- Record: found
- Abstract: found
- Article: found
Genomics of periodontal disease and tooth morbidity
review-article
Publication date (Electronic):
18 December 2019
Journal:
Periodontology 2000
Publisher:
John Wiley and Sons Inc.
Read this article at
There is no author summary for this article yet. Authors can add summaries to their articles on ScienceOpen to make them more accessible to a non-specialist audience.
Abstract
In this review we critically summarize the evidence base and the progress to date regarding the genomic basis of periodontal disease and tooth morbidity (ie, dental caries and tooth loss), and discuss future applications and research directions in the context of precision oral health and care. Evidence for these oral/dental traits from genome‐wide association studies first emerged less than a decade ago. Basic and translational research activities in this domain are now under way by multiple groups around the world. Key departure points in the oral health genomics discourse are: (a) some heritable variation exists for periodontal and dental diseases; (b) the environmental component (eg, social determinants of health and behavioral risk factors) has a major influence on the population distribution but probably interacts with factors of innate susceptibility at the person‐level; (c) sizeable, multi‐ethnic, well‐characterized samples or cohorts with high‐quality measures on oral health outcomes and genomics information are required to make decisive discoveries; (d) challenges remain in the measurement of oral health and disease, with current periodontitis and dental caries traits capturing only a part of the health‐disease continuum, and are little or not informed by the underlying biology; (e) the substantial individual heterogeneity that exists in the clinical presentation and lifetime trajectory of oral disease can be identified and leveraged in a precision medicine framework or, if unappreciated, can hamper translational efforts. In this review we discuss how composite or biologically informed traits may offer improvements over clinically defined ones for the genomic interrogation of oral diseases. We demonstrate the utility of the results of genome‐wide association studies for the development and testing of a genetic risk score for severe periodontitis. We conclude that exciting opportunities lie ahead for improvements in the oral health of individual patients and populations via advances in our understanding of the genomic basis of oral health and disease. The pace of new discoveries and their equitable translation to practice will largely depend on investments in the education and training of the oral health care workforce, basic and population research, and sustained collaborative efforts..
Related collections
Most cited references82
- Record: found
- Abstract: found
- Article: not found
Suppression of inflammation and acquired immunity by IL-37.
Charles Dinarello, Giulio Cavalli (2018)
IL-37 is a unique member of the IL-1 family of cytokines, which functions as a natural suppressor of inflammatory and immune responses. Immune and non-immune cells produce IL-37 precursor following pro-inflammatory stimuli. Following activating cleavage by caspase-1, mature IL-37 translocates to the nucleus, where it suppresses transcription of pro-inflammatory genes. Both precursor and mature IL-37 are also secreted in the extracellular space, where they bind IL-18Rα and recruit the IL-1R8 (formerly TIR8 or SIGIRR), which transduces anti-inflammatory signals by suppressing NF-kB and MAPK and by activating Mer-PTEN-DOK pathways. During inflammation, IL-37 restores the metabolism of the cell by reducing succinate, inhibiting mTOR, and activating AMPK. Transgenic mice expressing human IL-37 and wild type mice treated with recombinant human IL-37 are protected from several experimental models of inflammation, including endotoxin shock, colitis, lung and spinal cord injury, coronary artery disease, arthritis and inflammation-induced fatigue, while also exhibiting reduced adaptive immune responses. In humans, IL-37 likely functions to limit excessive inflammation: accordingly, IL-37 levels are abnormal in patients with inflammatory and autoimmune diseases. In this review, we provide an overview of the discovery and biology of IL-37, and discuss the potential for development of this cytokine as a therapeutic agent.
