Extensive Genotyping of the BDNF and NTRK2
Genes Define Protective Haplotypes Against
Obsessive-Compulsive Disorder
Pino Alonso, Mónica Gratacòs, José M. Menchón, Jerónimo Saiz-Ruiz, Cinto Segalàs,
Enrique Baca-García, Javier Labad, José Fernández-Piqueras, Eva Real, Concepción Vaquero,
Mercedes Pérez, Helen Dolengevich, Juan R. González, Mónica Bayés, Rafael de Cid, Julio Vallejo,
and Xavier Estivill
Background: Family, twin and molecular studies provide increasing evidence for the importance of genetic factors in obsessive-compulsive disorder (OCD). Recent work suggests that brain-derived neurotrophic factor (BDNF) may be involved in OCD pathophysiology. We used
a linkage disequilibrium (LD)-mapping approach to investigate the role that BDNF and its specific receptor neurotrophic tyrosine kinase
receptor type 2 (NTRK2) may play in increasing susceptibility to OCD.
Methods: Eight tag single nucleotide polymorphisms (tagSNPs) covering the BDNF gene region and 46 tagSNPs in the NTRK2 region were
genotyped in 215 OCD patients and 342 control subjects. Single nucleotide polymorphism association and haplotype analysis were
performed. The possible relationship between genetic factors and clinical characteristics including age of OCD onset, tic disorders, clinical
dimensions, and family history of OCD were investigated.
Results: Haplotype analysis revealed a significant association between OCD and a five-marker protective haplotype located toward the 5’
of the BDNF gene (odds ratio [OR] ⫽ .80; 95% confidence interval [CI] ⫽ .69 –.92; permutation p value ⫽ .006) containing the functional valine
(Val)66-to-methionine (Met) variant. A significant association between a NTRK2 intronic SNP (rs2378672) and OCD was identified (p ⬍ .0001)
in female patients under an additive model. A protective haplotype located in intron 19 of NTRK2 was also associated with OCD (OR ⫽ .76;
95% CI ⫽ .66 –.87; permutation p value ⫽ .001).
Conclusions: These findings support a role for the BDNF/NTRK2 signaling pathway in genetic susceptibility to OCD.
Key Words: Association, BDNF, haplotype, NTRK2, obsessivecompulsive disorder, tagSNPs
T
win and family studies indicate that genetic factors may
play a significant role in the transmission and expression
of obsessive-compulsive disorder (OCD). A higher concordance was reported in monozygotic twins (80% to 87%) than
in dizygotic twins (47% to 50%) (1) and a recent meta-analysis
indicated an aggregate risk of 8.3% for first-degree relatives of
OCD subjects (2).
Molecular genetic studies in OCD have mainly focused on
candidate genes from the serotonergic (5-HT) and dopaminergic
pathways. Various genes encoding for 5-HT receptors, 5-HT1D␤
(3,4) and 5HT2A (5), and for enzymes involved in the degradation
of 5-HT and dopamine, catechol-O-methyltransferase (6,7) and
monoamine oxidase A (6,8), have been implicated in the etiopathology of OCD. Nevertheless, the inconclusiveness of the
From the OCD Clinical and Research Unit (PA, JMM, CS, JL, ER, JV), Psychiatry
Department, Hospital Universitari de Bellvitge, Barcelona; CIBER en Epidemiología y Salud Pública (CIBERESP) (MG, JRG, XE), Spain; Genes and
Disease Program and CeGen Barcelona Genotyping Node (MG, JRG, MB,
RdC, XE), Center for Genomic Regulation (CRG), Barcelona; OCD Program
(JS-R, EB-G, JF-P, CV, MP, HD), Psychiatry Department, Hospital Ramon y
Cajal, University of Alcalá, Madrid; and Department of Life Sciences (XE),
Pompeu Fabra University, Barcelona, Spain.
Address reprint requests to Xavier Estivill, Ph.D., Genes and Disease Program, Center for Genomic Regulation, Plaça Charles Darwin s/n, PRBB
Building, E-08003 Barcelona, Catalunya, Spain; E-mail: xavier.estivill@
crg.es.
Received February 16, 2007; revised May 18, 2007; accepted June 14, 2007.
0006-3223/08/$34.00
doi:10.1016/j.biopsych.2007.06.020
results suggests that other currently unidentified genes of small to
moderate effect may contribute to the OCD phenotype.
Brain-derived neurotrophic factor (BDNF) promotes neuronal
proliferation, regeneration, and connectivity during development
and participates in the plasticity and maintenance of neurons
throughout adulthood (9). Neurotrophins bind with high affinity
(Kd ⬃10⫺9M) to members of the tyrosine kinase (Trk) receptor
family, mainly to neurotrophic tyrosine kinase receptor type 2
(NTRK2) in the case of BDNF (10). The effects of neurotrophins
depend not only on their levels but also on their binding affinity
to these transmembrane receptors and the duration and intensity
of downstream signaling cascades stimulated after receptor
activation (10). Only two studies have addressed BDNF involvement in OCD, with inconclusive outcomes. Hall et al. (11) used
164 trios with OCD to test four single nucleotide polymorphisms
(SNPs) and one microsatellite marker in the extended BDNF
locus. Single-locus transmission-distortion tests revealed significant evidence of an association with OCD for all the markers
tested in the subgroup of patients with childhood onset of OCD.
Nonetheless, Zai et al. (12) failed to replicate these findings on
testing for the association of two BDNF polymorphisms, BDNF(GT)n and the BDNF(valine [Val]66-to-methionine [Met]), with
OCD in 152 families with an affected proband.
Given these controversial results and recent findings suggesting the involvement of BDNF in the regulation of anxiety-related
behaviors (13), we hypothesized that BDNF and its high-affinity
receptor NTRK2 may contribute to genetic susceptibility to OCD.
We also wondered whether variants at these two genes may be
related to the phenotypic heterogeneity of the disorder and
whether genetic differences may arise when considering clinical
factors such as age at onset of OCD, comorbid tic disorders, clinical
BIOL PSYCHIATRY 2008;63:619 – 628
© 2008 Society of Biological Psychiatry
620 BIOL PSYCHIATRY 2008;63:619 – 628
symptom dimensions, or family history of OCD. To test this
hypothesis, we performed a case-control study through an extensive linkage disequilibrium (LD)-mapping approach to examine
both genes and analyzed the possible relationship between their
variants and OCD clinical characteristics.
