Haworth S Shungin D van der Tas J T Vucic S Medina-Gomez C Yakimov V Feenstra B Shaffer J R Lee M KStandl M Thiering E Wang C Boslashnnelykke K Waage J EyrichJessen L Noslashrrisgaard P E Joro R Seppaumllauml I Raitakari O Timpson N J (2018) Consortium-based genome-wide meta-analysisfor childhood dental caries traits Human Molecular Genetics[ddy237] httpsdoiorg101093hmgddy237

Publishers PDF also known as Version of recordLicense (if available)CC BYLink to published version (if available)101093hmgddy237

Link to publication record in Explore Bristol ResearchPDF-document

This is the final published version of the article (version of record) It first appeared online via Oxford UniversityPress at httpsacademicoupcomhmgadvance-articledoi101093hmgddy2375040780 Please refer to anyapplicable terms of use of the publisher

University of Bristol - Explore Bristol ResearchGeneral rights

This document is made available in accordance with publisher policies Please cite only thepublished version using the reference above Full terms of use are availablehttpwwwbristolacukpureuser-guidesexplore-bristol-researchebr-terms

A S S O C I A T I O N S T U D I E S A R T I C L E

Consortium-based genome-wide meta-analysis for

childhood dental caries traitsSimon Haworth1dagger Dmitry Shungin23dagger Justin T van der Tas4Strahinja Vucic4 Carolina Medina-Gomez567 Victor Yakimov8Bjarke Feenstra8 John R Shaffer910 Myoung Keun Lee10 Marie Standl11Elisabeth Thiering1112 Carol Wang13 Klaus Boslashnnelykke14Johannes Waage14 Leon Eyrich Jessen14 Pia Elisabeth Noslashrrisgaard14Raimo Joro15 Ilkka Seppala16 Olli Raitakari1718 Tom Dudding1 Olja Grgic45Edwin Ongkosuwito5 Anu Vierola15 Aino-Maija Eloranta15 Nicola X West19Steven J Thomas19 Daniel W McNeil20 Steven M Levy21 Rebecca Slayton22Ellen A Nohr23 Terho Lehtimaki16 Timo Lakka152425 Hans Bisgaard14Craig Pennell13 Jan Kuhnisch26 Mary L Marazita910 Mads Melbye82728Frank Geller8 Fernando Rivadeneira567 Eppo B Wolvius4Paul W Franks293031 Ingegerd Johansson2 and Nicholas J Timpson1

1Medical Research Council Integrative Epidemiology Unit at Bristol Medical School University of Bristol BristolBS8 2BN UK 2Department of Odontology Umea University Umea 901 87 Sweden 3Broad Institute of theMassachusetts Institute of Technology and Harvard University Cambridge MA 02142 USA 4Department ofOral and Maxillofacial Surgery Special Dental Care and Orthodontics 5The Generation R Study Group6Department of Internal Medicine 7Department of Epidemiology Erasmus Medical Center University MedicalCenter Rotterdam Rotterdam 3015 CN The Netherlands 8Department of Epidemiology Research StatensSerum Institut Copenhagen DK-2300 Denmark 9Department of Human Genetics Graduate School of PublicHealth University of Pittsburgh Pittsburgh PA 15261 USA 10Center for Craniofacial and Dental GeneticsDepartment of Oral Biology School of Dental Medicine University of Pittsburgh Pittsburgh PA 15213 USA11Institute of Epidemiology I Helmholtz Zentrum Munchen - German Research Center for EnvironmentalHealth Neuherberg D-85764 Germany 12Division of Metabolic and Nutritional Medicine Dr von HaunerChildrenrsquos Hospital University of Munich Medical Center Munich 80337 Germany 13Division of Obstetrics andGynaecology The University of Western Australia Perth WA 6009 Australia 14COPSAC CopenhagenProspective Studies on Asthma in Childhood Herlev and Gentofe Hospital University of CopenhagenCopenhagen 2730 Denmark 15Institute of Biomedicine School of Medicine University of Eastern Finland

dagger

The authors wish it to be known that in their opinion the first 2 authors should be regarded as joint First AuthorsReceived March 1 2018 Revised May 29 2018 Accepted June 14 2018

VC The Author(s) 2018 Published by Oxford University PressThis is an Open Access article distributed under the terms of the Creative Commons Attribution License (httpcreativecommonsorglicensesby40)which permits unrestricted reuse distribution and reproduction in any medium provided the original work is properly cited

1

Human Molecular Genetics 2018 Vol 0 No 0 1ndash15

doi 101093hmgddy237Advance Access Publication Date 20 June 2018Association Studies Article

Downloaded from httpsacademicoupcomhmgadvance-article-abstractdoi101093hmgddy2375040780by University of Bristol Library useron 20 July 2018

Kuopio Campus 70211 Kuopio Finland 16Department of Clinical Chemistry Fimlab Laboratories and FinnishCardiovascular Research Center Tampere - Faculty of Medicine and Life Sciences University of TampereTampere 33520 Finland 17Department of Clinical Physiology and Nuclear Medicine Turku University HospitalTurku 20520 Finland 18Research Centre of Applied and Preventive Cardiovascular Medicine University ofTurku Turku 20520 Finland 19Bristol Dental School University of Bristol Bristol BS1 2LY UK 20Department ofPsychology Eberly College of Arts and Sciences West Virginia University Morgantown WA 26506-6286 USA21Department of Preventive and Community Dentistry College of Dentistry University of Iowa Cedar RapidsIA 52242-1010 USA 22Department of Pediatric Dentistry (Retired) School of Dentistry University ofWashington Seattle WA 98195 USA 23Research Unit for Gynaecology and Obstetrics Department of ClinicalResearch University of Southern Denmark Odense 5000 Denmark 24Department of Clinical Physiology andNuclear Medicine Kuopio University Hospital Kuopio 70210 Finland 25Kuopio Research Institute of ExerciseMedicine Kuopio 70100 Finland 26Department of Conservative Dentistry and Periodontology UniversityHospital Ludwig-Maximilians-Universitat Munchen Munich 80336 Germany 27Department of ClinicalMedicine University of Copenhagen Copenhagen 2200 Denmark 28Department of Medicine StanfordUniversity School of Medicine Stanford CA 94305 USA 29Department of Clinical Sciences Genetic andMolecular Epidemiology Unit Lund University Malmo 202 13 Sweden 30Department of Public Health andClinical Medicine Umea University Umea 901 85 Sweden and 31Department of Nutrition Harvard T H ChanSchool of Public Health Boston MA 02115 USA

To whom correspondence should be addressed at MRC Integrative Epidemiology Unit Oakfield House Oakfield Grove Bristol BS8 2BN UKTel thorn44 (0) 1173310083 Fax thorn44 (0) 1179287325 Email simonhaworthbristolacuk

AbstractPrior studies suggest dental caries traits in children and adolescents are partially heritable but there has been no large-scaleconsortium genome-wide association study (GWAS) to date We therefore performed GWAS for caries in participants aged25ndash180 years from nine contributing centres Phenotype definitions were created for the presence or absence of treated oruntreated caries stratified by primary and permanent dentition All studies tested for association between caries and geno-type dosage and the results were combined using fixed-effects meta-analysis Analysis included up to 19 003 individuals(7530 affected) for primary teeth and 13 353 individuals (5875 affected) for permanent teeth Evidence for association with car-ies status was observed at rs1594318-C for primary teeth [intronic within ALLC odds ratio (OR) 085 effect allele frequency(EAF) 060 P 413e-8] and rs7738851-A (intronic within NEDD9 OR 128 EAF 085 P 163e-8) for permanent teeth Consortium-wide estimated heritability of caries was low [h2 of 1 (95 CI 0 7) and 6 (95 CI 0 13) for primary and permanentdentitions respectively] compared with corresponding within-study estimates [h2 of 28 (95 CI 9 48) and 17 (95 CI2 31)] or previously published estimates This study was designed to identify common genetic variants with modesteffects which are consistent across different populations We found few single variants associated with caries status underthese assumptions Phenotypic heterogeneity between cohorts and limited statistical power will have contributed these find-ings could also reflect complexity not captured by our study design such as genetic effects which are conditional on environ-mental exposure

IntroductionDental caries remains a prevalent public health problem in bothchildren and adults Untreated dental caries was estimated toaffect 621 million children worldwide in 2010 with little changein prevalence or incidence between 1990 and 2010 (1) This prob-lem is not unique to lower income countries around 50 ofchildren have evidence of caries by age 5 in industrializednations (2ndash4) Dental caries results from reduced mineral satura-tion of fluids surrounding teeth driven by ecological shifts inthe oral microbiome (5) Many different factors predisposetoward dental caries of which high sugar consumption poororal hygiene and low socio-economic status are the most noto-rious (6ndash8) Over the last decades there has been increasing ap-preciation for the role of genetic influences in dental caries Theimportance of genetic susceptibility for dental caries experience

was demonstrated in an animal model over 50 years ago a find-ing since substantiated in twin studies in humans (9ndash11) Of par-ticular relevance to caries traits in children and adolescentsBretz et al (10) analysed longitudinal rates of change in cariesstatus in children and found that caries progression and sever-ity were highly heritable in the primary and permanentdentition It has also been suggested that heritability for dentalcaries does not depend entirely on genetic predisposition tosweet food consumption (12) Despite evidence of a genetic con-tribution to caries susceptibility few specific genetic loci havebeen identified

Shaffer et al (13) performed the first GWAS for dental cariesin 2011 studying the primary dentition of 1305 children Theyfound evidence for association at novel and previously studiedcandidate genes (ACTN2 MTR EDARADD MPPED2 and LPO) but

2 | Human Molecular Genetics 2018 Vol 00 No 00

Downloaded from httpsacademicoupcomhmgadvance-article-abstractdoi101093hmgddy2375040780by University of Bristol Library useron 20 July 2018

no individual single-nucleotide polymorphisms (SNPs)exceeded the genome-wide significance threshold (P 50e-08)possibly as a consequence of the modest sample size (13)The first GWAS for dental caries in the permanent dentition inadults was performed at a similar time by Wang et al (14) Theyincluded 7443 adults from five different cohorts and identifiedseveral suggestive loci (P-value 10e-05) for dental caries(RPS6KA2 PTK2B RHOU FZD1 ADMTS3 and ISL1) differentloci from those mentioned above for the primary dentitionand again with no single variants reaching genome-widesignificance

The next wave of GWAS of caries suggested association at arange of different loci Two GWAS used separate phenotype defi-nitions for pit-and-fissure and smooth tooth surfaces and identi-fied different loci associated with dental caries susceptibility inboth primary and permanent dentition (1516) The GWAS in pri-mary dentition used a sample of approximately 1000 children andfound evidence for association at loci reported in previous stud-ies including MPPED2 RPS6KA2 and AJAP1 (13ndash16) The largestGWAS for dental caries in permanent dentition was performed ina Hispanic and Latino sample of 11 754 adults (17) This studyidentified unique genetic loci (NAMPT and BMP7) compared withprevious GWAS in individuals of European ancestry To date it isunclear whether the variability in nominated loci reflects true var-iability in the genetic architecture of dental caries across differentpopulations age periods and sub-phenotypic definitions ormerely represent chance differences between studies given themodest power in the studies performed to date

Dental caries is a complex and multifactorial disease causedby a complex interplay between environmental behaviouraland genetic factors Until now there has been a lack of large-scale studies of dental caries traits in children and the geneticbasis of these traits remains poorly characterized This investi-gation set out to examine the hypothesis that common geneticvariants influence dental caries with modest effects on suscep-tibility We anticipated that (a) caries in both primary and per-manent teeth would be heritable in children and adolescentsaged 25ndash18 years and (b) common genetic variants are likely toonly have small effects on the susceptibility of a complex dis-ease such as dental caries Therefore the aim of this large-scale consortium-based GWAS is to examine novel genetic lociassociated with dental caries in primary and permanent denti-tion in children and adolescents

ResultsSingle variant results

Meta-analysis of caries in primary teeth in individuals ofEuropean ancestry included 17 037 individuals (6922 affected)from 22 results files representing all nine coordinating centresAfter final quality control (QC) this meta-analysis included8 640 819 variants with mild deflation (genomic inflation factork frac14 0994) (Supplementary Material Fig S1) Meta-analysis ofcaries in primary teeth which included individuals of multipleethnicities in the Generation R (GENR) study included 19 003individuals (7530 affected) from 22 results files representing all9 coordinating centres There were 8 699 928 variants after finalQC with mild deflation in summary statistics (k frac14 0986)(Supplementary Material Fig S2) Analysis of caries status inpermanent teeth included 13 353 individuals (5875 affected)from 14 results files representing 7 coordinating centres Thesample size was smaller for permanent teeth as two coordinat-ing centres did not have phenotype data for permanent teeth