- Record: found
- Abstract: found
- Article: not found
A Genome-Wide Association Study Identifies Five Loci Influencing Facial Morphology in Europeans
Fan Liu, Fedde van der Lijn, Claudia Schurmann … (2012)
Introduction The morphogenesis and patterning of the face is one of the most complex events in mammalian embryogenesis. Signaling cascades initiated from both facial and neighboring tissues mediate transcriptional networks that act to direct fundamental cellular processes such as migration, proliferation, differentiation and controlled cell death. The complexity of human facial development is reflected in the high incidence of congenital craniofacial anomalies, and almost certainly underlies the vast spectrum of subtle variation that characterizes facial appearance in the human population. Facial appearance has a strong genetic component; monozygotic (MZ) twins look more similar than dizygotic (DZ) twins or unrelated individuals. The heritability of craniofacial morphology is as high as 0.8 in twins and families [1], [2], [3]. Some craniofacial traits, such as facial height and position of the lower jaw, appear to be more heritable than others [1], [2], [3]. The general morphology of craniofacial bones is largely genetically determined and partly attributable to environmental factors [4]–[11]. Although genes have been mapped for various rare craniofacial syndromes largely inherited in Mendelian form [12], the genetic basis of normal variation in human facial shape is still poorly understood. An appreciation of the genetic basis of facial shape variation has far reaching implications for understanding the etiology of facial pathologies, the origin of major sensory organ systems, and even the evolution of vertebrates [13], [14]. In addition, it is feasible to speculate that once the majority of genetic determinants of facial morphology are understood, predicting facial appearance from DNA found at a crime scene will become useful as investigative tool in forensic case work [15]. Some externally visible human characteristics, such as eye color [16]–[18] and hair color [19], can already be inferred from a DNA sample with practically useful accuracies. In a recent candidate gene study carried out in two independent European population samples, we investigated a potential association between risk alleles for non-syndromic cleft lip with or without cleft palate (NSCL/P) and nose width and facial width in the normal population [20]. Two NSCL/P associated single nucleotide polymorphisms (SNPs) showed association with different facial phenotypes in different populations. However, facial landmarks derived from 3-Dimensional (3D) magnetic resonance images (MRI) in one population and 2-Dimensional (2D) portrait images in the other population were not completely comparable, posing a challenge for combining phenotype data. In the present study, we focus on the MRI-based approach for capturing facial morphology since previous facial imaging studies by some of us have demonstrated that MRI-derived soft tissue landmarks represent a reliable data source [21], [22]. In geometric morphometrics, there are different ways to deal with the confounders of position and orientation of the landmark configurations, such as (1) superimposition [23], [24] that places the landmarks into a consensus reference frame; (2) deformation [25]–[27], where shape differences are described in terms of deformation fields of one object onto another; and (3) linear distances [28], [29], where Euclidean distances between landmarks instead of their coordinates are measured. Rationality and efficacy of these approaches have been reviewed and compared elsewhere [30]–[32]. We briefly compared these methods in the context of our genome-wide association study (GWAS) (see Methods section) and applied them when appropriate. We extracted facial landmarks from 3D head MRI in 5,388 individuals of European origin from Netherlands, Australia, and Germany, and used partial Procrustes superimposition (PS) [24], [30], [33] to superimpose different sets of facial landmarks onto a consensus 3D Euclidean space. We derived 48 facial shape features from the superimposed landmarks and estimated their heritability in 79 MZ and 90 DZ Australian twin pairs. Subsequently, we conducted a series of GWAS separately for these facial shape dimensions, and attempted to replicate the identified associations in 568 Canadians of European (French) ancestry with similar 3D head MRI phenotypes and additionally sought supporting evidence in further 1,530 individuals from the UK and 2,337 from Australia for whom facial phenotypes were derived from 2D portrait images. Results Characteristics of the study cohorts from The Netherlands (RS1, RS2), Australia (QTIMS, BLTS), Germany (SHIP, SHIP-TREND), Canada (SYS) and the United Kingdom (TwinsUK) are provided in Table 1. All participants included in this study were of European ancestry. Facial landmarks in the discovery cohorts RS1, RS2, QTIMS, SHIP, and SHIP-TREND were derived from directly comparable 3D head MRIs analyzed using the very same method (Figure 1). Similar 3D MRIs were available in SYS but the phenotyping method used here was slightly different (see Method). For the BLTS and TwinsUK cohorts, facial landmarks were derived from 2D portrait photos. The SYS (mean age 15 years), QTIMS (mean age 23 years), and BLTS (mean age 23 years) cohorts were on average much younger than other cohorts considered (mean age over 50 years, Table 1). The majority of the TwinsUK cohort was female (95.5%). 10.1371/journal.pgen.1002932.g001 Figure 1 Nine facial landmarks extracted via image registration tools from 3D MRIs. An MRI of one of the authors (MK) is used for illustration. A, with the landmark for left zygion (ZygL) highlighted, where a clipping plane was used to uncover the bone; B, with the landmarks for left (EyeL) and right pupils (EyeR) highlighted, where a clipping plane was used to uncover the vitreous humor; C, with the four nasal landmarks highlighted, including the left alare, nasion (Nsn), pronasale (Prn), and subnasale (Sbn). 10.1371/journal.pgen.1002932.t001 Table 1 Characteristics of the study subjects (N = 9,823). Cohort Country Individual For Image N Male% Age ± sd RS1 Netherlands Unrelated discovery 3D head MRI 2,470 46.4 59.7 ± 8.0 RS2 Netherlands Unrelated discovery 3D head MRI 745 43.1 59.0 ± 7.9 QTIMS Australia Twins discovery 3D head MRI 545 39.6 23.7 ± 2.3 SHIP Germany Unrelated discovery 3D head MRI 797 47.3 46.0 ± 12.8 SHIP-TREND Germany Unrelated discovery 3D head MRI 831 44.8 50.4 ± 13.6 SYS Canada Siblings replication 3D head MRI 568 48.1 15.1 ± 1.9 TwinsUK UK Twins replication 2D portrait photo 1,530 9.5 58.4 ± 12.9 BLTS Australia Twins replication 2D portrait photo 2,337 47.8 23.6 ± 4.6 For 3D MRI based phenotyping, we focused on the nine most well-defined landmarks from the upper part of the face, including Zygion Right (ZygR), Zygion Left (ZygL), Eyeball Right (EyeR), Eyeball Left (EyeL), Alare Right (AlrR), Alare Left (AlrL), Nasion (Nsn), Pronasale (Prn), and Subnasale (Sbn) (Figure 1). The lower part of the face i.e., from underneath the nose further down was not available due to brain-focussed MRI scanning. Raw landmark coordinates from 5,388 individuals in the five discovery cohorts (RS1, RS2, QTIMS, SHIP, and SHIP-TREND) showed systematic differences in position and orientation (Figure 2A). They were superimposed onto a consensus 3D Euclidean space based on partial PS (Figure 2B). A total of 27 principal components (PCs), and the centroid size parameter, were derived from the superimposed landmarks. Eleven PCs, each explaining >1% and all together explaining 96.0% of the total shape variance, were selected for further genetic association analysis (Table S1). Furthermore, we derived 36 Euclidean distances between each pair of landmarks. The partial PS had no effect on inter-landmark distances i.e., the distances remain the same after the superimposition. We derived the phenotypic correlations in discovery cohorts containing only adults or young adults. The SYS cohort was excluded from this correlation analysis because the changes through adolescence may confound the effect of age. Centroid size was highly correlated with the first PC (r = 0.96, Table S1) as well as with all 36 inter-landmark distances (mean r = 0.76; minimal r = 0.56 for Prn-Sbn; maximal r = 0.94 for ZygR-AlrL; Table S2). Inter-landmark distances were also correlated with each other (mean r = 0.56; minimal r = 0.10 between EyeL-AlrL and ZygL-EyeL; maximal r = 0.96 between AlrR-Nsn and AlrL-Nsn; Table S2). The distances between symmetric landmarks all showed the highest correlations (Table S2), consistent with general knowledge about facial symmetry. Compared with females, males on average had greater centroid size (P 3%); further investigations of less common and/or quite rare variants as for instance available from next generation sequencing data may provide a more complete figure on the genetic basis of facial morphology. In spite of these limitations we have been able to demonstrate that a phenotype as complex as human facial morphology can be successfully investigated via the GWAS approach with a moderately large sample size. Three of the five loci highlighted here map to known developmental genes with a previously demonstrated role in craniofacial patterning, one of which has been unequivocally associated with nasion position in a recent independent GWAS [34]. The remaining