Methods and Materials
Subjects
Two hundred fifteen consecutive Spanish Caucasian outpatients with OCD (103 men and 112 women) were included in the
study. Obsessive-compulsive disorder subjects were recruited
from the OCD Clinics of Bellvitge University Hospital (Barcelona,
Spain) (n ⫽ 132) and the Ramón y Cajal University Hospital
(Madrid, Spain) (n ⫽ 83) between 2003 and 2005. All patients
met DSM-IV criteria for OCD (14) and had had OCD symptoms
for at least 1 year. Diagnoses were independently assigned by
two psychiatrists with extensive clinical experience in OCD, who
separately interviewed the patients using the Structured Clinical
Interview for DSM-IV Axis I Disorders-Clinician Version (SCIDCV) (15). Exclusion criteria were a past or present history of
psychoactive substance abuse, a history of psychotic disorders,
age under 18 or over 65 years, mental retardation, and severe
organic or neurological pathology except tic disorder. Comorbidity with other DSM-IV Axis I disorders was not considered an
exclusion criterion, provided that OCD was the primary reason
for seeking medical assistance. During the selection period, 283
outpatients were assessed and fulfilled DSM-IV criteria for OCD.
Of these patients, 56 were ruled out in accordance with the
exclusion criteria and 12 refused to take part in the study.
The control group consisted of 342 unrelated Caucasian
subjects (202 men and 140 women, mean age: 39.8 years), who
were recruited from a group of blood donors and were not
psychiatrically screened.
Written informed consent was obtained from each subject
after complete description of the study, which was approved by
both hospitals’ ethics committees.
Clinical assessment included information on age at onset of
OCD, defined as the age when symptoms became a significant
source of distress and interfered with the patient’s social functioning. We decided to consider 10 years old as a reasonable
threshold to define the early-onset subgroup of patients (16). As
only 12.0% of our patients reported onset before this age, we
repeated our analysis using 17 years old as a less restrictive age
of onset (17) and defined three groups (below 10 years, between
10 and 17 years, and after 17 years old). Many OCD patients
suffer from obsessions and compulsions years before they meet
diagnostic criteria for OCD. To take this into account, we
considered age of appearance of any obsessive symptom remembered by the patient or a family member, and defined
groups of early versus late subclinical OCD onset. A diagnosis of
subclinical OCD implied that the subject met all criteria for
definite OCD except that symptoms were reported to occur for
less than 1 hour a day or were not reported as causing interference or distress. Presence of subclinical obsessive symptoms
before 10 years old was reported by 17.3% of our patients,
between 10 and 17 years old by 34.6%, and after 17 years old by
48.1%. A clinician-administered version of the Yale-Brown Obsessive-Compulsive Scale (Y-BOCS) (18) and the 21-item Hamilton
Depression Rating Scale (HDRS) (19) were used to assess the
severity of obsessive-compulsive and depressive symptoms, respectively. The clinician-administered version of the Y-BOCS Symptom
Checklist (18) was employed to ascertain scores on five previously
www.sobp.org/journal
P. Alonso et al.
identified symptom dimensions: symmetry/ordering, hoarding, contamination/cleaning, aggression/checking, and sexual/religious obsessions, classified as absent, past, or present (20). Presence of
family psychiatric history in first-degree relatives was established
according to the Family History Research Diagnostic Criteria (FHRDC) (21). Diagnostic information was obtained on 670 first-degree
relatives (on average, 3.1 relatives per proband). Direct interview
was possible in 474 of them. For those first-degree relatives who
were unavailable for assessment or were unwilling to be assessed,
information was gathered from two knowledgeable informants. In
these cases, relatives were considered affected if they had received
a formal diagnosis by a psychiatrist, had a history of psychiatric
hospitalization, or had been maintained on diagnosis-specific medication.
Tag Single Nucleotide Polymorphism Selection
From the HapMap Project dataset, we used genotypes from
public release 16 (Phase I data freeze, Single Nucleotide Polymorphism database [dbSNP] b124), corresponding to the 60
individuals from 30 Centre d’Etude du Polymorphisme Humain
(CEPH) trios of European descent (http://www.hapmap.org).
From the gene location, we extended the search for 5 to 10
kilobases (kb) upstream and downstream of BDNF and NTRK2
genes. Only SNPs with a unique mapping location on the
National Center for Biotechnology Information (NCBI) B34 assembly and a minor allele frequency (MAF) higher than 10%
were considered for further analysis. In the BDNF region (RefSeq:
NM_170731; chromosome 11: 27629852-27717072, NCBI B34
assembly), covering 87.2 kb, genotypes for 27 SNPs were available. Twenty-one of these had a MAF ⬎ 10% (average spaced 4.4
kb). In the NTRK2 gene region, covering 387 kb (RefSeq:
NM_006180; chromosome 9: 82724497-83111512, NCBI B34 assembly), information from 238 SNPs was available. Two hundred
six of these had a MAF ⬎ 10% (average spaced 1.6 kb).
Bins of common SNPs in strong LD, as defined by an r2 higher
than .85, were identified within both datasets by use of HapMapLDSelect-Processor. This uses the LD Select method (22) to
process HapMap genotype dump format data corresponding to
the locus region. Eight tag single nucleotide polymorphisms
(tagSNPs) were selected to cover all bins in the case of BDNF,
including a nonsynonymous variant in the coding exon (rs6265,
corresponding to the functional Val66Met amino acid change).
Forty-six tagSNPs were selected to cover all bins in the case of
NTRK2 (see Figure 1 and Supplements 1 and 2 for a graphic
representation of selected variants for BDNF and NTRK2, respectively, and Supplement 3 for SNP details). Information about
SNPlex (Applied Biosystems, Foster City, California) design,
quality control of genotypes, and analysis of population stratification is provided in Supplement 4.
Sample Power Calculation
First, we computed power calculations using the Genetic Power
Calculator (http://pngu.mgh.harvard.edu/⬃purcell/gpc/). Thus,
we determined that the case-control sample had 93% power for
detecting a risk allele with 10% frequency and a dominant
genotype relative risk of 2.0. To assess the power for detecting
association due to LD with a causal loci, we also carried out
power calculations for an indirect association study that uses
tagSNPs (23). We estimated that our study was able to detect a
susceptibility locus (80% power) with an odds ratio (OR) of 2.0 if
the MAF is .15 and an OR of 1.7 if the MAF is .3, assuming a
codominant effect at an unobserved locus, an alpha (␣) value of
BIOL PSYCHIATRY 2008;63:619 – 628 621
P. Alonso et al.
Figure 1. (A) Scaled diagram showing the BDNF gene structure. The single coding exon is shown as a black box. Noncoding exons and 3’UTR region are shown
in grey boxes. Estrogen response element is shown with an arrow and the NGF protein domain is shown above the gene structure. TagSNPs genotyped in the
genomic region in our study are shown, and markers with an asterisk correspond to those SNPs previously genotyped in OCD patients in other studies (11,12).
