436 VOLUME 42 | NUMBER 5 | MAY 2010 Nature GeNetics
l e t t e r s
Smoking is a leading global cause of disease and mortality1.
We established the Oxford-GlaxoSmithKline study (Ox-GSK) to
perform a genome-wide meta-analysis of SNP association with
smoking-related behavioral traits. Our final data set included
41,150 individuals drawn from 20 disease, population and control
cohorts. Our analysis confirmed an effect on smoking quantity
at a locus on 15q25 (P = 9.45 × 10−19) that includes CHRNA5,
CHRNA3 and CHRNB4, three genes encoding neuronal nicotinic
acetylcholine receptor subunits. We used data from the 1000
Genomes project to investigate the region using imputation, which
allowed for analysis of virtually all common SNPs in the region and
offered a fivefold increase in marker density over HapMap2 (ref. 2)
as an imputation reference panel. Our fine-mapping approach
identified a SNP showing the highest significance, rs55853698,
located within the promoter region of CHRNA5. Conditional
analysis also identified a secondary locus (rs6495308) in CHRNA3.
Smoking behavior and nicotine dependence are multifactorial traits
with substantial genetic influences3. There is an urgent need to better
understand the molecular neurobiology of nicotine dependence in order
to design targeted, more effective therapies4. Recently, genome-wide asso-
ciation studies (GWAS) have established one locus associated with nico-
tine dependence and smoking quantity, which implicates a cluster of three
genes, CHRNA5, CHRNA3 and CHRNB4 on chromosome 15q25, which
encode neuronal nicotinic acetylcholine receptor subunits5–9. This locus
is also associated with lung cancer8,10,11, peripheral arterial disease8 and
chronic obstructive pulmonary disease and lung function12.
We initially performed a GWAS meta-analytic study of smoking-
related traits in a total sample of 41,150 individuals of European
descent, sourced from several disease, population and control cohorts
(Table 1, Supplementary Table 1 and Online Methods). As the cohorts
were genotyped on a variety of different genome-wide SNP arrays
(Table 1 and Supplementary Table 1), we first imputed genotypes for
all data sets13 for all SNPs in the HapMap version release 22 (ref. 2).
The main focus of our analysis was on smoking quantity within
current and past smokers, defined as a semiquantitative trait based
on the self-reported variable of cigarettes smoked per day (CPD)8. We
performed association analyses separately within each cohort under
Meta-analysis and imputation refines the association of
15q25 with smoking quantity
Jason Z Liu1*, Federica Tozzi2, Dawn M Waterworth3, Sreekumar G Pillai3, Pierandrea Muglia2, Lefkos Middleton4,
Wade Berrettini5, Christopher W Knouff 6, Xin Yuan3, Gérard Waeber7,8, Peter Vollenweider7,8, Martin Preisig7,9,
Nicholas J Wareham10, Jing Hua Zhao10, Ruth J F Loos10, Inês Barroso11, Kay-Tee Khaw12, Scott Grundy13,
Philip Barter14, Robert Mahley15,16, Antero Kesaniemi17,18, Ruth McPherson19, John B Vincent20, John Strauss20,
James L Kennedy20, Anne Farmer21, Peter McGuffin21, Richard Day22, Keith Matthews22, Per Bakke23,
Amund Gulsvik23, Susanne Lucae24, Marcus Ising24, Tanja Brueckl24, Sonja Horstmann24, H-Erich Wichmann25–27,
Rajesh Rawal25, Norbert Dahmen28, Claudia Lamina25,29, Ozren Polasek30, Lina Zgaga31, Jennifer Huffman32,
Susan Campbell32, Jaspal Kooner33, John C Chambers34, Mary Susan Burnett35, Joseph M Devaney35,
Augusto D Pichard35, Kenneth M Kent35, Lowell Satler35, Joseph M Lindsay35, Ron Waksman35, Stephen Epstein35,
James F Wilson31, Sarah H Wild31, Harry Campbell31, Veronique Vitart32, Muredach P Reilly36,37, Mingyao Li38,
Liming Qu38, Robert Wilensky36, William Matthai36, Hakon H Hakonarson39, Daniel J Rader36,37, Andre Franke40,
Michael Wittig40, Arne Schäfer40, Manuela Uda41, Antonio Terracciano42, Xiangjun Xiao43, Fabio Busonero41,
Paul Scheet43, David Schlessinger42, David St Clair44, Dan Rujescu45, Gonçalo R Abecasis46, Hans Jörgen Grabe47,
Alexander Teumer48, Henry Völzke49, Astrid Petersmann50, Ulrich John51, Igor Rudan52,31, Caroline Hayward32,
Alan F Wright32, Ivana Kolcic30, Benjamin J Wright53, John R Thompson53, Anthony J Balmforth54,
Alistair S Hall54, Nilesh J Samani55, Carl A Anderson11, Tariq Ahmad56, Christopher G Mathew57, Miles Parkes58,
Jack Satsangi59, Mark Caulfield60, Patricia B Munroe60, Martin Farrall61, Anna Dominiczak62, Jane Worthington63,
Wendy Thomson63, Steve Eyre63, Anne Barton63, The Wellcome Trust Case Control Consortium65, Vincent Mooser3,
Clyde Francks2,64 & Jonathan Marchini1
*Complete lists of author affiliations appear at the end of this paper.
Received 19 October 2009; accepted 18 March 2010; published online 25 April 2010; doi:10.1038/ng.572
©
2
01
0
N
at
ur
e
A
m
er
ic
a,
In
c.
A
ll
ri
gh
ts
r
es
er
ve
d.

Nature GeNetics VOLUME 42 | NUMBER 5 | MAY 2010 437
l e t t e r s
an additive model using covariate effects for age, sex, disease case or
control status where applicable, and other cohort-specific covariates
(Supplementary Table 1). A meta-analysis was then carried out by
combining study-specific β (regression coefficient) estimates using
a fixed effects model14. In total, 15,574 subjects reported CPD values
over zero and were used for the meta-analysis of smoking quantity
(Table 1 and Supplementary Table 1). We followed up our most
promising association findings by comparing them with results from
two concurrent GWAS meta-analyses of smoking: the ENGAGE study
of 46,481 subjects15 and the TAG study of 74,035 subjects16. We also
made our meta-analysis results available to the authors of those stud-
ies to check their top findings for replication.
