SEARCH

SEARCH BY CITATION

Keywords:

  • Attention deficit hyperactivity disorder;
  • CDH13 gene;
  • executive functioning;
  • haplotype analysis;
  • working memory

Abstract

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments
  8. Supporting Information

Different analytic strategies, including linkage, association and meta-analysis support a role of CDH13 in the susceptibility to attention deficit/hyperactivity disorder (ADHD). CDH13 codes for cadherin 13 (or H-cadherin), which is a member of a family of calcium-dependent cell–cell adhesion proteins and a regulator of neural cell growth. We tested the association between CDH13 on three executive functioning tasks that are promising endophenotypes of ADHD. An adjusted linear regression analysis was performed in 190 ADHD-affected Dutch probands of the IMAGE project. Three executive functions were examined: inhibition, verbal and visuo-spatial working memory (WM). We tested 2632 single nucleotide polymorphisms (SNPs) within CDH13 and 20 kb up- and downstream of the gene (capturing regulatory sequences). To adjust for multiple testing within the gene, we applied stringent permutation steps. Intronic SNP rs11150556 is associated with performance on the Verbal WM task. No other SNP showed gene-wide significance with any of the analyzed traits, but a 72-kb SNP block located 446 kb upstream of SNP rs111500556 showed suggestive evidence for association (P-value range 1.20E-03 to 1.73E-04) with performance in the same Verbal WM task. This study is the first to examine CDH13 and neurocognitive functioning. The mechanisms underlying the associations between CDH13 and the clinical phenotype of ADHD and verbal WM are still unknown. As such, our study may be viewed as exploratory, with the results presented providing interesting hypotheses for further testing.

Multiple studies point the influence of genetic factors in the etiology of attention deficit/hyperactivity disorder (ADHD; heritability ∼70%) (Faraone et al. 2005; Nikolas et al. 2010). Its etiology is complex: combinations of genetic factors with small effect size interacting with each other and with environmental factors seem to contribute (Faraone et al. 2005). So far, only a small part of the genetic component of the complex clinical phenotype of ADHD has been explained (Franke et al. 2009). With the recent advent of genome-wide association studies (GWAS), the identification of new genes for multifactorial diseases and traits has become very successful (Manolio et al. 2008). However, the performance of GWAS in the psychiatric disorders, including ADHD, has been particularly poor (Franke et al. 2009; Manolio et al. 2009). One of the reasons probably is the largely suboptimal nature of the clinical psychiatric phenotypes, which are based on a categorization of symptom clusters with no proven biological significance.

Endophenotypes, heritable traits that are associated with a disorder, are hypothesized to be more suitable for detecting risk genes than the clinical phenotypes because they are genetically less complex by being etiologically closer to disease genes (Aron et al. 2005; Gottesman et al. 2003). Systematic reviews and meta-analyses indicate that response inhibition and both verbal and visuo-spatial working memory (WM) are impaired in subjects with ADHD (Oosterlaan and Sergeant 1998; Oosterlaan et al. 1998; Willcutt et al. 2005). We and others have found similar impairments in unaffected siblings of ADHD probands and significant between-sibling correlations, indicating these measures to be useful as endophenotypes of ADHD (McInnes et al. 2003; Rommelse et al. 2008b,d).

ADHD linkage and association studies position Cadherin 13 (CDH13; OMIM = 601364) as an interesting candidate gene for ADHD (Franke et al. 2009). In a recent ADHD GWAS, single nucleotide polymorphism (SNP) rs6565113, an intronic SNP in CDH13, was found to be associated with symptom count variables (Franke et al. 2009; Lasky-Su et al. 2008). Moreover, CDH13 is located in the only genome-wide significant locus identified in a meta-analysis of linkage studies using ADHD-affected status as a phenotype (Asherson et al. 2008; Zhou et al. 2008) and a SNP near this gene also is part of the top-25 in the ADHD-affected status GWAS (Neale et al. 2008). A meta-analysis of four ADHD GWAS also implicated the CDH13 gene in ADHD (Neale et al. 2010).

CDH13 codes for cadherin 13, a member of a family of calcium-dependent cell–cell adhesion proteins (Patel et al. 2003) and a regulator of neural cell growth. The broad distribution of H-cadherin in midbrain and telencephalon suggests that it may play an important role in building and maintaining neural circuitry (Takeuchi et al. 2000).