(RAINE and GENR) whilst the COPSAC group only had data forparticipants in the earlier birth cohort (COPSAC 2000)There were 8 734 121 variants after final QC with milddeflation in summary statistics (k frac14 0999) (SupplementaryMaterial Fig S3)

The strongest evidence for association with caries in pri-mary teeth was seen at rs1594318 [odds ratio (OR) 085 forC allele EAF 060 Pfrac14 413e-08] in the European ancestry meta-analysis (Figs 1 2 and 3 Table 1) This variant is intronic withinALLC on 2p25 a locus which has not previously been reportedfor dental caries traits In the meta-analysis combining individ-uals of all ancestries this variant no longer reached genome-wide significance although suggestive evidence persisted atrs1594318 (OR 0868 for C allele EAF 060 Pfrac14 378e-07) and otherintronic variants within ALLC in high linkage disequilibrium(LD) (Fig 3) For the permanent dentition the strongest statisti-cal evidence for association was seen between caries statusand rs7738851 (OR 128 for A allele EAF 085 Pfrac14 163e-08) (Figs 12 and 4 Table 1) This variant is intronic within NEDD9 on 6p24

Estimated heritability

Using participant level data in ALSPAC heritability was esti-mated at 028 (95 CI 009 048) and 017 (95 CI 002 031) forprimary and permanent teeth respectively Using summarystatistics at the meta-analysis level produced point estimatesnear zero heritability with wide confidence intervals (Table 2)

Cross-phenotype comparisons

Genome-wide mean chi-squared was too low to undertakegenome-wide genetic correlation using the linkage disequilib-rium score regression (LDSR) method for caries in either primaryor permanent teeth Hypothesis-free phenome-wide lookup forrs1594318 included 885 GWAS where either rs1594318 or a proxywith r2 gt 08 was present None of these traits showed evidenceof association with rs1594318 at a Bonferroni-corrected alpha of005 Lookup of rs7738851 and its proxies was performed against662 traits where similarly no traits reached a Bonferroni-corrected threshold Hypothesis-driven lookup in adult cariestraits revealed no strong evidence for persistent genetic effectsinto adulthood (Table 3)

Gene prioritization gene set enrichment and associationwith predicted gene transcription

Gene-based tests identified association between caries status inthe primary dentition and a region of 7q35 containing TCAF1OR2F2 and OR2F1 (Pfrac14 191e-06 158e-06 and 129e-06 respec-tively) There were insufficient independently associated loci toperform gene set enrichment analysis using DEPICT for either ofthe principal meta-analyses Association with predicted genetranscription was tested but no genes met the threshold for asso-ciation after accounting for multiple testing The single greatestevidence for association was seen between increased predictedtranscription of CDK5RAP3 and increased liability for permanentcaries (Pfrac14 394e-05) CDK5RAP3 is known to interact with PAK4and p14ARF with a potential role in oncogenesis (1819)

DiscussionDental caries in children and adolescents has not been studiedto date using a large-scale consortium-based genome-wide

3Human Molecular Genetics 2018 Vol 00 No 00 |

Downloaded from httpsacademicoupcomhmgadvance-article-abstractdoi101093hmgddy2375040780by University of Bristol Library useron 20 July 2018

meta-analysis approach Based on previous knowledge of theheritability of caries in young populations and from our under-standing of other complex diseases we anticipated that com-mon genetic variants would be associated with dental caries

risk with consistent effects across different cohorts We foundevidence for association between rs1594318 and caries in pri-mary teeth This variant showed weaker evidence for associa-tion in the multi-ethnic meta-analysis potentially relating to

Figure 1 Manhattan plots for each principal meta-analysis (A) Caries in primary teeth (European ancestry) n samples frac14 17 036 n variants frac14 8 640 819 k frac14 09944

Variants within 500Kb of rs1594318 are highlighted in green (B) Caries in primary teeth (multi-ethnic analysis) n samples frac14 19 003 n variants frac14 8 699 928 k frac1409861

(C) Caries in permanent teeth (European ancestry) n samples frac14 13 353 n variants frac14 8 734 121 k frac1409991 Variants within 500Kb of rs7738851 are highlighted in green

4 | Human Molecular Genetics 2018 Vol 00 No 00

Downloaded from httpsacademicoupcomhmgadvance-article-abstractdoi101093hmgddy2375040780by University of Bristol Library useron 20 July 2018

different allele frequencies across the different ethnic groupsincluded in analysis Frequency of the G allele is reported tovary between 024 in Asian populations and 042 in populationsof European ancestry based on 1KGP allele frequencies ALLC(Allantoicase) codes the enzyme allantoicase which is involvedin purine metabolism and whose enzymatic activity is believedto have been lost during vertebrate evolution Mouse studiessuggest that this loss of activity relates to low expression levelsand low substrate affinity rather than total non-functionality(20) Although there is some evidence that ALLC polymorphismsare associated with response to asthma treatment (21) there islimited understanding of the implications of variation in ALLCfor human health and it is possible that rs1594318 tags func-tionality elsewhere in the same locus

For permanent teeth we found evidence for associationbetween caries status and rs7738851 an intronic variant withNEDD9 (neural precursor cell-expressed developmentallydown-regulard gene 9) NEDD9 is reported to mediate integrin-initiated signal transduction pathways and is conserved from

gnathostomes into mammals (2223) NEDD9 appears to play anumber of functional roles in disease and normal develop-ment including regulation of neuronal differentiation devel-opment and migration (2224ndash28) One such function involvesregulation of neural crest cell migration (26) Disruption ofneural crest signalling is known to lead to enamel and dentindefects in animal models (2930) and might provide a mecha-nism for variation at rs7738851 to influence dental cariessusceptibility

Traditionally risk assessment for dental caries in childhoodhas concentrated on dietary behaviours and other modifiablerisk factors (31) with little focus on tooth quality Although ourunderstanding of the genetic risk factors for dental caries is in-complete authors have noted that the evidence from previousgenetic association studies tends to support a role for innatetooth structure and quality in risk of caries (3233) If validatedby future studies the association with rs7738851 would providefurther evidence for this argument and may in the future en-hance risk assessment in clinical practice

Figure 2 Regional association plots (A) Regional association plot for rs1594318 and caries in primary teeth (European ancestry meta-analysis) (B) Regional association

plot for rs7738851 and caries in permanent teeth

5Human Molecular Genetics 2018 Vol 00 No 00 |

Downloaded from httpsacademicoupcomhmgadvance-article-abstractdoi101093hmgddy2375040780by University of Bristol Library useron 20 July 2018

different allele frequencies across the different ethnic groupsincluded in analysis Frequency of the G allele is reported tovary between 024 in Asian populations and 042 in populationsof European ancestry based on 1KGP allele frequencies ALLC(Allantoicase) codes the enzyme allantoicase which is involvedin purine metabolism and whose enzymatic activity is believedto have been lost during vertebrate evolution Mouse studiessuggest that this loss of activity relates to low expression levelsand low substrate affinity rather than total non-functionality(20) Although there is some evidence that ALLC polymorphismsare associated with response to asthma treatment (21) there islimited understanding of the implications of variation in ALLCfor human health and it is possible that rs1594318 tags func-tionality elsewhere in the same locus

For permanent teeth we found evidence for associationbetween caries status and rs7738851 an intronic variant withNEDD9 (neural precursor cell-expressed developmentallydown-regulard gene 9) NEDD9 is reported to mediate integrin-initiated signal transduction pathways and is conserved from

gnathostomes into mammals (2223) NEDD9 appears to play anumber of functional roles in disease and normal develop-ment including regulation of neuronal differentiation devel-opment and migration (2224ndash28) One such function involvesregulation of neural crest cell migration (26) Disruption ofneural crest signalling is known to lead to enamel and dentindefects in animal models (2930) and might provide a mecha-nism for variation at rs7738851 to influence dental cariessusceptibility

Traditionally risk assessment for dental caries in childhoodhas concentrated on dietary behaviours and other modifiablerisk factors (31) with little focus on tooth quality Although ourunderstanding of the genetic risk factors for dental caries is in-complete authors have noted that the evidence from previousgenetic association studies tends to support a role for innatetooth structure and quality in risk of caries (3233) If validatedby future studies the association with rs7738851 would providefurther evidence for this argument and may in the future en-hance risk assessment in clinical practice

Figure 2 Regional association plots (A) Regional association plot for rs1594318 and caries in primary teeth (European ancestry meta-analysis) (B) Regional association

plot for rs7738851 and caries in permanent teeth

5Human Molecular Genetics 2018 Vol 00 No 00 |

Downloaded from httpsacademicoupcomhmgadvance-article-abstractdoi101093hmgddy2375040780by University of Bristol Library useron 20 July 2018

The lookup of lead associated variants against adult cariestraits provided no strong evidence for persistent association inadulthood This might imply genetic effects which are specificto the near-eruption timepoint An alternative explanation isthat the variants identified in the present study represent falsepositive signals as the statistical evidence presented is not irre-futable and there is no formal replication stage in our studyyet we see good consistency of effects across studies

The meta-analysis heritability estimates were lower thananticipated from either previous within-study heritability esti-mates (34) or the new within-study heritability estimatesobtained for this analysis There are several possible explana-tions for this phenomenon First the methods used in the pre-sent analysis are SNP based which consistently underestimate

heritability of complex traits relative to twin and family studies(35) Second meta-analysis heritability represents the heritabil-ity of genetic effects which are consistent across populations Inthe event of genuine differences in genetic architecture of den-tal caries across strata of age geography environmentalexposure or subtly different phenotypic meanings the meta-heritability estimate is not the same conceptually as theweighted average of heritability within each study

More intuitively genetic influences might be importantwithin populations with relatively similar environments but notdetermine much of the overall differences in risk when compar-ing groups of people in markedly different environments Thisview is consistent with existing literature from family basedand candidate gene association studies suggesting the genetic

Figure 3 Forest plot for rs1594318 and caries in primary teeth Effect sizes are expressed on a log OR scale grouped by geographical location The summary estimate is

from the fixed-effect meta-analysis of participants of European ancestry

6 | Human Molecular Genetics 2018 Vol 00 No 00

Downloaded from httpsacademicoupcomhmgadvance-article-abstractdoi101093hmgddy2375040780by University of Bristol Library useron 20 July 2018

architecture of dental caries is complex with multiple interac-tions For example genendashsex interactions are reported whichchange in magnitude between the primary and permanent den-tition (36) genetic variants may have heterogeneous effects onthe primary and permanent dentition (37) and environmentalexposures such as fluoride may interact with genetic effects(38) Finally the aetiological relevance of specific microbiomegroups appears to vary between different populations (39) sug-gesting genetic effects acting through the oral microbiomemight also vary between populations Unfortunately this studylacks statistical power to perform meta-analyses stratified onthese exposures so does not resolve this particular question

In line with any consortium-based approach the need toharmonize analysis across different collections led to some

compromises The phenotypic definitions used in this study donot contain information on disease extent or severity Loss ofinformation in creating these definitions may have contributedto the low statistical power of analysis Our motivation for usingsimple definitions was based on the facts that (a) case-controlstatus simply represents a threshold level of an underlying con-tinuum of disease risk (b) simple binary classifications facilitatecomparison of studies with different assessment protocols andpopulation risks and (c) simple classifications have been usedsuccessfully in a range of complex phenotypes

Between participating centres there are differences in char-acteristics such as age at participation phenotypic assessmentand differences in the environment (such as nutrition oral hy-giene and the oral microbiome) which might influence dental

Figure 4 Forest plot for rs7738851 and caries in permanent teeth Effect sizes are expressed on a log OR scale grouped by geographical location The summary estimate

is from fixed-effect meta-analysis

7Human Molecular Genetics 2018 Vol 00 No 00 |

Downloaded from httpsacademicoupcomhmgadvance-article-abstractdoi101093hmgddy2375040780by University of Bristol Library useron 20 July 2018

architecture of dental caries is complex with multiple interac-tions For example genendashsex interactions are reported whichchange in magnitude between the primary and permanent den-tition (36) genetic variants may have heterogeneous effects onthe primary and permanent dentition (37) and environmentalexposures such as fluoride may interact with genetic effects(38) Finally the aetiological relevance of specific microbiomegroups appears to vary between different populations (39) sug-gesting genetic effects acting through the oral microbiomemight also vary between populations Unfortunately this studylacks statistical power to perform meta-analyses stratified onthese exposures so does not resolve this particular question