Our meta-analysis of smoking quantity identified the CHRNA5–
CHRNA3 locus on 15q25 as the single significant locus of note in the
genome (Fig. 1, Table 2 and Supplementary Table 2), with a minimum
P = 9.45 × 10−19 for rs1051730, a SNP which has been previously
reported to be associated with traits related to smoking5–9; we also
found highly significant P values for many other SNPs in the region
(Supplementary Fig. 1 and Supplementary Table 2). All cohorts in
the analysis contributed at least somewhat to the 15q25 association
(Supplementary Fig. 1). Each copy of the A allele (34% frequency) had
a quantitative effect size on smoking quantity of 0.079 (95% confidence
interval 0.070–0.088), which is in line with previous estimates8. A joint
analysis of our total data set, together with the TAG and ENGAGE data
sets, for rs1051730 yielded P = 1.71 × 10−66 (Table 2).
Multiple variants at the 15q25 locus have been suggested to underlie
its effect on smoking quantity, including a nonsynonymous SNP
in CHRNA5 and variants that affect mRNA expression levels17–19.
We utilized our very large sample, in combination with data from the
1000 Genomes Project (see URLs), to perform fine mapping and mod-
eling of the 15q25 locus in relation to smoking quantity. We reasoned
that with the near complete information on common SNPs derived
from the 1000 Genomes data set, it might be possible to pinpoint a
variant or combination of variants that can explain the entirety of
the signal of association at 15q25. We used data from 108 estimated
CEU European-ancestry haplotypes from the April 2009 release of
the 1000 Genomes Pilot 1 data. This data set contained 2,189 SNPs
in our region of interest (Online Methods), which was approximately
a fivefold increase in density compared to the 437 SNPs in release 22
of HapMap. By imputing genotypes for all SNPs across this locus
from 1000 Genomes and by repeating the meta-analysis, we found
that the most significant association was with a new and previously
untested SNP which is not in the HapMap and is located within the
5′ untranslated region of CHRNA5; this location makes it a candidate
for affecting mRNA transcription (rs55853698, P = 1.31 × 10−16;
Fig. 2). The P value for the commonly reported SNP rs1051730 in this
table 1 summary information for the cohorts used in meta-analysis
Label Description Genotyping
Sample Sizes
All CPD > 0 Ever Never Current Non-current
WTCCC-RA Rheumatoid arthritis cases Affymetrix 500K 1,860 NA NA NA 262 558
EPIC Obesity case-control Affymetrix 500K 3,516 NA 1,927 1,589 353 1,574
WTCCC-HT Hypertension cases Affymetrix 500K 1,952 830 NA NA 1,274 672
GEMS Dyslipidemia case-control Affymetrix 500K 1,847 862 910 793 268 642
GSK-COPD COPD case-control Illumina 550 1,633 1,632 NA NA 725 905
GSK-Bipolar Bipolar depression case-control Illumina 550 1,805 944 1,008 790 498 510
GSK-UPD Unipolar depression case-control Illumina 550 1,792 899 935 856 503 432
WTCCC-IBD Crohn′s disease cases Affymetrix 500K 1,748 NA 713 540 713 420
KORA Population-based Affymetrix 500K 1,644 253 811 831 217 1,425
KORCULA Population-based Illumina 300 827 NA 376 451 179 654
LOLIPOP Population-based Affymetrix 500K 1,288 650 653 635 258 395
MedStar Coronary artery disease case-control Affymetrix 6.0 1,322 820 853 469 300 553
ORCADES Population-based Illumina 300 692 NA 288 404 60 632
PENNCATH Coronary artery disease case-control Affymetrix 6.0 1,401 NA NA NA 464 612
POPGEN Population-based Affymetrix 6.0 1,107 573 495 608 NA NA
CoLaus Population-based Affymetrix 500K 5,636 3,132 3,357 2,275 1,485 1,872
SardiNIA Population-based Affymetrix 500+10K 4,305 1,731 1,743 2,562 873 3,432
SHIP Population-based Affymetrix 6.0 4,080 2,011 2,631 1,449 1,240 2,840
VIS Population-based Illumina 300 769 NA 441 328 212 557
WTCCC-CAD Coronary artery disease cases Affymetrix 500K 1,926 1,237 1,457 461 239 1,218
TOTALS 41,150 15,574 18,598 15,041 10,123 19,903
Further details are given in Online Methods and supplementary table 1; NA, not applicable.
15
10
–
lo
g 1
0(P
)
5
1 2 3 4 65 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Chromosomes
Figure 1 Plot showing the significance of association of all SNPs in the
genome-wide smoking quantity meta-analysis. SNPs are plotted on
the x axis according to their positions on each chromosome against
association with smoking quantity on the y axis (−log10 P value). SNPs
with P values < 1.0 × 10−5 are highlighted in green.
©
2
01
0
N
at
ur
e
A
m
er
ic
a,
In
c.
A
ll
ri
gh
ts
r
es
er
ve
d.

438 VOLUME 42 | NUMBER 5 | MAY 2010 Nature GeNetics
l e t t e r s
analysis was similar but slightly less significant (P = 1.47 × 10−15). The
P values for our 1000 Genomes analysis were generally higher than
those from our HapMap-based analysis because not all of our study
cohorts were included in the 1000 Genomes imputation (see Online
Methods). rs55853698 is a G/T substitution, where the G allele has a
frequency ranging from 0.313–0.378 across the various cohorts.
To investigate whether the association to smoking quantity at 15q25
can be explained completely by rs55853698, we carried out tests of asso-
ciation for all SNPs spanning the CHRNA5-CHRNA3 locus conditional
upon this SNP (Fig. 2). Residual association was still detected at
many SNPs in the region, with the most significant signal occurring
at rs6495308 (P = 3.96 × 10−5), which is located within an intron of
CHRNA3 (Fig. 2). In the unconditioned analysis, rs6495308 has a
significance of P = 3.30 × 10−10. Further conditioning on rs6495308
after conditioning on rs55853698 leaves no obvious signal of association
in the region (Supplementary Fig. 2), suggesting that these two SNPs
together could be sufficient to explain the genetic effect.