two loci map to or close to C5orf50 and COL17A1, neither of which have previously been implicated in facial development. The associated DNA variants may either affect neighboring genes, or alternatively identify C5orf50, and COL17A1 as potential new players in the molecular regulation of facial patterning. Overall, we have uncovered five genetic loci that contribute to normal differences in facial shape, representing a significant advance in our knowledge of the genetic determination of facial morphology. Our findings may serve as a starting point for future studies, which may test for allele specific expression of these candidate genes and re-sequence their coding regions to identify possible functional variants. Moreover, our data also highlight that the high heritability of facial shape phenotypes (as estimated here and elsewhere), similar to that of adult body height [53], involves many common DNA variants with relatively small phenotypic effects. Future GWAS on the facial phenotype should therefore employ increased sample sizes as this has helped to identify more genes for many other complex human phenotypes such as height [53] and various human diseases. Combined with the emerging advances in 3D imaging techniques, this offers the poteintial to advance our understanding of the complex molecular interactions governing both normal and pathological variations in facial shape. Materials and Methods Rotterdam Study (RS), The Netherlands The RS is an ongoing population-based prospective study including a main cohort RS-I [54] and two extensions RS-II and RS-III [55], [56], including 15,000 participants altogether, of whom 12,000 have GWAS data. Collection and purification of DNA have been described in detail previously [57]. A subset of participants were scanned on a 1.5 T General Electric MRI unit (GE Healthcare, Milwaukee, WI, USA), using an imaging protocol including a 3D T1-weighted fast RF gradient recalled acquisition in steady state with an inversion recovery prepulse. The following parameters were used: 192 slices, a resolution of 0.49×0.49×0.8 mm3 (up sampled from 0.6×0.7×0.8 mm3 using zero padding in the frequency domain), a repetition time (TR) of 13.8 ms, an echo time (TE) of 2.8 ms, an inversion time (TI) of 400 ms, and a flip angle of 20°. More details on image acquisition can be found elsewhere [58]. The Medical Ethics Committee of the Erasmus MC, University Medical Center Rotterdam, approved the study protocol, and all participants provided written informed consent. Principal components analysis of SNP microarray data was used to identify ancestry outliers. These were removed before further analyses, and the present sample is of exclusively northern/western European origin. The current study included 3,215 RS participants who had both SNP microarray data and 3D MRI. These participants were considered here as two cohorts (RS1 N = 2,470 and RS2 N = 745) as they were scanned and genotyped at different times. Brisbane Longitudinal Twin Study (BLTS) and Queensland Twin Imaging Study (QTIMS), Australia Adolescent twins and their siblings were recruited over a period of sixteen years into the BLTS at 12, 14 and 16 years, as detailed elsewhere [59] and as young adults into the QTIMS [60], [61]. The present study includes a sub-sample of 545 young adults (aged 20–30 years, M = 23.7±2.3 years; 79 MZ and 90 DZ pairs, 110 unpaired twins, and 97 singletons, from a total of 332 families) from QTIMS with 3D MRIs, and 2,137 adolescents (aged 10–22 years, M = 15.6±1.5 years; 311 MZ and 530 DZ pairs, 44 unpaired twins and 411 singletons, from a total of 1,038 families) from BLTS with 2D portrait photos. 3D T1-weighted MR images were collected at the Centre for Advanced Imaging, University of Queensland, using a 4T Bruker Medspec whole body scanner (Bruker Medical, Ettingen, Germany) [61]. 2D portrait photos were taken from a distance of 1–2 meters for identification, with no specific instruction given to smile. Those who had both 3D MRI scans and 2D photos were included in discovery GWAS and excluded from the replication analysis in 2D photos. Over 70% were digital with the remainder being scanned from high quality film. The study was approved by the Human Research Ethics Committee, Queensland Institute of Medical Research. Informed consent was obtained from all participants (or parent/guardian for those aged less than 18 years). Study of Health In Pomerania (SHIP), Germany The SHIP is a longitudinal population based cohort study assessing the prevalence and incidence of common, population relevant diseases and their risk factors with examinations at baseline (SHIP-0, 1997–2001), 5-year-follow-Up (SHIP-1, 2002–2006) and an ongoing 10-year-follow-Up (SHIP-2, 2008–2012) [62], [63]. Data