In line with any consortium-based approach the need toharmonize analysis across different collections led to some

compromises The phenotypic definitions used in this study donot contain information on disease extent or severity Loss ofinformation in creating these definitions may have contributedto the low statistical power of analysis Our motivation for usingsimple definitions was based on the facts that (a) case-controlstatus simply represents a threshold level of an underlying con-tinuum of disease risk (b) simple binary classifications facilitatecomparison of studies with different assessment protocols andpopulation risks and (c) simple classifications have been usedsuccessfully in a range of complex phenotypes

Between participating centres there are differences in char-acteristics such as age at participation phenotypic assessmentand differences in the environment (such as nutrition oral hy-giene and the oral microbiome) which might influence dental

Figure 4 Forest plot for rs7738851 and caries in permanent teeth Effect sizes are expressed on a log OR scale grouped by geographical location The summary estimate

is from fixed-effect meta-analysis

7Human Molecular Genetics 2018 Vol 00 No 00 |

Downloaded from httpsacademicoupcomhmgadvance-article-abstractdoi101093hmgddy2375040780by University of Bristol Library useron 20 July 2018

caries or its treatment as reflected in the wide range of cariesprevalence between different study centres Varying phenotypiccharacteristics do not necessarily result in heterogeneous ge-netic effects as this variability may be uncorrelated with ge-netic effects There was little evidence for heterogeneity in thetop associated loci reported however the test for heterogeneityin genetic effects (I2) is limited by the small number of

participating studies in meta-analysis (40) and wide confidenceintervals for within-study genetic effect estimates Given theselimitations it is possible that heterogeneity contributed to lowstudy power and prevented more comprehensive single variantfindings

In the ALSPAC study we made extensive use of question-naire derived data This will systematically under-report true

Table 1 Lead associated single variants

Phenotype Variant Position Effectallele

Otherallele

EAF Beta (SE) Oddsratio

P-value N I2 P-value forheterogeneity

Annotation

Caries in primaryteeth (Europeanancestry analysis)

rs1594318 chr2 3733944 C G 060 0165 (0030) 0848 413e-08 16 994 00 069 IntronicALLC

Caries in primaryteeth (multi-ethnic analysis)a

rs1594318 chr2 3733944 C G 060 0142 (0028) 0868 378e-07 18 960 00 061 IntronicALLC

Caries in primaryteeth(multi-ethnicanalysis)a

rs872877 chr2 3735826 A G 059 0142 (0028) 0868 418e-07 18 958 175 068 IntronicALLC

Caries in permanentteeth

rs7738851 chr6 11241788 A T 085 0248 (0044) 128 163e-08 13 353 133 020 IntronicNEDD9

aNo single variants were associated with dental caries status at the genome-wide level in the multi-ethnic analysis of primary teeth however two variants are dis-

cussed in Results section and are included here for reference

Table 2 Within-sample and meta-analysis heritability estimates

Phenotype Method Estimated h2 (95 CI) N

Caries in primary teeth GCTA GREML 028 (009 048) 7230LDSR All participants 001 (000 006) 19 003

European ancestry only 001 (000 007) 17 036Caries in permanent teeth GCTA GREML 017 (002 031) 6657

LDSR 006 (000 012) 13 353

Table 3 Lookup of lead associated variants

Variant Discovery trait Risk increasingallele (discovery)

Cross trait lookup P-value Effect per cariesrisk increasingallele (se)

N

rs1594318 Caries in primaryteeth (Europeanancestrymeta-analysis)

G Adult cariestraits

DMFS (standard deviation of residualsof caries-affected surfaces)

087 00015 (00092) 26 790

Number of teeth (inverse normaltransformed residuals)

060 00051 (00098) 27 947

Standardized DFS (inverse normaltransformed residuals)

0033 00195 (00091) 26 532

Hypothesis free (No traits meeting threshold for multiple testing)rs7738851 Caries in

permanent teethA Adult caries

traitsDMFS (standard deviation of residuals

of caries-affected surfaces)057 0007 (0011) 26 791

Number of teeth (inverse normaltransformed residuals)

063 00064 (0013) 27 949

Standardized DFS (inverse normaltransformed residuals)

065 00054 (0012) 26 531

Hypothesis free (No traits meeting threshold for multiple testing)

Adult caries traits were defined as follows DMFSmdasha count of the number of decayed missing or filled tooth surfaces This count was residualized after regression on

age and age-squared and standard deviations of residuals calculated Number of teethmdasha count of the number of teeth in the mouth This count was residualized after

regression on age and age-squared and residuals underwent inverse normal transformation Standardized DFS The number of decayed and filled surfaces was divided

by the total number of tooth surfaces in the mouth This ratio was residualized after regression on age and age-squared and residuals underwent inverse normal

transformation

8 | Human Molecular Genetics 2018 Vol 00 No 00

Downloaded from httpsacademicoupcomhmgadvance-article-abstractdoi101093hmgddy2375040780by University of Bristol Library useron 20 July 2018

caries exposure compared with other studies as children ortheir parents are unlikely to be aware of untreated dental carieswhich would be evident to a trained assessor We have exploredsome of these issues previously and shown that self-reportmeasures at scale can be used to make meaningful inferenceabout dental health in childhood (41) We believe that misclassi-fication and under-reporting in questionnaire data would tendto bias genetic effect estimates and heritability toward the nullDespite this we show evidence for heritability using these defi-nitions and effect sizes at lead variants are comparable with ef-fect sizes obtained using clinically assessed data (Figs 3 and 4)

As our power calculations showed the sample size was suf-ficient to detect the identified variants associated at a genomewide significant level with caries in the primary teeth(rs1594318) and in permanent teeth (rs872877) where we ob-served relatively large effect sizes For smaller effect sizes wewere underpowered to identify association and did not detectany variants with effect sizes (expressed as per-allele increasedodds) smaller than 15 or 17 in the primary and permanentteeth respectively Caries is highly influenced by environmentalfactors and it is likely that its susceptibility is polygenic in na-ture (32) with individual genetic variants conferring small effectsizes as seen in other comparable complex traits (42)Furthermore some of the included studies had major differen-ces in their caries prevalence likely acting as a proxy for fea-tures affecting risk of caries This may have introducedheterogeneity and reduced power to detect association as dis-cussed further below

One area of interest in the literature is the ability of geneticsto guide personalized decisions on risk screening or identifyingtreatment modalities and this is also true in dentistry The ge-netic variants identified in this study are unlikely to be usefulon their own in this context given the modest effect sizes andlow total heritability observed in our meta-analysis We wouldsuggest clinicians should continue to consider environmentand aggregate genetic effects (eg knowledge of disease pat-terns of close relatives) rather than specific genetic variants atthis moment in time Nevertheless the findings of our studycontribute to a better understanding of the genetic and biologi-cal mechanisms underlying caries susceptibility

Materials and MethodsStudy samples

We performed genome-wide association (GWA) analysis fordental caries casecontrol status in a consortium including ninecoordinating centres Study procedures differed between thesecentres We use the term lsquoclinical dental assessmentrsquo to meanthat a child was examined in person whether this was in a den-tal clinic or a study centre We use the term lsquoexaminerrsquo to referto a dental professional and use the term lsquoassessorrsquo to refer toan individual with training who is not a dental professional forexample a trained research nurse

The Avon Longitudinal Study of Parents and Children(ALSPAC) is a longitudinal birth cohort which recruited pregnantwomen living near Bristol UK with an estimated delivery date be-tween 1991 and 1992 Follow-up has included clinical assessmentand questionnaires and is ongoing (43) A subset of childrenattended clinics including clinical dental assessment by a trainedassessor at age 31 43 and 61 months of age Parents were askedto complete questionnaires about their childrenrsquos health regu-larly including comprehensive questions at a mean age of 54and 64 years Parents and children were asked to complete

questionnaires about oral health at a mean age of 75 107 and178 years Please note that the study website contains detailsof all the data that are available through a fully searchable datadictionary (wwwbristolacukalspacresearchersaccess date lastaccessed June 2018) Both clinical and questionnaire derived datawere included in this analysis with priority given to clinical datawere available (Supplementary Material Table S3)

The Copenhagen Prospective Studies on Asthma inChildhood includes two population-based longitudinal birthcohorts in Eastern Denmark COPSAC2000 recruited pregnantwomen with a history of asthma between 1998 and 2001 (44)Children who developed wheeze in early life were consideredfor enrolment in a nested randomized trial for asthma preven-tion COPSAC2010 recruited pregnant women between 2008 and2010 and was not selected on asthma status Both COPSAC2000and COPSAC2010 studies included regular clinical follow-upWithin Denmark clinical dental assessment is routinely offeredto children and adolescents until the age of 18 years and sum-mary data from these examinations are stored in a national reg-ister These data were obtained via index linkage forparticipants of COPSAC2000 and COPSAC2010 and used to per-form joint analysis across both cohorts

The Danish National Birth Cohort (DNBC) is a longitudinalbirth cohort which recruited women in mid-pregnancy from1996 onwards (45) For this analysis index linkage was per-formed to obtain childhood dental records for mothers partici-pating in DNBC As with the COPSAC studies these data wereoriginally obtained by a qualified dentist and included surfacelevel dental charting

The Generation R study (GENR) recruited women in earlypregnancy with expected delivery dates between 2002 and 2006living in the city of Rotterdam the Netherlands The cohort ismulti-ethnic with representation from several non-Europeanethnic groups Follow-up has included clinical assessment visitsand questionnaires and is ongoing (46) Intra-oral photographywas performed as a part of their study protocol with surfacelevel charting produced by a dental examiner (a specialist inpaediatric dentistry) (47) Analysis in GENR included (a) a multi-ethnic association study including all individuals with geneticand phenotypic data (48) and (b) analysis including only individ-uals of European ancestry

The GENEVA consortium is a group of studies which under-take coordinated analysis across several phenotypes (49)Within GENEVA the Center for Oral Health Research in ruralAppalachia West Virginia and Pennsylvania USA (COHRA) theIowa Fluoride Study in Iowa USA (IFS) and the Iowa Head Start(IHS) study participated in analysis of dental traits in children(15) COHRA recruited families with at least one child aged be-tween 1 and 18 years of age with dental examination performedat baseline (50) IFS recruited mothers and new-born infants inIowa between 1992 and 1995 with a focus on longitudinal fluo-ride exposures and dental and bone health outcomes Clinicaldental examination in IFS was performed by trained assessorsaged 5 9 13 and 17 years (51) IHS recruited children participat-ing in an early childhood education program which included aone-time clinical dental examination (13)

The lsquoGerman Infant study on the influence of NutritionIntervention plus air pollution and genetics on allergy devel-opmentrsquo (GINIplus) is a multi-centre prospective birth cohortstudy which has an observational and interventional arm whichconducted a nutritional intervention during the first 4 monthsof life The study recruited new born infants with and withoutfamily history of allergy in the Munich and Wesel areasGermany between 1995 and 1998 (5253) The lsquoLifestyle-related

9Human Molecular Genetics 2018 Vol 00 No 00 |

Downloaded from httpsacademicoupcomhmgadvance-article-abstractdoi101093hmgddy2375040780by University of Bristol Library useron 20 July 2018

caries exposure compared with other studies as children ortheir parents are unlikely to be aware of untreated dental carieswhich would be evident to a trained assessor We have exploredsome of these issues previously and shown that self-reportmeasures at scale can be used to make meaningful inferenceabout dental health in childhood (41) We believe that misclassi-fication and under-reporting in questionnaire data would tendto bias genetic effect estimates and heritability toward the nullDespite this we show evidence for heritability using these defi-nitions and effect sizes at lead variants are comparable with ef-fect sizes obtained using clinically assessed data (Figs 3 and 4)

As our power calculations showed the sample size was suf-ficient to detect the identified variants associated at a genomewide significant level with caries in the primary teeth(rs1594318) and in permanent teeth (rs872877) where we ob-served relatively large effect sizes For smaller effect sizes wewere underpowered to identify association and did not detectany variants with effect sizes (expressed as per-allele increasedodds) smaller than 15 or 17 in the primary and permanentteeth respectively Caries is highly influenced by environmentalfactors and it is likely that its susceptibility is polygenic in na-ture (32) with individual genetic variants conferring small effectsizes as seen in other comparable complex traits (42)Furthermore some of the included studies had major differen-ces in their caries prevalence likely acting as a proxy for fea-tures affecting risk of caries This may have introducedheterogeneity and reduced power to detect association as dis-cussed further below