It has previously been suggested18 that a nonsynonymous SNP,
rs16969968, in CHRNA5 is associated with nicotine dependence risk
and lung cancer risk, but also that variants that cause high expression
of CHRNA5 mRNA, tagged by rs588765, increase the risk for nicotine
dependence independently. The marginal P values of rs16969968 and
rs588765 in our meta-analysis were P = 1.64 × 10−18 and P = 1.74 ×
10−3. Conditional analysis on rs16969968 within our cohorts still left
residual association within the region (Supplementary Fig. 2), with the
most significant signal again occurring at rs6495308 (P = 1.54 × 10−5).
Conditioning on both rs16969968 and rs588765, that is, the combina-
tion previously proposed18, leaves no obvious signal of association
(Supplementary Fig. 2). To further investigate which pair of SNPs best
explains the signal of association, we used the Bayesian information
criteria (BIC) measure of model fit, in which smaller values indicate
a better fit20. For the previous model18, that is, conditioning on both
rs16969968 and rs588765, we obtained BIC = 22,719.87 and a posterior
probability 0.15. For the model conditioning on the new promoter
SNP rs55853698 and rs6495308, we obtained BIC = 22,716.49 and a
posterior probability 0.85, which indicates a better model fit.
Examination of the linkage disequilibrium (LD) structure between
the SNPs considered here shows that rs1051730, rs16969968 and
rs55853698 are all close-tagging proxies of each other (all pairwise
r2 > 0.96). These variants either tag or potentially cause the princi-
pal risk for high smoking quantity attributable to the 15q25 locus,
but the high LD makes it difficult to assign specific causality. The
SNPs that show residual association, rs588765 and rs6495308, are
in low LD with each other (r2 = 0.21) and are both in only modest
LD with the principal SNPs (maximum r2 = 0.47). It is not therefore
clear that this locus can be completely understood in the way previ-
ously proposed18. Although the nonsynonymous SNP in CHRNA5,
rs16969968, may be important, we have identified a new and poten-
tially functional SNP in the 5′ untranslated region of this gene that
is a close proxy for the nonsynonymous SNP in terms of LD, but
which shows a slightly more significant association in our meta-
analysis. Furthermore, although rs588765 can explain much of the
secondary or residual association at this locus, we find that a largely
independent variant within CHRNA3, rs6495308, is the best tagger
of the residually associated variation; this variant also contributes to
a better-fitting two-SNP model and has a much stronger marginal
significance in our unconditioned analysis (P = 3.30 × 10−10 for
rs6495308 as compared to P = 1.74 × 10−3 for rs588765).
To our knowledge, our analysis has, for the first time, surveyed
virtually all of the common SNPs in the 15q25 region and provides
15
10
5–
lo
g 1
0(P
)
0
5
4
3
2
1
–
lo
g 1
0(P
)
0
CRABP1 IREB2
76.5 76.6 76.7
Chromosomal position (Mb)
76.8 76.9 77.0
AGPHD1
PSMA4
CHRNA5
CHRNA3 CHRNB4 ADAMTS7
MORF4L1
CTSH RASGRF1G
en
es
100
80
40
20
60
cM
/M
b
0
100
80
40
20
60
cM
/M
b
0
rs55853698
rs72740955
rs55781567
rs931794
rs1051730
rs1317286
rs2036527
rs16969968
rs9788721
rs6495308
rs481134
rs8023822
rs621849
rs680244
rs1051730
rs1317286
rs2036527
rs16969968
rs9788721
Figure 2 Chromosome 15q25 signal plots.
Signal plot based on the 1000 Genomes
imputation and meta-analysis of smoking
quantity association (top). SNPs are plotted
by their positions on the chromosome against
association with smoking quantity (−log10
P value) on the left y axis. The five SNPs
with the lowest P values from the HapMap
imputation are highlighted in red. The five
SNPs with the lowest P values from the 1000
Genomes imputation are highlighted in green
(unless already colored red). The rs identities
of highlighted SNPs are given in the box.
Recombination rates across the region are
shown by the red line plotted against the right
y axis. Chromosome 15q25 signal plot based
on the 1000 Genomes imputation and meta-
analysis of smoking quantity association,
conditional on rs55853698 (middle). The
five SNPs with the lowest P values from
the conditional analysis are highlighted in green. The five SNPs with the lowest P values from the unconditioned HapMap imputation analysis are
highlighted in red. Genes and the positions of exons using data from the UCSC genome browser (bottom; see URLs).
table 2 summary information for selected sNPs at 15q25 from meta-analysis of association with the smoking Quantity (sQ) phenotype
SNP Chr. Position Coded allele Coded allele freq.
Ox-GSK TAG ENGAGE Combined
P Phet P P P β s.e.m.
rs588765 15 76,652,480 T 0.43 1.74 × 10−3 0.50 NA NA NA NA NA
rs16969968 15 76,669,980 G 0.65 1.64 × 10−18 0.86 1.85 × 10−27 1.53 × 10−23 4.29 × 10−65 −0.078 0.0046
rs1051730 15 76,681,394 G 0.66 9.45 × 10−19 0.68 3.62 × 10−27 9.98 × 10−25 1.71 × 10−66 −0.079 0.0046
rs6495308 15 76,694,711 T 0.77 3.30 × 10−10 0.10 7.99 × 10−24 1.60 × 10−13 5.82 × 10−44 0.073 0.0052
Our study is referred to as Ox-GSK. Information for all SNPs spanning the 15q25 locus in our genome-wide analysis is given in supplementary table 2. Chr., chromosome; Freq.,
frequency; Phet, heterozygosity P value; NA, not applicable.
©
2
01
0
N
at
ur
e
A
m
er
ic
a,
In
c.
A
ll
ri
gh
ts
r
es
er
ve
d.

Nature GeNetics VOLUME 42 | NUMBER 5 | MAY 2010 439
l e t t e r s
one of the first examples of how data from the 1000 Genomes Project
can contribute new information to mapping and characterizing
loci for complex traits. We recommend that further analysis of this
locus should not be limited in focus to CHRNA5, nor particularly to
the nonsynonymous SNP rs16969968. It is notoriously difficult to dis-
tinguish functional variation when there is high LD across a region21.