(B) Scaled diagram showing the NTRK2 gene structure. The coding exons are shown as black boxes. Noncoding exons are shown as grey boxes. The protein
domains are shown above the gene structure as grey lines. TagSNPs genotyped in the genomic region in our study are shown. The grey bars with arrows
indicate the genomic region where the significant haplotypes were identified in the study. NGF, nerve growth factor; OCD, obsessive-compulsive disorder;
tagSNP, tag single nucleotide polymorphisms; UTR, untranslated region.
.05, and r2 ⫽ .85 for the ability of haplotypes to predict the allele
count at the causal locus.
Statistical Analysis
Departure from Hardy-Weinberg equilibrium (HWE) for all
the biallelic SNP markers was tested using a chi-square test
(1 df ). For individual SNP association analyses, genotype frequencies were assessed by means of multivariate methods based
on logistic regression analyses and analyzed under codominant,
dominant, recessive, overdominant, and additive models. The
best model was selected using the Akaike information criteria.
We estimated the crude OR and 95% confidence intervals (95%
CI). These analyses were performed using the SNPassoc R
package (24). To avoid false-positive results due to multiple
testing, we applied the Bonferroni correction for 54 independent
loci genotyped. Significant p values were raised to p ⫽ .0009.
The robustness of these individual findings was investigated
by a bootstrap analysis (25). We repeated the selection procedure on 300 bootstrap datasets generated from the original
sample and recorded the number of times that a particular
association was statistically significant (at Bonferroni significance
level) between each polymorphism and the phenotype.
Multinomial regression was used to assess the association
between cases, control subjects, and SNPs when cases were
categorized by clinical status: symptom dimensions, age at onset
of OCD, comorbid tic disorders, family history of OCD, and
comorbid affective disorders.
Linkage disequilibrium between polymorphisms and haplotype block structures was evaluated by Haploview software
version 3.2 (http://www.broad.mit.edu/mpg/haploview). Haplotype blocks were generated by the algorithm of four-gamete
rules (26). For the haplotype estimations, we used a sliding
window technique using also the haplo.stats R package. Furthermore, the permutation procedure was used to estimate the
significance of the best result (1000 permutations). Haplotypes
with frequencies lower than 1% were excluded. Interaction
effects between SNPs and haplotypes of BDNF and NTRK2 genes
were tested using the likelihood ratio test.
Results
The demographic and clinical characteristics of the patients
and control subjects are summarized in Table 1. Patients from
Barcelona and Madrid did not differ significantly in terms of
either sociodemographic data or OCD features (data not shown).
A higher proportion of male subjects was detected in the control
group (␹2 ⫽ .01), and control subjects were significantly older
than OCD patients (t ⫽ ⫺5.75; df ⫽ 548; p ⫽ .01).
Single-SNP Association Analysis
The genotype distribution was in Hardy-Weinberg equilibrium both in control subjects and patients (alpha ⫽ .01) (Supplement 3).
No significant differences in genotype distributions for BDNF
were detected between OCD patients and control subjects under
the five different inheritance models tested (Figure 2). The
possibility that the lack of association was due to the confounding effect of sex was ruled out by a stratified analysis.
When we analyzed NTRK2, four SNPs showed nominal
p-values lower than .05: rs1201363 in 5’ upstream of the gene;
rs1439050 in intron 4; rs11140783 in intron 14, and rs11795386
and rs2378672, both in intron 19. After Bonferroni correction,
only rs2378672 remained significant (p ⬍ .0001) (Figure 2). A
stratified analysis by sex revealed that significance was restricted
to the female subgroup under an additive model (Table 2). The
association of SNP rs2378672 and OCD remains statistically
significant according to the bootstrap p-values (p ⬍ .0009).
Comparisons considering age at onset of OCD detected no
significant differences regarding family history of OCD, gender,
or OCD severity between patients with early- and late-onset of
the disorder. We could not replicate Hall et al.’s (11) finding in
any of the onset subgroups of OCD or subclinical OCD in the
BDNF gene and found no positive signals in the NTRK2 gene
(data not shown). Likewise, we could not find any evidence of
the association of BDNF or NTRK2 polymorphisms with other
clinical variables, including symptom dimensions, tic disorders,
and comorbid affective disorders.
www.sobp.org/journal
622 BIOL PSYCHIATRY 2008;63:619 – 628
P. Alonso et al.
Table 1. Sociodemographic and Clinical Characteristics of 215 Patients with Obsessive-Compulsive Disorder
Characteristics
Age, Years, Mean ⫾ SD, (range)
Male/Female
Years of Education, Years, Mean ⫾ SD
Marital Status, Single, n, %
Age at Onset of OCD, Years, Mean ⫾ SD
Age at Onset of Subclinical OCD, Years, Mean ⫾ SD
Duration of Illness, Years, Mean ⫾ SD
Previous Treatments:
SRI trials completed, n (%)
0
1
2
⬎2
Basal Y-BOCS Score, Mean ⫾ SD
Global
Obsessions
Compulsions
Basal HDRS, Score, Mean ⫾ SD
Comorbid Diagnosis, n (%)
Any comorbid diagnosis
Affective disorders
Major depressive disorder (n ⫽ 26)
Dysthymia (n ⫽ 11)
Depressive disorder NEC (n ⫽ 9)
Bipolar disorder (n ⫽ 7)
Anxiety disorders other than OCD
Tic disorder or GT
Eating disorders
Psychiatric Family History, n (%)
Any psychiatric diagnosis
OCD
Anxiety disorders other than OCD
Affective disorders
Tic disorder or GT
Symptom Dimensions (present, n, %)
Contamination/cleaning
Aggressive/checking
Sexual/religious
Hoarding
Symmetry/ordering
Subjects with OCD (n ⫽ 215)
Control Subjects (n ⫽ 342)
34.1 ⫾ 10.8 (18–60)
103/112
11.6 ⫾ 3.0
136, 63.2%
19.5 ⫾ 8.3
17.5 ⫾ 8.5
14.6 ⫾ 10.0
39.8 ⫾ 11.9 (18–65)
202/140
24 (11.1%)
120 (55.8%)
40 (18.6%)
31 (14.4%)
25.8 ⫾ 6.0
13.0 ⫾ 3.1
12.9 ⫾ 3.6
12.4 ⫾ 6.2
109 (50.6%)
53 (24.6%)
16 (7.4%)
24 (11.1%)
16 (7.4%)
136 (63.2%)
39 (18.1%)
32 (14.8%)
35 (16.2 %)
30 (13.9%)
99 (46.0%)
146 (67.9%)
40 (18.6%)
58 (26.9%)
75 (34.8%)
GT, Gilles de la Tourette syndrome; HDRS, Hamilton Depression Rating Scale; NEC, not elsewhere classified; OCD,
obsessive-compulsive disorder; SRI, serotonin reuptake inhibitors; Y-BOCS, Yale-Brown Obsessive Compulsive Scale.
