Progress in Neuro-Psychopharmacology & Biological Psychiatry 46 (2013) 36–42

Contents lists available at SciVerse ScienceDirect

Progress in Neuro-Psychopharmacology & Biological

Psychiatry
journal homepage: www.elsevier.com/locate/pnp

Genetic association studies between SNPs and suicidal behavior:

A meta-analytical field synopsis
Anne H.E. Schild ⁎, Jakob Pietschnig, Ulrich S. Tran, Martin Voracek
University of Vienna, School of Psychology, Department of Basic Psychological Research and Research Methods, Liebiggasse 5, 1010 Vienna, Austria

a r t i c l e i n f o a b s t r a c t

Article history: The large number of published meta-analyses on the associations between single-nucleotide polymorphisms
Received 15 April 2013 (SNPs) and suicidal behavior mirrors the enormous research interest in this topic. Although meta-analytic ev-
Received in revised form 20 June 2013 idence is abundant and certain patterns are apparent, those have not been integrated into a general frame-
Accepted 24 June 2013
work as of yet. In a systematic review, genetic association studies between SNPs and suicidal behavior
Available online 2 July 2013
were identified. Previously published meta-analyses for eight SNPs were updated and the results of the dif-
Keywords:
ferent meta-analyses were compared. Meta-analyses for 15 SNPs, which had not been subjected to meta-
Association study analysis before, were conducted. The present meta-analytical field synopsis showed five major similarities
Genetics of suicide between new and published analyses: 1) Summary effect sizes were small and rarely statistically significant,
Meta-analysis 2) heterogeneity between studies was often substantial, 3) there were no time trends, 4) effects were easily
Single nucleotide polymorphism swayed and were largely dependent on individual studies, and 5) publication bias does not play a role in this
Suicidal behavior field of research. Meta-analytic data show once more that major contributions of single genes are unlikely.
Systematic review However, association studies and corresponding meta-analyses have been an important and necessary stepping
stone in the development of modern and more complex approaches in the genetics of suicidal behavior.
© 2013 Elsevier Inc. All rights reserved.

1. Introduction
Family studies (Brent and Mann, 2005), adoption studies (Voracek,
2007), and twin studies (Voracek and Loibl, 2007) have kindled a
large research interest in the genetics of suicidal behavior. The first ge-
netic association study in this field was published nearly 20 years ago
(Nielsen et al., 1994) and since then a large number of association stud-
ies have been conducted.
The serotonergic system has been the main focus of the majority of
these studies (Antypa et al., 2013), although a variety of neurotransmit-
ters have been implicated in suicidal behavior (see Ernst et al., 2009,
for a review). Its apparent implication in the pathogenesis of suicide-
related phenotypes, the positive effects of drugs targeting the seroto-
nergic system, and altered 5-HIAA levels (5-Hydroxyindoleacetic acid,
the main metabolite of serotonin) have made it a prime candidate in
the search for the genetic underpinnings of suicide (Ernst et al., 2009).
Published meta-analyses on SNPs implicated in the serotonergic
system have covered the serotonin synthesis (e.g., TPH1: Bellivier
et al., 2004; Lalovic and Turecki, 2002; Li and He, 2006; Rujescu
Abbreviations: SNP, Single nucleotide polymorphism; TPH, Tryptophan hydroxylase;
5-HT, 5-hydroxytryptamine (serotonin); 5-HIAA, 5-Hydroxyindoleacetic acid; 5-HTR1A,
5-HT1A receptor; 5-HTT, Serotonin transporter; 5-HTTLPR, Serotonin-transporter-linked
polymorphic region; GWAS, Genome-wide association study.
⁎ Corresponding author. Tel.: +43 1 4277 471 24; fax: +43 1 4277 9 471.
E-mail addresses: anne.schild@univie.ac.at (A.H.E. Schild), J.Pietschnig@mdx.ac
(J. Pietschnig), ulrich.tran@univie.ac.at (U.S. Tran), martin.voracek@univie.ac.at
(M. Voracek).
et al., 2003), reception (e.g., 5-HTR1A: Anguelova et al., 2003; Li and
He, 2006), and re-uptake (e.g., 5-HTT: Anguelova et al., 2003; Li and
He, 2007; Lin and Tsai, 2004). Close scrutiny of those meta-analyses
reveals certain patterns and similarities which may be generalized
to other neurotransmitter systems which have received far less atten-
tion (e.g., the dopaminergic system).
In the last decade, this large research interest has culminated in the
publication of a considerable number of meta-analyses investigating
associations between single-nucleotide polymorphisms (SNPs) and
suicidal behavior. While it is clear that suicidal behavior has a consid-
erable genetic component (Bondy et al., 2006; Brent and Mann, 2005),
meta-analyses on single SNP association studies show no or marginal
effects. This issue has been discussed in the framework of the missing
heritability (Manolio et al., 2009; Zuk et al., 2012) and numerous ex-
planations have been proposed.
While meta-analytic evidence is abundant in the genetics of sui-
cide, a number of questions remain unanswered and no attempts
have been made to review the field as a whole. In this regard, the
present review aims at presenting a meta-meta-analysis in order to
integrate the evidence into a general framework.
2. Methods
2.1. Literature search and study selection
The search string suic* AND gene* was entered into ISI Web of
Knowledge, Medline, and Scopus. The initial search included all records