There are many ways in which variants can be functional, including
expression regulatory changes that affect either close or distant genes,
epigenetic changes, splicing effects, alterations to microRNA bind-
ing sites, or noncoding RNAs21. It is also conceivable that association
with common variants can arise through the effects of multiple, rarer
variants that happen to be relatively restricted to specific haplotype
backgrounds. In addition, common insertions or deletions can have
functional effects, and the 1000 Genomes data will allow for analysis
of this class of variant via an imputation framework.
The second-strongest association with smoking quantity within the
genome in our meta-analysis was at a locus on 8p21 that received modest
support from the TAG and ENGAGE studies (Supplementary Table 2
and Supplementary Fig. 3; P = 5.26 × 10−7 for rs11782673). This locus
would not remain significant after correcting for genome-wide multiple
testing; however, it is noteworthy that the locus spans CHRNA2, another
gene that encodes a neuronal nicotinic acetylcholine receptor subunit.
In addition to our analysis of smoking quantity, we also performed
a genome-wide test for allelic differences between those who reported
currently smoking or having smoked in the past versus those who said
they had never been smokers (the ever/never phenotype; sample sizes
are shown in Table 1 and Supplementary Table 1). This test aimed to
identify genetic effects on the establishment of a smoking habit. No locus
achieved genome-wide significance in this analysis, and none of the top
15 loci showed evidence of replication (Supplementary Table 2 and
Supplementary Fig. 4). Likewise, no consistent results emerged when
we tested for allelic differences between those who reported smoking at
present versus those who had smoked in the past but had stopped at the
time of interview (Supplementary Table 2 and Supplementary Fig. 4).
When age-adjusted, this is a rough measure of smoking cessation.
Our study identified association at some loci that, although not
reaching genome-wide significance in our own meta-analysis, sup-
ported findings from the concurrent TAG and ENGAGE studies15,16.
These include new loci on chromosomes 8 and 19 for smoking quan-
tity, on chromosome 11 for ever/never and on chromosome 9 for cur-
rent versus non-current smokers15,16. These findings have provided
further new insights into the biology of smoking behavior.
URLs. ProbABEL software, http://mga.bionet.nsc.ru/~yurii/ABEL/;
SNPTEST, IMPUTE and SNPMETA software, http://www.stats.ox.ac.
uk/~marchini/software/gwas/gwas.html; 1000 Genomes Project:
http://www.1000genomes.org/; April 2009 release of the 1000 Genomes
Pilot 1 data, ftp://ftp-trace.ncbi.nih.gov/1000genomes/ftp/pilot_data/
release/2009_04/; UCSC Genome Browser, http://genome.ucsc.edu/;
MERLIN, http://www.sph.umich.edu/csg/abecasis/merlin/; R, http://
www.r-project.org/.
MetHOdS
Methods and any associated references are available in the online version
of the paper at http://www.nature.com/naturegenetics/.
Note: Supplementary information is available on the Nature Genetics website.
ACKNOWLEDGMENTS
GlaxoSmithKline (GSK), a pharmaceuticals company that is interested in
developing new cessation therapies for smoking, funded a postdoctoral fellowship
for J.Z.L. at Oxford University. GSK also funded the collection, characterization,
and, in some cases, the genotyping and genotype data preparation for several of
the cohorts used in this study. A. Roses and P. Matthews played crucial roles in
establishing and funding the Medical Genetics activities at GSK. Acknowledgments
that are specific to individual cohorts are given in the Supplementary Note.
AUTHOR CONTRIBUTIONS
J.Z.L. carried out most of the analysis for this study. J.M. and C.F. conceived and
directed this study and wrote the manuscript. F.T., D.M.W. and V.M. were involved
in study design and helped to coordinate the inclusion of many of the GSK
cohorts. S.G.P., P. Muglia, L.M., W.B., C.W.K., X.Y., G.W., P.V., M. Preisig, N.J.W.,
J.H.Z., R.J.F.L., I.B., K.-T.K., S.G., P. Barter, R. Mahley, A.K., R. McPherson, J.B.V.,
J. Strauss, J.L.K., A. Farmer, P. McGuffin, R.D., K.M., P. Bakke, A.G., S.L., M.I., T.B.,
S.H., H.-E.W., R.R., N.D., C.L., O.P., L.Z., J.H., S.C., J.K., J.C.C., M.S.B., J.M.D.,
A.D.P., K.M.K.. L.S., J.M.L., R. Waksman, S. Epstein, J.F.W., S.H.W., H.C., V.V.,
M.P.R., M.L., L.Q., R. Wilensky, W.M., H.H.H., D.J.R., A. Franke, M.W., A.S., M.U.,
A. Terracciano, X.X., F.B., P.S., D.S., D.St.C., D.R., G.R.A., H.J.G., A. Teumer, H.V.,
A.P., U.J., I.R., C.H., A.F.W., I.K., B.J.W., J.R.T., A.J.B., A.S.H., N.J.S., C.A.A., T.A.,
C.G.M., M. Parkes, J. Satsangi, M.C., P.B.M., M.F., A.D., J.W., W.T., S. Eyre, A.B.
and W.T.C.C.C. prepared and shared data sets and, in some cases, cohort-specific
results from their own primary analysis.
COMPETING FINANCIAL INTERESTS
The authors declare competing financial interests: details accompany the full-text
HTML version of the paper at http://www.nature.com/naturegenetics/.
Published online at http://www.nature.com/naturegenetics/.
Reprints and permissions information is available online at http://npg.nature.com/
reprintsandpermissions/.
1. Ezzati, M., Lopez, A.D., Rodgers, A., Vander Hoorn, S. & Murray, C.J. Selected
major risk factors and global and regional burden of disease. Lancet 360,
1347–1360 (2002).
2. Frazer, K.A. et al. A second generation human haplotype map of over 3.1 million
SNPs. Nature 449, 851–861 (2007).
3. Li, M.D. The genetics of nicotine dependence. Curr. Psychiatry Rep. 8, 158–164
(2006).
4. Benowitz, N.L. Neurobiology of nicotine addiction: implications for smoking
cessation treatment. Am. J. Med. 121, S3–S10 (2008).
5. Berrettini, W. et al. α-5/α-3 nicotinic receptor subunit alleles increase risk for heavy
smoking. Mol. Psychiatry 13, 368–373 (2008).