In this study, we aimed to examine the relationship of CDH13 with selected neurocognitive endophenotypes of ADHD and evaluate the association with variation in three tasks of executive functioning in ADHD-affected children from the Dutch subsample of the International Multicentre ADHD Genetics (IMAGE) project (Kuntsi et al. 2006).

Methods

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments
  8. Supporting Information

Participants

The participating Dutch ADHD-affected probands were part of a larger sample of the International IMAGE study, which were also administered to neurocognitive tasks. This is an international collaborative study in seven European countries (Belgium, Germany, Ireland, Spain, Switzerland, the Netherlands and the United Kingdom) and Israel that aims to identify genes that increase the risk for ADHD using linkage and association strategies. Ethical approval for the study was obtained from National Institutes of Health recognized local ethical review boards and all families gave written informed consent prior to participation. All participants were aged 5–17 and of European Caucasian descent. Exclusion criteria included IQ < 70, presence of autism, epilepsy, brain disorders and any genetic or medical disorder associated with externalizing behaviors that might mimic ADHD. Details of the sample collection for this study and screening procedures are described elsewhere (Brookes et al. 2006; Kuntsi et al. 2006). In short, the children were screened with the following rating scales: parent and teacher Conners' long version rating scales (Conners et al. 2003) and the Strength and Difficulties Questionnaire (Goodman 2001). Only clinical cases with an average T-score of the DSM-IV total symptom score (N-scale) greater than 63 on the Conners scales and scores >90th percentile on the SDQ-hyperactivity scale were recruited into IMAGE. Subsequently, the clinical diagnosis of the probands was verified with the Parental Account of Childhood Symptoms, PACS (Taylor et al. 1986). PACS is a semi-structured, standardized, investigator-based, parent-informed interview developed as an instrument to provide an objective measure of children's behavior. A standardized algorithm for PACS was applied to all raw PACS data to yield a diagnosis based on operational DSM-IV criteria for ADHD.

Neuropsychological measures

The participants were tested at one of the three sites (Karakter Child and Adolescent University Centre in Nijmegen, Accare Child and Adolescent University Centre Groningen or at the Vrije Universiteit in Amsterdam). Testing required approximately 2–3 h per child and took place in a quiet test room by experienced child and adolescent psychologists and trained undergraduate students. The same protocol at each site was used to reduce variability in instructions. The child and adolescent psychologists and trained undergraduate students were trained by the same chief investigators and every person who applied the test was given a code. During data-cleaning, we checked for extreme outliers which could be because of wrong administration or computer problems. Siblings of the same family were tested simultaneously. The administration of neuropsychological tests was counterbalanced to rule out possible effects of fatigue on the tests. Psychostimulants (e.g. methylphenidate) were required to be withdrawn for at least 2 days prior to neurocognitive assessment, as they may positively influence a variety of neuropsychological functions (Kempton et al. 1999). The withdrawal of other medication was carried out depending on plasma half life.

The three measures investigated in this work are the stop signal reaction time and performance on Visuo-Spatial Sequencing and Digit Span backwards. The tasks have been fully described elsewhere (Rommelse et al. 2007a,b,c; Rommelse et al. 2008b,d). In short, The Stop Task aims to measure motor inhibition of an ongoing response (Logan et al. 1984). Go-trials consisted of a drawing of a plane that was either pointing to the right or to the left (Scheres et al. 2006). Children pressed a response button that corresponded to the direction of the stimulus as quickly and as accurately as possible. Stop-trials were identical to the go-stimulus, but a stop-signal was presented (drawing of a cross that was superimposed on the plane). During the stop-signal trials, children were required to withhold their response. The latency of the stop-process, the stop signal reaction time (SSRT), we used as the independent variable in the current analysis.

The Visuo-Spatial Sequencing task was used to measure accuracy of visuo-spatial WM (Rommelse et al. 2008c). Nine circles were symmetrically organized in a square (3 × 3). On each trial, a sequence of circles was pointed at by a computer-driven hand. Children were instructed to replicate the exact same sequence of circles, by pointing to them with the small, self-driven hand. Difficulty increased with the number of targets to remember and the complexity of the spatial pattern. The dependent measure we used was the total number of correct targets in the correct order.