When comparing genotype distributions among patients according to the presence of family history of OCD, a significant
difference emerged in SNP rs6265, corresponding to the functional Val66Met variant, under an overdominant model (p ⫽
.008). Patients with family history of OCD were more frequently
heterozygous for this SNP than patients without this family
background. No significant differences in genotype distribution
for NTRK2 polymorphisms were detected among patients according to this criterion.
Haplotype Association Analysis
In BDNF, we identified two blocks with high LD (Supplement 2),
a short one of 1 kb and a large block of 79 kb containing the
functional Val66Met variant (rs6265). The results from the sliding
window analyses in BDNF pointed to a five-marker region
located toward the 5’ of the gene (Supplement 5). The association test with the common reconstructed haplotypes showed that
a significant haplotype was associated with OCD (OR ⫽ .80; 95%
CI ⫽ .69 –.92; permutation p value ⫽ .006). Thus, this composite
www.sobp.org/journal
haplotype confers a protective effect and contains the Val allele
of the functional Val66Met variant (Table 3).
In NTRK2, we identified 11 blocks with high LD (Supplement 1),
which ranged in size from 2 to 29 kb. Single nucleotide polymorphism rs2378672, significant in the univariate analysis and
associated with a protective effect, was located in block 10,
spanning 29 kb, and located in intron 19 of the gene. We further
used a sliding window approach (Supplement 6) and performed
the case-control haplotype association analysis of this particular
block in patients and control subjects as shown in Table 4. A
specific haplotype (G-C-G-G-A-C-A) was found to be protective
(OR ⫽ .21; 95% CI ⫽ .09 –.49; permutation p value ⫽ .006).
Interaction Analysis
The interaction effects between BDNF and NTRK2 were
assessed for all possible SNP combinations under an additive
model. No significant interaction was detected either between
SNPs or between the protective haplotypes of the two genes
(Figure 3).
P. Alonso et al.
BIOL PSYCHIATRY 2008;63:619 – 628 623
Figure 2. Log p values from likelihood ratio test for each SNP (in
BDNF and NTRK2 genes) depending
on different genetic models in obsessive-compulsive disorder patients. The eight first SNPs are those
of BDNF and in chromosome position order. The rest of the SNPs correspond to those of NTRK2 and are
also in chromosome position order.
The bold dotted line indicates significance upon Bonferroni correction. SNP, single nucleotide polymorphism.
Discussion
The aim of the present study was to explore whether genetic
variants at the BDNF and/or NTRK2 loci constitute risk factors for
OCD. While there have been two previous studies of specific
variants in BDNF, this is, to our knowledge, the first attempt to
evaluate common nucleotide variation across the genomic region
that encompasses both BDNF and its receptor NTRK2.
We could not find any single-marker association between the
BDNF SNPs under investigation and OCD. However, haplotype
analysis revealed a significant association between OCD and a
five-marker protective composite haplotype. This haplotype
extends 73 kb upstream of exon 1 and includes the functional
Val66Met variant (rs6265, located in exon 6). This variant, located
in the 5’ pro-BDNF sequence, affects intracellular trafficking and
secretion of BDNF. Studies assessing the involvement of the
Val66Met variant in mental disorders have produced conflicting
results. However, a recent meta-analysis suggests that this association is confined to substance-related disorders, eating disorders, and schizophrenia (27). The BDNF Met variant has been
associated with smaller hippocampal and occipital lobar gray
matter volumes and poor medial temporal lobe-related memory
performance both in schizophrenic patients and healthy subjects
(28). Studies in animal models recently reported that transgenic
BDNF⫹/Met and BDNFMet/Met mice show a deficit in activity-dependent release of BDNFMet from neurons and a decrease in dendritic
arbor complexity. Moreover, BDNFMet/Met mice display increases in
anxiety-related behaviors that did not respond to fluoxetine (13). So,
the Val allele may exert some kind of protective effect against
certain pathologies such as psychosis or adult OCD, while being
associated with increased genetic risk for other mental illnesses. In
fact, it has been postulated that the Val allele specifically affects the
liability to childhood psychiatric pathologies including juvenileonset mood disorders (29), attention-deficit/hyperactivity disorder
(ADHD) (30), or early-onset OCD (11). Moreover, our inability to
detect a single-marker association between the Val66Met polymorphism and OCD indicates that not only this SNP but other genetic
variants in tight LD with it are important in configuring an extended
functional haplotype, implicated in the susceptibility to OCD.
With respect to NTRK2, we found that SNP rs2378672, located
in intron 19, was significantly associated with the OCD phenotype in the subgroup of female patients. Gene deletion studies in