6. Bierut, L.J. et al. Novel genes identified in a high-density genome wide association
study for nicotine dependence. Hum. Mol. Genet. 16, 24–35 (2007).
7. Li, M.D. Identifying susceptibility loci for nicotine dependence: 2008 update based
on recent genome-wide linkage analyses. Hum. Genet. 123, 119–131 (2008).
8. Thorgeirsson, T.E. et al. A variant associated with nicotine dependence, lung cancer
and peripheral arterial disease. Nature 452, 638–642 (2008).
9. Caporaso, N. et al. Genome-wide and candidate gene association study of cigarette
smoking behaviors. PLoS One 4, e4653 (2009).
10. Amos, C.I. et al. Genome-wide association scan of tag SNPs identifies a susceptibility
locus for lung cancer at 15q25.1. Nat. Genet. 40, 616–622 (2008).
11. Hung, R.J. et al. A susceptibility locus for lung cancer maps to nicotinic acetylcholine
receptor subunit genes on 15q25. Nature 452, 633–637 (2008).
12. Pillai, S.G. et al. A genome-wide association study in chronic obstructive pulmonary
disease (COPD): identification of two major susceptibility loci. PLoS Genet. 5,
e1000421 (2009).
13. Marchini, J., Howie, B., Myers, S., McVean, G. & Donnelly, P. A new multipoint
method for genome-wide association studies by imputation of genotypes. Nat.
Genet. 39, 906–913 (2007).
14. Normand, S.L. Meta-analysis: formulating, evaluating, combining, and reporting.
Stat. Med. 18, 321–359 (1999).
15. Thorgeirsson, T. et al. Sequence variants at CHRNB3–CHRNA6 and CYP2A6 affect
smoking behavior. Nat. Genet. 42, 448–453 (2010).
16. Tobacco and Genetics Consortium. Genome-wide meta-analyses identify multiple
loci associated with smoking behavior. Nat. Genet. 42, 441–447 (2010).
17. Falvella, F.S. et al. Transcription deregulation at the 15q25 locus in association
with lung adenocarcinoma risk. Clin. Cancer Res. 15, 1837–1842 (2009).
18. Wang, J.C. et al. Risk for nicotine dependence and lung cancer is conferred by
mRNA expression levels and amino acid change in CHRNA5. Hum. Mol. Genet.
18, 3125–3135 (2009).
19. Wang, J.C. et al. Genetic variation in the CHRNA5 gene affects mRNA levels and is
associated with risk for alcohol dependence. Mol. Psychiatry 14, 501–510
(2008).
20. Schwarz, G. Estimating the dimension of a model. Ann. Stat. 6, 461–464
(1978).
21. Ioannidis, J.P., Thomas, G. & Daly, M.J. Validating, augmenting and refining
genome-wide association signals. Nat. Rev. Genet. 10, 318–329 (2009).
©
2
01
0
N
at
ur
e
A
m
er
ic
a,
In
c.
A
ll
ri
gh
ts
r
es
er
ve
d.

440 VOLUME 42 | NUMBER 5 | MAY 2010 Nature GeNetics
1Department of Statistics, University of Oxford, Oxford, UK. 2Genetics Division, GlaxoSmithKline, Verona, Italy. 3Genetics Division, GlaxoSmithKline, Upper Merion,
Pennsylvania, USA. 4Division of Neurosciences and Mental Health, Imperial College London, UK. 5Department of Psychiatry, University of Pennsylvania School of
Medicine, Philadelphia, Pennsylvania, USA. 6Genetics Division, GlaxoSmithKline, Research Triangle Park, North Carolina, USA. 7University Hospital Center, University
of Lausanne, Lausanne, Switzerland. 8Department of Internal Medicine, University of Lausanne, Lausanne, Switzerland. 9Department of Psychiatry, University of
Lausanne, Lausanne, Switzerland. 10Medical Research Council Epidemiology Unit, Institute of Metabolic Science, Cambridge, UK. 11Wellcome Trust Sanger Institute,
Hinxton, UK. 12Department of Public Health and Primary Care, University of Cambridge, Cambridge, UK. 13Center for Human Nutrition, University of Texas
Southwestern Medical Center, Dallas, Texas, USA. 14The Heart Research Institute, Sydney, New South Wales, Australia. 15Gladstone Institute of Cardiovascular
Disease, University of California, San Francisco, California, USA. 16American Hospital, Istanbul, Turkey. 17Department of Internal Medicine, University of Oulu, Oulu,
Finland. 18Biocenter Oulu, University of Oulu, Oulu, Finland. 19Division of Cardiology, University of Ottawa Heart Institute, Ottawa, Ontario, Canada. 20Centre for
Addiction and Mental Health, University of Toronto, Toronto, Ontario, Canada. 21Medical Research Council Social, Genetic and Developmental Psychiatry Centre,
Institute of Psychiatry, King′s College London, London, UK. 22Center for Neuroscience, Division of Medical Sciences, University of Dundee, Dundee, UK. 23Institute
of Medicine, University of Bergen, Bergen, Norway. 24Max Planck Institute of Psychiatry, Munich, Germany. 25Institute of Epidemiology, Helmholtz Zentrum München,
German Research Center for Environmental Health, Neuherberg, Germany. 26Institute of Medical Informatics, Biometry and Epidemiology, Ludwig-Maximilians-
Universität, Munich, Germany. 27Klinikum Grosshadern, Munich, Germany. 28Psychiatrische Klinik und Poliklinik University of Mainz, Mainz, Germany. 29Division
of Genetic Epidemiology, Department of Medical Genetics, Molecular and Clinical Pharmacology, Innsbruck Medical University, Innsbruck, Austria. 30School of Public
Health, School of Medicine, University of Zagreb, Croatia. 31Centre for Population Health Sciences, University of Edinburgh, Edinburgh, UK. 32Institute of Genetics
and Molecular Medicine, MRC Human Genetics Unit, Edinburgh, UK. 33National Heart and Lung Institute, Imperial College London, London, UK. 34Division of
Epidemiology, Imperial College London, London, UK. 35Cardiovascular Research Institute, MedStar Research Institute, Washington Hospital Center, Washington DC,
USA. 36The Cardiovascular Institute, University of Pennsylvania, Philadelphia, Pennsylvania, USA. 37The Institute for Translational Medicine and Therapeutics, School
of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, USA. 38Biostatistics and Epidemiology, University of Pennsylvania, Philadelphia, Pennsylvania,
USA. 39The Center for Applied Genomics, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania, USA. 40Institute of Clinical Molecular Biology, Christian-