One area of interest in the literature is the ability of geneticsto guide personalized decisions on risk screening or identifyingtreatment modalities and this is also true in dentistry The ge-netic variants identified in this study are unlikely to be usefulon their own in this context given the modest effect sizes andlow total heritability observed in our meta-analysis We wouldsuggest clinicians should continue to consider environmentand aggregate genetic effects (eg knowledge of disease pat-terns of close relatives) rather than specific genetic variants atthis moment in time Nevertheless the findings of our studycontribute to a better understanding of the genetic and biologi-cal mechanisms underlying caries susceptibility

Materials and MethodsStudy samples

We performed genome-wide association (GWA) analysis fordental caries casecontrol status in a consortium including ninecoordinating centres Study procedures differed between thesecentres We use the term lsquoclinical dental assessmentrsquo to meanthat a child was examined in person whether this was in a den-tal clinic or a study centre We use the term lsquoexaminerrsquo to referto a dental professional and use the term lsquoassessorrsquo to refer toan individual with training who is not a dental professional forexample a trained research nurse

The Avon Longitudinal Study of Parents and Children(ALSPAC) is a longitudinal birth cohort which recruited pregnantwomen living near Bristol UK with an estimated delivery date be-tween 1991 and 1992 Follow-up has included clinical assessmentand questionnaires and is ongoing (43) A subset of childrenattended clinics including clinical dental assessment by a trainedassessor at age 31 43 and 61 months of age Parents were askedto complete questionnaires about their childrenrsquos health regu-larly including comprehensive questions at a mean age of 54and 64 years Parents and children were asked to complete

questionnaires about oral health at a mean age of 75 107 and178 years Please note that the study website contains detailsof all the data that are available through a fully searchable datadictionary (wwwbristolacukalspacresearchersaccess date lastaccessed June 2018) Both clinical and questionnaire derived datawere included in this analysis with priority given to clinical datawere available (Supplementary Material Table S3)

The Copenhagen Prospective Studies on Asthma inChildhood includes two population-based longitudinal birthcohorts in Eastern Denmark COPSAC2000 recruited pregnantwomen with a history of asthma between 1998 and 2001 (44)Children who developed wheeze in early life were consideredfor enrolment in a nested randomized trial for asthma preven-tion COPSAC2010 recruited pregnant women between 2008 and2010 and was not selected on asthma status Both COPSAC2000and COPSAC2010 studies included regular clinical follow-upWithin Denmark clinical dental assessment is routinely offeredto children and adolescents until the age of 18 years and sum-mary data from these examinations are stored in a national reg-ister These data were obtained via index linkage forparticipants of COPSAC2000 and COPSAC2010 and used to per-form joint analysis across both cohorts

The Danish National Birth Cohort (DNBC) is a longitudinalbirth cohort which recruited women in mid-pregnancy from1996 onwards (45) For this analysis index linkage was per-formed to obtain childhood dental records for mothers partici-pating in DNBC As with the COPSAC studies these data wereoriginally obtained by a qualified dentist and included surfacelevel dental charting

The Generation R study (GENR) recruited women in earlypregnancy with expected delivery dates between 2002 and 2006living in the city of Rotterdam the Netherlands The cohort ismulti-ethnic with representation from several non-Europeanethnic groups Follow-up has included clinical assessment visitsand questionnaires and is ongoing (46) Intra-oral photographywas performed as a part of their study protocol with surfacelevel charting produced by a dental examiner (a specialist inpaediatric dentistry) (47) Analysis in GENR included (a) a multi-ethnic association study including all individuals with geneticand phenotypic data (48) and (b) analysis including only individ-uals of European ancestry

The GENEVA consortium is a group of studies which under-take coordinated analysis across several phenotypes (49)Within GENEVA the Center for Oral Health Research in ruralAppalachia West Virginia and Pennsylvania USA (COHRA) theIowa Fluoride Study in Iowa USA (IFS) and the Iowa Head Start(IHS) study participated in analysis of dental traits in children(15) COHRA recruited families with at least one child aged be-tween 1 and 18 years of age with dental examination performedat baseline (50) IFS recruited mothers and new-born infants inIowa between 1992 and 1995 with a focus on longitudinal fluo-ride exposures and dental and bone health outcomes Clinicaldental examination in IFS was performed by trained assessorsaged 5 9 13 and 17 years (51) IHS recruited children participat-ing in an early childhood education program which included aone-time clinical dental examination (13)

The lsquoGerman Infant study on the influence of NutritionIntervention plus air pollution and genetics on allergy devel-opmentrsquo (GINIplus) is a multi-centre prospective birth cohortstudy which has an observational and interventional arm whichconducted a nutritional intervention during the first 4 monthsof life The study recruited new born infants with and withoutfamily history of allergy in the Munich and Wesel areasGermany between 1995 and 1998 (5253) The lsquoLifestyle-related

9Human Molecular Genetics 2018 Vol 00 No 00 |

Downloaded from httpsacademicoupcomhmgadvance-article-abstractdoi101093hmgddy2375040780by University of Bristol Library useron 20 July 2018

factors Immune System and the development of Allergies inEast and West Germanyrsquo study (LISA) is a longitudinal birth co-hort which recruited between 1997 and 1999 across four sites inGermany (5254) For participants living in the Munich area fol-low-up used similar protocols in both GINIplus and LISA withquestionnaire and clinic data including clinical dental examina-tion by trained examiners at age 10 and 15 years Analysis forcaries in GINIplus and LISA was therefore performed acrossboth studies for participants at the Munich study centre

The Physical Activity and Nutrition in Children (PANIC)Study is an ongoing controlled physical activity and dietary in-tervention study in a population of children followed retrospec-tively since pregnancy and prospectively until adolescenceAltogether 512 children 6ndash8 years of age were recruited in 2008ndash2009 (55) The main aims of the study are to investigate risk fac-tors and pathophysiological mechanisms for overweight type 2diabetes atherosclerotic cardiovascular diseases musculoskel-etal diseases psychiatric disorders dementia and oral healthproblems and the effects of a long-term physical activity and di-etary intervention on these risk factors and pathophysiologicalmechanisms Clinical dental examinations were performed by aqualified dentist with tooth level charting

The Cardiovascular Risk in Young Finns Study (YFS) is amulti-centre investigation which aimed to understand thedeterminants of cardiovascular risk factors in young people inFinland The study recruited participants who were aged 3 6 912 15 and 18 years old in 1980 Eligible participants living in spe-cific regions of Finland were identified at random from a na-tional population register and were invited to participateRegular follow-up has been performed through physical exami-nation and questionnaires (56) Clinical dental examination wasperformed by a qualified dentist with tooth level charting

The Western Australian Pregnancy Cohort (RAINE) study is abirth cohort which recruited women between 16th and 20thweek of pregnancy living in the Perth area Western AustraliaRecruitment occurred between 1989 and 1991 with regular fol-low-up of mothers and their children through research clinicsand questionnaires (57) The presence or absence of dental car-ies was recorded by a trained assessor following clinical dentalexamination at the year 3 clinic follow-up

Further details of study samples are provided in SupplementaryMaterial Table S1

Medical EthicsWithin each participating study written informed consent wasobtained from the parents of participating children after receiv-ing a full explanation of the study Children were invited to giveassent where appropriate All studies were conducted in accor-dance with the Declaration of Helsinki

Ethical approval for the ALSPAC study was obtained fromthe ALSPAC Ethics and Law Committee and the Local ResearchEthics Committee Full details of ethical approval policies andsupporting documentation are available online (httpwwwbristolacukalspacresearchersresearch-ethics date last accessedJune 2018) Approval to undertake analysis of caries traits wasgranted by the ALSPAC executive committee (B2356)

The COPSAC2000 cohort was approved by the RegionalScientific Ethical Committee for Copenhagen and Frederiksberg(KF 01-28996) and the Danish Data Protection Agency (2008-41-1574) The 2010 cohort (COPSAC2010) was approved by theDanish Ethics Committee (H-B-2008-093) and the Danish DataProtection Agency (2008-41-2599)

The DNBC study of caries was approved by the Scientific EthicsCommittee for the Capital City Region (Copenhagen) the DanishData Protection Agency and the DNBC steering committee

Each participating site in the GENEVA consortium cariesanalysis received approval from the local university institu-tional review board (federal wide assurance number forGENEVA caries project FWA00006790) Within the COHRA armlocal approval was provided by the University of Pittsburgh(0207030506048) and West Virginia University (15620B) whilstthe IFS and IHS arms received local approval from theUniversity of Iowarsquos Institutional Review Board

The GENR study design and specific data acquisition wereapproved by the Medical Ethical Committee of the ErasmusUniversity Medical Center Rotterdam The Netherlands (MEC-2007-413)

The GINIplus and LISA studies were approved by the ethicscommittee of the Bavarian Board of Physicians (10 year follow-up 05100 for GINIplus and 07098 for LISA 15 year follow-up10090 for GINIplus 12067 for LISA)

The PANIC study protocol was approved by the ResearchEthics Committee of the Hospital District of Northern Savo Allparticipating children and their parents gave informed writtenconsent

The YFS study protocol was approved by local ethics com-mittees for contributing sites

The RAINE study was approved by the University of WesternAustralia Human Research Ethics Committee

Phenotypes

Primary teeth exfoliate and are replaced by permanent teethbetween 6 and 12 years of age We aimed to separate caries sta-tus in primary and permanent teeth wherever possible usingclinical information or age criteria in line with our expectationthat the genetic risk factors for dental caries might differ be-tween primary and permanent dentition For children in themixed dentition we created two parallel case definitions whilstin younger or older children a single case definition wassufficient

All study samples included a mixture of children with dentalcaries and children who were caries-free with varying degreesof within-mouth or within-tooth resolution To facilitate com-parison across these differing degrees of resolution all analysiscompared children who were caries-free (unaffected) or haddental caries (affected) Missing teeth could represent exfolia-tion or delayed eruption rather than the endpoint of dental car-ies and therefore missing teeth were not included in classifyingchildren as caries-free or caries affected

In children aged 250 years to 599 years any individual with1 or more decayed or filled tooth was classified as caries af-fected with all remaining individuals classified as unaffectedIn children aged 600 years to 1199 years of age parallel defini-tions were determined for the primary dentition and permanentdentition respectively Any individual with at least 1 decayedor filled primary tooth was classified as caries affected for pri-mary teeth while all remaining participants were classified asunaffected In parallel any individual with at least 1 decayed orfilled permanent tooth was classified as caries affected for per-manent teeth while all remaining individuals were classified asunaffected In children and adolescents aged 1200 to 1799years of age any individual with 1 or more decayed or filledtooth or tooth surface (excluding third molar teeth) was classi-fied as caries affected with remaining individuals classified asunaffected

10 | Human Molecular Genetics 2018 Vol 00 No 00

Downloaded from httpsacademicoupcomhmgadvance-article-abstractdoi101093hmgddy2375040780by University of Bristol Library useron 20 July 2018

Analysis was conducted in cross-section meaning a singleparticipant could only be represented in a single phenotype defi-nition once Where multiple sources of dental data were availablefor a single participant within a single phenotype definition win-dow the first source of data was selected (reflecting the youngestage at participation) in line with our expectation that caries sta-tus would be most heritable in the near-eruption period

The sources of data used to create these phenotypic defini-tions are given in Supplementary Material Table S3 WithinALSPAC only questionnaire responses were used to supple-ment data from clinical examination The questions asked didnot distinguish between primary and permanent teeth Basedon the age at questionnaire response we derived variableswhich prioritized responses from questionnaires before 600years of age (thought to predominantly represent caries in pri-mary teeth) and responses after 1000 years of age (which mightpredominantly represent caries in permanent teeth) The finaldata sweep considered in this analysis targeted adolescents atage 1750 years Some participants responded to this after their18th birthday Data derived from this final questionnaire sweepwere not included in the principal meta-analyses but were in-cluded in the GCTA heritability analysis

Genotypes and imputation

All participating studies used genetic data imputed to a compre-hensive imputation panel The 1000 genomes phase 1 version 3panel (1KG phase 1 v3) was used as a common basis across sixcentres (GINIplusLISA GENR GENEVA YFS PANIC RAINE)(Supplementary Material Table S1) In ALSPAC DNBCCOPSAC2000 and COPSAC 2010 the haplotype reference consor-tium (HRC v10 and v11) imputation panels were used(Supplementary Material Table S1)