A subtest of the WISC-III or WAIS-III, Digit Span (DS), was used to measure verbal WM (Wechsler 2000; Wechsler 2002). It has two presentations, DS-Forward and DS-Backward, the latter of which is a robust indicator of executive function, requiring the manipulation of information in memory. Forward digit span requires the individual to store and reproduce a digit sequence in its correct serial order, where the number of digits to be remembered is progressively increased over successive trials. Children were instructed to reproduce sequences as accurately as possible. In the backward condition, the child repeated the numbers in the opposite order. The highest number of digits recalled in the backward condition (maximum span backwards) was used as the dependent variable, because this aspect of the task places highest demand on the WM system.

A Van der Waerden transformation was applied (SPSS version 16) to the performance-derived variables for each task for normalization and standardization purposes. The z-scores of the performance measures on Digit Span and Visuo-Spatial Sequencing were mirrored, so that all dependent measures would have the same meaning, with a higher score reflecting a worse performance.

DNA collection, genotyping and association analysis

Directly after collection, blood samples were sent to Rutgers University Cell and DNA Repository, NJ, USA, where DNA was extracted from part of the blood or from immortalized cell lines. Details of the genotyping and data cleaning process have been reported elsewhere (Neale et al. 2008). In short, genome-wide genotyping was performed by Perlegen Sciences using the Perlegen platform. The Perlegen Array has 600 000 tagging SNPs designed to be in high linkage disequilibrium (LD) with untyped SNPs for three HapMap populations (CEU, YRB and CHB/JPN). Genotype data cleaning and quality control procedures were performed by The National Center for Biotechnology Information (NCBI) using the GAIN QA/QC Software Package (version 0.7.4) developed by Gonçalo Abecasis and Shyam Gopalakrishnan at the University of Michigan. Data were excluded on the basis of the following quality control metrics: (1) call rate < 95%; (2) gender discrepancy; (3) per-family Mendelian errors > 2; (4) sample heterozygosity < 32%; (5) genotype call quality score cut-off < 10; (6) a combination of SNP call rate and minor allele frequency (MAF) [(a) 0.01 ≤ MAF < 0.05 and call rate ≥ 99%; (b) 0.05 ≤ MAF < 0.10 and call rate ≥ 97% and (c) 0.10 ≥ MAF and call rate ≥ 95%]; (7) Hardy–Weinberg equilibrium P-value < 0.0001 and (8) duplicate sample discordance. With this filtering, 438 784 SNPs were retained in the final dataset.

To increase coverage in the targeted genomic areas, we used the imputation approach implemented in MACH software (http://www.sph.umich.edu/csg/yli/mach/download/) and imputed Hapmap II release 22 genotypes into our dataset. The imputed data underwent an extra QC step in which SNPs with an imputation score (RSQR in MACH) < 0.3 and MAF < 0.05 were excluded. After this step, we ended up with a total of 2 182 904 SNPs across the genome. Genotypes for the CDH13 gene were extracted from this SNP set. Genotypes and neuropsychological measures were available for 190 ADHD-affected children.

Descriptive information on age, gender, Conner's scores and the three measures investigated was analyzed using SPSS v16 for windows. Association for the QC-approved 190 datasets was performed using the linear regression option in PLINK v1.07 (http://pngu.mgh.harvard.edu/~purcell/plink/) adjusting by age and gender. We tested 2632 SNPs within CDH13 and 20 kb up- and downstream of the gene (aiming to capture many of the relevant regulatory sequences that might be involved with the CDH13 gene). To adjust for multiple testing, we ran 10 000 max(T) permutation tests for all SNPS using the -mperm command in PLINK and used a threshold for significance = 0.05 for the empirical P value. This is achieved by comparing the observed test statistic against the maximum of all permuted statistics for each replicate. The P-value now controls for multiple comparisons because it indicates the probability of detecting a test statistic this large, given the total number of tests performed.

We also performed a haplotype analysis for the CDH13 SNPs. Given the analytical limitations of estimating haplotypes for 2632 SNPs, we restricted our analysis to SNP rs11150556 and its 10-kb flanking region. Haplotypes were estimated using the ‘haplo.em’ functions implemented in the haplo.stats package (Schaid et al. 2002). Haplotype association analyses were carried out in a 3-SNP sliding window design scanning the effect of the selected CDH13 SNPs using the haplo.score.slide function implemented in haplo.stats. This approach allows adjustment for covariates. This analysis was corrected for multiple testing by applying the simulate = TRUE parameter in haplo.score.slide (Schaid et al. 2002).