Albrechts-University, Kiel, Germany. 41Istituto di Neurogenetica e Neurofarmacologia, Consiglio Nazionale delle Ricerche, Monserrato, Cagliari, Italy. 42National
Institute on Aging, Baltimore, Maryland, USA. 43Department of Epidemiology, University of Texas M.D. Anderson Cancer Center, Houston, Texas, USA. 44Department
of Mental Health, University of Aberdeen, Aberdeen, UK. 45Division of Molecular and Clinical Neurobiology, Department of Psychiatry, Ludwig-Maximilians-University,
Munich, Germany. 46Center for Statistical Genetics, Department of Biostatistics, University of Michigan, Ann Arbor, Michigan, USA. 47Department of Psychiatry and
Psychotherapy, University of Greifswald, Greifswald, Germany. 48Interfacultary Institute for Genetics and Functional Genomics, University of Greifswald, Greifswald,
Germany. 49Institute for Community Medicine, University of Greifswald, Greifswald, Germany. 50Institute of Clinical Chemistry and Laboratory Medicine, University
of Greifswald, Greifswald, Germany. 51Department of Social Medicine and Epidemiology, University of Greifswald, Greifswald, Germany. 52Croatian Centre for Global
Health, University of Split, Split, Croatia. 53Department of Health Sciences, University of Leicester, Leicester, UK. 54Mulitdisciplinary Cardiovascular Research Centre
(MCRC), Leeds Institute of Genetics, Health and Therapeutics (LIGHT), University of Leeds, Leeds, UK. 55Department of Cardiovascular Sciences, University of
Leicester, Glenfield Hospital, Leicester, UK. 56Peninsula College of Medicine and Dentistry, Exeter, UK. 57Department of Medical and Molecular Genetics, King’s
College London School of Medicine, Guy’s Hospital, London, UK. 58Gastroenterology Research Unit, Addenbrooke’s Hospital, Cambridge, UK. 59Gastrointestinal Unit,
Molecular Medicine Centre, University of Edinburgh, Western General Hospital, Edinburgh, UK. 60Clinical Pharmacology and Barts and the London Genome Centre,
William Harvey Research Institute, Barts and the London School of Medicine, Queen Mary University of London, London, UK. 61Department of Cardiovascular
Medicine, University of Oxford, Wellcome Trust Centre for Human Genetics, Oxford, UK. 62British Heart Foundation Glasgow Cardiovascular Research Centre, Division
of Cardiovascular and Medical Sciences, University of Glasgow, Western Infirmary, Glasgow, UK. 63arc Epidemiology Research Unit, School of Translational Medicine,
Faculty of Medical and Human Sciences, University of Manchester, UK. 64Wellcome Trust Centre for Human Genetics, University of Oxford, Oxford, UK. 65A full list
of members is provided in the supplementary Note. Correspondence should be addressed to J.M. (marchini@stats.ox.ac.uk) or C.F. (clyde.francks@well.ox.ac.uk).
l e t t e r s
©
2
01
0
N
at
ur
e
A
m
er
ic
a,
In
c.
A
ll
ri
gh
ts
r
es
er
ve
d.

Nature GeNeticsdoi:10.1038/ng.572
ONLINe MetHOdS
Study samples. Study collections and their basic characteristics are listed in
Table 1 and Supplementary Table 1. Subjects used in our analysis were adults
of European descent. Summary descriptions of the collections are given below,
together with primary citations that describe the collections fully. Data were
used in accordance with the ethical permissions and consents relating to each
collection.
GEMS22: The Genetic Epidemiology of Metabolic Syndrome (GEMS)
study consists of dyslipidemic case individuals (age 20–65 years) matched
with normolipidemic controls by sex and recruitment site, drawn from
non-Mediterranean subjects of the GEMS study (from Finland, Switzerland,
Canada, Australia and the United States).
CoLaus23: The Cohorte Lausannoise (CoLaus) is a single-center,
cross-sectional population-based study, including individuals aged
35–75 years randomly selected from the list of residents of the city of
Lausanne, Switzerland.
GSK COPD12: This collection includes case individuals with chronic
obstructive pulmonary disease, diagnosed according to Global Initiative for
Chronic Obstructive Lung Disease (GOLD) criteria, and unaffected controls
recruited from Bergen, Norway.
GSK UPD24: This collection includes case individuals with recurrent
major depression according to DSM-IV criteria and age- and gender-
matched unaffected controls, recruited at the Max-Planck Institute of
Psychiatry in Munich, Germany. Subjects were also recruited at two sat-
ellite recruiting hospitals (Bezirkskrankenhaus Augsburg and Klinikum
Ingolstadt) in the Munich area.
GSK Bipolar25: The Bipolar collection included DSM-IV-diagnosed
bipolar case individuals and controls from subjects recruited at three
study sites: the Institute of Psychiatry (IOP) in London, UK; the Centre for
Addiction and Mental Health in Toronto, Canada; and the University of
Dundee, UK.
GSK LOLIPOP26: The London Life Sciences Prospective Population
(LOLIPOP) was a population based study including Indian Asian and
European white men and women recruited from the lists of 58 general prac-
titioners in West London.
GSK MedStar27: The MedStar cohort included case individuals with
acute coronary syndrome or chronic coronary artery disease (CAD) from
Washington DC, together with unaffected controls.
Penn-CATH27: The Penn-CATH cohort was a University of Pennsylvania
Medical Center-based angiographic study from which case individuals with
CAD and controls with no evidence of CAD at the coronary angiography
were derived.
EPIC28: The EPIC-Obesity cohort was a case-control cohort for obesity
drawn from the EPIC-Norfolk cohort which included men and women of
European ancestry aged 39–79 years recruited in Norfolk, UK.
KORA29: The Cooperative Health Research in the Region of Augsburg
(KORA) study was an epidemiological survey of the general population living
in the city of Augsburg, southern Germany, and two adjacent counties.
WTCCC HT30: The WTCCC-HT collection comprised severely hyperten-
sive probands ascertained from families with multiple affected members in
the UK as part of the BRIGHT study.