mice reveal that defects in NTRK2 signaling can have detrimental
www.sobp.org/journal
624 BIOL PSYCHIATRY 2008;63:619 – 628
P. Alonso et al.
Table 2. Stratified Analysis Covariation by Sex of the NTRK2 TagSNP rs2378672 in Patients with OCD
Males
Codominant
AA
AG
GG
Dominant
AA
AG-GG
Recessive
AA-AG
GG
Overdominant
AA-GG
GG
Additive
Females
Codominant
AA
AG
GG
Dominant
AA
AG-GG
Recessive
AA-AG
GG
Overdominant
AA-GG
GG
Additive
p Value
AICa
1.00
.33 (.11–.98)
.05
381.9
1.00
.31 (.10–.93)
.02
380.3
103 (100.00)
0 (.00)
1.00
.35
384.8
171 (89.06)
21 (10.94)
192 (65.1)
99 (96.12)
4 (3.88)
103 (34.9)
1.00
.33 (.11–.99)
.32 (.11–.93)
.03
380.8
.02
380.0
114 (85.71)
18 (13.53)
1 (.75)
109 (97.32)
3 (2.68)
0 (.00)
1.00
.17 (.05–.61)
.003
332.3
114 (85.71)
19 (14.29)
109 (97.32)
3 (2.68)
1.00
.17 (.05–.57)
.0008b
330.6
132 (99.25)
1 (.75)
112 (100.00)
0 (.00)
1.00
.27
340.6
115 (86.47)
18 (13.53)
133 (54.3)
109 (97.32)
3 (2.68)
112 (45.7)
1.00
.18 (.05–.61)
.17 (.05–.59)
.001
331.6
.0007b
330.3
Control Subjects n (%)
Patients n (%)
OR (95% CI)
170 (88.54)
21 (10.94)
1 (.52)
99 (96.12)
4 (3.88)
0 (.00)
170 (88.54)
22 (11.46)
99 (96.12)
4 (3.88)
191 (99.48)
1 (.52)
AIC, Akaike Information Criterion; CI, confidence interval; OCD, obsessive-compulsive disorder; OR, odds ratio.
AIC, Akaike Information Criterion, which attempts to find the minimal model that correctly explains the data.
b
Significant value at p ⬍ .0009 after Bonferroni correction for 54 independent comparisons.
a
effects on the viability and differentiation of neocortical neurons
(31) and increase impaired learning behavior and inappropriate
coping responses when facing complex and/or stressful learning
paradigms (32). NTRK2 contains an extracellular domain consisting of three tandem leucine-rich motifs flanked by two cysteine
clusters and two immunoglobulin (Ig)-like domains, a single
transmembrane segment and an intracellular domain possessing
a tyrosine-kinase domain. The positive susceptibility region that
we identified lies in intron 19, located in the region that contains
the exons coding for the tyrosine kinase domain of the protein
(exons 17 to 21). We investigated the putative functional effects
of rs2378672 through the interactive web-based SNP analysis tool
Pupasuite (http://pupasuite.bioinfo.cipf.es/). The results show
that rs2378672 lies in a mouse conserved region but does not
Table 3. Haplotype Association Results for a Five-Marker Window Across BDNF in OCD Patients
TagSNP
Haplotype
rs6265
rs12273363
H1
H2
H3
H4
H5
H6
H7
H8
H9
Rare (⬍1%)
G
A
A
G
G
G
G
G
G
T
T
T
C
C
T
T
T
T
rs908867
G
G
G
G
G
A
A
G
G
—
rs1491850
rs1491851
Frequency
OR
95% CI
p Valuea
T
C
C
C
C
T
T
C
T
T
C
T
C
T
C
T
T
C
.32
.18
.03
.10
.06
.05
.02
.03
.16
.02
Referent
.94
.93
1.10
.80
.96
1.08
1.22
.93
.85
(.86–1.02)
(.77–1.13)
(.99–1.23)
(.69–.92)
(.83–1.10)
(.88–1.34)
(1.01–1.47)
(.67–1.08)
(.67–1.08)
.14
.46
.08
.001 (.006)b
.51
.45
.04
.16
.18
Odds ratios (OR) and confidence intervals (CI) for tagSNP haplotypes using the most common haplotype (haplotype 1) as referent are shown.
BDNF, brain-derived neurotrophic factor; CI, confidence interval; OCD, obsessive-compulsive disorder; OR, odds ratio; tagSNP, tag single nucleotide
polymorphism.
a
p value adjusted covariation by sex.
b
p value obtained in the permutation analysis.
www.sobp.org/journal
BIOL PSYCHIATRY 2008;63:619 – 628 625
P. Alonso et al.
Table 4. Haplotype Combination of the NTRK2 LD Block Containing rs2378672, Significant in the Univariate Analysis, in OCD Patients
TagSNP
Haplotype
H1
H2
H3
H4
H5
H6
H7
H8
Rare (⬍1%)
rs10512159 rs1948308 rs2378672 rs1387926 rs1387924 rs3739570 rs1490403 Frequency
G
A
C
G
C
G
G
G
T
T
C
C
C
C
C
C
A
A
A
A
A
A
A
G
G
G
A
G
G
G
G
G
—
C
C
C
A
C
C
C
A
C
C
C
C
C
C
T
C
T
T
T
A
A
T
A
A
.42
.16
.12
.12
.01
.04
.07
.05
.014
OR
CI 95%
p Valuea
Referent
.81
.95
1.03
1.33
.59
.89
.21
.68
(.52–1.25)
(.63–1.44)
(.69–4.54)
(.47–3.79)
(.29–1.20)
(.54–1.47)
(.09–.49)
(.38–1.23)
.33
.82
.89
.59
.14
.65
.0003 (.0006)b
.20
Odds ratios (OR) and confidence intervals (CI) for tagSNP haplotypes using the most common haplotype (haplotype 1) as referent are shown.
CI, confidence interval; LD, linkage disequilibrium; OCD, obsessive-compulsive disorder; OR, odds ratio; tagSNP, tag single nucleotide polymorphism.
a
p value adjusted covariation by sex.
b
p value obtained in the permutation analysis.
alter splicing points that constitute the splicing signal, nor does it
disrupt the suggested regulatory regions for controlling gene
expression. The overall LD pattern in the region is low. Even
when the most recent HapMap data (Phase III, release 22; April
2007) are considered, no tightly linked markers are associated to
rs2378672 (r2 ⬍ .5). This suggests that, in the absence of a direct
causal implication of this SNP, the observed association of the
intron 19 polymorphism should be in linkage disequilibrium with
another unraveled functionally important sequence variant adjacent to the investigated NTRK2 gene region, which may be
population specific. Thus, confirmation of the association should
be corroborated by resequencing of the region to characterize
additional genetic information in our population. Furthermore,
analyses of NTRK2 messenger RNA (mRNA) levels and replication in other populations should be considered before definitive
conclusions can be derived.
Two previous reports have also shown the positive association of the NTRK2 gene with psychiatric disorders. In the first
one, a mutational screening in eating disorders patients found
that two three-marker haplotypes were associated with purging
anorexia nervosa and bulimia nervosa (33). In the second study,
a family-based analysis showed a positive association between
allelic variants of the NTRK2 gene and three nicotine dependence
measures (34). Interestingly, involvement of NTRK2 in the development of eating disorders reported by Ribases et al. (33)
appears to depend not only on its direct participation in food
intake and body weight regulation, but also on modulation of
harm avoidance, a personality trait highly correlated with OCD
(35). Further studies should clarify whether the contribution of
NTRK2 to the shared genetic susceptibility between OCD and
eating disorders depends on its influence on genetic variants of
personality dimensions commonly associated with both disorders.
The gender-selective nature of the association between
NTRK2 polymorphisms and OCD is congruent with the previously reported sexually dimorphic pattern of genetic susceptibility to OCD (8,36). Although published results are highly controversial (6,37,38), our results suggest that this dimorphic pattern
may extend to genes outside the monoamine pathways. Estrogens have recently been shown to modify BDNF influence in the
brain serotonergic system—the neurotransmitter pathway most
clearly related to OCD—via its NTRK2 receptor. Female serotonin transporter (SERT) and BDNF-deficient mice show significantly lesser reductions in serotonin concentration in the hypothalamus and other brain regions and no increase in anxiety-like
behaviors compared with male SERT ⫻ BDNF-deficient mice,
suggesting that estrogen availability exerts a protective effect
against the monoamine depletion associated with these genetic
deficiencies (39). Therefore, women who inherit certain NTRK2
polymorphisms may somehow lose part of this estrogen protective effect, which, together with other as yet unknown genetic
factors, may influence BDNF expression, resulting in an increased susceptibility to OCD. Nevertheless, our results must be
considered with caution since the significant difference on
gender distribution between OCD patients and control subjects
in our sample may have influenced this specific outcome.
We were not able to detect any significant association between BDNF or NTRK2 variants and age at onset of OCD.
Although greater familial loading for OCD has been associated
with early onset of the disorder, other studies have described, as
in our case, a lack of any difference between adults with earlyand late-onset OCD in terms of their family histories of obsessive
illness (40). It has recently been emphasized that although
familial aggregation is largely concentrated in families with
early-onset OCD probands, between 43% and 66% of these
early-onset patients are not familial cases (16,41). So, age of
onset does not appear to be a sufficient clinical index for
individualizing the familial character of OCD and its genetic
basis.
Since BDNF has been postulated to contribute to genetic risk
to affective disorders, the most frequent comorbid condition in
OCD, we decided to analyze our results considering the presence
of comorbid affective diagnosis. Our negative results suggest that
the influence of BDNF and NTRK2 on OCD susceptibility is not
related to comorbid affective disorders. Similar analyses considering comorbid eating disorders would have been very interesting but could not be performed due to the limited number of
patients who met both conditions.
Significant genetic differences emerged when considering
familiarity of OCD. Our results for the association between the
heterozygosity of the Val66Met variant and family OCD may
suggest a case of positive molecular heterosis related to family
forms of the disorder. Molecular heterosis occurs when subjects
heterozygous for a polymorphism associated with a single gene
show a significantly greater or lesser effect for a quantitative or
dichotomous trait than homozygotes. Examples of molecular
heterosis are quite frequent in humans and have been described
in genes of the serotonergic and dopaminergic pathways involved in the pathophysiology of several psychiatric disorders
(42). Brain-derived neurotrophic factor is known to exist as a
www.sobp.org/journal
626 BIOL PSYCHIATRY 2008;63:619 – 628
P. Alonso et al.
Figure 3. Analysis of interaction
(epistasis) between SNPs at BDNF
and NTRK2 in obsessive-compulsive disorder patients using an additive model. The plot contains the
p values obtained from different
likelihood ratio tests. Different colors indicate different statistical significant levels. The diagonal contains the p values from likelihood
ratio test for the crude effect of
each SNP. The upper triangle in matrix contains the p values for the
interaction log-likelihood ratio test.
Finally, the lower triangle contains
the p values from LRT comparing the
two-SNP additive likelihood with the
best of the single-SNP models. LRT,
likelihood ratio test; SNP, single nucleotide polymorphism.
tightly associated homodimer, which is required for high affinity
binding to its dimeric receptor (43). In this particular case, the
two structurally different BDNF subunits may distort the ability of
the dimer to interact with the NTRK2 receptor.
Published studies suggest that genetic factors may be specially
implicated in susceptibility to certain clinical obsessive-compulsive
dimensions, including aggressive/sexual/religious (44), symmetry/
ordering (44), and, in particular, hoarding symptoms (36,45). Although we tested for such an association in the case of BDNF and
NTRK2 polymorphisms, we were not able to detect any significant
genetic variant. Hoarding, symmetry/ordering, and sexual/religious obsessions and compulsions were present in less than one
third of our sample; therefore, our negative results may be
related to a type II error. Recent work in the OCD field has led to
the development of the Dimensional Yale-Brown ObsessiveCompulsive Scale (DY-BOCS), an instrument that permits meawww.sobp.org/journal
surement of the presence and severity of obsessive-compulsive
symptoms within distinct dimensions (46). The use of a dimensional approach of this kind may constitute a useful tool for
analyzing the genetic basis of the different obsessive-compulsive
symptoms, avoiding their consideration as categorically excluding subtypes. Alternatively, our negative results may suggest that
BDNF and NTRK2 polymorphisms increase overall susceptibility
to OCD, while other coexisting genetic or environmental factors
determine the specific pattern of presentation of the disorder.
Some of the shortcomings of the study include the fact that
our patients were recruited from specialized OCD clinics and
may not be representative of the community. The information
regarding age of onset was obtained retrospectively, introducing
the possibility that patient recall may have been inaccurate. As in
all case-control designs, population stratification may constitute a
confounding factor. On one hand, the Spanish population is
P. Alonso et al.
highly homogeneous (47), and on the other, structure convincingly identified only one population stratum among cases and
control subjects. Nevertheless, subtle or weak admixture might
have not been identified with the limited number of genotyped
ancestry informative markers (AIMs) and thus cannot be completely discarded. Finally, we used a group of psychiatrically
unscreened blood donors as the control group, which reduces
the power to detect associations. Moskvina et al. (48) have
recently concluded that for real-world situations in complex
genetics, the use of unscreened control subjects is potentially
cost-effective and can be considered for disorders with population prevalence lower than 20%. So, since OCD has a population
lifetime risk of approximately 1%, the use of unscreened control
subjects, although not ideal, can be considered to have a
negligible effect on power.
In sum, our results add further evidence to the complex
pattern of inheritance of OCD and suggest that BDNF/NTRK2
signaling may be one of the molecular pathways involved in its
pathogenesis. Additional studies that examine the role of the
genes involved in the neurotrophic systems and their interaction
with other neurotransmitter pathways or hormonal factors are
required to determine whether variant alleles at these loci might
play an etiologic role in the expression of OCD. Further dissection of the samples using clinical phenotypes may also help to
disentangle the genetic heterogeneity of OCD. Finally, possible
protective BDNF and NTRK2 allelic combinations against OCD
should be identified and their effect on other neurotransmitter
systems explored, since they could constitute a promising pharmacotherapy target.