0278-5846/$ – see front matter © 2013 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.pnpbp.2013.06.014
 A.H.E. Schild et al. / Progress in Neuro-Psychopharmacology & Biological Psychiatry 46 (2013) 36–42 37

available in these databases up until May 2012 and yielded more than
40,000 hits, which were then screened for eligibility. Exclusion of irrele-
vant studies was performed in several steps (see Fig. 1 for an overview of
the process of literature search and selection): 1) Duplicates were auto-
matically identified, 2) studies referring to non-human subjects were
automatically identified (using keywords such as mouse, mice, monkey*,
or transgenic), 3) studies pertaining to cancer research were automatically
identified (using keywords such as tumor*, oncolog*, metastas* or cancer).
All automatic literature screening processes were conducted with the
open-source software JabRef 2.6 (JabRef Development Team, 2010).
From the remaining 5548 records, studies were selected first by
screening titles and abstracts and, if the study appeared to be relevant,
full texts were reviewed. Studies were checked for inclusion by three re-
searchers (AHES, UST, and JP). Reference lists of relevant meta-analyses
were then screened manually for missing records.
2.2. Inclusion criteria
In order to be included in the meta-analyses four eligibility criteria
had to be fulfilled. Studies had to a) be published in peer-reviewed
journals, b) report genetic association studies of suicidal behavior using
a case–control design, c) report data for suicidal cases (i.e., attempted
or completed suicide), and d) report independent data.
2.3. Data extraction
Data on allele and genotype frequencies were extracted by three
researchers (AHES, JP, and UST). Variables extracted as potential
moderators included country of origin, ethnicity, phenotype of cases
(attempted or completed suicide, non-violent or violent) and type
of controls (clinical or healthy).
2.4. Statistical analysis
Data analyses were performed using the package metafor
(Viechtbauer, 2010) in the open-source software R (R Development
Core Team, 2012). The strength of association in individual two-by-
two tables was calculated using odds ratios (ORs) with an OR greater
than 1 indicating a positive association between the risk allele and
suicidal behavior.
Individual allele-wise and genotype-wise meta-analyses were con-
ducted for all SNPs. Heterogeneity was assessed using the Q statistic
and I2, with I2 b 25 considered to be indicating low, 25 ≥ I2 ≤ 75 medi-
um, and I2 N 75 high heterogeneity, respectively (Huedo-Medina et al.,
2006).
Random-effect models (DerSimonian and Laird) were adopted for
all analyses, as the theoretical assumption of no heterogeneity was

PRISMA 2009 Flow Diagram

Records identified through

database searching after removal
Identification

of studies relating to cancer

research Additional records identified
(n = 18311) through other sources
(n = 0)

Records after duplicates removed

(n = 5548)
Screening

Records screened Records excluded

(n = 5548) (n = 5183)

Full-text articles assessed Full-text articles excluded

Eligibility

for eligibility (n = 258)

(n = 365)

Studies included in
qualitative synthesis
(n = 0)
Included

Studies included in
quantitative synthesis
(meta-analysis)
(n = 107)

Fig. 1. Overview of the process of participant selection and inclusion.

38 A.H.E. Schild et al. / Progress in Neuro-Psychopharmacology & Biological Psychiatry 46 (2013) 36–42

considered to be highly unlikely considering the complex pathogenesis
of suicidal behavior. Furthermore, previously published meta-analyses
suggest large and statistically significant amounts of between-studies
heterogeneity (e. g., Li and He, 2007; Saetre et al., 2009). Moreover,
from a statistical point of view, the reliable detection of true heteroge-
neity with the Q-test in meta-analyses based on a small number of
included studies is difficult, as these tests typically have low power
(see Viechtbauer, 2007, for a review).
Publication bias was assessed using Egger's regression test (Egger
et al., 1997) and Duval and Tweedie's trim-and-fill analysis (Duval
and Tweedie, 2000). Publication bias was tested for only in meta-
analyses which included a minimum of 11 independent studies due
to extremely low power of these methods in small meta-analyses
(Macaskill et al., 2001). Leave-one-out sensitivity analyses and sub-
group analyses (whenever possible) were conducted for all meta-
analyses which included a minimum of 11 independent studies.
3. Results
The present meta-analyses include a total of 107 studies (see Sup-
plement 1 for a reference list). The complete set provided sufficient in-
formation for meta-analysis for 23 individual SNPs. Comparisons
between previously published meta-analyses and updated meta-
analyses (see Table 1) reveal six common patterns within and across
individual SNPs.
While there was a general increase in included studies over time,
comparisons between meta-analyses on the same SNP revealed incon-
sistent patterns of study inclusion (compare, for example, study inclu-
sion for TPH1 A218C/A779C). These concerned not only decisions on
which studies to include, but also methodological considerations
such as the handling of non-independent data (e. g., Li and He, 2006).
Heterogeneity between studies was significant in the majority of
the meta-analyses, but had not been quantified as of yet. The present
I2 values showed a large range and differed widely between SNPs.
The observed pooled ORs were all very small and rarely statistical-
ly significant. Neither of the SNPs revealed a consistent pattern of sta-
tistical significance across the meta-analyses, which suggests that
effects rely strongly on individual studies and may easily be swayed.
Fig. 2A shows the results of a sensitivity analysis (leave-one-out) for
TPH1 A218C/A779C which identified three out of 25 studies whose
exclusion would be influential enough to change the overall result
to statistical significance. Fig. 2B, on the other hand, identified 15
out of 21 studies whose exclusion had enough leverage to turn the
overall result for 5-HTTLPR L/S statistically insignificant.