WTCCC CAD, WTCCC CD and WTCCC RA30: These studies included
individuals with CAD, Crohn’s disease and rheumatoid arthritis from the
Wellcome Trust Case Control Consortium Study.
POPGEN study31: The Population Genetic Cohort (POPGEN) was a
cross sectional epidemiological survey of regional German populations from
Schleswig-Holstein, northern Germany.
SHIP study32: The Study of Health in Pomerania (SHIP) was a longitudi-
nal, population-based survey from West Pomerania, Germany. Data from the
baseline cohort were used for this study.
VIS study33: This population cohort comprised Croatians aged 18–93
years recruited from the villages of Vis and Komiza on the Dalmatian
island of Vis.
ORCADES study34: The Orkney Complex Disease Study (ORCADES) was
a family-based, cross-sectional study that sought to identify genetic factors
influencing cardiovascular and other disease risk in the population isolate of
the Orkney Isles in northern Scotland.
KORCULA study35: The KORCULA study included healthy volunteers aged
18 and over from the villages of Lumbarda, Žrnovo, and Račišće on the Island
of Korcula, Croatia.
SardiNIA study36: The SardiNIA was a population-based longitudinal cohort
study that included male and female related individuals, aged 14 years and
above, from a cluster of four towns in the Ogliastra province of Sardinia, Italy.
Genotyping, quality control and imputation. Supplementary Table 1 lists the
various genotype platforms used for each cohort, the genotype calling algo-
rithms, SNP and sample quality control measures and details of the imputation
and association analysis software used. The quality control measures from
previous analyses of each cohort were adopted for this study and are detailed
in the table. We used NCBI build 36 coordinates for SNP base-pair positions
so that all the cohorts could be successfully combined.
We imputed all SNPs reported in the CEU sample in HapMap Phase II
using various imputation algorithms13,37 (see URLs for a link to ProbABEL).
Imputations were performed after excluding samples and SNPs that did not
meet the study-specific quality control criteria. Genotypes were imputed for
SNPs not present in the genome-wide arrays or for those where genotyping
had failed to meet the quality control criteria.
Only imputed SNPs with good imputation quality were included in the
meta-analysis. This was defined as proper_info ≥ 0.5 (a software-specific sta-
tistic for the studies analyzed with IMPUTE/SNPTEST13) or rsq-hat ≥ 0.5
(a statistic used for studies analyzed using MACH37) and Imp_info ≥ 0.5 (a
statistic used for studies analyzed using ProbABEL).
Derivation of smoking phenotypes. We used the categorical smoking quan-
tity levels previously defined8. The smoking quantity levels were 0 (defined
as 1–10 CPD), 1 (11–20 CPD), 2 (21–30 CPD) and 3 (31 or more CPD). Each
increment represents an increase in smoking quantity of 10 cigarettes per day.
Most of the cohorts in our study have maximal CPD recorded on each sample,
but a few collected average CPD (Supplementary Table 1). We examined the
distributions of CPD across cohorts and found no large differences between
those cohorts using average CPD and those using maximal CPD. The mean
and standard deviation of the CPD measurements in each cohort are given in
Supplementary Table 1. The ever/never and current/non-current phenotypes
used were those collected by the individual cohorts. Not all cohorts had all
three phenotypes (smoking quantity, ever/never and current/non-current)
collected. Precise details of the phenotypes collected in each cohort are given
in Supplementary Table 1. An assessment would typically be questionnaire-
based, following a structure such as the following:
Tick the option that best describes you:
- I smoke now
- I don’t smoke now. I have stopped for … years.
- I have never smoked
About how many cigarettes do you or did you smoke per day?
List the number of years you have smoked.
Statistical analysis and meta-analysis. Each cohort was analyzed separately
for each of the three phenotypes considered. The majority of the analysis was
carried out on the raw genotype data at the Department of Statistics, University
of Oxford, but some cohorts (SardiNIA, VIS, KORCULA, ORCADES and SHIP)
carried out their own analysis and submitted results for the meta-analysis.
For the binary traits (ever/never and current/non-current) tests for additive
genetic effects on the log-odds scale were carried out using logistic regression.
For the categorical smoking quantity phenotype, tests for additive genetic
effects were carried out on a linear scale using linear regression. The programs
SNPTEST, ProbABEL and MERLIN were used on the various cohorts to fit
these models, taking account of the genotype uncertainty at imputed SNPs.
All tests conditioned on sex and age, and for some cohorts, other covariates
of self-reported ancestry, country of origin or principal components analysis-
derived covariates were included (a complete list of covariates is given in
Supplementary Table 1). A genomic control inflation factor (λ) estimate was
calculated for each phenotype and each cohort (Supplementary Table 3).
The meta-analysis was carried out by combining study-specific β estimates
using a fixed effects model14, which used the inverse of the variance of the
©
2
01
0
N
at
ur
e
A
m
er
ic
a,
In
c.
A
ll
ri
gh
ts
r
es
er
ve
d.

Nature GeNetics doi:10.1038/ng.572
study-specific β estimates to give weight to the contribution of each study. The
variance of each cohort’s β estimate was multiplied by the genomic control λ
estimate to correct for observed inflation38. Specifically,
b
b l s
l s
s
l s
b
sMETA META META
META=
( )
( ) = ( ) =
∑
∑ ∑
i i i
i
i i
i
i i
i
Z
2
2 21
1
1
, ,
META
,
where βi, σi2 and λi are the β estimate, β-estimate variance and genomic con-
trol λ estimate for the ith cohort. This method is appropriate when the same
phenotype and measurement scale are used in each cohort, and it has the
advantage that measures of effect size (eβ is an estimate of the odds ratio of
the risk allele) and its standard error can be calculated. We also repeated the
analysis of smoking quantity by combining z-scores from each cohort weighted
by their sample size38 and obtained almost identical results. All meta-analysis
was carried out using the SNPMETA program (see URLs). After performing
each meta-analysis, the overall λ estimate for each phenotype was 1.0145 for
smoking quantity, 1.002 for ever/never and 0.998 for current/non-current.
For each SNP, we also calculated a P value for the heterogeneity across
the studies38.