Each study performed routine QC measures during genotyp-ing imputation and association testing (SupplementaryMaterial Table S2) Further pre-meta-analysis QC was per-formed centrally using the EasyQC R package and accompany-ing 1KG phase1 v3 reference data (58) Minor allele count (MAC)was derived as the product of minor allele frequency (MAF) andsite-specific number of alleles (twice the site-specific samplesize) Variants were dropped which had a per-file MAC of 6 orlower a site-specific sample size of 30 or lower or an imputeINFO score of less than 04 Sites which reported effect and non-effect alleles other than those reported in 1KG phase 1 v3 refer-ence data were dropped Following meta-analysis sites with aweighted MAF of less than 0005were dropped along with var-iants present in less than 50 of the total sample

Statistical analysis

Association testingEach cohort preformed GWA analysis using an additive geneticmodel Caries status was modelled against genotype dosagewhilst accounting for age at phenotypic assessment agesquared sex and cryptic relatedness Sex was accounted for byderiving phenotypic definitions and performing analysis sepa-rately within male and female participants or by including sexas a covariate in association testing Each study adoptedapproaches to account for cryptic relatedness and populationstratification as described in Supplementary Material Table S2In the GENR study parallel analyses were conducted for partici-pants of European ancestry (using the approach described inSupplementary Material Table S2) and the entire study

population using a previously published method (48) The soft-ware and exact approach used by each study is shown inSupplementary Material Table S2

Meta-analysisResults of GWA analysis within each study were combined intwo principal meta-analyses representing caries status in pri-mary teeth and caries status in permanent teeth For primaryteeth parallel meta-analyses were performed one using resultsof multi-ethnic analysis in the GENR study and the other usingresults of European ancestry analysis in the GENR study TheGENR study did not have phenotypic data for permanent teeththerefore the analysis of permanent teeth contained only indi-viduals of European ancestry Fixed-effects meta-analyses wasperformed using METAL (59) with genomic control of inputsummary statistics enabled and I2 test for heterogeneityMeta-analysis was run in parallel in two centres and resultscompared All available studies with genotype and phenotypicinformation were included in a one-stage design thereforethere was no separate replication stage

Meta-analysis heritability estimatesFor each principal meta-analysis population stratification andheritability were assessed using LDSR (60) Reference LD scoreswere taken from HapMap3 reference data accompanying theLDSR package

Within-sample heritability estimatesFor comparison heritability within the ALSPAC study wasassessed using the GREML method (61) implemented in theGCTA software package (62) using participant level phenotypedata and a genetic relatedness matrix estimated from commongenetic variants (with MAFgt 50) present in HapMap3

Hypothesis-free cross-trait lookupWe used PLINK 20 (63) to clump meta-analysis summary statis-tics based on LD structure in reference data from the UK10Kproject We then performed hypothesis-free cross-trait lookupof independently associated loci using the SNP lookup functionin the MRBase catalogue (64) Proxies with an r2 of 08 or higherwere included where the given variant was not present in anoutcome of interest We considered performing hypothesis-freecross-trait genetic correlation analysis using bivariate LD scoreregression implemented in LDhub (65)

Lookup in previously published paediatric caries GWASPreviously published caries GWAS was performed within theGENEVA consortium which is also represented in our meta-analysis We therefore did not feel it would be informative to under-take lookup of associated variants in previously published results

Lookup in GWAS for adult caries traitsThis analysis was planned and conducted in parallel withanalysis of quantitative traits measuring lifetime caries expo-sure in adults (manuscript in draft)The principal trait studiedin the adult analysis was an index of decayed missing and filledtooth surfaces (DMFS) This index was calculated from results ofclinical dental examination excluding third molar teeth TheDMFS index was age-and-sex standardized within each partici-pating adult study before GWAS analysis was undertakenStudy-specific results files were then combined in a fixed-effects meta-analysis In addition to DMFS two secondary

11Human Molecular Genetics 2018 Vol 00 No 00 |

Downloaded from httpsacademicoupcomhmgadvance-article-abstractdoi101093hmgddy2375040780by University of Bristol Library useron 20 July 2018

caries traits were studied in adults namely number of teeth(a count of remaining natural teeth at time of study participa-tion) and standardized DFS (derived as the number of decayedand filled surfaces divided by the number of natural tooth surfa-ces remaining at time of study participation) After age-and-sexstandardization these secondary traits had markedly non-normal distribution and were therefore underwent rank-basedinverse normal transformation before GWAS analysis andmeta-analysis We performed cross-trait lookup of lead associ-ated variants in the paediatric caries meta-analysis againstthese three adult caries traits As the unpublished analysis alsocontains samples which contributed to previously publishedGWAS we did not feel it would be informative to undertake ad-ditional lookup in published data

Gene prioritization gene set enrichment and association withpredicted gene transcriptionGene-based testing of summary statistics was performed usingMAGMA (66) with reference data for LD correction taken fromthe UK10K project and gene definitions based on a 50 kb windoweither side of canonical gene start stop positions Gene set en-richment analysis was considered using the software packageDEPICT (67) Tests for association between phenotype and pre-dicted gene transcription were performed using S-PrediXcan (68)which is a summary-statistic implementation of the PrediXcanmethod This method aims to assess the effects of tissue-specificgene transcription on phenotypes Gene transcription models aretrained in datasets with transcriptomic data then used to predictgene expression in datasets with phenotypic data This methodwas applied using the MetaXcan standalone software (httpsgithubcomhakyimlabMetaXcan date last accessed June 2018)and a transcription prediction model trained in whole blood(obtained from the PedictDB data repository at httppredictdborg date last accessed June 2018) Bonferroni correction was ap-plied on the basis of approximately 7000 independent gene-basedtests for two caries traits giving an experiment-wide significancelevel of approximately Plt 36e-06

Power calculationsPost-hoc power calculations were performed using the freeweb-based tool Genetic Association Study (GAS) PowerCalculator and the software utility Quanto (v124) (httpscsgsphumicheduabecasisgas_power_calculatorindexhtmlhttpbiostatsusceduQuantohtml date last accessed June2018) (69) Using the sample size and caries prevalence of the fi-nal meta-analysis samples we calculated the minimum effectsize required to have 80 discovery power at a significance levelof 50e-08 for variants with MAF between 005 and 050 For pri-mary teeth (17 037 individuals 6922 caries affected prevalence406) we were able to detect variants with a minimal effectsize (OR) between 113 and 137 for variants with MAF of 050and 005 respectively (115 for MAF of 040) (SupplementaryMaterial Figs S4 and S5) For permanent teeth (13 353 individu-als of which 5875 were caries-affected prevalence 440) wehad 80 power to detect variants with a minimal effect size(OR) between 115 and 143 for variants with MAF of 050 and005 respectively (117 for MAF of 040) (SupplementaryMaterial Figs S4 and S5)

Supplementary MaterialSupplementary Material is available at HMG online

Acknowledgements

This work was supported by Wellcome (grant number 202802Z16Z to NT 201237Z16Z to SH) and the UK Medical ResearchCouncil (grant number MC_UU_120133) NT works in a unitwhich receives support from the University of Bristol and in abiomedical research centre which receives support from theNational Institute for Health Research The Swedish ResearchCouncil provides support to DS in the form of an InternationalFellowship (grant number 41-2016-00416)ALSPAC receives core support from the UK Medical ResearchCouncil and Wellcome (grant number 1022152132) and theUniversity of Bristol This publication is the work of the authorsand Nicholas Timpson will serve as guarantor for the contentsof this article A comprehensive list of grants funding availableon the ALSPAC website (httpwwwbristolacukalspacexternaldocumentsgrant-acknowledgementspdf) Collection ofphenotype data was supported by Wellcome and the UKMedical Research Council (grant number 076467Z05Z) GWASdata was generated by Sample Logistics and GenotypingFacilities at Wellcome Sanger Institute and LabCorp (LaboratoryCorporation of America) using support from 23andMeThe Young Finns Study has been financially supported by theAcademy of Finland (grant numbers 286284 134309 [Eye] 126925121584 124282 129378 [Salve] 17787 [Gendi] and 41071 (Skidi)] theSocial Insurance Institution of Finland Competitive State ResearchFinancing of the Expert Responsibility area of Kuopio Tampereand Turku University Hospitals (grant number X51001) Juho VainioFoundation Paavo Nurmi Foundation Finnish Foundation forCardiovascular Research Finnish Cultural Foundation TampereTuberculosis Foundation Emil Aaltonen Foundation YrjoJahnsson Foundation Signe and Ane Gyllenberg Foundation andDiabetes Research Foundation of Finnish Diabetes Association

Analysis within the GENEVA consortium was supported by thefollowing USA National Institutes of Health (NIH) grants from theNational Institute of Dental and Craniofacial Research (NIDCR)(grant numbers R01-DE014899 U01-DE018903 R03-DE024264 R01-DE09551 R01-DE12101 P60-DE-013076) and a National Institutesfor Health contract (contract number HHSN268200782ndash096C)

Analysis within Raine was supported by the National Healthand Medical Research Council of Australia (grant numbers572613 and 40398) and the Canadian Institutes of HealthResearch (grant number MOP-82893) The authors are grateful tothe Raine Study participants and their families and to the RaineStudy research staff for cohort coordination and data collectionThe authors gratefully acknowledge the NHampMRC for their long-term funding to the study over the last 25 years and also the fol-lowing institutes for providing funding for Core Management ofthe Raine Study The University of Western Australia (UWA)Curtin University the Raine Medical Research Foundation theUWA Faculty of Medicine Dentistry and Health Sciences theTelethon Kids Institute the Womenrsquos and Infantrsquos ResearchFoundation (King Edward Memorial Hospital) MurdochUniversity The University of Notre Dame (Australia) and EdithCowan University The authors gratefully acknowledge the as-sistance of the Western Australian DNA Bank (National Healthand Medical Research Council of Australia National EnablingFacility) We would also like to acknowledge the Raine Studyparticipants for their ongoing participation in the study and theRaine Study Team for study coordination and data collectionThis work was supported by resources provided by the PawseySupercomputing Centre with funding from the AustralianGovernment and Government of Western Australia

12 | Human Molecular Genetics 2018 Vol 00 No 00

Downloaded from httpsacademicoupcomhmgadvance-article-abstractdoi101093hmgddy2375040780by University of Bristol Library useron 20 July 2018

We are very grateful to the children and families who agreed toparticipate in the contributing studies without whom this re-search would not be possible We would like to acknowledgethe role of Mark McCarthy and the Early Growth Genetics con-sortium in recruiting studies which contributed to this analysisFor ALSPAC we are extremely grateful to all the families whotook part in this study the midwives for their help in recruitingthem and the whole ALSPAC team which includes interviewerscomputer and laboratory technicians clerical workers researchscientists volunteers managers receptionists and nursesThe authors are grateful to the Raine Study participants and theirfamilies and to the Raine Study research staff for cohort coordina-tion and data collection The authors gratefully acknowledge theassistance of the Western Australian DNA Bank (National Healthand Medical Research Council of Australia National EnablingFacility) We would also like to acknowledge the Raine Study par-ticipants for their ongoing participation in the study and theRaine Study Team for study coordination and data collectionThe Generation R Study is conducted by the Erasmus MedicalCenter in close collaboration with the School of Law and Facultyof Social Sciences of the Erasmus University Rotterdam theMunicipal Health Service Rotterdam area Rotterdam theRotterdam Homecare Foundation Rotterdam and the StichtingTrombosedienst amp Artsenlaboratorium Rijnmond (STAR-MDC)Rotterdam We acknowledge the contribution of children andparents general practitioners hospitals midwives and phar-macies in Rotterdam The generation and management ofGWAS genotype data for the Generation R Study was done atthe Genetic Laboratory of the Department of Internal MedicineErasmus MC The Netherlands We thank Pascal Arp MilaJhamai Marijn Verkerk Manoushka Ganesh Lizbeth Herreraand Marjolein Peters for their help in creating managing andQC of the GWAS database The general design of Generation RStudy was made possible by financial support from the ErasmusMedical Center Rotterdam the Erasmus University Rotterdamthe Netherlands Organization for Health Research andDevelopment (ZonMw) the Netherlands Organisation forScientific Research (NWO) the Ministry of Health Welfare andSport and the Ministry of Youth and Families Additionally theNetherlands Organization for Health Research andDevelopment supported the Generation R Study (ZonMw90700303 ZonMw 91610159 ZonMw VIDI 016136361 andZonMw VIDI 016136367) to FR and CM-G of this manuscriptThis project also received funding from the European UnionrsquosHorizon 2020 research and innovation programme under thefollowing grant agreements [No 633595 (DynaHEALTH) and No733206 (LIFECYCLE)] Furthermore Generation R received addi-tional funding from the European Research Council (ERCConsolidator Grant ERC-2014-CoG-648916)