collection from the baseline sample included 4,308 subjects. A new cohort targeted 5,000 participants (SHIP-TREND) has been started parallel to the SHIP-2-Follow-Up. In addition to the baseline examinations, participants of SHIP-2 and SHIP-TREND also had a whole-body MRI scan [64]. MRI examinations were performed on a 1.5T MR imager (Magnetom Avanto; Siemens Medical Systems, Erlangen, Germany). Head scans were taken with an axial ultra-fast gradient echo sequence (T1 MPRage, TE 1900.0, TR 3.4, Flip angle 15°, 1.0×1.0×1.0 mm voxel size). The present study includes 797 SHIP as well as 831 SHIP-TREND participants which had both SNP genotyping data and 3D MRI. The medical ethics committee of University of Greifswald approved the study protocol, and oral and written informed consents were obtained from each of the study participants. Saguenay Youth Study (SYS), Canada Adolescent sibpairs (12 to 18 years of age) were recruited from a French-Canadian population with a known founder effect living in the Saguenay-Lac-Saint-Jean region of Quebec, Canada in the context of the ongoing Saguenay Youth Study [65]. Local ethics committee approved the study; the parents and adolescent participants gave informed consent and assent, respectively. MRI was acquired on a Phillips 1.0-T magnet using the following parameters: 3D radio frequency spoiled gradient-echo scan with 140–160 slices, an isotropic resolution of 1 mm, TR 25 ms, TE 5 ms, and flip angle of 30°. Outlying individuals including those with putative Indigenous American admixture were excluded based on genetic outlier analysis [66]. The current study contains 568 adolescents with MRI and SNP data. St. Thomas' UK Adult Twin Registry (TwinsUK), United Kingdom The TwinsUK cohort is unselected for any disease and is representative of the general UK population [67]. All were volunteers, recruited through national media campaigns. Written informed consent was obtained from every participant. Population substructure and admixture was excluded using eigenvector analysis on SNP microarray data. The current study included 1,366 individuals with 2D portrait photos and SNP microarray data. Facial landmarks from 3D head MRI In our discovery cohorts, since the lower part of the face was not available from the MRIs, we focused on nine landmarks of the upper face (Figure 1). These included Right (ZygR) and left (ZygL) zygion: the most lateral point located on the cortex of the zygomatic arches; right (EyeR) and left (EyeL) eyeball: the middle point of the eyeball; right (AlrR) and left (AlrL) alare: the most lateral point on the surface of the ala nasi; nasion (Nsn): the skin point where the bridge of the nose meets the forehead; pronasale (Prn): the most anterior tip of the nose; subnasale (Sbn): the point where the base of the nasal septum meets the philtrum. Although these landmarks provide only a very sparse representation of the facial shape, they cover most prominent facial features and are easy to interpret and compare to other studies [22], [34], [51]. Furthermore, these landmarks could be measured with higher accuracy from images than most semilandmarks [22]. The coordinates of these nine landmarks were derived with an automated technique as described previously [20], which uses image registration to transfer predefined landmarks from a limited set of annotated images to an unmarked image. The manual annotation was based on landmark definitions from the anthropology literature [68], which were adapted for application to T1-weighted MR images of the head. None of the MRI showed distortions in a visual inspection. Furthermore, the automatically localized landmark positions were robust against the number of samples included. The test-retest correlations based on a subset of 40 subjects from QTIMS who were scanned twice were high (r>0.99). In the SYS cohort, in total 56 facial landmarks were available from a previous quantitative analysis of craniofacial morphology using 3D MRI [22]. In brief, an average MRI was constructed using non-rigid image registration. The surface of this average image represents the mean facial features and was then annotated with 56 landmarks and semi-landmarks. These landmarks were then warped using the nonlinear transformation that maps each subject to the average. This allows for automatic identification of the different craniofacial landmarks. Facial landmarks from 2D portrait photos We defined eight landmarks in 2D portrait photos that approximately correspond to the respective landmarks ascertained from our 3D MRIs. These include EyeL, EyeR, Prn, AlrL, AlrR, Nsn, earlobe left (EarL), and earlobe right (EarR) (Figure 5A). Note the Sbn, ZygL and ZygR landmarks available in 3D MRIs could not derived in 