Table 1
Characteristics of published and updated meta-analyses.

SNP Study k Ethnicitya N case/N ctrl Casesb Controlsc Analysisd Modele OR (95% CI)f p (Q)g I2

TPH1 A218C/A779C Lalovic and Turecki (2002) 15 Cauc, Asian 1290/2295 Att, compl Heal Aw Na 1.14 (0.97, 1.34) 0.61 Na
Rujescu et al. (2003) 7 Cauc 898/1179 Att, compl Heal, clin Aw Na 1.33 (1.17, 1.5)⁎ 0.15 Na
Bellivier et al. (2004) 9 Cauc 861/1485 Att, compl Heal, clin Aw, gw FE, RE 1.27 (1.12, 1.43)⁎ 0.04 Na
Li and He (2006) 34h Cauc, Asian 3922/6700h Att, compl Heal, clin Aw, gw FE, RE 1.12 (1.02,1.23)⁎ 0.002 Na
A218C: Clayden et al. (2012) 21 Na 4829/7945 Att, compl Na Na RE 1.13 (0.99, 1.28) 0.06 62.5
A779C: Clayden et al. (2012) 8 Na 1512/3408 Att, compl Na Na RE 1.1 (0.85, 1.43) 0.15 Na
Schild 2013 25 Cauc, Asian 2957/4413 Att, compl Heal Aw, gw RE 1.09 (0.98, 1.22) 0.001 52.48
Schild 2013 13 Cauc, Asian 1334/1802 Att, compl clin Aw, gw RE 1.03 (0.81, 1.31) b 0.001 77.42
5-HTT 5-HTTLPR Anguelova et al. (2003) 12 Cauc, Asian 1168/1371 Att, compl Heal, clin Aw, gw FE 1.17 (1.04, 1.32)⁎ N 0.05 Na
Lin and Tsai (2004) 18 Cauc, Asian 1521/2429 Att, compl Heal Aw, gw Na 1.08 (Na) b 0.05 Na
Lin and Tsai (2004) 8 Cauc, Asian 511/831 Att, compl clin Aw, gw Na 1.27 (Na) 0.06 Na
Li and He (2007) 39h Cauc, Asian 3096/5936h Att, compl Heal, clin Awi, gw RE 0.88 (0.8, 0.97)⁎ b 0.01 Na
Clayden et al. (2012) 31 Na 6324/10285 Att, compl Na Na RE 1.06 (0.9, 1.26) 0.47 95.8
Schild 2013 21 Cauc, Asian 2536/3984 Att, compl Heal Aw, gw RE 1.09 (1.00, 1.19)⁎ 0.04 23.48
Schild 2013 8 Cauc, Asian 986/1715 Att, compl clin Aw, gw RE 1.10 (0.95, 1.26) 0.15 10.36
5-HTR2A T102C Anguelova et al. (2003) 9 Cauc, Asian 596/1003 Att, compl, SI Heal, clin Aw FE 1.09 (0.93, 1.27) N 0.05 Na
Li and He (2006) 25h Cauc, Asian 1954/2860 Att, compl, SI Heal, clin Aw, gw FE 0.94 (0.87, 1.03) 0.05 Na
Clayden et al. (2012) 18 Na 3759/5692 Att, compl Na Na RE 0.92 (0.81, 1.04) 0.11 Na
Schild 2013 11 Cauc, Asian 1193/1683 Att, compl Heal Aw, gw RE 1.10 (0.99, 1.23) 0.64 0
Schild 2013 6 Cauc, Asian 738/962 Att, compl clin Aw, gw RE 0.98 (0.85, 1.14) 0.68 0
5-HTR1A A6526G Li and He (2006) 7h Cauc, Asian Na Na Na Aw, gw RE 0.92 (0.72, 1.18) 0.01 Na
Schild 2013 5 Cauc, Asian 709/1088 Att, compl Heal, clin Aw, gw RE 1.01 (0.88, 1.17) 0.56 0
5-HTR1A C1019G Angles et al. (2012) 4 Cauc, Asian 957/957 Na Heal Aw RE 1.08 (0.8, 1.45) 0.28 Na
Clayden et al. (2012) 6 Na 2022/2135 Att, compl Na Na RE 1.14 (0.89, 1.47) 0.3 Na
Schild 2013 5 Cauc, Asian 1265/1423 Att, compl Heal, clin Aw, gw RE 0.60 (0.35, 1.03) b 0.001 95.05
COMT Val158Met Kia-Keating et al. (2007) 6 Cauc, Asian 519/933 Att, compl Heal, clin Aw Na 1.25 (1.01, 1.56)⁎ Na Na
Tovilla-Zarate et al. (2011) 12k, j
Cauc, Asian 2587/1777 Att, compl Heal Aw RE 1.09 (0.97,1.23) b 0.001 Na
Calati et al. (2011) 10j Cauc, Asian 1324/1415 Att, compl Heal, clin Aw RE 1.10 (0.92, 1.32) 0.008 60
Clayden et al. (2012) 9 Na 3226/3055 Att, compl Na Na RE 1.13 (0.89, 1.42) 0.32 Na
Schild 2013 11 Cauc, Asian 1797/1982 Att, compl Heal, clin Aw, gw RE 0.92 (0.72, 1.16) b 0.001 80.59
5-HTR1B G861C Kia-Keating et al. (2007) 7 Cauc, Asian 789/1247 Att, compl Heal, clin Aw Na 1.06 (0.92, 1.23) Na Na
Clayden et al. (2012) 10 Na 2947/4066 Att, compl Na Na RE 1.07 (0.85, 1.35) 0.76 Na
Schild 2013 9 Cauc, Asian 1161/1826 Att, compl Heal, clin Aw, gw RE 1.06 (0.94, 1.20) 0.63 0.001
BDNF Val66Met Zai et al. (2012) 11 Cauc, Asian 3352/1202 Att, compl Heal, clin Aw, gw RE 1.16 (1.01, 1.32)⁎ b 0.05 Na
Clayden et al. (2012) 7 Na 1700/2548 Att, compl Na Na RE 1.14 (0.83, 1.56) 0.63 Na
a
Cauc (Caucasian), Asian.
b
Att (attempted), compl (completed), SI (suicidal ideation).
c
Heal (healthy), clin (clinical).
d
Aw (allele-wise), gw (genotype-wise).
e
FE (fixed effect), RE (random effects).
f
Result of allele-wise analysis.
g
Significance level for Q-statistic.
h
Non-independent groups.
i
For L allele.
j
Includes at least one study with cases only.
k
Includes one study with patients without suicidal behavior but considered to be suicidal based on clinical assessment.
⁎ Significant at p ≤ 0.05.
 A.H.E. Schild et al. / Progress in Neuro-Psychopharmacology & Biological Psychiatry 46 (2013) 36–42 39