PA and MG contributed equally to this article.
Financial support was received from the Psychiatry Genetics
Network (G03/184) and Red de Enfermedades Mentales (REMTAP Network) (Spanish Ministry of Health, Carlos III Research
Institute), the Ministry of Education and Science (SAF200501005), the Fondo de Investigaciones Sanitarias de la Seguridad
Social (FIS PI40632 and FIS PI052307; CIBER-CB06/03/0034),
the Department of Health (Generalitat de Catalunya), and the
Department of Universities, Research and Information Society
(2005SGR00008; 2005SGR322) (Generalitat de Catalunya).
We thank all the study participants and the staff from the
Department of Psychiatry of Hospital Universitari de Bellvitge and
Hospital Ramón y Cajal who collaborated to obtain the sample of
this study, and Michael Maudsley for revising the manuscript.
The authors have no conflict of interests.
Supplementary material cited in this article is available
online.
1. Carey G, Gottesman II (1981): Twin and family studies of anxiety, phobic
and obsessive disorders. In: Klein DF, Rabkin JG, editors. Anxiety: New
Research and Changing Concepts. New York: Raven.
2. Hettema JM, Neale MC, Kendler KS (2001): A review and meta-analysis of
the genetic epidemiology of anxiety disorders. Am J Psychiatry 158:
1568 –1578.
3. Mundo E, Richter MA, Sam F, Macciardi F, Kennedy JL (2000): Is the
5-HT(1Dbeta) receptor gene implicated in the pathogenesis of obsessive-compulsive disorder? Am J Psychiatry 157:1160 –1161.
4. Mundo E, Richter MA, Zai G, Sam F, McBride J, Macciardi F, et al. (2002):
5HT1Dbeta receptor gene implicated in the pathogenesis of obsessivecompulsive disorder: Further evidence from a family-based association
study. Mol Psychiatry 7:805– 809.
5. Enoch MA, Greenberg BD, Murphy DL, Goldman D (2001): Sexually
dimorphic relationship of a 5-HT2A promoter polymorphism with obsessive-compulsive disorder. Biol Psychiatry 49:385–388.
BIOL PSYCHIATRY 2008;63:619 – 628 627
6. Karayiorgou M, Sobin C, Blundell ML, Galke BL, Malinova L, Goldberg P,
et al. (1999): Family-based association studies support a sexually dimorphic effect of COMT and MAOA on genetic susceptibility to obsessivecompulsive disorder. Biol Psychiatry 45:1178 –1189.
7. Niehaus DJ, Kinnear CJ, Corfield VA, du Toit PL, van Kradenburg J,
Moolman-Smook JC, et al. (2001): Association between a catechol-omethyltransferase polymorphism and obsessive-compulsive disorder
in the Afrikaner population. J Affect Disord 65:61– 65.
8. Camarena B, Rinetti G, Cruz C, Gomez A, de La Fuente JR, Nicolini H
(2001): Additional evidence that genetic variation of MAO-A gene supports a gender subtype in obsessive-compulsive disorder. Am J Med
Genet 105:279 –282.
9. Huang EJ, Reichardt LF (2001): Neurotrophins: Roles in neuronal development and function. Annu Rev Neurosci 24:677–736.
10. Chao MV (2003): Neurotrophins and their receptors: A convergence
point for many signalling pathways. Nat Rev Neurosci 4:299 –309.
11. Hall D, Dhilla A, Charalambous A, Gogos JA, Karayiorgou M (2003):
Sequence variants of the brain-derived neurotrophic factor (BDNF)
gene are strongly associated with obsessive-compulsive disorder. Am J
Hum Genet 73:370 –376.
12. Zai G, Arnold P, Strauss J, King N, Burroughs E, Richter MA, et al. (2005):
No association between brain-derived neurotrophic factor gene and
obsessive-compulsive disorder. Psychiatr Genet 15:235.
13. Chen ZY, Jing D, Bath KG, Ieraci A, Khan T, Siao CJ, et al. (2006): Genetic
variant BDNF (Val66Met) polymorphism alters anxiety-related behavior.
Science 314:140 –143.
14. American Psychiatric Association (1994): Diagnostic and Statistical Manual
of Mental Disorders, 4th ed. Washington, DC: American Psychiatric Press.
15. First MB, Spitzer RL, Gibbon M, Williams JBW (1997): Structured Clinical
Interview for DSM-IV Axis I Disorders-Clinician Version (SCID-CV). Washington, DC: American Psychiatric Press, Inc.
16. Rosario-Campos MC, Leckman JF, Curi M, Quatrano S, Katsovitch L,
Miguel EC, et al. (2005): A family study of early-onset obsessive-compulsive disorder. Am J Med Genet B Neuropsychiatr Genet 136:92–97.
17. Hemmings SM, Kinnear CJ, Lochner C, Niehaus DJ, Knowles JA,
Moolman-Smook JC, et al. (2004): Early- versus late-onset obsessivecompulsive disorder: Investigating genetic and clinical correlates.
Psychiatry Res 128:175–182.
18. Goodman WK, Price LH, Rasmussen SA, Mazure C, Fleischmann RL, Hill
CL, et al. (1989): The Yale-Brown Obsessive Compulsive Scale. I. Development, use, and reliability. Arch Gen Psychiatry 46:1006 –1011.
19. Hamilton M (1960): A rating scale for depression. J Neurol Neurosurg
Psychiatry 23:56 – 61.
20. Mataix-Cols D, Rauch SL, Manzo PA, Jenike MA, Baer L (1999): Use of
factor-analyzed symptom dimensions to predict outcome with serotonin reuptake inhibitors and placebo in the treatment of obsessivecompulsive disorder. Am J Psychiatry 156:1409 –1416.
21. Andreasen NC, Endicott J, Spitzer RL, Winokur G (1977): The family
history method using diagnostic criteria. Reliability and validity. Arch
Gen Psychiatry 34:1229 –1235.
22. Carlson CS, Eberle MA, Rieder MJ, Yi Q, Kruglyak L, Nickerson DA (2004):
Selecting a maximally informative set of single-nucleotide polymorphisms for association analyses using linkage disequilibrium. Am J Hum
Genet 74:106 –120.