Conflict of Interest statement None declared

FundingFunding to pay the Open Access publication charges for this ar-ticle was provided by the Medical Research Council IntegrativeEpidemiology Unit at the University of Bristol which is sup-ported by the Medical Research Council and University ofBristol

References1 Kassebaum NJ Bernabe E Dahiya M Bhandari B

Murray CJL and Marcenes W (2015) Global burden of

untreated caries a systematic review and metaregressionJ Dent Res 94 650ndash658

2 Vernazza CR Rolland SL Chadwick B and Pitts N (2016)Caries experience the caries burden and associated factorsin children in England Wales and Northern Ireland 2013 BrDent J 221 315ndash320

3 van der Tas JT Kragt L Elfrink MEC Bertens LCMJaddoe VWV Moll HA Ongkosuwito EM and WolviusEB (2017) Social inequalities and dental caries insix-year-old children from the Netherlands J Dentistry 6218ndash24

4 Schuller AA van Dommelen P and Poorterman JHG(2014) Trends in oral health in young people in theNetherlands over the past 20 years a study in a changingcontext Commun Dent Oral Epidemiol 42 178ndash184

5 Philip N Suneja B and Walsh LJ (2018) Ecologicalapproaches to dental caries prevention paradigm shift orshibboleth Caries Res 52 153ndash165

6 Peres MA Sheiham A Liu P Demarco FF Silva AEAssunc~ao MC Menezes AM Barros FC and Peres KG(2016) Sugar consumption and changes in dental caries fromchildhood to adolescence J Dent Res 95 388ndash394

7 Sheiham A and James WP (2015) Diet and dental cariesthe pivotal role of free sugars reemphasized J Dent Res 941341ndash1347

8 Lynch RJ (2013) The primary and mixed dentitionpost-eruptive enamel maturation and dental caries a re-view Int Dent J 63 3ndash13

9 Boraas JC Messer LB and Till MJ (1988) A genetic contri-bution to dental caries occlusion and morphology as dem-onstrated by twins reared apart J Dent Res 67 1150ndash1155

10 Bretz WA Corby PM Schork NJ Robinson MT CoelhoM Costa S Melo Filho MR Weyant RJ and Hart TC(2005) Longitudinal analysis of heritability for dental cariestraits J Dent Res 84 1047ndash1051

11 Chung CS and Larson RH (1967) Factors and inheritanceof dental caries in the rat J Dent Res 46 559ndash564

12 Bretz WA Corby PMA Melo MR Coelho MQ CostaSM Robinson M Schork NJ Drewnowski A and HartTC (2006) Heritability estimates for dental caries and su-crose sweetness preference Arch Oral Biol 51 1156ndash1160

13 Shaffer JR Wang X Feingold E Lee M Begum FWeeks D Cuenco KT Barmada MM Wendell SKCrosslin DR et al (2011) Genome-wide association scan forchildhood caries implicates novel genes J Dent Res 901457ndash1462

14 Wang X Shaffer JR Zeng Z Begum F Vieira AR NoelJ Anjomshoaa I Cuenco KT Lee MK Beck J et al (2012)Genome-wide association scan of dental caries in the per-manent dentition BMC Oral Health 12 57

15 Zeng Z Feingold E Wang X Weeks DE Lee M CuencoDT Broffitt B Weyant RJ Crout R McNeil DW et al(2014) Genome-wide association study of primary dentitionpit-and-fissure and smooth surface caries Caries Res 48330ndash338

16 Zeng Z Shaffer JR Wang X Feingold E Weeks DE LeeM Cuenco KT Wendell SK Weyant RJ Crout R et al(2013) Genome-wide association studies of pit-and-fissure-and smooth-surface caries in permanent dentition J DentRes 92 432ndash437

17 Morrison J Laurie CC Marazita ML Sanders AEOffenbacher S Salazar CR Conomos MP Thornton TJain D Laurie CA et al (2016) Genome-wide associationstudy of dental caries in the Hispanic Communities Health

13Human Molecular Genetics 2018 Vol 00 No 00 |

Downloaded from httpsacademicoupcomhmgadvance-article-abstractdoi101093hmgddy2375040780by University of Bristol Library useron 20 July 2018

StudyStudy of Latinos (HCHSSOL) Hum Mol Genet 25807ndash816

18 Mak GWY Lai WL Zhou Y Li MT Ng IOL and ChingYP (2012) CDK5RAP3 is a novel repressor of p14(ARF) in he-patocellular carcinoma cells PLoS One 7 e42210

19 Mak GWY Chan MML Leong VYL Lee JMF YauTO Ng IOL and Ching YP (2011) Overexpression of anovel activator of PAK4 the CDK5 kinase-associated proteinCDK5RAP3 promotes hepatocellular carcinoma metastasisCancer Res 71 2949ndash2958

20 Vigetti D Pollegioni L Monetti C Prati M Bernardini G andGornati R (2002) Property comparison of recombinant amphib-ian and mammalian allantoicases FEBS Lett 512 323ndash328

21 Park TJ Park JS Cheong HS Park BL Kim LH HeoJS Kim YK Kim KU Uh ST Lee HS et al (2014)Genome-wide association study identifies ALLC polymor-phisms correlated with FEV1 change by corticosteroid ClinChim Acta 436 20ndash26

22 Tikhmyanova N Little JL and Golemis EA (2010) CASproteins in normal and pathological cell growth control CellMol Life Sci 67 1025ndash1048

23 Singh MK Dadke D Nicolas E Serebriiskii IGApostolou S Canutescu A Egleston BL and Golemis EA(2008) A novel Cas family member HEPL regulates FAK andcell spreading Mol Biol Cell 19 1627ndash1636

24 Kumar S Tomooka Y and Noda M (1992) Identification ofa set of genes with developmentally down-regulated expres-sion in the mouse-brain Biochem Biophys Res Commun 1851155ndash1161

25 Latasa MJ Jimenez-Lara AM and Cosgaya JM (2016)Retinoic acid regulates Schwann cell migration via NEDD9induction by transcriptional and post-translational mecha-nisms Biochim Biophys Acta Mol Cell Res 1863 1510ndash1518

26 Aquino JB Lallemend F Marmigere F Adameyko IIGolemis EA and Ernfors P (2009) The retinoic acid induc-ible Cas-family signaling protein Nedd9 regulates neuralcrest cell migration by modulating adhesion and actin dy-namics Neuroscience 162 1106ndash1119

27 Nikonova AS Gaponova AV Kudinov AE and GolemisEA (2014) CAS proteins in health and disease an updateIUBMB Life 66 387ndash395

28 Riccomagno MM Sun LO Brady CM AlexandropoulosK Seo S Kurokawa M and Kolodkin AL (2014) Cas adap-tor proteins organize the retinal ganglion cell layer down-stream of integrin signaling Neuron 81 779ndash786

29 Wang SK Komatsu Y and Mishina Y (2011) Potential con-tribution of neural crest cells to dental enamel formationBiochem Biophys Res Commun 415 114ndash119

30 Duverger O Zah A Isaac J Sun HW Bartels AK LianJB Berdal A Hwang J and Morasso MI (2012) Neuralcrest deletion of Dlx3 leads to major dentin defects throughdown-regulation of Dspp J Biol Chem 287 12230ndash12240

31 Divaris K (2016) Predicting dental caries outcomes in chil-dren a ldquoriskyrdquo concept J Dent Res 95 248ndash254

32 Chapple ILC Bouchard P Cagetti MG Campus GCarra M-C Cocco F Nibali L Hujoel P Laine MLLingstrom P et al (2017) Interaction of lifestyle behaviouror systemic diseases with dental caries and periodontal dis-eases consensus report of group 2 of the joint EFPORCAworkshop on the boundaries between caries and periodontaldiseases J Clin Periodontol 44 S39ndashS51

33 Nibali L Di Iorio A Tu Y-K and Vieira AR (2017) Host ge-netics role in the pathogenesis of periodontal disease andcaries J Clin Periodontol 44 S52ndashS78

34 Wang X Shaffer JR Weyant RJ Cuenco KT DeSensiRS Crout R McNeil DW and Marazita ML (2010) Genesand their effects on dental caries may differ between pri-mary and permanent dentitions Caries Res 44 201ndash284

35 Docherty AR Moscati A Peterson R Edwards ACAdkins DE Bacanu SA Bigdeli TB Webb BT Flint Jand Kendler KS (2016) SNP-based heritability estimates ofthe personality dimensions and polygenic prediction of bothneuroticism and major depression findings fromCONVERGE Transl Psychiatry 6 e926

36 Shaffer JR Wang XJ McNeil DW Weyant RJ Crout Rand Marazita ML (2015) Genetic susceptibility to dental car-ies differs between the sexes a family-based study CariesRes 49 133ndash140

37 Bayram M Deeley K Reis MF Trombetta VM RuffTD Sencak RC Hummel M Dizak PM Washam KRomanos HF et al (2015) Genetic influences on dentalenamel that impact caries differ between the primary andpermanent dentitions Eur J Oral Sci 123 327ndash334

38 Shaffer JR Carlson JC Stanley BOC Feingold ECooper M Vanyukov MM Maher BS Slayton RLWilling MC Reis SE et al (2015) Effects of enamel matrixgenes on dental caries are moderated by fluoride exposuresHum Genet 134 159ndash167

39 Johansson I Witkowska E Kaveh B Holgerson PL andTanner ACR (2016) The microbiome in populations with alow and high prevalence of caries J Dent Res 95 80ndash86

40 von Hippel PT (2015) The heterogeneity statistic I(2) can bebiased in small meta-analyses BMC Med Res Methodol 1535

41 Haworth S Dudding T Waylen A Thomas SJ andTimpson NJ (2017) Ten years on is dental general anaes-thesia in childhood a risk factor for caries and anxiety BrDent J 222 299ndash304

42 Kemp JP Morris JA Medina-Gomez C Forgetta VWarrington NM Youlten SE Zheng J Gregson CLGrundberg E Trajanoska K et al (2017) Identification of 153new loci associated with heel bone mineral density andfunctional involvement of GPC6 in osteoporosis Nat Genet49 1468

43 Boyd A Golding J Macleod J Lawlor DA Fraser AHenderson J Molloy L Ness A Ring S and Smith GD(2013) Cohort profile the lsquochildren of the 90srsquo-the index off-spring of the Avon Longitudinal Study of Parents andChildren Int J Epidemiol 42 111ndash127

44 Bisgaard H (2004) The Copenhagen Prospective Study onAsthma in Childhood (COPSAC) design rationale and base-line data from a longitudinal birth cohort study Ann AllergyAsthma Immunol 93 381ndash389

45 Olsen J Melbye M Olsen SF Sorensen TIA Aaby PAndersen AMN Taxbol D Hansen KD Juhl M SchowTB et al (2001) The Danish National Birth Cohort - its back-ground structure and aim Scand J Public Health 29300ndash307

46 Kooijman MN Kruithof CJ van Duijn CM Duijts LFranco OH van Ijzendoorn MH de Jongste JC KlaverCCW van der Lugt A Mackenbach JP et al (2016) TheGeneration R Study design and cohort update 2017 Eur JEpidemiol 31 1243ndash1264

47 van der Tas JT Kragt L Veerkamp JJS Jaddoe VWVMoll HA Ongkosuwito EM Elfrink MEC and WolviusEB (2016) Ethnic disparities in dental caries amongsix-year-old children in the Netherlands Caries Res 50489ndash497

14 | Human Molecular Genetics 2018 Vol 00 No 00

Downloaded from httpsacademicoupcomhmgadvance-article-abstractdoi101093hmgddy2375040780by University of Bristol Library useron 20 July 2018

48 Medina-Gomez C Felix JF Estrada K Peters MJHerrera L Kruithof CJ Duijts L Hofman A van DuijnCM Uitterlinden AG et al (2015) Challenges in conductinggenome-wide association studies in highly admixedmulti-ethnic populations the Generation R Study Eur JEpidemiol 30 317ndash330