A A
1.20
2.0

1.15

log OR
1.5
OR

1.10

1.05 1.0

1.00

0.5

0 5 10 15 20 25 1996 2000 2004 2008

Study Publication year

1.20
B B

2
1.15

log OR
OR

1.10

1
1.05

1.00

1995 2000 2005 2010

0 5 10 15 20
Study
Fig. 2. Results from leave-one-out sensitivity analyses. Note. A: Results for TPH1
A218C/A779C for healthy controls. B: Results for 5-HTTLPR L/S for healthy controls.
Estimates shown in red indicate studies whose exclusion would change the (non-)
significance of the overall result. (For interpretation of the references to color in this
figure legend, the reader is referred to the web version of this article.)
In a similar vein, no general time trends were apparent (see Fig. 3A
and B), but pooled effects decreased slightly over time. This phenom-
enon appears to be rooted in growing study inclusion with time (see
Fig. 4) which leads to a stabilization of swaying effects.
Subgroup analyses followed the same general patterns (see Fig. 5 for
an example): 1) Effects rarely reached statistical significance, 2) hetero-
geneity remained considerable, even within the groups, and 3) groups
did not differ significantly from each other.
Contour-enhanced funnel plots (see Fig. 6) show symmetry around
the summary effect and symmetry regarding the significance contours.
Finally, both Egger's test and the trim and fill-analyses indicated ab-
sence of publication bias, which is in line with previous meta-analyses.
The large interest in SNPs of the serotonergic system is mirrored in
Table 2, which provides data for SNPs not subjected to meta-analysis
as of yet. Furthermore, it shows obvious parallels to previously pub-
lished analyses: again, pooled effects were small and rarely statistical-
ly significant and heterogeneity was often substantial.
4. Discussion
Genetic association studies have covered a large spectrum of
different SNPs and this area of research has attracted a lot of meta-
analytical interest (Antypa et al., 2013). In this regard, it is especially
Publication year
Fig. 3. Development of effect over time. Note: results from regression analyses between
size of effect and publication year for TPH1 A218C/A779C for healthy controls (A) and
clinical controls (B).
interesting to note that interest in this topic remained high and unfal-
tering for nearly a decade even though meta-analytic evidence sug-
gested no or only marginal effects. The serotonergic system remained
the prime focus of this research (Antypa et al., 2013).
Apparently, the observed pooled effects are mostly not statistically
significant and are very small in size. Keeping in mind how easily ef-
fects are swayed and how strongly they depend on the number and
nature of included studies, the explanatory value of any given statisti-
cally significant summary effect is limited. Partly inconsistent study
inclusion across different meta-analyses on the same topic further
emphasizes the volatile nature of significant summary effects. De-
clines in effects overtime (e.g., Li and He, 2006) are another example
of undue influences of individual studies rather than reflections of
genuine trends.
Publication bias is regarded as one of the main threats to the valid-
ity of results from meta-analysis and has been found in medical
research time and again (e.g., Dwan et al., 2008; McGauran et al.,
2010). Interestingly, publication bias does not appear to be an issue
in this particular field of application. Possible explanations include,
but are not limited to, the inherently exploratory character of genet-
ic studies and the study of hard and objective data. Furthermore,
economic considerations, such as the cost of genetic studies and re-
quirements by funding agencies may play a role in this positive devel-
opment. All in all, motivation to publish results irrespective of their
outcome seems to be high both for authors and editors.
40 A.H.E. Schild et al. / Progress in Neuro-Psychopharmacology & Biological Psychiatry 46 (2013) 36–42