23. Chapman JM, Cooper JD, Todd JA, Clayton DG (2003): Detecting disease
associations due to linkage disequilibrium using haplotype tags: A class
of tests and the determinants of statistical power. Hum Hered 56:18 –31.
24. Gonzalez JR, Armengol L, Sole X, Guino E, Mercader JM, Estivill X, et al.
(2007): SNPassoc: An R package to perform whole genome association
studies. Bioinformatics 23:644 – 645.
25. Efron B, Tibshirani RJ (1993): An Introduction to the Bootstrap. New York:
Chapman & Hall.
26. Wang N, Akey JM, Zhang K, Chakraborty R, Jin L (2002): Distribution of
recombination crossovers and the origin of haplotype blocks: The interplay of population history, recombination, and mutation. Am J Hum
Genet 71:1227–1234.
27. Gratacòs M, Gonzalez JR, Mercader JM, de Cid R, Urretavizcaya M, Estivill
X (2007): BDNF val66met and psychiatric disorders: Meta-analysis of
case-control studies confirm association to substance-related disorders, eating disorders and schizophrenia. Biol Psychiatry 61:911–922.
28. Ho BC, Milev P, O’Leary DS, Librant A, Andreasen NC, Wassink TH (2006):
Cognitive and magnetic resonance imaging brain morphometric correlates of brain-derived neurotrophic factor Val66Met gene polymor-
www.sobp.org/journal
628 BIOL PSYCHIATRY 2008;63:619 – 628
29.
30.
31.
32.
33.
34.
35.
36.
37.
phism in patients with schizophrenia and healthy volunteers. Arch Gen
Psychiatry 63:731–740.
Strauss J, Barr CL, George CJ, Devlin B, Vetro A, Kiss E, et al. (2005):
Brain-derived neurotrophic factor variants are associated with childhood-onset mood disorder: Confirmation in a Hungarian sample. Mol
Psychiatry 10:861– 867.
Kent L, Green E, Hawi Z, Kirley A, Dudbridge F, Lowe N, et al. (2005):
Association of the paternally transmitted copy of common valine allele
of the val66met polymorphism of the brain-derived neurotrophic factor
(BDNF) gene with susceptibility to ADHD. Mol Psychiatry 10:939 –943.
Gates MA, Tai CC, Macklis JD (2000): Neocortical neurons lacking the
protein-tyrosine kinase B receptor display abnormal differentiation and
process elongation in vitro and in vivo. Neuroscience 98:437– 447.
Minichiello L, Korte M, Wolfer D, Kuhn R, Unsicker K, Cestari V, et al.
(1999): Essential role for TrkB receptors in hippocampus-mediated
learning. Neuron 24:401– 414.
Ribases M, Gratacos M, Badia A, Jimenez L, Solano R, Vallejo J, et al.
(2005): Contribution of NTRK2 to the genetic susceptibility to anorexia
nervosa, harm avoidance and minimum body mass index. Mol Psychiatry 10:851– 860.
Beuten J, Ma JZ, Payne TJ, Dupont RT, Lou XY, Crews KM, et al. (2007):
Association of specific haplotypes of neurotrophic tyrosine kinase receptor 2 gene (NTRK2) with vulnerability to nicotine dependence in
African-Americans and European-Americans. Biol Psychiatry 61:48 –55.
Alonso P, Menchón JM, Jiménez S, Segalàs J, Mataix-Cols D, Jaurrieta N,
et al. (in press): Personality dimensions in obsessive-compulsive disorder: Relation with clinical variables. Psychiatry Res.
Lochner C, Kinnear CJ, Hemmings SM, Seller C, Niehaus DJ, Knowles JA,
et al. (2005): Hoarding in obsessive-compulsive disorder: Clinical and
genetic correlates. J Clin Psychiatry 66:1155–1160.
Karayiorgou M, Altemus M, Galke BL, Goldman D, Murphy DL, Ott J, et al.
(1997): Genotype determining low catechol-O-methyltransferase activity
as a risk factor for obsessive-compulsive disorder. Proc Natl Acad Sci U S A
94:4572– 4575.
www.sobp.org/journal
P. Alonso et al.
38. Alsobrook JP 2nd, Zohar AH, Leboyer M, Chabane N, Ebstein RP, Pauls
DL (2002): Association between the COMT locus and obsessive-compulsive disorder in females but not males. Am J Med Genet 114:116 –120.
39. Ren-Patterson RF, Cochran LW, Holmes A, Lesch KP, Lu B, Murphy DL
(2006): Gender-dependent modulation of brain monoamines and anxiety-like behaviors in mice with genetic serotonin transporter and BDNF
deficiencies. Cell Mol Neurobiol 26:753–778.
40. Sobin C, Blundell ML, Karayiorgou M (2000): Phenotypic differences in
early- and late-onset obsessive-compulsive disorder. Compr Psychiatry
41:373–379.
41. Chabane N, Delorme R, Millet B, Mouren MC, Leboyer M, Pauls D (2005):
Early-onset obsessive-compulsive disorder: A subgroup with a specific
clinical and familial pattern? J Child Psychol Psychiatry 46:881– 887.
42. Comings DE, MacMurray JP (2000): Molecular heterosis: A review. Mol
Genet Metab 71:19 –31.
43. Radziejewski C, Robinson RC, DiStefano PS, Taylor JW (1992): Dimeric
structure and conformational stability of brain-derived neurotrophic
factor and neurotrophin-3. Biochemistry 31:4431– 4436.
44. Alsobrook IJ, Leckman JF, Goodman WK, Rasmussen SA, Pauls DL (1999):
Segregation analysis of obsessive-compulsive disorder using symptombased factor scores. Am J Med Genet 88:669 – 675.
45. Zhang H, Leckman JF, Pauls DL, Tsai CP, Kidd KK, Campos MR (2002):
Genomewide scan of hoarding in sib pairs in which both sibs have Gilles
de la Tourette syndrome. Am J Hum Genet 70:896 –904.
46. Rosario-Campos MC, Miguel EC, Quatrano S, Chacon P, Ferrao Y, Findley
D, et al. (2006): The Dimensional Yale-Brown Obsessive-Compulsive
Scale (DY-BOCS): An instrument for assessing obsessive-compulsive
symptom dimensions. Mol Psychiatry 11:495–504.
47. Calafell F, Bertranpetit J (1994): Principal component analysis of gene
frequencies and the origin of Basques. Am J Phys Anthropol 93:201–215.
48. Moskvina V, Holmans P, Schmidt KM, Craddock N (2005): Design of
case-control studies with unscreened controls. Ann Hum Genet 69:566 –
576.