2D photos. We developed an algorithm to locate these landmarks in 2D portrait images and implemented it in an in-house C++ program. Briefly, the algorithm first recognizes the face layout within an image by matching a face template. It then recognizes eyes, nose, and ears by matching corresponding templates. The template matching routines were based on external open source computer vision library, OpenCV 2.3.1 (http://sourceforge.net/projects/opencvlibrary/). The automatically identified landmarks were then manually adjusted by 5 research assistants on a standard computer screen. Facial shape phenotypes We used un-scaled PS, or partial PS [24], [69], to superimpose the landmarks from the 3D MRIs in the discovery cohorts onto a consensus 3D Euclidean space. Unlike full PS, partial PS only re-positions and re-orientates but does not rescale the landmark configurations; thus, it has no effect on the Euclidean distances between landmarks as measured in terms of millimeters from MRIs. Keeping the absolute inter-landmark distances allows us to interpret the association results more directly. Furthermore, the full PS has been criticized for introducing artificial correlations between landmarks [70]. We considered the centroid size as a measurement of face size, and it was significantly correlated with absolute head volume (r = 0.95). We derived 11 principal components (PCs) from the superimposed landmarks, each explaining at least 1% of the total phenotypic variance. We also derived 36 Euclidean distances between all pairs of landmarks. Thus 48 phenotypes were included in our GWAS, including centroid size, 11 PCs, and 36 inter-landmark distances. All phenotypes were approximately normally distributed and outliers (>3sd) were removed. Deformation approaches including the use of transformational grids provide an alternate way to study shape difference. Thin plate splines (TPS) [27] depicts the deformation geometrically, where the total deformation is decomposed into several orthogonal components to localize and illustrate the shape differences. We used TPS to illustrate the facial shape differences between males and females using the tpsgrid function in R library shapes. For 3D MRI data in SYS, we used the 56 landmarks derived in a previous study and calculated 1,540 Euclidean distances between all pairs of landmarks. These distances were considered as phenotypes in our replication analysis of GWAS findings. We also chose a subset of nine landmarks most closely resembled those used for the current study for exact replication. Since the size of the face vary substantially between 2D portrait images, we used the full PS [24] to also remove the scaling differences between landmark configurations. Note that the inter-landmark distances from 2D photos do not represent the absolute distances in terms of millimeters regardless of whether full or partial PS was used. After superimposition, we calculated 28 Euclidean distances between all pairs of the 8 landmarks, which were considered as phenotypes in the replication analysis. The PS analyses were performed with CRAN package shapes developed by Ian Dryden [30]. Heritability analysis By clarifying which facial features are under strong genetic control, we should be better able to identify specific genes that influence facial variation. Heritability estimates are also important indicators of the phenotype quality. Using QTIMS (79 MZ pairs, 90 DZ pairs) heritability analysis was carried out in Mx [71] using full information maximum likelihood estimation of additive genetic variance (i.e. heritability), common environmental variance, and unique environmental variance. Sex and age were included as covariates. Phenotypic correlations were estimated in BLTS (311 MZ pairs, 90 DZ pairs) and in TwinsUK (93 MZ pairs, 352 DZ pairs) where the facial shape phenotypes were derived from 2D photos. Genotyping, imputation, quality control Details of SNP microarray genotyping, quality control and genotype imputation are described in prior GWAS conducted in RS [16], QTIMS and BLTS [72], SHIP [62], SYS [73], and TwinsUK [74]. In brief, DNA samples from the RS, BTNS, SYS and TwinsUK cohorts were genotyped using the Human 500–610 Quad Arrays of Illumina and samples from SHIP were genotyped using the Genome-Wide Human SNP Array 6.0 of Affymetrix and HumanOmni2.5 of Illumina, respectively. Genotyping of the SHIP-TREND probands (n = 986) was performed using the Illumina HumanOmni2.5-Quad, which has not been reported previously and described here as follows. DNA from whole blood was prepared using the Gentra Puregene Blood Kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol. Purity and concentration of DNA was determined using a NanoDrop ND-1000 UV-Vis