Results

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments
  8. Supporting Information

A total of 238 ADHD children were available in our study. Complete data: performance on the WM tasks and the inhibition task, and SNP genotypes, were available for 190 ADHD-affected probands of the Dutch IMAGE sample. Table 1 shows the descriptive statistics of our study sample including the performance measures on WM tasks, the SSRT and the clinical ADHD variables. A total of 98.7% of our sample was affected with the ADHD combined subtype and 84.9% was male.

Table 1.  Descriptive information of the study group
Dutch IMAGE sample
  1. *One hundred and ninety children have available GWAS data.

Probands (n)238*
% combined ADHD type98.7
Mean Inattention Conner's Score (SD)71.12 (8.43)
Mean Hyperactive-Impulsive Conner's Score (SD)79.11 (9.25)
Mean Verbal working memory Task Score (SD)0.034 (0.86)
Mean Visio-spatial working memory Task Score (SD)0.123 (0.86)
Mean Inhibition Task Score (SD)0.069 (1.00)
Mean age in years (SD)11.99 (2.50)
% males84.90

Table 2 shows the correlation measures between the ADHD variables and the three measures investigated: as expected, the Conner's Hyperactive/Impulsive and the Conner's Inattentive scores showed the highest correlation (0.486, P < 0.01). The correlation between the neurocognitive endophenotypes of ADHD ranged from 0.405 to 0.261, P value <0.01. Interestingly, the correlations between the verbal WM task and ADHD were low and non-significant (Table 2).

Table 2.  Correlation between working memory tasks and ADHD scores and symptom count
Trait Conner's Hyperactive/Impulsive ScoreVerbal working memory TaskVisio-spatial working memory TaskInhibition TaskHyperactive-impulsive symptom countInattentive symptom count
  1. *Correlations significant at the 0.05 level (in bold).

  2. **Correlations significant at the 0.01 level (in bold).

Conner's Inattentive Score ρ (N) 0.486 ** (238)0.27 (238)0.24 (238) 0.036 (210)−0.080 (237) 0.174 ** (237)
Conner's Hyperactive- Impulsive Score ρ (N) −0.68 (238)−0.064 (238) 0.183 ** (210) 0.226 ** (237)0.022 (237)
Verbal working memory Task ρ (N)   0.405 ** (238) 0.261 ** (210)0.103 (237)−0.040 (237)
Visio-spatial working memory Task ρ (N)    0.238 ** (210) 0.104 * (237)−0.66 (237)
Inhibition Task ρ (N)    0.039 (209)−0.24 (209)
Hyperactive-impusive symptom count ρ (N)      0.155 * (237)

All SNPs showing association with at least one neurocognitive task at uncorrected P-values <1.00E-02 (1 df linear test from the – linear command in plink) plus SNPs associated with ADHD status from the meta-analysis of Neale et al. (2010) as well as SNP rs6565113 associated with ADHD quantitative phenotype (Lasky-Su et al. 2008) are shown in Table 3. Analysis of the 2632 CDH13 SNPs identified SNP rs11150556 to be gene-wide significantly associated with performance on the Verbal WM task (after 10 000 Max(T) permutations). Carriers of the C/C genotype showed a significantly worse performance compared to C/T and T/T carriers (TEST STAT = 4.177, β = 0.3523, additive modelP = 4.58E − 05, Fig. 1). This intronic SNP is located less than 100 bp downstream of SNP rs7184058, which showed nominally significant association with ADHD status (P = 0.001121) in the meta-analysis by Neale et al. (2010). Permutation analysis showed that the association of this SNP with Verbal WM was significant (EmpP10000 permutations = 0.036, Table 3). This association signal remained significant after including Conner's inattentive and hyperactive-impulsive scores in the analysis (TEST STAT = 4.031, β = 0.3426, additive model P = 8.21E − 05).

Table 3.  Association between SNPs in CDH13 and working memory in ADHD-affected children Thumbnail image of
image

Figure 1. Mean values of Verbal working memory Task by SNP rs1115056 genotypes.