0.0
0.1 > p > 0.05
Rujescu 2003 7 0.05 > p > 0.01
< 0.01

0.1
Standard error
Bellivier 2004 9

0.2
Saetre 2010 12

0.3
Schild 2012b 13

0.4
Lalovic 2002 15 0.5 1.0 1.5 2.0 2.5
Odds Ratio

Fig. 6. Funnel plot for TPH1 A218C/A779C for healthy controls. Note: contours (gray)
Li 2006 22
indicate significance levels of included studies, dotted lines show summary effect
and 95% confidence interval for the summary effect.

Schild 2012a 25 between groups has been ignored in the analysis by Li and He (2006)
both for the subjects and in regard to the two polymorphisms. As
A218C and A779C are in full linkage equilibrium (Li and He, 2006),
0.8 1 1.2 1.4
they are generally analyzed together. This, however, does not justify
Fig. 4. Development of pooled effect for TPH1 A218C/A779C for number of included stud- the inclusion of data on both SNPs based on the same subjects, as
ies. Note: white numbers within summary diamonds indicate number of included studies. these are perfectly correlated. Although the effects of the violation of
independence of groups in meta-analysis may be acceptable in some
When comparing results from different meta-analyses, it is important cases (Hedges et al., 2010), this is an extreme example where the
to keep methodological considerations in mind. In contrast to the other same groups may be included up to four times in the same analysis.
published meta-analyses on TPH1 A218C/A779C, non-independence In the majority of meta-analyses, random-effects models have
been applied. Quantifications of true between-study heterogeneity
have not been published up until now. However, assumptions of low
or non-significant heterogeneity are theoretically incompatible with
Healthy (k = 25)
the complex nature of suicidal behavior. Even if one was to ignore dif-
52.48
ferences between study designs, ethnicities of participants, or types of
suicidal behavior, the phenotype remains strongly intertwined with
genetic components of the underlying psychiatric disorders.
Clinical (k = 13) 77.42 Although there is a distinct genetic component in suicidal behavior
(e.g., Brent and Mann, 2005; Egeland and Sussex, 1985; Fu et al., 2002;
Statham et al., 1998), genetic association studies do not allow for def-
inite conclusions. Due to the substantial comorbidity between suicidal
Asian (k = 8) 41.94
behavior and a large range of psychiatric disorders, it is not possible to
disentangle genetic components for suicide and those associated with
underlying disorders. Neither comparisons between suicidal persons
Caucasian (k = 21) 55.8 and healthy controls can answer that question, nor can comparisons
between cases and controls affected by the same disorder. The possi-
bility that the associations merely reflect distinct or more extreme
subtypes within the specific framework of the same disorder can
Completed (k = 6) 0 never be ruled out.
One might argue that such subtypes or effects at the far end of the
spectrum of suicidal behavior (e.g., violent suicide completers) have
Attempted (k = 21) 56.99 not yet received enough attention. Although this may be true, it is
important to keep in mind that suicide is, statistically speaking, still a
very rare event. Considering that effects of single SNPs are usually
very small (especially in connection with rare frequencies), the predic-
Violent (k = 4) 38.95 tive value of any given association is marginal. Although further specifi-
cation of the phenotype may elucidate effects within circumscribed
groups, their explanatory value may be negligible due to the extreme
rarity of such cases.
Non−violent (k = 3) 54.93
It has been argued that significant effects in meta-analyses for sin-
gle SNPs can be observed for individual subgroups (Clayden et al.,
2012). While the association between 5-HTTLPR and suicide attempts
1 1.5 2
was found to be statistically significant, no effect for suicide com-
Fig. 5. Subgroup results for TPH1 A218C/A779C for healthy controls. Note: white num- pleters was reported. However, the difference between the two
bers within summary diamonds indicate true heterogeneity between studies (I2). groups was not statistically significant (Clayden et al., 2012). It cannot
 A.H.E. Schild et al. / Progress in Neuro-Psychopharmacology & Biological Psychiatry 46 (2013) 36–42 41

Table 2
Characteristics of previously unpublished meta-analyses.