Spectrophotometer (Thermo Scientific). The integrity of all DNA preparations was validated by electrophoresis using 0.8% agarose-1x TBE gels stained with ethidium bromide. Subsequent sample processing and array hybridization was performed as described by the manufacturer (Illumina) at the Helmholtz Zentrum München. Genotypes were called with the GenCall algorithm of GenomeStudio Genotyping Module v1.0. Arrays with a call rate below 94%, duplicate samples as identified by estimated IBD as well as individuals with reported and genotyped gender mismatch were excluded. The final sample call rate was 99.51%. Imputation of genotypes in the SHIP-TREND cohort was performed with the software IMPUTE v2.1.2.3 against the HapMap II (CEU v22, Build 36) reference panel. 667,024 SNPs were excluded before imputation (HWE p-value≤0.0001, call rate ≤0.95, monomorphic SNPs) and 366 SNPs were removed after imputation due to duplicate RSID but different positions. The total number of SNPs after imputation and quality control was 3,437,411. The genetic data analysis workflow was created using the Software InforSense. Genetic data were stored using the database Caché (InterSystems). After SNP imputation to the HapMap Phase II CEU reference panel (Build 36) and quality control, 2,558,979 autosomal SNPs were common in all discovery cohorts and used for analyses. GWAS and replication We conducted discovery phase GWAS in a combined set of all discovery cohorts (RS1, RS2, QTIM, SHIP, SHIP-Trend) for 48 facial shape phenotypes. Imputed GWAS data in all discovery cohorts were merged according to the positive strand. We tested 2,558,979 autosomal SNPs with linear regression (adjusted for sex, age, EIGENSTRAT-derived ancestry informative covariates [75], plus any additional ancestry informative covariates as appropriate) in GenABEL [76]. The centroid size was adjusted in the analysis of inter-landmark distances. SNPs with MAF 3% (results not shown). All SNPs with P values<5×10−8 in our discovery phase GWAS were sought for replication in SYS, TwinsUK, and BLTS. Promising SNPs were tested for association with 1,054 inter-landmark distances in SYS and 28 inter-landmark distances in a combined sample of TwinsUK and BLTS assuming additive allelic effect adjusted for sex and age using MERLIN [79], which also takes into account family relationships. We report the association results for the same phenotypes as discovered in GWAS as exact replication. In addition, for each SNP we report the percentage of significantly (P<0.05) associated phenotypes, which is expected to be lower than 5% under the null hypothesis of no association. For the analysis of 11 NSCL/P associated SNPs in our discovery cohorts, we additionally Bonferroni corrected the P values for 48 correlated phenotypes since no specific facial phenotypes were selected for replication. Supporting Information Figure S1 Thin plate spline deformation illustrating facial shape differences in males compared to females in discovery cohorts (N = 5388). The pixel information obtained from the mean shape of males was mapped to that of females. The deformed images illustrate the difference between the mean shpae of males (the curved plates) compared to that of females (imaginary flat plates). A. a 3D view of the mean facial shape of all individuals in the discovery cohorts before deformation; B. side projection of the deformed grid; C. front projection of the deformed grid; D. up-down projection of the deformed grid. (TIF) Click here for additional data file. Table S1 Correlation between PC and Euclidean distances in discovery cohorts. The color shade is independent between columns. (XLSX) Click here for additional data file. Table S2 Correlation matrix between pair-wise Euclidean distances and size in discovery cohorts. (XLSX) Click here for additional data file. Table S3 Effect of sex and age on facial shape in discovery cohorts. Distances are presented in millimeters. P values were adjusted for centroid size. (XLSX) Click here for additional data file. Table S4 Heritability of facial shape phenotypes derived from 3D MRI in QTIMS. Twin correlations and proportions of variance due to additive genetic (A), common environmental (C), and unique environmental (E) influences, shown with 95% confidence intervals (age and sex adjusted). (XLSX) Click here for additional data file. Table S5 SNPs (n = 102) associated (P<5e-7) with facial shape phenotypes in discovery phase GWAS. Trait, the phenotype for which the minimal P value was obtained. MinP, the minimal P value. (XLSX) Click here for additional data file. Table S6 Raw genotype and respective phenotype data for all SNPs that revealed genome-wide significant and suggestive evidence. (XLSX) Click here for additional data file.