49 Cornelis MC Agrawal A Cole JW Hansel NN BarnesKC Beaty TH Bennett SN Bierut LJ Boerwinkle EDoheny KF et al (2010) The gene environment associationstudies consortium (GENEVA) maximizing the knowledgeobtained from GWAS by collaboration across studies of mul-tiple conditions Genet Epidemiol 34 364ndash372

50 Polk DE Weyant RJ Crout RJ McNeil DW Tarter REThomas JG and Marazita ML (2008) Study protocol of theCenter for Oral Health Research in Appalachia (COHRA) eti-ology study BMC Oral Health 8 18

51 Levy SM Warren JJ Broffitt B Hillis SL and KanellisMJ (2003) Fluoride beverages and dental caries in the pri-mary dentition Caries Res 37 157ndash165

52 Taal HR St Pourcain B Thiering E Das S Mook-Kanamori DO Warrington NM Kaakinen M Kreiner-Moller E Bradfield JP Freathy RM et al (2012) Commonvariants at 12q15 and 12q24 are associated with infant headcircumference Nat Genet 44 532

53 von Berg A Kramer U Link E Bollrath C Heinrich JBrockow I Koletzko S Grubl A Filipiak-Pittroff BWichmann HE et al (2010) Impact of early feeding on child-hood eczema development after nutritional interventioncompared with the natural course - the GINIplus study up tothe age of 6 years Clin Exp Allergy 40 627ndash636

54 Zutavern A Brockow I Schaaf B Bolte G von BergA Diez U Borte M Herbarth O Wichmann HEHeinrich J et al (2006) Timing of solid food introductionin relation to atopic dermatitis and atopic sensitizationresults from a prospective birth cohort study Pediatrics117 401ndash411

55 Eloranta AM Lindi V Schwab U Tompuri T KiiskinenS Lakka HM Laitinen T and Lakka TA (2012) Dietaryfactors associated with overweight and body adiposity inFinnish children aged 6-8 years the PANIC Study Int JObesity 36 950ndash955

56 Raitakari OT Juonala M Ronnemaa T Keltikangas-Jarvinen L Rasanen L Pietikainen M Hutri-Kahonen NTaittonen L Jokinen E Marniemi J et al (2008) Cohort pro-file the cardiovascular risk in young Finns study Int JEpidemiol 37 1220ndash1226

57 Straker L Mountain J Jacques A White S Smith ALandau L Stanley F Newnham J Pennell C andEastwood P (2017) Cohort profile the Western AustralianPregnancy Cohort (RAINE) Study-generation 2 Int JEpidemiol 5 1384ndash1385

58 Winkler TW Day FR Croteau-Chonka DC Wood ARLocke AE Magi R Ferreira T Fall T Graff M Justice

AE et al (2014) Quality control and conduct ofgenome-wide association meta-analyses Nat Protoc 91192ndash1212

59 Willer CJ Li Y and Abecasis GR (2010) METAL fast andefficient meta-analysis of genomewide association scansBioinformatics 26 2190ndash2191

60 Bulik-Sullivan BK Loh P-R Finucane HK Ripke SYang J Patterson N Daly MJ Price AL Neale BM andSchizophrenia Working Group (2015) LD score regressiondistinguishes confounding from polygenicity ingenome-wide association studies Nat Genet 47 291ndash291thorn

61 Yang JA Benyamin B McEvoy BP Gordon S HendersAK Nyholt DR Madden PA Heath AC Martin NGMontgomery GW et al (2010) Common SNPs explain a largeproportion of the heritability for human height Nat Genet42 565ndashU131

62 Yang JA Lee SH Goddard ME and Visscher PM (2011)GCTA a tool for genome-wide complex trait analysis Am JHum Genet 88 76ndash82

63 Chang CC Chow CC Tellier LCAM Vattikuti S PurcellSM and Lee JJ (2015) Second-generation PLINK rising to thechallenge of larger and richer datasets Gigascience 4 7

64 Hemani G Zheng J Wade KH Laurin C Elsworth BBurgess S Bowden J Langdon R Tan V Yarmolinsky Jet al (2016) MR-Base a platform for systematic causal infer-ence across the phenome using billions of genetic associa-tions eLife doi107554eLife34408

65 Zheng J Erzurumluoglu AM Elsworth BL Kemp JPHowe L Haycock PC Hemani G Tansey K Laurin C StPourcain B et al (2017) LD Hub a centralized database andweb interface to perform LD score regression that maxi-mizes the potential of summary level GWAS data for SNPheritability and genetic correlation analysis Bioinformatics33 272ndash279

66 de Leeuw CA Mooij JM Heskes T and Posthuma D(2015) MAGMA generalized gene-set analysis of GWAS dataPLoS Comput Biol 11 e1004219

67 Pers TH Karjalainen JM Chan Y Westra HJ Wood ARYang J Lui JC Vedantam S Gustafsson S Esko T et al(2015) Biological interpretation of genome-wide associationstudies using predicted gene functions Nat Commun 6 9

68 Barbeira AN Dickinson SP Torres JM Bonazzola RZheng J Torstenson ES Wheeler HE Shah KPEdwards T Garcia T et al (2017) Exploring the pheno-typic consequences of tissue specific gene expression var-iation inferred from GWAS summary statistics NatCommun doi101038s41467-018-03621-1

69 Skol A Scott L Abecasis G and Boehnke M (2006) Jointanalysis is more efficient than replication-based analysis fortwo-stage genome-wide association studies Nat Genet 38209ndash213

15Human Molecular Genetics 2018 Vol 00 No 00 |

Downloaded from httpsacademicoupcomhmgadvance-article-abstractdoi101093hmgddy2375040780by University of Bristol Library useron 20 July 2018

• ddy237-TF1
• ddy237-TF2

StudyStudy of Latinos (HCHSSOL) Hum Mol Genet 25807ndash816

18 Mak GWY Lai WL Zhou Y Li MT Ng IOL and ChingYP (2012) CDK5RAP3 is a novel repressor of p14(ARF) in he-patocellular carcinoma cells PLoS One 7 e42210

19 Mak GWY Chan MML Leong VYL Lee JMF YauTO Ng IOL and Ching YP (2011) Overexpression of anovel activator of PAK4 the CDK5 kinase-associated proteinCDK5RAP3 promotes hepatocellular carcinoma metastasisCancer Res 71 2949ndash2958

20 Vigetti D Pollegioni L Monetti C Prati M Bernardini G andGornati R (2002) Property comparison of recombinant amphib-ian and mammalian allantoicases FEBS Lett 512 323ndash328

21 Park TJ Park JS Cheong HS Park BL Kim LH HeoJS Kim YK Kim KU Uh ST Lee HS et al (2014)Genome-wide association study identifies ALLC polymor-phisms correlated with FEV1 change by corticosteroid ClinChim Acta 436 20ndash26

22 Tikhmyanova N Little JL and Golemis EA (2010) CASproteins in normal and pathological cell growth control CellMol Life Sci 67 1025ndash1048

23 Singh MK Dadke D Nicolas E Serebriiskii IGApostolou S Canutescu A Egleston BL and Golemis EA(2008) A novel Cas family member HEPL regulates FAK andcell spreading Mol Biol Cell 19 1627ndash1636

24 Kumar S Tomooka Y and Noda M (1992) Identification ofa set of genes with developmentally down-regulated expres-sion in the mouse-brain Biochem Biophys Res Commun 1851155ndash1161

25 Latasa MJ Jimenez-Lara AM and Cosgaya JM (2016)Retinoic acid regulates Schwann cell migration via NEDD9induction by transcriptional and post-translational mecha-nisms Biochim Biophys Acta Mol Cell Res 1863 1510ndash1518

26 Aquino JB Lallemend F Marmigere F Adameyko IIGolemis EA and Ernfors P (2009) The retinoic acid induc-ible Cas-family signaling protein Nedd9 regulates neuralcrest cell migration by modulating adhesion and actin dy-namics Neuroscience 162 1106ndash1119

27 Nikonova AS Gaponova AV Kudinov AE and GolemisEA (2014) CAS proteins in health and disease an updateIUBMB Life 66 387ndash395

28 Riccomagno MM Sun LO Brady CM AlexandropoulosK Seo S Kurokawa M and Kolodkin AL (2014) Cas adap-tor proteins organize the retinal ganglion cell layer down-stream of integrin signaling Neuron 81 779ndash786

29 Wang SK Komatsu Y and Mishina Y (2011) Potential con-tribution of neural crest cells to dental enamel formationBiochem Biophys Res Commun 415 114ndash119

30 Duverger O Zah A Isaac J Sun HW Bartels AK LianJB Berdal A Hwang J and Morasso MI (2012) Neuralcrest deletion of Dlx3 leads to major dentin defects throughdown-regulation of Dspp J Biol Chem 287 12230ndash12240

31 Divaris K (2016) Predicting dental caries outcomes in chil-dren a ldquoriskyrdquo concept J Dent Res 95 248ndash254

32 Chapple ILC Bouchard P Cagetti MG Campus GCarra M-C Cocco F Nibali L Hujoel P Laine MLLingstrom P et al (2017) Interaction of lifestyle behaviouror systemic diseases with dental caries and periodontal dis-eases consensus report of group 2 of the joint EFPORCAworkshop on the boundaries between caries and periodontaldiseases J Clin Periodontol 44 S39ndashS51

33 Nibali L Di Iorio A Tu Y-K and Vieira AR (2017) Host ge-netics role in the pathogenesis of periodontal disease andcaries J Clin Periodontol 44 S52ndashS78

34 Wang X Shaffer JR Weyant RJ Cuenco KT DeSensiRS Crout R McNeil DW and Marazita ML (2010) Genesand their effects on dental caries may differ between pri-mary and permanent dentitions Caries Res 44 201ndash284

35 Docherty AR Moscati A Peterson R Edwards ACAdkins DE Bacanu SA Bigdeli TB Webb BT Flint Jand Kendler KS (2016) SNP-based heritability estimates ofthe personality dimensions and polygenic prediction of bothneuroticism and major depression findings fromCONVERGE Transl Psychiatry 6 e926

36 Shaffer JR Wang XJ McNeil DW Weyant RJ Crout Rand Marazita ML (2015) Genetic susceptibility to dental car-ies differs between the sexes a family-based study CariesRes 49 133ndash140

37 Bayram M Deeley K Reis MF Trombetta VM RuffTD Sencak RC Hummel M Dizak PM Washam KRomanos HF et al (2015) Genetic influences on dentalenamel that impact caries differ between the primary andpermanent dentitions Eur J Oral Sci 123 327ndash334

38 Shaffer JR Carlson JC Stanley BOC Feingold ECooper M Vanyukov MM Maher BS Slayton RLWilling MC Reis SE et al (2015) Effects of enamel matrixgenes on dental caries are moderated by fluoride exposuresHum Genet 134 159ndash167

39 Johansson I Witkowska E Kaveh B Holgerson PL andTanner ACR (2016) The microbiome in populations with alow and high prevalence of caries J Dent Res 95 80ndash86

40 von Hippel PT (2015) The heterogeneity statistic I(2) can bebiased in small meta-analyses BMC Med Res Methodol 1535

41 Haworth S Dudding T Waylen A Thomas SJ andTimpson NJ (2017) Ten years on is dental general anaes-thesia in childhood a risk factor for caries and anxiety BrDent J 222 299ndash304

42 Kemp JP Morris JA Medina-Gomez C Forgetta VWarrington NM Youlten SE Zheng J Gregson CLGrundberg E Trajanoska K et al (2017) Identification of 153new loci associated with heel bone mineral density andfunctional involvement of GPC6 in osteoporosis Nat Genet49 1468

43 Boyd A Golding J Macleod J Lawlor DA Fraser AHenderson J Molloy L Ness A Ring S and Smith GD(2013) Cohort profile the lsquochildren of the 90srsquo-the index off-spring of the Avon Longitudinal Study of Parents andChildren Int J Epidemiol 42 111ndash127

44 Bisgaard H (2004) The Copenhagen Prospective Study onAsthma in Childhood (COPSAC) design rationale and base-line data from a longitudinal birth cohort study Ann AllergyAsthma Immunol 93 381ndash389

45 Olsen J Melbye M Olsen SF Sorensen TIA Aaby PAndersen AMN Taxbol D Hansen KD Juhl M SchowTB et al (2001) The Danish National Birth Cohort - its back-ground structure and aim Scand J Public Health 29300ndash307

46 Kooijman MN Kruithof CJ van Duijn CM Duijts LFranco OH van Ijzendoorn MH de Jongste JC KlaverCCW van der Lugt A Mackenbach JP et al (2016) TheGeneration R Study design and cohort update 2017 Eur JEpidemiol 31 1243ndash1264