SNP selection for replication. In collaboration with two other groups carrying
out similar meta-analyses of smoking related traits (ENGAGE15 and TAG16),
we agreed to an in silico replication strategy in which for each phenotype
(smoking quantity, ever/never, current/non-current) each group would select
15 regions of the genome showing evidence for association, and summary
data (P values, β estimate, β-estimate variances, sample sizes, genomic con-
trol λ estimates and sample sizes) would be shared across groups to facilitate
replication. We selected the top 15 regions for each phenotype on the basis
of the P values we obtained in our own meta-analysis. We excluded regions
in which only a small number of cohorts contributed to the study because
the information measure at the SNPs in the excluded cohorts were below
our thresholds. We also excluded regions where the heterogeneity between
the studies was high. Each selected region consisted of several SNPs showing
evidence of association in our meta-analysis with P values below 1 × 10−5.
For each of the three phenotypes, the results from all the cohorts in all three
concurrent studies were combined together using the same genomic-control–
corrected inverse-variance meta-analysis method described above. A full list of
the selected regions and the summary information from all three phenotypes
is given in Supplementary Table 2.
1000 Genomes imputation analysis. We used 108 estimated CEU haplotypes
from the April 2009 release of the 1000 Genomes Pilot 1 data to carry out
our fine-mapping experiments at the 15q25 locus (see URLs for a link to
the data source). We used these haplotypes to carry out imputation in the
interval 76.4–77.0 Mb on chromosome 15 in 12 of the cohorts (GSK-Bipolar,
GSK-UPD, GSK-COPD, KORA, POPGEN, Lausanne, GSK-LOLIPOP, GSK-
GEMS, MedStar, SHIP, WTCCC-CAD and WTCCC-HT) using the program
IMPUTE13. This release contains 2,189 SNPs in this interval, compared to 437
SNPs in release 22 of the HapMap data. Meta-analysis of the imputed data was
then carried out in the same way as described above. An important techni-
cal detail when carrying out imputation using the 1000 Genomes haplotype
data is how to align it with the genotype data from genome-wide studies. The
program IMPUTE aligns SNPs between the haplotype and genotype database
on the basis of base-pair position (rather than using SNP identifiers such as rs
identities) so that as long as the same coordinate system is used for both the
haplotype and genotype data, the alignment is automatic.
Conditional analysis and modeling. The analysis conditional upon the SNPs
was carried out using all of the centrally analyzed cohorts (Bipolar, UPD,
COPD, KORA, POPGEN, Lausanne, LOLIPOP, GEMS, MEDSTAR, SHIP,
WTCCC-CAD and WTCCC-HT). At the SNP being conditioned upon, we
used expected genotype counts, as this allowed us to combine data from
cohorts which had imputed the SNP and cohorts which had genotyped the
SNP. These expected counts were included in the baseline null model as an
additional covariate, along with the other covariates such as age, sex and cov-
ariates coding for population structure. The same method was used when
conditioning upon two SNPs. The model selection analysis of the two pairs of
SNPs in the 15q25 region was carried out using the expected genotype counts.
Analysis was carried out using the R statistical package.
22. Stirnadel, H. et al. Genetic and phenotypic architecture of metabolic syndrome-
associated components in dyslipidemic and normolipidemic subjects: the GEMS
Study. Atherosclerosis 197, 868–876 (2008).
23. Firmann, M. et al. The CoLaus study: a population-based study to investigate the
epidemiology and genetic determinants of cardiovascular risk factors and metabolic
syndrome. BMC Cardiovasc. Disord. 8, 6 (2008).
24. Muglia, P. et al. Genome-wide association study of recurrent major depressive
disorder in two European case-control cohorts. Mol. Psychiatry published online,
doi:10.1038/mp.2008.131 (23 December 2008).
25. Scott, L.J. et al. Genome-wide association and meta-analysis of bipolar disorder in
individuals of European ancestry. Proc. Natl. Acad. Sci. USA 106, 7501–7506
(2009).
26. Chahal, N.S. et al. Ethnicity-related differences in left ventricular function, structure
and geometry: a population study of UK Indian Asians and European whites. Heart
96, 466–471 (2009).
27. Kathiresan, S. et al. Genome-wide association of early-onset myocardial infarction
with single nucleotide polymorphisms and copy number variants. Nat. Genet. 41,
334–341 (2009).
28. Day, N. et al. EPIC-Norfolk: study design and characteristics of the cohort. European
Prospective Investigation of Cancer. Br. J. Cancer 80 (suppl. 1), 95–103 (1999).
29. Wichmann, H.E., Gieger, C. & Illig, T. KORA-gen–resource for population genetics,
controls and a broad spectrum of disease phenotypes. Gesundheitswesen 67,
S26–S30 (2005).
30. Wellcome Trust Case-Control Consortium. Genome-wide association study of 14,000
cases of seven common diseases and 3,000 shared controls. Nature 447, 661–678
(2007).
31. Krawczak, M. et al. PopGen: population-based recruitment of patients and controls
for the analysis of complex genotype-phenotype relationships. Community Genet.
9, 55–61 (2006).
32. John, U. et al. Study of Health in Pomerania (SHIP): a health examination survey
in an East German region: objectives and design. Soz. Praventivmed. 46, 186–194
(2001).
33. Vitart, V. et al. SLC2A9 is a newly identified urate transporter influencing serum
urate concentration, urate excretion and gout. Nat. Genet. 40, 437–442 (2008).
34. McQuillan, R. et al. Runs of homozygosity in European populations. Am. J. Hum.
Genet. 83, 359–372 (2008).
35. Zemunik, T. et al. Genome-wide association study of biochemical traits in Korcula
Island, Croatia. Croat. Med. J. 50, 23–33 (2009).
36. Pilia, G. et al. Heritability of cardiovascular and personality traits in 6,148
Sardinians. PLoS Genet. 2, e132 (2006).
37. Li, Y. & Abecasis, G.R. Mach 1.0: rapid haplotype reconstruction and missing
genotype inference. Am. J. Hum. Genet. s79, 2290 (2006).
38. de Bakker, P.I. et al. Practical aspects of imputation-driven meta-analysis of
genome-wide association studies. Hum. Mol. Genet. 17, R122–R128 (2008).
©
2
01
0
N
at
ur
e
A
m
er
ic
a,
In
c.
A
ll
ri
gh
ts
r
es
er
ve
d.