System Gene SNP Allele ka N case/N ctrl OR (95% CI) p (Q)b I2

5-HTc 5-HTR1A Pro16Leu C 3 439/540 0.71 (0.47, 1.08) 0.15 73.29

5-HTR1A Gly272Asp G 2 333/358 0.99 (0.76, 1.23) 0.72 0
5-HTR1B A161T A 3 276/564 1.00 (0.77, 0.30) 0.43 0
5-HTR1F C78T T 2 271/345 1.04 (0.63, 1.71) 0.32 0
5-HTR2A A1438G A 7 867/1555 1.10 (0.98, 1.24) 0.54 0
5-HTR2A His452Tyr His 3 609/612 1.33 (1.02, 1.72)⁎ 0.63 0
5-HTR2C Cys23Ser C 5 537/786 1.06 (0.72, 1.61) 0.07 59.08
5-HTR6 C267T C 3 409/466 1.23 (0.91, 1.68) 0.2 40.44
5-HTTLPR VNTR 10 6 769/1308 1.19 (1.01, 1.41)⁎ 0.41 22.31
TPH1 G5806T T 3 480/803 0.95 (0.75, 1.20) 0.19 38.39
TPH2 rs1386494 A 2 235/350 0.56 (0.07, 4.57) b 0.001 97.61
BDNF d
p75NTR s205l C 2 148/265 2.27 (1.15, 4.47)⁎ 0.61 0
DAe DRD2 Ins/Del Ins 2 225/385 1.13 (0.42, 3.00) 0.004 87.69
NAf ADRA C219G C 2 272/256 0.85 (0.61, 1.21) 0.23 31.26
RAg ACE Ins/Del Ins 2 597/805 0.46 (0.14, 1.44) b 0.001 97.56
a
Number of included studies.
b
Significance level for Q-statistic.
c
Serotonin.
d
Neurotrophins.
e
Dopamine.
f
Noradrenaline.
g
Renin-angiotensin.
⁎ Significant at p ≤ 0.05.