47 van der Tas JT Kragt L Veerkamp JJS Jaddoe VWVMoll HA Ongkosuwito EM Elfrink MEC and WolviusEB (2016) Ethnic disparities in dental caries amongsix-year-old children in the Netherlands Caries Res 50489ndash497

14 | Human Molecular Genetics 2018 Vol 00 No 00

Downloaded from httpsacademicoupcomhmgadvance-article-abstractdoi101093hmgddy2375040780by University of Bristol Library useron 20 July 2018

48 Medina-Gomez C Felix JF Estrada K Peters MJHerrera L Kruithof CJ Duijts L Hofman A van DuijnCM Uitterlinden AG et al (2015) Challenges in conductinggenome-wide association studies in highly admixedmulti-ethnic populations the Generation R Study Eur JEpidemiol 30 317ndash330

49 Cornelis MC Agrawal A Cole JW Hansel NN BarnesKC Beaty TH Bennett SN Bierut LJ Boerwinkle EDoheny KF et al (2010) The gene environment associationstudies consortium (GENEVA) maximizing the knowledgeobtained from GWAS by collaboration across studies of mul-tiple conditions Genet Epidemiol 34 364ndash372

50 Polk DE Weyant RJ Crout RJ McNeil DW Tarter REThomas JG and Marazita ML (2008) Study protocol of theCenter for Oral Health Research in Appalachia (COHRA) eti-ology study BMC Oral Health 8 18

51 Levy SM Warren JJ Broffitt B Hillis SL and KanellisMJ (2003) Fluoride beverages and dental caries in the pri-mary dentition Caries Res 37 157ndash165

52 Taal HR St Pourcain B Thiering E Das S Mook-Kanamori DO Warrington NM Kaakinen M Kreiner-Moller E Bradfield JP Freathy RM et al (2012) Commonvariants at 12q15 and 12q24 are associated with infant headcircumference Nat Genet 44 532

53 von Berg A Kramer U Link E Bollrath C Heinrich JBrockow I Koletzko S Grubl A Filipiak-Pittroff BWichmann HE et al (2010) Impact of early feeding on child-hood eczema development after nutritional interventioncompared with the natural course - the GINIplus study up tothe age of 6 years Clin Exp Allergy 40 627ndash636

54 Zutavern A Brockow I Schaaf B Bolte G von BergA Diez U Borte M Herbarth O Wichmann HEHeinrich J et al (2006) Timing of solid food introductionin relation to atopic dermatitis and atopic sensitizationresults from a prospective birth cohort study Pediatrics117 401ndash411

55 Eloranta AM Lindi V Schwab U Tompuri T KiiskinenS Lakka HM Laitinen T and Lakka TA (2012) Dietaryfactors associated with overweight and body adiposity inFinnish children aged 6-8 years the PANIC Study Int JObesity 36 950ndash955

56 Raitakari OT Juonala M Ronnemaa T Keltikangas-Jarvinen L Rasanen L Pietikainen M Hutri-Kahonen NTaittonen L Jokinen E Marniemi J et al (2008) Cohort pro-file the cardiovascular risk in young Finns study Int JEpidemiol 37 1220ndash1226

57 Straker L Mountain J Jacques A White S Smith ALandau L Stanley F Newnham J Pennell C andEastwood P (2017) Cohort profile the Western AustralianPregnancy Cohort (RAINE) Study-generation 2 Int JEpidemiol 5 1384ndash1385

58 Winkler TW Day FR Croteau-Chonka DC Wood ARLocke AE Magi R Ferreira T Fall T Graff M Justice

AE et al (2014) Quality control and conduct ofgenome-wide association meta-analyses Nat Protoc 91192ndash1212

59 Willer CJ Li Y and Abecasis GR (2010) METAL fast andefficient meta-analysis of genomewide association scansBioinformatics 26 2190ndash2191

60 Bulik-Sullivan BK Loh P-R Finucane HK Ripke SYang J Patterson N Daly MJ Price AL Neale BM andSchizophrenia Working Group (2015) LD score regressiondistinguishes confounding from polygenicity ingenome-wide association studies Nat Genet 47 291ndash291thorn

61 Yang JA Benyamin B McEvoy BP Gordon S HendersAK Nyholt DR Madden PA Heath AC Martin NGMontgomery GW et al (2010) Common SNPs explain a largeproportion of the heritability for human height Nat Genet42 565ndashU131

62 Yang JA Lee SH Goddard ME and Visscher PM (2011)GCTA a tool for genome-wide complex trait analysis Am JHum Genet 88 76ndash82

63 Chang CC Chow CC Tellier LCAM Vattikuti S PurcellSM and Lee JJ (2015) Second-generation PLINK rising to thechallenge of larger and richer datasets Gigascience 4 7

64 Hemani G Zheng J Wade KH Laurin C Elsworth BBurgess S Bowden J Langdon R Tan V Yarmolinsky Jet al (2016) MR-Base a platform for systematic causal infer-ence across the phenome using billions of genetic associa-tions eLife doi107554eLife34408

65 Zheng J Erzurumluoglu AM Elsworth BL Kemp JPHowe L Haycock PC Hemani G Tansey K Laurin C StPourcain B et al (2017) LD Hub a centralized database andweb interface to perform LD score regression that maxi-mizes the potential of summary level GWAS data for SNPheritability and genetic correlation analysis Bioinformatics33 272ndash279

66 de Leeuw CA Mooij JM Heskes T and Posthuma D(2015) MAGMA generalized gene-set analysis of GWAS dataPLoS Comput Biol 11 e1004219

67 Pers TH Karjalainen JM Chan Y Westra HJ Wood ARYang J Lui JC Vedantam S Gustafsson S Esko T et al(2015) Biological interpretation of genome-wide associationstudies using predicted gene functions Nat Commun 6 9

68 Barbeira AN Dickinson SP Torres JM Bonazzola RZheng J Torstenson ES Wheeler HE Shah KPEdwards T Garcia T et al (2017) Exploring the pheno-typic consequences of tissue specific gene expression var-iation inferred from GWAS summary statistics NatCommun doi101038s41467-018-03621-1

69 Skol A Scott L Abecasis G and Boehnke M (2006) Jointanalysis is more efficient than replication-based analysis fortwo-stage genome-wide association studies Nat Genet 38209ndash213

15Human Molecular Genetics 2018 Vol 00 No 00 |

Downloaded from httpsacademicoupcomhmgadvance-article-abstractdoi101093hmgddy2375040780by University of Bristol Library useron 20 July 2018

• ddy237-TF1
• ddy237-TF2

48 Medina-Gomez C Felix JF Estrada K Peters MJHerrera L Kruithof CJ Duijts L Hofman A van DuijnCM Uitterlinden AG et al (2015) Challenges in conductinggenome-wide association studies in highly admixedmulti-ethnic populations the Generation R Study Eur JEpidemiol 30 317ndash330

49 Cornelis MC Agrawal A Cole JW Hansel NN BarnesKC Beaty TH Bennett SN Bierut LJ Boerwinkle EDoheny KF et al (2010) The gene environment associationstudies consortium (GENEVA) maximizing the knowledgeobtained from GWAS by collaboration across studies of mul-tiple conditions Genet Epidemiol 34 364ndash372

50 Polk DE Weyant RJ Crout RJ McNeil DW Tarter REThomas JG and Marazita ML (2008) Study protocol of theCenter for Oral Health Research in Appalachia (COHRA) eti-ology study BMC Oral Health 8 18

51 Levy SM Warren JJ Broffitt B Hillis SL and KanellisMJ (2003) Fluoride beverages and dental caries in the pri-mary dentition Caries Res 37 157ndash165

52 Taal HR St Pourcain B Thiering E Das S Mook-Kanamori DO Warrington NM Kaakinen M Kreiner-Moller E Bradfield JP Freathy RM et al (2012) Commonvariants at 12q15 and 12q24 are associated with infant headcircumference Nat Genet 44 532

53 von Berg A Kramer U Link E Bollrath C Heinrich JBrockow I Koletzko S Grubl A Filipiak-Pittroff BWichmann HE et al (2010) Impact of early feeding on child-hood eczema development after nutritional interventioncompared with the natural course - the GINIplus study up tothe age of 6 years Clin Exp Allergy 40 627ndash636

54 Zutavern A Brockow I Schaaf B Bolte G von BergA Diez U Borte M Herbarth O Wichmann HEHeinrich J et al (2006) Timing of solid food introductionin relation to atopic dermatitis and atopic sensitizationresults from a prospective birth cohort study Pediatrics117 401ndash411

55 Eloranta AM Lindi V Schwab U Tompuri T KiiskinenS Lakka HM Laitinen T and Lakka TA (2012) Dietaryfactors associated with overweight and body adiposity inFinnish children aged 6-8 years the PANIC Study Int JObesity 36 950ndash955

56 Raitakari OT Juonala M Ronnemaa T Keltikangas-Jarvinen L Rasanen L Pietikainen M Hutri-Kahonen NTaittonen L Jokinen E Marniemi J et al (2008) Cohort pro-file the cardiovascular risk in young Finns study Int JEpidemiol 37 1220ndash1226

57 Straker L Mountain J Jacques A White S Smith ALandau L Stanley F Newnham J Pennell C andEastwood P (2017) Cohort profile the Western AustralianPregnancy Cohort (RAINE) Study-generation 2 Int JEpidemiol 5 1384ndash1385

58 Winkler TW Day FR Croteau-Chonka DC Wood ARLocke AE Magi R Ferreira T Fall T Graff M Justice

AE et al (2014) Quality control and conduct ofgenome-wide association meta-analyses Nat Protoc 91192ndash1212

59 Willer CJ Li Y and Abecasis GR (2010) METAL fast andefficient meta-analysis of genomewide association scansBioinformatics 26 2190ndash2191

60 Bulik-Sullivan BK Loh P-R Finucane HK Ripke SYang J Patterson N Daly MJ Price AL Neale BM andSchizophrenia Working Group (2015) LD score regressiondistinguishes confounding from polygenicity ingenome-wide association studies Nat Genet 47 291ndash291thorn

61 Yang JA Benyamin B McEvoy BP Gordon S HendersAK Nyholt DR Madden PA Heath AC Martin NGMontgomery GW et al (2010) Common SNPs explain a largeproportion of the heritability for human height Nat Genet42 565ndashU131

62 Yang JA Lee SH Goddard ME and Visscher PM (2011)GCTA a tool for genome-wide complex trait analysis Am JHum Genet 88 76ndash82

63 Chang CC Chow CC Tellier LCAM Vattikuti S PurcellSM and Lee JJ (2015) Second-generation PLINK rising to thechallenge of larger and richer datasets Gigascience 4 7

64 Hemani G Zheng J Wade KH Laurin C Elsworth BBurgess S Bowden J Langdon R Tan V Yarmolinsky Jet al (2016) MR-Base a platform for systematic causal infer-ence across the phenome using billions of genetic associa-tions eLife doi107554eLife34408

65 Zheng J Erzurumluoglu AM Elsworth BL Kemp JPHowe L Haycock PC Hemani G Tansey K Laurin C StPourcain B et al (2017) LD Hub a centralized database andweb interface to perform LD score regression that maxi-mizes the potential of summary level GWAS data for SNPheritability and genetic correlation analysis Bioinformatics33 272ndash279

66 de Leeuw CA Mooij JM Heskes T and Posthuma D(2015) MAGMA generalized gene-set analysis of GWAS dataPLoS Comput Biol 11 e1004219

67 Pers TH Karjalainen JM Chan Y Westra HJ Wood ARYang J Lui JC Vedantam S Gustafsson S Esko T et al(2015) Biological interpretation of genome-wide associationstudies using predicted gene functions Nat Commun 6 9

68 Barbeira AN Dickinson SP Torres JM Bonazzola RZheng J Torstenson ES Wheeler HE Shah KPEdwards T Garcia T et al (2017) Exploring the pheno-typic consequences of tissue specific gene expression var-iation inferred from GWAS summary statistics NatCommun doi101038s41467-018-03621-1

69 Skol A Scott L Abecasis G and Boehnke M (2006) Jointanalysis is more efficient than replication-based analysis fortwo-stage genome-wide association studies Nat Genet 38209ndash213

15Human Molecular Genetics 2018 Vol 00 No 00 |

Downloaded from httpsacademicoupcomhmgadvance-article-abstractdoi101093hmgddy2375040780by University of Bristol Library useron 20 July 2018

• ddy237-TF1
• ddy237-TF2