2009 Vienna - European Society of Human Genetics
2009 Vienna - European Society of Human Genetics
2009 Vienna - European Society of Human Genetics
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Concurrent Sessions<br />
c04.6<br />
Duplication <strong>of</strong> the EFNB1 gene in familial hypertelorism:<br />
imbalance in ephrin-b1 expression and abnormal phenotypes in<br />
humans and mice<br />
C. Babbs 1 , H. Stewart 2 , L. Williams 3 , L. Connell 3 , A. Goriely 1 , S. R. F. Twigg 1 , K.<br />
Smith 3 , T. Lester 3 , A. O. M. Wilkie 1 ;<br />
1 Weatherall Institute <strong>of</strong> Molecular Medicine, University <strong>of</strong> Oxford, Oxford, United<br />
Kingdom, 2 Department <strong>of</strong> Clinical <strong>Genetics</strong>, Churchill Hospital, Oxford, United<br />
Kingdom, 3 <strong>Genetics</strong> Laboratories, Churchill Hospital, Oxford, United Kingdom.<br />
Familial Hypertelorism, characterised by widely spaced eyes, classically<br />
shows autosomal dominant inheritance (Teebi type), but some<br />
pedigrees are compatible with X-linkage. No pathogenic mechanism<br />
has been described previously, but clinical similarity has been noted to<br />
crani<strong>of</strong>rontonasal syndrome (CFNS), which is caused by mutations in<br />
the X-linked EFNB1 gene.<br />
Here we report a family in which females in three generations presented<br />
with hypertelorism, but lacked either craniosynostosis or a<br />
grooved nasal tip, excluding CFNS. DNA sequencing <strong>of</strong> EFNB1 was<br />
normal, but MLPA indicated a duplication <strong>of</strong> all 5 exons <strong>of</strong> EFNB1,<br />
which segregated with the hypertelorism. We characterised the duplication<br />
breakpoint sequence, revealing a direct duplication <strong>of</strong> 937 kb<br />
including EFNB1 and two flanking genes, PJA1 and STARD8. By use<br />
<strong>of</strong> Pyrosequencing, we measured imbalance <strong>of</strong> EFNB1 expression in<br />
the affected grandmother. After correction for skewed X-inactivation,<br />
we show that the X chromosome bearing the duplicated EFNB1 genes<br />
produces approximately twice as much EFNB1 transcript as the normal<br />
X chromosome.<br />
We propose that in the context <strong>of</strong> X-inactivation, the difference in expression<br />
level <strong>of</strong> EFNB1 between the normal and duplicated X chromosomes<br />
results in abnormal cell sorting during embryogenesis, leading<br />
to hypertelorism. To support this hypothesis we provide evidence<br />
from a mouse model carrying a hypomorphic Efnb1 allele, that abnormal<br />
cell sorting occurs in the cranial region. Hence we propose<br />
that X-linked cases resembling Teebi hypertelorism may have a similar<br />
pathogenesis to CFNS, and that cellular mosaicism for different levels<br />
<strong>of</strong> ephrin-b1 (as well as simple presence/absence) leads to crani<strong>of</strong>acial<br />
abnormalities.<br />
c05.1<br />
combined analysis <strong>of</strong> 19 common validated type 2 diabetes<br />
susceptibility gene variants show moderate discriminative value<br />
and no evidence <strong>of</strong> gene-gene interaction<br />
T. Sparso 1 , N. Grarup 1 , C. Andreasen 1 , A. Albrechtsen 2 , J. Holmkvist 1 , G. Andersen<br />
1 , T. Jørgensen 3,4 , K. Borch-Johnsen 1,5 , A. Sandbæk 6 , T. Lauritzen 6 , S.<br />
Madsbad 7 , T. Hansen 1,8 , O. Pedersen 1,7 ;<br />
1 Hagedorn Research, Gent<strong>of</strong>te, Denmark, 2 Department <strong>of</strong> Biostatistics, University<br />
<strong>of</strong> Copenhagen, Copenhagen, Denmark, 3 Research Centre for Prevention<br />
and Health, Glostrup University Hospital, Glostrup, Denmark, 4 Faculty <strong>of</strong> Health<br />
Science, University <strong>of</strong> Copenhagen, Denmark, 5 Faculty <strong>of</strong> Health Science, University<br />
<strong>of</strong> Aarhus, Aarhus, Denmark, 6 Department <strong>of</strong> General Practice, Institute<br />
<strong>of</strong> Public Health, Aarhus, Denmark, 7 Faculty <strong>of</strong> Health Science, University <strong>of</strong><br />
Copenhagen, Copenhagen, Denmark, 8 Faculty <strong>of</strong> Health Sciences, University<br />
<strong>of</strong> Southern Denmark, Denmark.<br />
Background: The list <strong>of</strong> validated type 2 diabetes susceptibility variants<br />
has recently been expanded from three to 19. The identified variants<br />
are common and have low penetrance in the general population. The<br />
aim <strong>of</strong> this study was to examine for gene-gene interactions and investigate<br />
the combined effect <strong>of</strong> the 19 variants by applying receiver operating<br />
characteristics (ROC) to demonstrate the discriminatory value<br />
between glucose-tolerant individuals and type 2 diabetes patients in a<br />
cross-sectional population <strong>of</strong> Danes.<br />
Methods: The 19 variants were genotyped in three study populations:<br />
The population-based Inter99 study, the ADDITION study, and in additional<br />
type 2 diabetic patients and glucose-tolerant individuals. The<br />
