Assessment of genotyping tools applied in genetic susceptibility studies of periodontal disease: A systematic review

Assessment of genotyping tools applied in genetic susceptibility studies of periodontal disease: a systematic review Alicia de Coo, MSa*, Inés Quintela, PhDb, Juan Blanco, MD, DDS, PhDc, Pedro Diz, MD, DDS, PhDd, Ángel Carracedo, MD, PhDe aGrupo de Medicina Xenómica, Universidad de Santiago de Compostela, España. bGrupo de Medicina Xenómica, Centro Nacional de Genotipado - Plataforma de Recursos Biomoleculares y Bioinformáticos - Instituto de Salud Carlos III (CeGen-PRB2-ISCIII), Universidad de Santiago de Compostela (USC), Santiago de Compostela, A Coruña, España. cGrupo de Odontología Médico-Quirúrgica (OMEQUI), Instituto de Investigación Sanitaria de Santiago de Compostela (IDIS), Universidad de Santiago de Compostela (USC), Santiago de Compostela, A Coruña, España. dGrupo de Odontología Médico-Quirúrgica (OMEQUI), Instituto de Investigación Sanitaria de Santiago de Compostela (IDIS), Universidad de Santiago de Compostela (USC), Santiago de Compostela, A Coruña, España. eGrupo de Medicina Xenómica, Centro Nacional de Genotipado - Plataforma de Recursos Biomoleculares y Bioinformáticos - Instituto de Salud Carlos III (CeGen-PRB2-ISCIII), Universidad de Santiago de Compostela (USC), Santiago de Compostela, A Coruña, España; Grupo de Medicina Xenómica, CIBERER, Fundación Pública Galega de Medicina Xenómica - SERGAS Santiago de Compostela, A Coruña, España. CORRESPONDENCE: *Alicia de Coo Diz Grupo de Medicina Xenómica Department of Surgery and Medical-Surgical Specialties Faculty of Medicine and Odontology, University of Santiago de Compostela c/ Entrerríos sn, 15782 Santiago de Compostela, Spain. Tel: +34 881812344, Fax: +34 981562226, E-mail: [email protected] *Manuscript text Click here to view linked References ABSTRACT Objective: A systematic review to evaluate the various genotyping tools and study strategies employed to define genetic susceptibility to periodontitis. Methods: The review was performed in accordance with Preferred Reporting Items for Systematic Review and Meta-Analysis guidelines. The search for publications referring to the genetic bases of periodontal disease was performed on the MEDLINE-PubMed and Cochrane Library databases, on trials registers, and on the web pages of regulatory agencies. Results: We found 2439 potentially eligible articles, of which only 25 satisfied the established inclusion criteria and were processed for data extraction. The review revealed marked heterogeneity between studies, caused in part by the lack of a universally accepted definition for periodontitis phenotypes and by the variety of genotyping tools available. The most commonly used technique was genotyping candidate genes. Conclusion: The few rigorous studies that have been published on genetic susceptibility to periodontitis are subject to severe methodological bias due to their design and the genotyping tools employed. Despite their limitations, candidate gene studies continue to be the predominant methodological approach, rather than genome-wide association studies. Further studies must be designed using a universally accepted, validated diagnostic criterion for periodontitis, analysing multiple genes and polymorphisms in combination with rare variants. KEYWORDS: Periodontal disease, Genetics, Gene polymorphism INTRODUCTION The existence of a genetic component to periodontal disease (PD) has been confirmed in twin studies, which estimate that 38% to 82% of populational variability in the clinical parameters of periodontal disease is attributable to genetic factors (Michalowicz et al., 1991a, 19991b, 2000). It has been suggested that the genetic component of chronic periodontitis (CP) might have been overestimated (Torres de Heens, Loos, & van der Velden, 2010), but that it is more relevant in aggressive periodontitis (AgP), as has been demonstrated in familial aggregation studies (Benjamin & Baer, 1967; Marazita et al., 1994; Saxén & Nevanlinna, 1984). As with diabetes, certain types of cancer, Alzheimer's disease, Crohn's disease, and schizophrenia, PD is considered to be genetically complex, resulting from the interaction of genes with the environment (Laine, Crielaard, & Loos, 2012; Stabholz, Soskolne, & Shapira, 2010). Two main hypotheses have been proposed regarding the genetic basis of complex diseases. The first is the common disease/common variant (CD/CV) hypothesis (Reich & Lander, 2001), in which it is assumed that genetic variants common in the general population but which individually have a weak effect are those that have the greatest influence on genetic susceptibility to complex disorders. These genetic variants are called polymorphisms when they are present with a frequency of at least 1% in the population (Hart, Marazita, & Wright, 2000). They include single nucleotide polymorphisms (SNPs), which are very common in the genome and consist of a change of one nucleotide base for another. The second hypothesis is the common disease/rare variant (CD/RV) hypothesis, in which the main contributors to susceptibility to complex diseases are rare variants (minor allele frequency <1%) present in the genome (Pritchard, 2001). The hypothesis that is deemed valid will thus determine the strategy applied to detect variants that favour disease. Most PD studies assume the CD/CV hypothesis; thus, the most common strategy consists of the search for polymorphisms that could affect the periodontal disease phenotype. One of the first studies to identify genetic markers for PD was published in 1997 by Kornman et al (Kornman et al., 1997). Those authors analysed several polymorphisms of the interleukin-1 (IL-1) family (polymorphisms of IL-1A at position-889, of IL-1B at position -511 and +3953, and of IL-1RN intron 2) and found evidence of an association between IL-1 and the severity of PD in nonsmokers, differentiating between individuals with mild and severe PD. These encouraging results led to further genetic studies on PD that applied distinct methodological designs. One of those was the genome-wide linkage analysis, which has been widely used to study complex diseases such as schizophrenia (Stefansson et al., 2002) and type 2 diabetes (Nisticò et al., 1996); however, in contrast to its notable efficacy in the genetic mapping of Mendelian diseases, the technique has limited usefulness in the detection of alleles with a weak effect, which are common in complex diseases (Plomin, Haworth, & Davis, 2010). Another design is the association study, which enables relationships to be established between a specific polymorphism and a disease or phenotypic trait. This approach determines differences in the frequency of a genetic marker in cases and controls to indicate a relationship between that marker and the disorder under investigation. The two principal methods in this context are the candidate gene association study (CGAS) and the genome-wide association study (GWAS). The CGAS is based on an analysis of genes that are linked to a certain disease through their function or their position in the genome. It is therefore essential for investigators to understand the pathophysiology of a disorder before performing this type of study. Based on the promising results from what was considered the first CGAS (Kornman et al., 1997), this type of design became the most widely used to analyse the genetics of PD. However, the very rapid development of genotyping technologies in recent years has led to substitution of CGASs with GWASs, because these latter studies theoretically enable us to identify new genetic markers. The first GWAS to study periodontitis was published in 2010 and established an association between SNP rs1537415, present in gene GLT6D1 (glycosyltransferase), and AgP (Schaefer et al., 2010a). Given the complexity of PD, the role of genetics is significant in its onset, progression, and severity (Albandar & Tinoco, 2002). The identification of genetic risk factors associated with PD is therefore essential to improving prevention and treatment strategies for this disease (Song, Yao, He, & Xu, 2015). The objective of this systematic review was to evaluate the genotyping tools and study strategies used in the literature to establish the genetic bases of PD and to determine their main limitations. MATERIALS AND METHODS The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) review process was used to perform the present systematic review (Moher, Liberati, Tetzlaff, Altman, & PRISMA Group, 2009).The PRISMA checklist was followed in both the planning and the reporting of the review. Focused Question The research question formulated was based on PECO framework: If in the studies performed on patients with periodontitis reported in the literature (Population) in which the genetic basis of susceptibility to periodontal disease was determined (Exposition), we then analyse the limitations and potential biases of the genotyping tools used (Control), are the validity and reliability of the results confirmed (Outcome)? Eligibility criteria The following inclusion criteria were applied: studies published in English; case-control design; specification of the PD phenotype; a sample size equal to or greater than 100 individuals in either the case or control group, or in both groups. No limitations were placed on the year of publication or the age, population group, or country of origin of the participants. Publications with any of the following potential biases were excluded: a study group that included individuals with systemic diseases that could favour the onset or augment the severity of PD; patients treated with drugs that provoke gingival hyperplasia; and pregnant women (pregnancy gingivitis). Reviews and meta-analyses were also excluded, even if they referred to genetic susceptibility to PD. We used the number of SNPs analysed as inclusion criteria for candidate gene studies. There have been many studies of this type that have analysed a reduced number of SNPs; they are generally lacking in quality control and have a lack of appropriate statistical corrections, HardyWeinberg equilibrium analysis or correspondence of cases and controls with a population background. Thus, due to the criteria that define the copy number variations (CNVs) (Lin, Naj, & Wang, 2013; Merikangas, Corvin, & Gallagher, 2009) and the limitations observed in those studies with a reduced number of SNPs, only the studies that analysed ≥10 SNPs were included. Information sources The search for articles was performed using the electronic databases of the National Library of Medicine, Washington DC, USA (MEDLINE, PubMed), and of the Cochrane Library, in addition to trial registers and the web pages of regulatory agencies. A manual search was performed of the most relevant journals in the field of periodontics of the last 10 years. In addition, "snowball" methods were also applied such as pursuing references of references from all selected full-text papers. The search was performed up to Jan 2018. Study selection was performed in accordance with the most recent recommendations of the Centre for Reviews and Dissemination of the University of York (Centre for Reviews and Dissemination, University of York, 2008).The following keywords were used: “periodontitis” and “periodontal disease”, in combination with “genetic polymorphism”, “single nucleotide polymorphism”, “mutation”, “genes”, “new genome sequencing”, “exome sequencing”, “whole genome sequencing”, “CNV”, and “genome wide association”. The “humans” filter was selected for all searches. Search strategy The search strategy used a combination of Medical Subject Headings terms and keywords for MEDLINE and Cochrane Library. The search strategy used is described in the supporting information (appendix S2). Study selection Study selection was performed using a two-stage screening process performed by two independent reviewers (P. D. and A. de C.). Disagreement on the inclusion or exclusion of a specific article was resolved by consensus. In the first phase, studies that did not satisfy the inclusion criteria were eliminated, as were those considered irrelevant based on their title or the content of the abstract. A full-text review of the remaining articles was then performed to confirm the inclusion criteria and to check quality. The most relevant data were then extracted. Concordance between the two reviewers was calculated using the Kappa coefficient for the first and second screenings. Data collection process/data items The eligibility criteria were applied after agreement had been achieved between the two reviewers for the studies to be included in the review. Data were gathered on the general characteristics of the study (year of publication, clinical parameters recorded, number of SNPs studied, genotyping method) and on their populational characteristics (population, age, number of participants, smoking, diabetes). The extracted data were presented in the same form in which they were described in the original studies, except for those studies with the same design but different forms of presentation of the results, in which case the results were processed to facilitate extraction of the data of interest. Risk of bias in individual studies We aimed to assess the risk of bias of the included studies through sensitivity analysis. The risk of bias/quality assessment for the individual studies was assessed according to the suggested scale (score of 0 to 20) adapted to genetic analysis of periodontal genetic association studies (Nibali, 2013). Summary measures/synthesis of results No meta-analysis was attempted for this systematic review; thus, no formal statistical analyses were performed. RESULTS Study selection The search provided a total of 2439 potentially selectable articles (MEDLINE PubMed, n=2392; Cochrane, n=47; other sources, n=0) (Figure 1). Duplicate articles were eliminated by comparing the results obtained with each database, leaving 2412 articles potentially useful for the review. The titles and abstracts of these studies were then examined, and 87 articles were selected for full-text review; of these, 62 were excluded because they did not satisfy the inclusion criteria. This left a final count of only 25 studies for analysis in this review. The Kappa value for inter-reviewer concordance was 0.95 for the first screening (by title and abstract) and 0.89 for the second screening (full-text review). We detected several publications in which samples derived from the same study group were analysed. In 2010, Chai et al. analysed samples gathered between May 2005 and August 2007 from a primary care clinic in the Prince Philip Dental Hospital of the Faculty of Dentistry of the University of Hong Kong; they analysed the SNPs present in C5 (Chai, Song, Zee, & Leung, 2010a) and FCGR (Chai, Song, Zee, & Leung, 2010b). In 2010, Schaefer et al. conducted the first GWAS to study periodontitis (Schaefer et al., 2010a). They analysed the German population with AgP and replicated the association in a panel of Dutch patients with AgP. Subsequently, these cohorts were examined in a large candidate-gene association study by Schaefer et al. in 2014 and 2015 (Schaefer et al., 2014; Schaefer et al., 2015). Also in 2010, Schaefer et al. analysed 12 SNPs in the gene DEFB1 in individuals from Germany and the Netherlands with CP and AgP (Schaefer et al., 2010), and in 2011 performed an analysis of 51 SNPs in genes CDKN2A and CDKN2B for individuals with AgP and of 37 SNPs in the same genes but in individuals with CP (Schaefer et al., 2011). Also in 2011, Scapoli et al. studied 14 candidate genes, including some belonging to the IL-1 cluster that had already been studied a year earlier in the same population sample (with the exception of 25 new participants from the University of Turin) (Scapoli et al., 2010, 2011) and with the same results. In 2014, de Jong et al. (de Jong et al., 2014) analysed cases and controls described and studied previously by Schaefer et al. in 2010 (Schaefer et al., 2010a). The difference between the two studies was that de Jong et al. only investigated SNPs in genes SLC23A1 and SLC23A2 instead of the whole genome, as the earlier study had done. One year later, in 2015, Schaefer et al. (Schaefer et al., 2015), investigating the genetic risk factors shared between coronary artery disease and periodontitis, analysed the same German/Austrian, Dutch, and Turkish cohorts as Schaefer et al. in 2010 and 2013 (Schaefer et al., 2010a, 2013). In addition, patients from the Netherlands and from Turkey were studied at a whole genome level by Munz et al. in 2017 (Munz et al. 2017). Finally, Offenbacher et al. (Offenbacher et al., 2016) and Munz et al. (Munz et al., 2017) analysed the same German population whose cases were previously described by Schaefer 2015 and whose controls were recruited from the Competence Network FOCUS (Food Chain Plus) (Müller et al., 2015), the Dortmunder Gesundheits studie (Berger, 2012), and the Heinz Nixdorf Recall Studies 1-3 (Schmermund et al., 2002). Study characteristics The principal characteristics of the 25 studies selected for this review are summarised in Table 1 and in the following sections: Diagnosis of periodontal disease, Demographics, Quality assessment, Clinical parameters, Study methodology, Genotyping method, and Genetic results. Diagnosis of periodontal disease Most of the studies (20 of 25) were published in 2010 or later; nevertheless, the diagnostic criteria for the PD phenotype in ten of these studies were those proposed by Armitage in 1999 (Armitage, 1999), and another three used other sources from that same year (Flemmig, 1999; Lang et al., 1999). In the remaining 12 studies, the diagnostic criteria individuals) greatly contribute to the risk of false-positive or false-negative results (Burgner, Jamieson, & Blackwell, 2006; Newton-Cheh & Hirschhorn, 2005). Common or rare Copy Number Variations (CNVs) have been associated with genetic susceptibility for many diseases (Merikangas et al., 2009), and as such remain one of many viable categories of genomic variation to be explored for possible genetic contributions to disease (Lin et al., 2013). Threshold parameters for calling a CNV recommend analysing CNVs spanning at least 10 SNPs (Lin et al., 2013). However, the exclusion of articles with fewer than 10 SNPs could also constitute a methodological limitation of the present study. The predominant methodological design in the research evaluated was the candidate gene approach. In these association studies, a hypothesis is established a priori not only on the effect of certain genes on the risk of developing the disease but also on the presence of functional variants of these genes (Wilkening, Chen, Bermejo, & Canzian, 2009). The genes targeted in this analysis were selected based on a previous understanding of the mechanisms involved in the molecular biology of the disease. This study design thus presents limitations, given that there could be genes that have an influence on the development of the disease but whose function or involvement in key pathways of the disease process remain undefined. The number of variants analysed in studies that use this design is therefore limited. GWASs also have a series of limitations. As occurs with other diseases, some of the genes associated with PD do not express a known biological function that could explain this association. In GWASs, it is essential to guarantee the power of the study, because the analysis of thousands of SNPs in a small sample will lead to a high risk of finding false associations, mainly because of random fluctuations in allele frequency between cases and controls (Schaefer, Jepsen, & Loos, 2011). In addition, GWASs need to be automated procedures for determining genotypes, because the assigned genotypes are not manually curated; thus, erroneous genotypes could go unnoticed thus introducing false positive associations. To date, 13 studies using GWAS methodology have been published in the databases selected, although only four satisfied the inclusion criteria for the present review (Schaefer et al., 2010a; Feng et al., 2014; Munz et al., 2017; Offenbacher et al., 2016). The first GWAS was performed by Schaefer et al. in 2010 (Schaefer et al., 2010a). Those authors analysed the genomes of 283 white individuals of German nationality with AgP and found a strong association between polymorphism rs1537415, located in gene GLT6D1, and AgP; this association was confirmed on repeating the study in a group of Dutch patients. Four years later, Feng et al. (Feng et al., 2014) used the GWAS technique to study the DNA of a group of 866 individuals from the dental register of the University of Pittsburgh. They analysed more than 500,000 SNPs and compared the 20 most relevant SNPs with a bank of samples from Brazilian patients, confirming that locus 16q22.3 contributed to the onset of CP. The authors of the two most recent GWASs agree in stating potential biases, such as the origin of the participant population, but they draw attention to the importance of GWAS in the design of future studies on the genetic bases of PD using multiple candidate genes. In 2016, Offenbacher et al. (Offenbacher et al., 2016) examined the association of more than 2.1 million markers with CP in a larger Dental Atherosclerosis Risk in Communities cohort (n=4766 Northern European). BEGAIN and UBE3D showed associations with severe CP and with moderate CP, respectively. These data were not included in the review because the control group was formed of individuals with mild periodontitis. They also analysed an independent German sample including 717 AgP cases, but no association was found (Offenbacher et al., 2016). The data corresponding to this cohort was included in the present review. One year later, Munz et al. conducted a GWAS using German and Dutch AgP samples (896 cases and 7104 controls), validated the association in a German CP sample (993 cases and 1416 controls), and replicated the positive findings in a Turkish AgP sample (223 cases and 564 controls). They found an association of SIGLEC5 and DEFA1A3 with periodontitis at a genome-wide level in the pooled samples (Munz et al., 2017). Of the four articles that used GWAS methodology and were included in the review, two were considered robust GWASs (Munz et al., 2017; Schaefer et al., 2010a), but the association of genes found with the PD was considered weak, because the results were not replicated in an independent GWAS or in a robust CGAS. Furthermore, association studies performed at a genome-wide level do not tag rare variants (Cirulli & Goldstein, 2010). As stated in the CD/RV hypothesis, rare variants can contribute substantially to the risk of developing a disease such as PD. Rare variants have modest-to-small effect sizes on phenotypic variation; thus, their analysis has the potential to reveal novel variants implicated in complex diseases (Auer & Lettre, 2015). GWASs performed to date in PD have been based on the CD/CV hypothesis and have not taken rare variants into account in the design of the arrays, due to their high cost. The rapid improvement in the technology for measuring genomic variation has allowed the appearance of DNA sequencing-based approaches, such as next generation sequencing (NGS) and custom arrays. These tools help ascribe the contribution of rare variants to complex traits and diseases, thus eliminating this recognised limitation of GWASs (Cirulli & Goldstein, 2010). The studies included in the present review show considerable heterogeneity with regard to the genotyping techniques employed, reflecting ever-advancing technological developments. Chronologically, PCR was the first genotyping method to be described and was followed by variants of the same process (PCR-SSP and PCR-RFLP). PCR-SSP uses specific probes to amplify the DNA. Two PCRs are necessary to type a single SNP using this method, one with the specific primer and another with the common primer; additionally, this PCR variant requires a larger amount of DNA than other methods (Bugert et al., 2003). PCR-RFLP is based on the selective amplification of certain restriction fragments of a DNA sample, which gives robust and unequivocal results as long as a suitable endonuclease is used. The problem with this technique is the need for post-PCR DNA processing, which increases the risk of contamination (Dorak, 2006). Other studies have used the TaqMan genotyping system (Applied Biosystems, Foster City, CA, USA). This technology consists of a fluorophore covalently bonded to the 5´ terminus of a probe, which enables genotyping of the samples to be performed. By using allele-specific probes in the PCR, the amplification and detection steps are combined so that, in contrast to other methods, no additional post-PCR processing is required. The estimated error rate with this technique is less than 1 in every 2000 genotyping assays (Ranade et al., 2001). When the aim is to analyse a larger number of SNPs, these basic methods are costly and time-consuming if used alone, making them uneconomical in studies demanding a high yield. A system used for massive SNP analysis is SNPlex (Tobler et al., 2005), a method based on capillary electrophoresis. This method is a good alternative to previous genotyping methods because it only requires a small amount of DNA and uses easily accessible electrophoresis instruments (Tobler et al., 2005). Another genotyping tool is the novel nano-Invader DNA chip system. This approach is a miniaturised genotyping technique, in which the maximum sample volume is of 45 nanolitres. The sample is deposited on a polycarbonate slide with a capacity to perform 5040 tests per chip; in contrast to the other genotyping tools, this system does not require the PCR purification and hybridisation steps. PCR amplification and other time-intensive processing phases that require special apparatus to hold the amplified products are tedious but necessary when employing other methods of DNA typing, such as Sequenom (Sequenom, San Diego, CA, USA), Affymetrix (Affymetrix, High Wycombe, UK) and Illumina (Illumina, San Diego, CA, USA). Sequenom's MassARRAY technology incorporates a genotyping method based on the mass of the nucleotide added to the end of the primer. The mass is measured in a mass spectrometer using the MALDI-TOF technique (Jurinke, van den Boom, Cantor, & Köster, 2002). Given that this technique is based on an intrinsic property of the molecules — their absolute mass (Brookes, 1999)— it avoids the introduction of modifications at a molecular level, such as the fluorescent marking characteristic of other methods (Meyer & Ueland, 2011). Sequenom's MassARRAY has certain drawbacks; it involves an initial purification stage that makes automation difficult and requires elaborate equipment, increasing its cost (Dorak, 2006). The Affymetrix technology employs probes synthesised directly on a quartz wafer. Two types of array were described in the selected publications: the Affymetrix Human Mapping 500K array set (including 500,000 SNPs) (Affymetrix & Inc, 2006) and the Affymetrix Genome-Wide Human SNP array 5.0 (including 500,568 SNPs) (Affymetrix, 2007). In addition, Affymetrix makes it possible to perform a genome-wide analysis in a single experiment (Kennedy et al., 2003; Thermo Fisher Scientific Inc., 2017). The Illumina Human OmniBeadChip, Illumina Human 610Quav1_B BeadChip, and Illumina Inmunochip arrays are based on the Illumina Bead Array technology, which uses a fibreoptic or silica substrate onto which small silica beads (3 microns) are applied (Shen et al., 2005). Each bead is covered by hundreds of thousands of copies of a specific oligonucleotide, so that the fragments of DNA analysed will bind to the oligonucleotides if they possess the complementary region (Illumina, 2015). The Illumina Immunochip, the most recent of the three arrays, is a custom-made Illumina Infinium genotyping array that includes 196,524 SNPs, designed to perform deep replication and fine mapping of established GWAS loci from major inflammatory and autoimmune diseases (Cortes & Brown, 2011). However, Immunochip has the following weaknesses: The chip depends on the power of the initial GWAS for its marker selection and does not cover the whole genome. Second, Immunochip is designed for use in white European populations. Third, many rare variants are not represented on the chip because they have yet to be identified. Another weakness is that a proportion of very rare variants will probably not be accurately genotyped by the chip, although based on early indications, the chip performs this genotyping well (Cortes & Brown, 2011). Another Illumina array used in the included studies was the Illumina Exome 12v1 (in which12 denotes the number of samples that can be run on a single exome chip and v1 denotes version 1), hereafter referred to as an exome chip. Markers on exome chips are primarily exome SNPs (over 92% of the SNPs are in the exome) (Exome Chip Consortia, 2011). The exonic content consists of more than 240,000 markers representing diverse populations (European, African, Chinese, and Hispanic). The exome chip includes 196,524 SNPs and is designed to concentrate on rare variants rather than on common ones. The Affymetrix and Illumina arrays increase the genomic cover for genome-wide analysis and for the detection of copy number variations (CNVs) compared with other methods (Illumina, 2015; Thermo Fisher Scientific Inc., 2017). In addition, both platforms have genotype calling algorithms. The selection of one or another tool is determined by the number of SNPs and samples that they can analyse, and by their yield. The genotyping platforms most frequently employed in the studies selected for this review were TaqMan, Sequenom, and Illumina. TaqMan was one of the most common techniques employed, although it has a moderate yield and is limited in terms of the number of SNPs and samples that can be analysed (Ranade et al., 2001). Another tool used was Sequenom, which enables a moderate number of SNPs and samples to be studied, with a high yield (thousands of genotypes per day) (Jurinke et al., 2002). Illumina SNP arrays have a high capacity for the analysis of markers (including rare variants) and samples, and with a high yield (one million genotypes per day) (Shen et al., 2005). PCR and its variants, PCR-RFLP and PCR-SSP, were the second-most frequently used tools, despite their great limitations (Dorak, 2006). These were followed by Affymetrix SNP arrays, enabled to analyse a high number of SNPs and samples, and with a high yield (one million genotypes per day) (Kennedy et al., 2003). Like Illumina, Affymetrix allows creation of customized arrays incorporating desire markers, including rare variants. Another tool used was the nano-Invader, a miniaturised genotyping technique that, despite achieving good precision, sensitivity, speed, and yield in genotyping without PCR amplification, presents pipetting and evaporation issues due to the small sample volume used (Nomura, Kondo, Nagano, Matsui, & Egashira, 2007). Only two reports included in this review describe the use of this tool; both studies were written by the same group (Kobayashi et al., 2009a, 2009b). Later studies have used other genotyping tools, suggesting that because of the rapid advances in genotyping techniques, the nano-Invader method is less effective than its successors. Last in the list was SNPlex, with a high yield (thousands of genotypes per day) that allows the analysis of a moderate number of markers and samples (Tobler et al., 2005). As described above, genotyping rare variants is a difficult process but an interesting approach to study risk of developing a disease such as PD. Today, the best approach to screen for novel variations and evaluate the contribution of rare variants to complex diseases is sequencing by NGS and custom arrays. However, conducting whole-genome sequencing in large cohorts using this sequencing technology remains economically prohibitive. Given that 