be ruled out, therefore, that such effects exist for one group, but not
for the other. It is possible that either there are the same underlying
effects altogether or suicide completers may be affected by a broader
range of genetic variations, i.e., this phenotype may be genetically
even more complex leading to a situation where the role of a single
SNP diminishes and the accumulation and interaction of several
SNPs become more pronounced.
Genome-wide association studies (GWAS) show no or little
genome-wide replicable evidence (Butler et al., 2010; Cheng et al.,
2006; Perlis et al., 2010; Schosser et al., 2011; Willour et al., 2012)
yet. Although a few associations have been reported (e.g., 3p14,
2p12; Butler et al., 2010), they are not consistent over different
GWAS. Keeping in mind that the different GWAS have focused on a
variety of underlying psychiatric disorders (e.g., bipolar disorders,
major depressive disorder, and affective disorders in general), it may
be that the reported associations mostly reflect the underlying psychi-
atric disorders rather than suicidal behavior as such. Moreover, the
lack of statistically significant associations are largely due to a lack of
power, as suggested from simulation studies (Spencer et al., 2009)
and meta-analyses of GWAS data (Butler et al., 2010; Perlis et al.,
2010; Ripke et al., 2013). Estimates suggest that future GWAS focusing
on major depressive disorder should analyze large numbers of partic-
ipants of at least 100,000 patients and the same amount of healthy
controls (Ripke et al., 2013). Such large samples may be acquired
with the use of data from electronic databases and health care organi-
zations (Allen et al., 2010; Ripke et al., 2013). A widespread and inter-
national effort to achieve such high patient rates may thus be the key
to unraveling the role of genetic associations in suicidal behavior.
The first meta-analysis on the association between suicidal behav-
ior and genetic data, published by Lalovic and colleagues (Lalovic and
Turecki, 2002), revealed a non-significant effect and already mirrors
the large variation between studies observed in subsequent meta-
analyses (for example regarding ethnicities, phenotypes, and underly-
ing psychiatric disorders). Just a year later, another meta-analysis was
published (Anguelova et al., 2003) and, although it dealt with two dif-
ferent SNPs, the same underlying results were apparent. During the
last decade, the research interest in the genetics of suicide has never
faltered, which is reflected in the large number of published meta-
analyses. These meta-analyses have, however, revealed similar results
focusing on small and volatile effects. Although these results may ap-
pear discouraging, it is important to acknowledge that this line of re-
search marks the beginning in the search of the genetic underpinnings
of suicide. While this straightforward approach may seem overly sim-
plistic from today's point of view, ruling out the existence of major con-
tributions of single genes was a necessary step in paving the way for the
developments seen today (e.g., Dwivedi, 2012).
The issue of the missing heritability (Manolio et al., 2009; Zuk et al.,
2012) has been discussed widely, especially in the framework of
GWAS. A number of explanations for this phenomenon have been pro-
posed including genetic interactions (Zuk et al., 2012), rare variants
with large effects, or structural variation (including copy neutral vari-
ations and copy number variations), and small sample sizes (Manolio
et al., 2009). Current works focus on areas such as epigenetics, gene-
environment interactions, gene expression profiling, or the study of
intermediate phenotypes or rare genetic variants (Dwivedi, 2012), to
name just a few.
The number of published meta-analyses gives an inkling of the
huge interest in the association between simple SNPs and suicidal be-
havior. It is high time, however, to accept that the genetics of suicide
are more complex than that and that large contributions of single
genes are unlikely. In moving beyond such simple unidirectional ex-
planations, we can fully embrace modern developments in this area
of research and begin a new chapter in the genetics of suicide.
Supplementary data to this article can be found online at http://
dx.doi.org/10.1016/j.pnpbp.2013.06.014.
Acknowledgments
Financial disclosures: the authors reported no biomedical financial
interests or potential conflicts of interests.
References
Allen HL, Estrada K, Lettre G, Berndt SI, Weedon MN, Rivadeneira F, et al. Hundreds of
variants clustered in genomic loci and biological pathways affect human height.
Nature 2010;467:832–8.
Angles MR, Ocana DB, Medellin BC, Tovilla-Zarate C. No association between the HTR1A
gene and suicidal behavior: a meta-analysis. Rev Bras Psiquiatr 2012;34:38–42.
Anguelova M, Benkelfat C, Turecki G. A systematic review of association studies inves-
tigating genes coding for serotonin receptors and the serotonin transporter: II. Sui-
cidal behavior. Mol Psychiatry 2003;8:646–53.
Antypa N, Serretti A, Rujescu D. Serotonergic genes and suicide: a systematic review.
Eur Neuropsychopharmacol 2013(0). [Advance online publication].
Bellivier F, Chaste P, Malafosse A. Association between the TPH gene A218C polymor-
phism and suicidal behavior: a meta-analysis. Am J Med Genet B Neuropsychiatr
Genet 2004;124B:87–91.
Bondy B, Buettner A, Zill P. Genetics of suicide. Mol Psychiatry 2006;11:336–51.
42 A.H.E. Schild et al. / Progress in Neuro-Psychopharmacology & Biological Psychiatry 46 (2013) 36–42
Brent DA, Mann JJ. Family genetic studies, suicide, and suicidal behavior. Am J Med
Genet B Neuropsychiatr Genet Semin Med Genet 2005;133C:13–24.
Butler AW, Breen G, Tozzi F, Craddock N, Gill M, Korszun A, et al. A genomewide
linkage study on suicidality in major depressive disorder confirms evidence
for linkage to 2p12. Am J Med Genet B Neuropsychiatr Genet 2010;153B:
1465–73.
Calati R, Porcelli S, Giegling I, Hartmann AM, Mller HJ, Ronchi DD, et al. Catechol-
o-methyltransferase gene modulation on suicidal behavior and personality traits:
review, meta-analysis and association study. J Psychiatr Res 2011;45:309–21.
Cheng R, Juo SH, Loth J, Nee J, Iossifov I, Blumenthal R, et al. Genome-wide linkage scan
in a large bipolar disorder sample from the National Institute of Mental Health ge-
netics initiative suggests putative loci for bipolar disorder, psychosis, suicide, and
panic disorder. Mol Psychiatry 2006;11:252–60.
Clayden RC, Zaruk M, Meyre D, Thabane L, Samaan Z. The association of attempted sui-
cide with genetic variants in the SLC6A4 and TPH genes depends on the definition of
suicidal behavior: a systematic review and meta-analysis. Transl Psychiatry 2012;2:
e166.
Duval S, Tweedie R. Trim and fill: a simple funnel-plot-based method of testing and
adjusting for publication bias in meta-analysis. Biometrics 2000;56:455–63.
Dwan K, Altman DG, Arnaiz JA, Bloom J, Chan AW, Cronin E, et al. Systematic review of
the empirical evidence of study publication bias and outcome reporting bias. PLoS
One 2008;3:e3081.
Dwivedi Y, editor. The neurobiological basis of suicide. Frontiers in neuroscienceCRC
Press; 2012.
Egeland JA, Sussex JN. Suicide and family loading for affective disorders. JAMA 1985;254(7):
915–8.
Egger M, Smith GD, Schneider M, Minder C. Bias in meta-analysis detected by a simple,
graphical test. BMJ 1997;315:629–34.
Ernst C, Mechawar N, Turecki G. Suicide neurobiology. Prog Neurobiol 2009;89:
315–33.
Fu Q, Heath AC, Bucholz KK, Nelson EC, Glowinski AL, Goldberg J, et al. A twin study of ge-
netic and environmental influences on suicidality in men. Psychol Med 2002;32(1):
11–24.
Hedges LV, Tipton E, Johnson MC. Robust variance estimation in meta-regression with
dependent effect size estimates. Res Synth Methods 2010;1:39–65.
Huedo-Medina TB, Sanchez-Meca J, Marin-Martinez F, Botella J. Assessing heterogene-
ity in meta-analysis: Q statistic or I2 index? Psychol Methods 2006;11:193–206.
JabRef Development Team. JabRef. URL:http://jabref.sourceforge.net, 2010.
Kia-Keating BM, Glatt SJ, Tsuang MT. Meta-analyses suggest association between COMT,
but not HTR1B, alleles, and suicidal behavior. Am J Med Genet B Neuropsychiatr
Genet 2007;144B:1048–53.
Lalovic A, Turecki G. Meta-analysis of the association between tryptophan hydroxylase
and suicidal behavior. Am J Med Genet 2002;114:533–40.
Li D, He L. Further clarification of the contribution of the tryptophan hydroxylase (TPH)
gene to suicidal behavior using systematic allelic and genotypic meta-analyses.
Hum Genet 2006;119:233–40.
Li D, He L. Meta-analysis supports association between serotonin transporter (5-HTT)
and suicidal behavior. Mol Psychiatry 2007;12:47–54.
Lin PY, Tsai GC. Association between serotonin transporter gene promoter polymor-
phism and suicide: results of a meta-analysis. Biol Psychiatry 2004;55:1023–30.
Macaskill P, Walter SD, Irwig L. A comparison of methods to detect publication bias in
meta-analysis. Stat Med 2001;20:641–54.
Manolio TA, Collins FS, Cox NJ, Goldstein DB, Hindorff LA, Hunter DJ, et al. Finding the
missing heritability of complex diseases. Nature 2009;461:747–53.
McGauran N, Wieseler B, Kreis J, Schler YB, Klsch H, Kaiser T. Reporting bias in medical
research — a narrative review. Trials 2010;11:37.
Nielsen DA, Goldman D, Virkkunen M, Tokola R, Rawling SR, Linnoila M. Suicidality and
5-hydroxyindoleacetic acid concentration associated with a tryptophan hydroxy-
lase polymorphism. Arch Gen Psychiatry 1994;51:34–8.
Perlis RH, Huang J, Purcell S, Fava M, Rush AJ, Sullivan PF, et al. Genome-wide asso-
ciation study of suicide attempts in mood disorder patients. Am J Psychiatry
2010;167:1499–507.
R Development Core Team. R: A language and environment for statistical, computing.
3-900051-07-0; 2012 [URL: http://www.R-project.org].
Ripke S, Wray NR, Lewis CM, Hamilton SP, Weissman MM, Breen G, et al. A mega-
analysis of genome-wide association studies for major depressive disorder. Mol
Psychiatry 2013;18:497–511.
Rujescu D, Giegling I, Sato T, Hartmann AM, Möller HJ. Genetic variations in tryptophan
hydroxylase in suicidal behavior: analysis and meta-analysis. Biol Psychiatry
2003;54:465–73.
Saetre P, Lundmark P, Hansen T, Rasmussen HB, Djurovic S, Melle I, et al. The tryptophan
hydroxylase 1 (TPH1) gene, schizophrenia susceptibility, and suicidal behavior: a
multi-centre case–control study and meta-analysis. Am J Med Genet B Neuropsychiatr
Genet 2009;153B:387–96.
Schosser A, Butler AW, Ising M, Perroud N, Uher R, Ng MY, et al. Genomewide associ-
ation scan of suicidal thoughts and behaviour in major depression. PLoS One
2011;6:e20690.
Spencer CCA, Su Z, Donnelly P, Marchini J. Designing genome-wide association studies:
sample size, power, imputation, and the choice of genotyping chip. PLoS Genet
2009;5(5):e1000477.
Statham DJ, Heath AC, Madden PA, Bucholz KK, Bierut L, Dinwiddie SH, et al. Suicidal be-
haviour: an epidemiological and genetic study. Psychol Med 1998;28(4):839–55.
Tovilla-Zarate C, Juarez-Rojop I, Ramon-Frias T, Villar-Soto M, Pool-Garcia S, Medellin
BC, et al. No association between COMT val158met polymorphism and suicidal be-
havior: meta-analysis and new data. BMC Psychiatry 2011;11:151.
Viechtbauer W. Hypothesis tests for population heterogeneity in meta-analysis. Br J
Math Stat Psychol 2007;60:29–60.
Viechtbauer W. Conducting meta-analyses in R with the metafor package. J Stat Softw
2010;36:1–48.
Voracek M. Genetic factors in suicide: reassessment of adoption studies and individ-
uals' beliefs about adoption study findings. Psychiatr Danub 2007;19:139–53.
Voracek M, Loibl LM. Genetics of suicide: a systematic review of twin studies. Wien
Klin Wochenschr 2007;119:463–75.
Willour VL, Seifuddin F, Mahon PB, Jancic D, Pirooznia M, Steele J, et al. A genome-wide
association study of attempted suicide. Mol Psychiatry 2012;17:433–44.
Zai CC, Manchia M, Luca VD, Tiwari AK, Chowdhury NI, Zai GC, et al. The brain-derived
neurotrophic factor gene in suicidal behaviour: a meta-analysis. Int J Neuropsycho-
pharmacol 2012;15:1037–42.
Zuk O, Hechter E, Sunyaev SR, Lander ES. The mystery of missing heritability: genetic
interactions create phantom heritability. Proc Natl Acad Sci 2012;109:1193–8.