- Record: found
- Abstract: found
- Article: found
Genome-wide analysis of dental caries and periodontitis combining clinical and self-reported data
Dmitry Shungin, Simon Haworth, Kimon Divaris … (2019)
Dental caries and periodontitis account for a vast burden of morbidity and healthcare spending, yet their genetic basis remains largely uncharacterized. Here, we identify self-reported dental disease proxies which have similar underlying genetic contributions to clinical disease measures and then combine these in a genome-wide association study meta-analysis, identifying 47 novel and conditionally-independent risk loci for dental caries. We show that the heritability of dental caries is enriched for conserved genomic regions and partially overlapping with a range of complex traits including smoking, education, personality traits and metabolic measures. Using cardio-metabolic traits as an example in Mendelian randomization analysis, we estimate causal relationships and provide evidence suggesting that the processes contributing to dental caries may have undesirable downstream effects on health.
Author and article information
Contributors
Kimon Divaris: kimon_divaris@unc.edu
Journal
Journal ID (nlm-ta): Periodontol 2000
Journal ID (iso-abbrev): Periodontol. 2000
Journal ID (doi): 10.1111/(ISSN)1600-0757
Journal ID (publisher-id): PRD
Title:
Periodontology 2000
Publisher:
John Wiley and Sons Inc.
(Hoboken
)
ISSN
(Print):
0906-6713
ISSN
(Electronic):
1600-0757
Publication date
(Electronic):
18
December
2019
Publication date
(Print):
February
2020
Volume: 82
Issue: 1
(
doiID:
10.1111/prd.v82.1
)
Pages: 143-156
Affiliations
[
1
]
Department of Periodontology
School of Dentistry
University of North Carolina at Chapel Hill
Chapel Hill
North Carolina, USA
[
2
]
Department of Oral and Craniofacial Health Sciences
School of Dentistry
University of North Carolina at Chapel Hill
Chapel Hill
North Carolina, USA
Author notes
[*]
[*
]
Correspondence
Kimon Divaris, Department of Pediatric Dentistry, UNC School of Dentistry, University of North Carolina‐Chapel Hill, 228 Brauer Hall, CB #7450, Chapel Hill, NC 27599, USA.
Email: kimon_divaris@ 123456unc.edu
Kimon Divaris, Department of Pediatric Dentistry, UNC School of Dentistry, University of North Carolina‐Chapel Hill, 228 Brauer Hall, CB #7450, Chapel Hill, NC 27599, USA.
Email: kimon_divaris@ 123456unc.edu
Article
Publisher ID: PRD12320DOI: 10.1111/prd.12320
PMC ID: 6972532
PubMed ID: 31850632
SO-VID: c0148387-0697-4edc-9c8a-70a02e83e533
Copyright © © 2019 The Authors.
Periodontology 2000 Published by John Wiley & Sons Ltd.
License:
This is an open access article under the terms of the http://creativecommons.org/licenses/by/4.0/ License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
History
Page count
Figures: 4,
Tables: 4,
Pages: 14,
Words: 10912
Funding
Funded by: NIH/NIDCR
Award ID: U01DE025046
Categories
Subject:
Review Article
Subject:
Review Articles
Custom metadata
source-schema-version-number 2.0
cover-date February 2020
details-of-publishers-convertor Converter:WILEY_ML3GV2_TO_JATSPMC version:5.7.5 mode:remove_FC converted:21.01.2020
Data availability: