enhanced if it is
shown that the conclusion is independent of the
methods. Thus, we have applied two methods that can
potentially differ in their conclusions. Specifically, we
used both the methods of Hedges and Olkin31and those
of Hunter and Schmidt.10The differences in estimates
are explained in the Results section.
For each study we computed d (Hunter & Schmidt’s
method) and d+ (Hedges & Olkin’s method) from the
DRD4-novelty seeking meta-analysis
AN Kluger et al
reported means and standard deviations. Whenever a
report did not provide means and standard deviations,
but provided a t-test score, we computed d directly
from t. Whenever a report provided means for sub-
groups of either people or sub-scales, we first averaged
the means and variances of NS across the sub groups,
or sub scales, to derive mean and standard deviation
NS for those with and those without the long allele.
We performed this aggregation with one exception
because it provided two large samples; one for
alcoholics and one for non-alcoholics (see coding
below). When a study dichotomized NS to high/low
score, we treated the proportion differences as mean
differences by calculating standard deviation for pro-
Finally, we coded for each study the following
demographic variables that according to the literature
could potentially moderate the association: mean age
of the sample, proportion of males in the sample,
whether the sample was based on alcoholics (1) or non-
alcoholic population (0), and whether the sample was
from continental Europe (1) or not (0). The later vari-
able was chosen because most of the negative reports
came from continental Europe.
Prior to performing the meta-analysis we explored the
distribution of the literature with a stem-and-leaf dis-
play (Table 2). The results in the literature appear to
be normally distributed without noticeable outliers
that can bias the meta-analysis estimates. This table
helps to obtain a bird’s eye view that suggests that the
entire literature is distributed between moderate-nega-
tive effects to moderate-positive effects (d ? ?0.2? is
considered weak, ?0.2? ? d ? ?0.5? is moderate and d ?
?0.8? is strong32).
The distribution of the results in this literature
(Table 2) suggests that this literature does not suffer
from a serious publication bias because many reports
were published despite reporting either negative
effects or non-significant effects. This observation was
further explored with a funnel plot9with d values on
ation between presence of long DRD4 allele and NS
Stem-and-leaf display of raw d values of the associ-
This display shows all the raw data (d values). For example,
the d value of Strobel et al (1999) is 0.63 and is represented
as 3 under the stem of 0.6, and the d value of Ebstein et al
(1996) is 0.58 and it is represented as 8 under the stem of 0.4
(that includes values between 0.40 and 0.59).
one axis and standard errors on the other axis. As can
be expected from the almost normal distribution
shown in Table 2, the funnel plot was largely sym-
metrical ruling out a serious publication bias. This
increases the confidence one can place in the meta-ana-
lytical results presented next.
Both meta-analytical methods suggested similar con-
clusions (Table 3). The mean difference in NS between
those who carry the DRD4 long allele and those who
do not carry it is negligible (about 1/20 of a standard
deviation). The confidence interval for the mean
includes zero (eg, the mean effect in the literature is not
statistically significantly different from zero). However,
both methods indicate that whereas the mean effect is
close to zero, the differences between studies are due
in part to true heterogeneity, ie, different studies esti-
mating different ‘true’ underlying association para-
meters. Specifically, some studies find positive associ-
ation and some studies find negative association and
this divergence in results is not likely to be random
(the Q-statistic is highly significant for the Hedges &
Olkin method and according to the Hunter & Schmidt’s
method only 24.8% of the variability in results across
studies could be attributed to random sampling error).
Hence both methods strongly indicate that there are
unknown moderators of the association between DRD4
polymorphism and NS.
Given the statistical evidence for the presence of
moderators, we calculated the weighted correlation
between the coded variables and d (Table 4). The
weights for the correlations are based on the sample
size of each effect.31using the WEIGHT option of the
readily available PROC CORR procedure in SAS.
All the demographic variables showed non-signifi-
cant trends (weighted r’s) of attenuating the results
(Table 4). The strongest trend is for samples from Con-
tinental Europe, which are most likely to yield negative
association between the long allele and NS. Similarly,
the older the mean age of the sample the weaker, or
even negative, were the associations between the long
allele and NS. The other potential moderators yielded
non-significant and extremely small correlations.
Our results are consistent with a recent trend found in
36 meta-analyses that led to the conclusion that ‘The
first study often suggests a stronger genetic effect than
is found by subsequent studies’ (p 1).33Indeed, the
mean association between the DRD4 long allele, on the
basis of data aggregated over almost 4000 individuals,
is negligible and statistically not significant. This con-
clusion is strengthened by noting that the association
between DRD4 polymorphism and NS is similar in
magnitude (Table 3) to the association between DRD4
polymorphism and other traits (HA, RD; −0.14, 0.10,
respectively). These were not the primary targets of
investigation and yet yielded similar and numerically
stronger effects. These could be considered as control
variables that should show weaker effects than the one
found for NS. Thus the pattern of results suggests, con-
DRD4-novelty seeking meta-analysis
AN Kluger et al
Meta-analytic estimates of the mean association between the presence of the DRD4 exon III long allele and person-
Hedges & Olkin’s methodHunter & Schmidt’s method
Traitnkd LCLUCLQPd LCL UCL% EV
n, Number of people across all studies; k, number of samples; LCL, 95% lower confidence limit; UCL, 95% upper confidence
limit; % EV, percent of variance among samples that can be attributed to random sampling error. When % EV is estimated to
be 100% it is assumed that all variance between studies is random and thus no variability in true d is expected, hence LCL
and UCL, for true d, are identical to d. Tomitaka et al (1999) mentioned data for HA and RD in their paper but did not report
it there. Dr Tomitaka generously provided us with these data. Q, Q-statistics of homogeneity; P, probability of observing Q
under the null hypothesis of homogeneity.
characteristics and strength of the association between DRD4
polymorphism and NS
Moderator analysis of the link between sample
Mean age of sample
Sample from Continental Europe
% Male in sample
Sample of alcoholics
k, Number of samples; r, sample weighted correlation coef-
trary to initial reports, and consistent with recent
reviews8that there is no simple association between
DRD4 polymorphism and NS.
However, the variability in the direction and magni-
tude of the associations between DRD4 polymorphism
and NS found in the literature indicates that there are
as yet unknown causes for finding weak to moderate
positive effects in some studies, null findings in some
other studies, and weak negative effects in yet other
studies. Some of this variability was estimated to be
due to true heterogeneity between studies (ie, the true
association between DRD4 polymorphism and NS var-
ies in the population due to unknown causes) but the
range in which the association is likely to vary is
mostly in the range of small effects (d ? ?0.2?). Thus,
the reasons for this variability may be very difficult
An important question is not whether an effect
exists, because both weak positive and weak negative
associations are likely to exist in the population, but
rather whether the range of likely effects estimated in
the current meta-analysis is meaningful. Specifically,
the statistically true variability falls in a range that sug-
gests that most true effects are very weak. For example,
even if we assume, as the data suggest, that in some
conditions the link between presence of long alleles
and NS is about d = 0.16 (Hunter & Schmidt method,
Table 3), or that the correlation coefficient between
presence of long alleles and NS is 0.08, it is question-
able whether practically these true effects justify a
search for moderators. Detecting moderators of such a
small effect size statistically requires very large data
sets. However, note that weak statistical effects could
be important but that the judgement is based not on
statistics but on substantive knowledge of meaning of
the effect.34For example, if one translates findings
regarding the association between smoking (yes/no)
and occurrence of a heart attack (yes/no), the d stat-
istics will be roughly 0.14. However, it corresponds to
an odds ratio of 6:1, where 6% of smokers are expected
to suffer from heart attack as opposed to 1% of non-
smokers.34In that case, the very small statistical effect
size (most people whether smokers or not will not suf-
fer heart attack) is very meaningful. The meaning-
fulness is independent of effect size. Thus, researchers
in the field have to assess whether or not variable small
true association between DRD4 and NS is theoretically
meaningful. This may become clearer, perhaps, with
the identification of other genetic correlates of NS.
We, nevertheless, attempted to identify potential
moderators in the data set analyzed in the current
meta-analysis. Four variables that might be contribu-
ting to differences in reports and that were identifiable
in the published studies included the mean age of sam-
ple, ethnic groups (European vs others), gender and
clinical characteristics of the sample (alcoholic vs non-
alcoholic). None of the four moderators we analyzed
proved significant although there was a trend for age
and ethnicity to affect the purported association
between the DRD4 polymorphism and NS. Given the
small number of studies in this meta-analysis (k = 20),
the moderator analysed results must be interpreted
with great caution due to second order sampling
error.10,35That is, unlike the estimate of the association
between DRD4 and NS that is based on approximately
4000 observations (high statistical power), the esti-
mates of the moderators are based on 20 samples
(extremely low statistical power).
One potential concern with this meta-analysis is the
comparison of the long vs short allele. First, there is
evidence of sequence variation within particular repeat
lengths that may also play a role in receptor func-
DRD4-novelty seeking meta-analysis
AN Kluger et al
tioning. At least 35 distinct repeat variant motifs were
observed for the DRD4 gene.36The two most common
haplotypes for the 4 and 7 repeat alleles respectively
account for 95% and 89% of the observed sequence
variants.36Due to the infrequence of the additional
haplotypes, it seems to us unlikely that they influenced
our estimates of the association between NS and allelic
type. However, this possibility cannot be excluded and
one particular sequence variant of the 7 or 4 repeat
allele may be responsible for the effect of one or both
of these alleles on NS. Although no information is cur-
rently available concerning possible functional differ-
ences among variant repeats of the same length, it
might be worthwhile to address this issue in future
genetic and physiological studies.
Second, on the basis of a worldwide sample it is
argued that the 7 repeat allele is evolutionarily younger
than the common 4 repeat allele. Furthermore, on the
basis of this report, it is suggested that the 7 repeat
allele is not simply related to other common alleles and
that it has increased to high frequency in humans by
positive selection.36Thus, the reports that collapse sev-
eral alleles into one category of ‘long’ may be missing
important information. However, in our meta-analysis
15 out of the 20 samples compared the specific effect
of 7 repeat presence/absence on NS and yielded practi-
cally the same results (eg, mean d of these 15 studies
is 0.01 and the heterogeneity estimates remain largely
unchanged also). Nevertheless, future research regard-
ing the DRD4 gene should consider these recent find-
ings regarding the 7 repeat allele.
The failure to find a moderator is not surprising
given the combination of small number of studies (k)
and small range of effects (all d ? ?1?). To discover
unknown moderators, future studies should search for
an interaction between the presence of the long allele
and other factors in affecting NS. These factors may be
either genetic (eg, allelic or locus heterogeneity),
environmental (eg, meaning of NS behavior in different
cultures), or measurement related (eg, NS response
may not be equivalent across cultures and languages).
Perhaps a more fruitful approach towards identifying
factors that truly contribute to the variability in associ-
ation studies between DRD4 and NS would be based
on the highly polymorphic nature of this gene. At least
10 polymorphisms (outside of the repeat region) of the
DRD4 receptor have been described to date and nine
are discussed in a recent review by Paterson et al.8The
most recently reported polymorphisms are a group of
SNPs (single nucleotide polymorphisms) in the pro-
moter region, one of which (−521 C/T) is observed to
regulate transcription of the receptor.37This plethora
of DRD4 receptor region polymorphisms raises the
possibility that the reported association between the
exon III polymorphism and NS is due to linkage dis-
equilibrium between the long repeat and another one
or more of the reported polymorphisms at this locus.
Allelic heterogeneity can create situations in which
multiple different alleles are associated with NS, rather
than a single specific allele. Studying haplotypes can
to some degree overcome hypothetical weak linkage
equilibrium between the 7 allele and a true functional
polymorphism.38For example, a haplotype approach
has been used to further analyze the role of the DRD4
receptor in attention deficit disorder. Indeed, such an
analysis strengthens the connection between the gene
and the disorder although the relationship between
DRD4 and attention deficit is not simple.39
Additionally, in the case of personality traits locus
heterogeneity is also likely to be present and many
genes dispersed in the genome may be partially con-
tributing to NS. For example, Cloninger et al40in a
linkage study of anxiety-related traits has identified a
number of epistatic loci on different chromosomes. We
have provisionally identified two additional functional
polymorphisms that may be contributing to NS.41It
should also be noted that in Chinese and Japanese
populations the long repeat alleles, including the 7
repeat, are extremely rare and cannot be contributing
to NS. Indeed, in the Japanese population an associ-
ation has been reported between NS and one of the pro-
moter region polymorphisms.37
Human personality is a complex phenotype in which
both genetic and environmental factors play a role.
Moreover, similar to other complex traits it is likely
that a number of genes, each yielding a small effect
size, contribute to the phenotype and any one polymor-
phism may neither be necessary nor sufficient to deter-
mine the trait. Since different investigators draw on
diverse populations in studies of personality, the gen-
etic background of these populations likely varies for
founding the identification of any specific genes of
small effect size that contributes to personality traits.
Validation of a role of polymorphisms in contributing
to a complex trait may well depend on knowledge of
additional genes contributing to this trait, and when
studies are carried out across ethnic groups, on the fre-
quency of these other polymorphisms across popu-
lations. The growing availability of powerful statistical
techniques, high-throughput genotyping and large
numbers of polymorphic markers such as single nucle-
otide polymorphisms makes such proposed studies
This research was supported by the Recanati Fund of
the School of Business Administration at the Hebrew
University (ANK) and a grant from the Israel Science
Foundation founded by the Israel Academy of Sciences
and Humanities (RPE). The authors wish to thank EM
Berry and two anonymous reviewers for their useful
comments on an earlier draft of this paper.
1 Loehlin JC. Genes and Environment in Personality Development.
Sage Publications: Newbury Park, 1992.
2 Cloninger CR. A systematic method for clinical description and
classification of personality variants: a proposal. Arch Gen Psy-
chiatry 1987; 44: 573–588.
3 Van Tol HH, Wu CM, Guan HC, Ohara K, Bunzow JR, Civelli O et
DRD4-novelty seeking meta-analysis
AN Kluger et al
al. Multiple dopamine D4 receptor variants in the human popu-
lation. Nature 1992; 358: 149–152.
4 Lichter JB, Barr CL, Kennedy JL, Van Tol HH, Kidd KK, Livak KJ. A
hypervariable segment in the human dopamine receptor D4 (DRD4)
gene. Hum Mol Genet 1993; 2: 767–773.
5 Ebstein RP, Nemanov L, Klotz I, Gritsenko I, Belmaker RH.
Additional evidence for an association between the dopamine D4
receptor (D4DR) exon III repeat polymorphism and the human per-
sonality trait of Novelty Seeking. Mol Psychiatry 1997; 2: 472–477.
6 Ebstein RP, Novick O, Umansky R, Priel B, Osher Y, Blaine D et
al. Dopamine D4 receptor (D4DR) exon III polymorphism associa-
ted with the human personality trait of novelty seeking. Nature
Genet 1996; 12: 78–80.
7 Lusher JM, Chandler C, Ball D. Dopamine D4 receptor gene (DRD4)
is associated with Novelty Seeking (NS) and substance abuse: the
saga continues. Mol Psychiatry 2001; 6: 497–499.
8 Paterson AD, Sunohara GA, Kennedy JL. Dopamine D4 receptor
gene: novelty or nonsense? Neuropsychopharmacology 1999; 21:
9 Sterne JAC, Egger M. Funnel plots for detecting bias in meta-analy-
sis: guidelines on choice of axis. J Clin Epidemiol 2001; 54:
10 Hunter JE, Schmidt FL. Methods of Meta-Analysis: Correcting Error
and Bias in Research Findings. Sage Publications: Newbury Park,
11 Field AP. Meta-analysis of correlation coefficients: a Monte Carlo
comparison of fixed- and random-effects methods. Psychol Meth
2001; 6: 161–180.
12 Persson ML, Wasserman D, Geijer T, Frisch A, Rockah R, Michae-
lovsky E et al. Dopamine D4 receptor gene polymorphism and per-
sonality traits in healthy volunteers. Eur Arch Psychiatry Clin Neu-
rosci 2000; 250: 203–206.
13 Pogue-Geile M, Ferrell R, Deka R, Debski T, Manuck S. Human
novelty-seeking personality traits and dopamine D4 receptor poly-
morphisms: a twin and genetic association study. Am J Med Genet
1998; 81: 44–48.
14 Strobel A, Wehr A, Michel A, Brocke B. Association between the
dopamine D4 receptor (DRD4) exon III polymorphism and meas-
ures of Novelty Seeking in a German population. Mol Psychiatry
1999; 4: 378–384.
15 Jonsson EG, Nothen MM, Gustavsson JP, Neidt H, Forslund K, Mat-
tila-Evenden M et al. Lack of association between dopamine D4
receptor gene and personality traits. Psychol Med 1998; 28: 985–
16 Sander T, Harms H, Dufeu P, Kuhn S, Rommelspacher H, Schmidt
LG. Dopamine D4 receptor exon III alleles and variation of novelty
seeking in alcoholics. Am J Med Genet 1997; 74: 483–487.
17 Bau CHD, Roman T, Almeida S, Hutz MH. Dopamine D4 receptor
gene and personality dimensions in Brazilian male alcoholics. Psy-
chiatric Genet 1999; 9: 139–143.
18 Benjamin J, Osher Y, Kotler M, Gritsenko I, Nemanov L, Belmaker
RH et al. TPQ personality trait of novelty seeking associated with
the dopamine receptor D4 exon III (DRD4), serotonin transporter
promoter region (5-HTTLPR) and catechol O-methyl transferase
(COMT val → met) polymorphisms. Mol Psychiatry 1999; 4: S23–
19 Ekelund J, Lichtermann D, Jarvelin MR, Peltonen L. Association
between novelty seeking and the type 4 dopamine receptor gene
in a large Finnish cohort sample. Am J Psychiatry 1999; 156:
20 Gelernter J, Kranzler H, Coccaro E, Siever L, New A, Mulgrew CL.
D4 dopamine-receptor (DRD4) alleles and novelty seeking in sub-
stance-dependent, personality-disorder, and control subjects. Am J
Hum Genet 1997; 61: 1144–1152.
21 Noble EP, Ozkaragoz TZ, Ritchie TL, Zhang XX, Belin TR, Sparkes
RS. D-2 and D-4 dopamine receptor polymorphisms and person-
ality. Am J Med Genet 1998; 81: 257–267.
22 Ono Y, Manki H, Yoshimura K, Muramatsu T, Mizushima H, Higu-
chi S et al. Association between dopamine D4 receptor (D4DR)
exon III polymorphism and novelty seeking in Japanese subjects.
Am J Med Genet 1997; 74: 501–503.
23 Parsian A, Zhang ZH. Human chromosomes 11p15 and 4p12 and
alcohol dependence: possible association with the GABRB1 gene.
Am J Med Genet 1999; 88: 533–538.
24 Tomitaka M, Tomitaka S, Otuka Y, Kim K, Matuki H, Sakamoto K
et al. Association between novelty seeking and dopamine receptor
D4 (DRD4) exon III polymorphism in Japanese subjects. Am J Med
Genet 1999; 88: 469–471.
25 Kuhn KU, Meyer K, Nothen MM, Gansicke M, Papassotiropoulos
A, Maier W. Allelic variants of dopamine receptor D4 (DRD4) and
serotonin receptor 5HT2c (HTR2c) and temperament factors: repli-
cation tests. Am J Med Genet 1999; 88: 168–172.
26 Gebhardt C, Leisch F, Schussler P, Fuchs K, Stompe T, Sieghart W
et al. Non-association of dopamine D4 and D2 receptor genes with
personality in healthy individuals. Psychiatr Genet 2000; 10:
27 Herbst JH, Zonderman AB, McCrae RR, Costa PT. Do the dimen-
sions of the temperament and character inventory map a simple
genetic architecture? Evidence from molecular genetics and factor
analysis. Am J Psychiatry 2000; 157: 1285–1290.
28 Swift G, Larsen B, Hawi Z, Gill M. Novelty seeking traits and D4
dopamine receptors. Am J Med Genet 2000; 96: 222–223.
29 Vandenbergh DJ, Zonderman AB, Wang J, Uhl GR, Costa PT. No
association between novelty seeking and dopamine D4 receptor
(D4DR) exon III seven repeat alleles in Baltimore Longitudinal
Study of Aging participants. Mol Psychiatry 1997; 2: 417–419.
30 Malhotra AK, Virkkunen M, Rooney W, Eggert M, Linnoila M,
Goldman D. The association between the dopamine D-4 receptor
(D4DR) 16 amino acid repeat polymorphism and Novelty Seeking.
Mol Psychiatry 1996; 1: 388–391.
31 Hedges L, Olkin I. Statistical Methods for Meta Analysis. 1985;
Academic Press: New York.
32 Cohen J. Statistical Power Analysis for the Behavioral Sciences.
Erlbaum Associates: Hillsdale, NJ, 1988.
33 Ioannidis JPA, Ntzani EE, Trikalinos TA, Contopoulos-Ioannidis
DG. Replication validity of genetic association studies. Nature
Genet 2001; 29: 306–309.
34 Cohen J. The earth is round (p?.05). Am Psychol 1994; 49: 997–
35 Kluger AN, DeNisi A. The effects of feedback interventions on per-
formance: a historical review, a meta-analysis, and a preliminary
feedback intervention theory. Psychol Bull 1996; 119: 254–284.
36 Ding Y-C, Chi H-C, Grady DL, Morishima A, Kidd JR, Kidd KK et
al. Evidence of positive selection acting at the human dopamine
receptor D4 gene locus. PNAS 2002; 99: 309–314.
37 Okuyama Y, Ishiguro H, Nankai M, Shibuya H, Watanabe A, Arin-
ami T. Identification of a polymorphism in the promoter region of
DRD4 associated with the human novelty seeking personality trait.
Mol Psychiatry 2000; 5: 64–69.
38 Schork NJ, Fallin D, Thiel B, Xu X, Broeckel U, Jacob HJ et al. The
future of genetic case-control studies. Adv Genet 2001; 42: 191–
39 Barr CL, Wigg KG, Wu J, Zai C, Bloom S, Tannock R et al. Linkage
study of two polymorphisms at the dopamine D3 receptor gene and
attention-deficit hyperactivity disorder. Am J Med Genet 2000; 96:
40 Cloninger CR. A unified biosocial theory of personality and its role
in the development of anxiety states: a reply to commentaries. Psy-
chiatr Dev 1988; 6: 83–120.
41 Benjamin J, Osher Y, Kotler M, Gritsenko I, Nemanov L, Belmaker
RH et al. Association between tridimensional personality question-
naire (TPQ) traits and three functional polymorphisms: dopamine
receptor D4 (DRD4), serotonin transporter promoter region (5-
HTTLPR) and catechol O-methyltransferase (COMT). Mol Psy-
chiatry 2000; 5: 96–100.