Download figure to PowerPoint

Our analysis also identified suggestive evidence of associations with Verbal WM performance for a SNP-cluster in high LD spanning 72 kb and located 446 kb upstream of rs11150556 (Table 3 and Fig. S1, Supporting Information). The minor allele frequencies in this cluster range from 8 to 38% and were not in apparent LD with rs11150556 (Fig. S1).

In a haplotype analysis of 62 3-SNP haplotypes (64 SNPs included by selecting SNP rs11150556 and the 10-kb flanking regions), we found a total of five haplotypes associated with WM (maximum simulated P≤ 0.05; Table S1 and Fig. S2, Supporting Information). Haplotypes containing SNP rs11150556 showed the most significant associations with WM, this result is consistent with the results of the single SNP analysis (Fig. S2).

We found no significant associations between CDH13 SNPs and visuo-spatial WM and the SSRT tasks.

Discussion

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments
  8. Supporting Information

CDH13 has been identified as a candidate gene for ADHD using different analytic strategies. We tested the association between CDH13 on three executive functioning tasks that are promising endophenotypes of ADHD. Our adjusted linear regression analysis of SNPs on CDH13 in 190 ADHD-affected Dutch children showed that SNP rs11150556 is gene-wide significantly associated (including haplotype analysis) with verbal WM (independently from ADHD severity) and that a 80-kb SNP block, 446 kb upstream of SNP rs111500556 showed suggestive evidence of association with the same Verbal WM task.

The relationship between ADHD and WM deficits and inhibition are well proven (Willcutt et al. 2005). The correlation values observed between neuropsychological and symptom measures were therefore surprising, however, not unexpected. A previous study from our group (Rommelse et al. 2008a) shows that: (1) an endophenotypic construct significantly predicts the diagnostic status (affected, non-affected and control) and (2) affected children portray a more severe ADHD phenotype than one would expect based on their cognitive dysfunctioning. In other words, group differences at an endophenotypic level and phenotypic level are not directly comparable for affected children (Nigg et al. 2005).

This study is the first to examine CDH13 in neurocognitive functioning and the first to help explain the mechanisms underlying the association between CDH13 and the clinical phenotype of ADHD. The broad expression of H-cadherin in the midbrain and telencephalon suggests that it plays an important role in building and maintaining neural circuitry (Takeuchi et al. 2000). More specifically, H-cadherin may be responsible for cell–cell adhesion (Patel et al. 2003) and the regulation of neural cell growth (Takeuchi et al. 2000). Deficient functioning of the H-cadherin system may therefore lead to a lower number of neurons and negatively affect neuronal growth affecting the structure and/or the number of neuronal connections (Poelmans et al., unpublished observations).

We found CDH13 genetic variation to specifically affect WM (single SNPs and haplotypes). WM is one of the major executive functions supported by the frontal lobes (Pennington et al. 1996) and seems to be mediated by a complex network of brain structures including fronto-striatal dopaminergic circuits (Frank et al. 2007; Goldman-Rakic 1996). Furthermore, the dorsolateral prefrontal cortex appears to be involved in tasks tapping the central executive (CE) (Collette et al. 2002; D’Esposito et al. 1995), as well as in verbal and spatial WM tasks.

The most influential and supported model of WM is Baddeley's multi-component model (Baddeley 2010) which postulates the existence of two short-term storage systems, one for visual material, the visuo-spatial sketchpad, and one for verbal-acoustic material, the phonological loop. The CE control system regulates the two storage systems (Baddeley 2010; Castellanos et al. 2006). The CE component of WM controls and manipulates the stored information, and acts on information retrieved from long-term memory to support complex cognitive activities (Martinussen et al. 2005). While the forward condition of the Digit Span task assesses only the phonological loop capacity (Baddeley 2010; Gathercole 1999), the backward condition used in this study requires both storage (phonological loop) and transformation (processing) of material within WM (Gathercole 1999), and has been extensively employed in the WM literature to index CE resources (Gathercole 1998; Gathercole et al. 2000; Thomason et al. 2009). The principal role of the CE system is to coordinate attention and not necessarily to hold information in mind, nevertheless it is considered a part of WM. In ADHD, it is thought that impairments observed in complex tasks of WM (Pennington et al. 1996) may be attributable to a dysfunction in the CE component rather than in the verbal or spatial buffers or rehearsal processes (Karatekin 2004). Consistent with this, children with ADHD perform worse than other children on Backward but not Forward Digit Span (McInnes et al. 2003). Since we found an association between CDH13 and the Digit Span backwards, it may be possible that CDH13 is related to the CE component of WM. This is also consistent with the fact that an association between CDH13 and lower WM performance was found for verbal WM but not for spatial WM, as the visuo-spatial WM task used in our study relies more on the maintenance of information and less on processing/manipulation (Crone et al. 2006). However, we explored the relationship between SNP rs11150556 and forward digit span by testing (post hoc) the association. Our results show a significant P value = 0.0134 with ADHD children carriers of the CC genotype also showing worse performance. This result might imply that CDH13 is associated with verbal WM in a global way. In other words, it may rely on both the maintenance and processing of verbal information. In addition, we could speculate that CDH13 may be associated with verbal WM in contrast to visuo-spatial WM.

This study is limited by the relatively small sample size, and replication of our findings in independent datasets is needed. We addressed the problem of multiple testing by performing very stringent permutation test to our three WM tasks association results. Also, the sample selection may affect results in different ways. A homogeneous group (combined type, almost all boys) is advantageous in finding an association between a gene and neurocognitive functioning in ADHD. On the other hand, the combined type may reflect the more severe phenotype which lies at the end of the ADHD continuum and therefore limiting the possibility to find an association. Furthermore, the literature is not straightforward whether ADHD subtypes differ on neurocognitive functioning. Some find no differences (O’Brien et al. 2010; Schweitzer et al. 2006; Seidman et al. 2005), while others do (Solanto et al. 2007).

CDH13 has been extensively studied in relation with cancer (Berx et al. 2009) and has been identified as a susceptibility locus for high blood pressure (Org et al. 2009) but, to our knowledge, our study represents the first attempt to investigate the relationship of the CDH13 gene with neuropsychological performance in children with ADHD. The association found between CDH13 and verbal WM is consistent with its expression pattern in the brain, especially in mature cerebral cortex and medulla (Takeuchi et al. 2000) and may help to increase our understanding of how this gene contributes to ADHD susceptibility, in particular, because WM is one of the main deficits and an important endophenotypes in ADHD (Arnsten 2011; Jacobson et al. 2011; Kofler et al. 2011; Pauli-Pott et al. 2011).

Intronic SNP associations are particularly difficult to relate with a specific clinically relevant trait but GWAS results found and replicated the association of several intronic SNPs in CDH13 with an ADHD quantitative phenotype (Lasky-Su et al. 2008), methamphetamine dependence (Uhl et al. 2008) and alcohol dependence (Treutlein et al. 2009). This further strengthens the evidence for a role of this gene in ADHD, as substance abuse/dependence is a common comorbidity of ADHD. Our study may be viewed as exploratory, with the results presented to be considered hypothesis-generating and our findings require caution and should be replicated in other cohorts for final confirmation.

References

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments
  8. Supporting Information

Acknowledgments

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments
  8. Supporting Information

We thank all the persons who kindly participated in this research. The IMAGE project was supported by National Institutes of Health (NIH) grants R01MH081803 and R01MH62873 to S.V.F. Site Principal Investigators are J.B. and J.S. Chief Investigators at each site are N.R., M.A. and B.F. The dataset(s) used for the analyses described in this manuscript were obtained from the dbGaP Database through dbGaP accession number phs000016.v2.p2. Samples and associated phenotype data for Whole Genome Association Study of Attention Deficit Hyperactivity Disorder were provided by S.F. Analyses were (partially) performed using the Genetics Computer Cluster (GCC, http://www.geneticcluster.org/) which is financially supported by the Netherlands Scientific Organization (NWO 480-05-003). The authors declare no conflict of interest.

Supporting Information

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Acknowledgments
  8. Supporting Information

Supporting Information

Additional Supporting Information may be found in the online version of this article:

Figure S1: Linkage disequilibrium plot of the CDH13 gene.

Figure S2: CDH13 3-SNP sliding window haplotype associations with WM.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer-reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

FilenameFormatSizeDescription
GBB_724_sm_fs1.doc435KSupporting info item
GBB_724_sm_fs2.doc50KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.