case-control studies involved 4,093 type 2 diabetic patients and 5,302<br />
glucose-tolerant individuals.<br />
Results: Single variant analyses demonstrated allelic odds ratios (OR)<br />
ranging from 1.04 (95%CI: 0.98,1.11) to 1.33 (95%CI: 1.22,1.45).<br />
When combining the 19 variants subgroups with extreme risk pr<strong>of</strong>iles<br />
showed 3-fold difference in risk <strong>of</strong> type 2 diabetes (lower 10% carriers<br />
with 22 risk alleles, OR<br />
2.93 (95%CI: 2.38,3.62,p=1.6×10 -25 ). We calculated the area under a<br />
ROC curve to estimate the discrimination rate between glucose-toler-<br />
ant individuals and type 2 diabetes patients based on the 19 variants.<br />
We found an area under the ROC curve <strong>of</strong> 0.60. Two-way gene-gene<br />
interaction showed few nominal interactions.<br />
Conclusion: The 19 validated variants enables detection <strong>of</strong> subgroups<br />
in substantial increased risk <strong>of</strong> type 2 diabetes, however the discrimination<br />
between glucose tolerance and type 2 diabetes is still too inaccurate<br />
to achieve clinical value.<br />
c05.2<br />
Joint re-analysis <strong>of</strong> twenty-nine correlated sNPs supports the<br />
role <strong>of</strong> PcLO/Piccolo as a causal risk factor for major depressive<br />
disorder<br />
Z. Bochdanovits1 , A. van der Vaart2 , M. Verhage2 , A. Smit2 , E. de Geus2 , D.<br />
Posthuma2 , D. Boomsma2 , B. Penninx1 , W. Hoogendijk1 , P. Heutink1 ;<br />
1 2 VU Medical Center, Amsterdam, The Netherlands, Vrije Universiteit, Amsterdam,<br />
The Netherlands.<br />
The first genome-wide association study (GWAS) for major depressive<br />
disorder (MDD) has implicated the pre-synaptic protein Piccolo,<br />
but results from multiple replication cohorts remained inconclusive. We<br />
propose a simple method for the joint (re-)analysis <strong>of</strong> multiple SNPs,<br />
based on published summary data. Our approach is based on two<br />
observations. Firstly, finemapping studies are focused, by design, on<br />
a limited number <strong>of</strong> moderately to strongly correlated SNPs. All tested<br />
SNPs are expected to reflect the true association <strong>of</strong> the unknown<br />
causal variant proportional to their LD with it, in concordance with the<br />
“Fundamental Theorem <strong>of</strong> the HapMap”. Secondly, given such correlated<br />
SNP data it has been suggested before that a joint analysis <strong>of</strong> all<br />
markers together is most powerful for detecting a true association. A<br />
closer examination <strong>of</strong> the results reported in the GWAS study reveals<br />
that the data indeed concur with the ”Theorem <strong>of</strong> the HapMap”. Based<br />
on the above we re-analyzed the replication data using a novel joint<br />
test <strong>of</strong> association and conclude and the results strongly favors Piccolo<br />
to be a causal risk factor for major depression. This study was performed<br />
within the framework <strong>of</strong> Top Institute Pharma project: number<br />
T5-203.<br />
c05.3<br />
Unified framework for epistasis detection in (un)relateds<br />
K. Van Steen1,2 , T. Cattaert1 , M. Calle3 ;<br />
1 2 3 Montefiore Institute, Liège, Belgium, GIGA, Liège, Belgium, University <strong>of</strong> Vic,<br />
Vic, Spain.<br />
When searching for epistatic patterns parametric regression approaches<br />
have severe limitations when there are too many independent<br />
variables in relation to the number <strong>of</strong> observed outcome events.<br />
Alternatively, the non-parametric Multifactor Dimensionality Reduction<br />
method, MDR (Ritchie et al. 2001), can be applied. The common feature<br />
<strong>of</strong> MDR and its extensions is that they are extremely computerintensive,<br />
that best models are evaluated on the basis <strong>of</strong> cross-validation<br />
(prediction accuracy measures) and permutations and that only<br />
one such best model is proposed when looking at interactions <strong>of</strong> a<br />
particular order.<br />
We propose a novel unified multifactor dimensionality reduction strategy<br />
for genetic interaction association analysis that can handle both<br />
unrelated individuals and families <strong>of</strong> any structure, different outcome<br />
types (e.g., categorical, continuous or survival type), easy covariate<br />
handling or adjustment for lower order interactions or confounding<br />
factors, all within the same framework. When applied to family data,<br />
we obtain a less computationally intensive method than current MDR<br />
adaptations to family-data and allow several clusters <strong>of</strong> markers to be<br />
proposed as showing significant association with the outcome under<br />
investigation. This better reflects locus heterogeneity and genetic heterogeneity,<br />
usually present in complex diseases.<br />
Our epistasis detection method is further evaluated and validated via a<br />
simulation study, by computing type I error and power under a variety<br />
<strong>of</strong> scenarios, and via application to a real-life data set.