98% of the human genome is noncoding, and therefore more difficult to interpret, whole-exome sequencing methods were developed to sequence only the coding regions of the genome (exome). The Affymetrix and Illumina platforms introduced the Affymetrix Axiom exome array and the exome chip, respectively, as an affordable alternative to exome sequencing. The Axiom exome array and the exome chip only enable researchers to test a predetermined set of variants. In the study performed by Schaefer et al. in 2014 (Schaefer et al., 2014), included in the present review, a significant association was found between the rare variant rs62481981 at the IRF5 gene after covariate adjustment (p=.0012; 801 AgP, 1476 controls). However, the association lost significance after correction for multiple testing in the replication. Schaefer et al. genotyped all candidate genes with the Illumina Immunochip. The authors noted that the lack of statistical power was not sufficient to provide evidence of the association of less frequent variants. Therefore, due to inappropriate sample recruitment (lack of a universally accepted diagnostic criterion for periodontitis, the small sample size studied, and not taking into account ethnicity differences and environmental effects) and the poor study designs used to investigate genetic susceptibility (CGAS approach with low sample size, not adjusting for covariates, no study of rare variants, and no replication of the results in an independent sample), few true genetic associations for PD have been identified. Conclusions There are a wide variety of definitions and variables analysed to establish a diagnosis of PD. In addition, studies on the genetic susceptibility of PD use different criteria for selecting participants. All these factors make the selection of a single methodological strategy for the analysis of genetic susceptibility to PD a challenging task. Despite their inability to identify new or unexpected loci associated with PD in the genome and their other limitations, candidate gene studies continue to be the most widely employed to analyse genetic susceptibility to PD. However, with advances in scientific knowledge and genotyping technologies, CGSs are gradually being replaced by GWASs. No optimal genotyping tool yet exists for the genetic study of PD. The genotyping technology to be used must be considered; but also, depending on the specific human population that will be studied, the SNPs selected by each platform must also be taken into consideration. Therefore, despite the intrinsic limitations of SNP arrays, those that enable evaluation of specific rare population variants could be the most suitable. However, no studies on PD with a GWAS design that take into account rare variants have been published to date. Future studies should therefore apply strict, validated diagnostic criteria for PD, using large study cohorts, adjusting the results according to the relevant risk factors, and including an analysis of multiple genes and polymorphisms in combination with rare variants. FUNDING This review was supported in part by University of Santiago de Compostela and by the Spanish Ministry of Education, Culture and Sport by a FPU grant (“Beca de Formación de Profesorado Universitario”). DECLARATIONS OF INTEREST: None REFERENCES Affymetrix, I. (2007). Data Sheet, Genome-Wide Human SNP Array 5.0. Retrieved from http://tools.thermofisher.com/content/sfs/brochures/genomewide_snp5_datasheet.pdf Affymetrix, & Inc. (2006). Data Sheet, GeneChip® Human Mapping 500K Array Set. Retrieved from https://tools.thermofisher.com/content/sfs/brochures/500k_datasheet.pdf Albandar, J. M., & Tinoco, E. M. B. (2002). Global epidemiology of periodontal diseases in children and young persons. Periodontology 2000, 29, 153–76. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12102707 American Academy of Periodontology Task Force Report on the Update to the 1999 Classification of Periodontal Diseases and Conditions. (2015). Journal of Periodontology, 86(7), 835–838. https://doi.org/10.1902/jop.2015.157001 Armitage, G. C. (1999). Development of a classification system for periodontal diseases and conditions. Northwest Dentistry, 79(6), 31–5. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/11413609 Atanasovska-Stojanovska, A., Popovska, M., Trajkov, D., & Spiroski, M. (2013). IL1 cluster gene polymorphisms in Macedonian patients with chronic periodontitis. Bratislavske Lekarske Listy, 114(7), 380–385. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/23822621 Auer, P. L., & Lettre, G. (2015). Rare variant association studies: considerations, challenges and opportunities. Genome Medicine, 7(1), 16. https://doi.org/10.1186/s13073-015-0138-2 Benjamin, S. D., & Baer, P. N. (1967). Familial patterns of advanced alveolar bone loss in adolescence (periodontosis). Periodontics, 5(2), 82–8. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/5228312 Berger, K. (2012). [DHS: The Dortmund health study]. Bundesgesundheitsblatt, Gesundheitsforschung, Gesundheitsschutz, 55(6–7), 816–21. https://doi.org/10.1007/s00103-012-1492-5 Brett, P. M., Zygogianni, P., Griffiths, G. S., Tomaz, M., Parkar, M., D’Aiuto, F., & Tonetti, M. (2005). Functional Gene Polymorphisms in Aggressive and Chronic Periodontitis. Journal of Dental Research, 84(12), 1149–1153. https://doi.org/10.1177/154405910508401211 Brookes, A. J. (1999). The essence of SNPs. Gene, 234(2), 177–86. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/10395891 Bugert, P., Lese, A., Meckies, J., Zieger, W., Eichler, H., & Klüter, H. (2003). Optimized sensitivity of allele-specific PCR for prenatal typing of human platelet alloantigen single nucleotide polymorphisms. BioTechniques, 35(1), 170–4. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12866418 Burgner, D., Jamieson, S. E., & Blackwell, J. M. (2006). Genetic susceptibility to infectious diseases: big is beautiful, but will bigger be even better? The Lancet. Infectious Diseases, 6(10), 653–63. https://doi.org/10.1016/S1473-3099(06)70601-6 Centre for Reviews and Dissemination, University of York. (2008). Retrieved November 16, 2017, from https://bmchealthservres.biomedcentral.com/articles/10.1186/s12913017-2703-4 Chai, L., Song, Y.-Q., Zee, K.-Y., & Leung, W. K. (2010a). Single nucleotide polymorphisms of complement component 5 and periodontitis. Journal of Periodontal Research, 45(3), 301–308. https://doi.org/10.1111/j.1600-0765.2009.01234.x Chai, L., Song, Y.-Q., Zee, K.-Y., & Leung, W. K. (2010b). SNPs of Fc-gamma Receptor Genes and Chronic Periodontitis. Journal of Dental Research, 89(7), 705–710. https://doi.org/10.1177/0022034510365444 Cirulli, E. T., & Goldstein, D. B. (2010). Uncovering the roles of rare variants in common disease through whole-genome sequencing. Nature Reviews. Genetics, 11(6), 415– 25. https://doi.org/10.1038/nrg2779 Cortes, A., & Brown, M. A. (2011). Promise and pitfalls of the Immunochip. Arthritis Research & Therapy, 13(1), 101. https://doi.org/10.1186/ar3204 de Jong, T. M. H., Jochens, A., Jockel-Schneider, Y., Harks, I., Dommisch, H., Graetz, C., … Schaefer, A. S. (2014). SLC23A1 polymorphism rs6596473 in the vitamin C transporter SVCT1 is associated with aggressive periodontitis. Journal of Clinical Periodontology, 41(6), 531–540. https://doi.org/10.1111/jcpe.12253 Divaris, K., Monda, K. L., North, K. E., Olshan, A. F., Lange, E. M., Moss, K., … Offenbacher, S. (2012). Genome-wide Association Study of Periodontal Pathogen Colonization. Journal of Dental Research, 91(7_suppl), S21–S28. https://doi.org/10.1177/0022034512447951 Divaris, K., Monda, K. L., North, K. E., Olshan, A. F., Reynolds, L. M., Hsueh, W.-C., … Offenbacher, S. (2013). Exploring the genetic basis of chronic periodontitis: a genome-wide association study. Human Molecular Genetics, 22(11), 2312–2324. https://doi.org/10.1093/hmg/ddt065 Dorak, T. M. (2006). Real-time PCR. (M. Tevfik Dorak, Ed.), Real-time PCR (ilustrada,). Oxford: Taylor & Francis. https://doi.org/10.4016/17251.01 Exome Chip Consortia. (2011). Exome Chip Design Wiki Site. Retrieved January 25, 2018, from https://genome.sph.umich.edu/wiki/Exome_Chip_Design Feng, P., Wang, X., Casado, P. L., Küchler, E. C., Deeley, K., Noel, J., … Vieira, A. R. (2014). Genome wide association scan for chronic periodontitis implicates novel locus. BMC Oral Health, 14(1), 84. https://doi.org/10.1186/1472-6831-14-84 Fiebig, A., Jepsen, S., Loos, B. G., Scholz, C., Schäfer, C., Rühling, A., … GrössnerSchreiber, B. (2008). Polymorphisms in the interleukin-1 (IL1) gene cluster are not associated with aggressive periodontitis in a large Caucasian population. Genomics, 92(5), 309–315. https://doi.org/10.1016/j.ygeno.2008.07.004 Flemmig, T. F. (1999). Periodontitis. Annals of Periodontology, 4(1), 32–37. https://doi.org/10.1902/annals.1999.4.1.32 Freitag-Wolf, S., Dommisch, H., Graetz, C., Jockel-Schneider, Y., Harks, I., Staufenbiel, I., … Schaefer, A. S. (2014). Genome-wide exploration identifies sex-specific genetic effects of alleles upstream NPY to increase the risk of severe periodontitis in men. Journal of Clinical Periodontology, 41(12), 1115–1121. https://doi.org/10.1111/jcpe.12317 Garlet, G. P., Horwat, R., Ray, H. L., Garlet, T. P., Silveira, E. M., Campanelli, A. P., … Silva, R. M. (2012). Expression Analysis of Wound Healing Genes in Human Periapical Granulomas of Progressive and Stable Nature. Journal of Endodontics, 38(2), 185–190. https://doi.org/10.1016/j.joen.2011.09.011 Gunji, T., Onouchi, Y., Nagasawa, T., Katagiri, S., Watanabe, H., Kobayashi, H., … Ishikawa, I. (2007). Functional polymorphisms of the FPR1 gene and aggressive periodontitis in Japanese. Biochemical and Biophysical Research Communications, 364(1), 7–13. https://doi.org/10.1016/j.bbrc.2007.09.105 Hart, T. C., Marazita, M. L., & Wright, J. T. (2000). The impact of molecular genetics on oral health paradigms. Critical Reviews in Oral Biology and Medicine: An Official Publication of the American Association of Oral Biologists, 11(1), 26–56. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/10682900 Hong, K.-W., Shin, M.-S., Ahn, Y.-B., Lee, H.-J., & Kim, H.-D. (2015). Genomewide association study on chronic periodontitis in Korean population: results from the Yangpyeong health cohort. Journal of Clinical Periodontology, 42(8), 703–710. https://doi.org/10.1111/jcpe.12437 Illumina. (2015). BeadArray Microarray Technology. Retrieved November 16, 2017, from https://www.illumina.com/science/technology/beadarray-technology.html Jurinke, C., van den Boom, D., Cantor, C. R., & Köster, H. (2002). The use of MassARRAY technology for high throughput genotyping. Advances in Biochemical Engineering/biotechnology, 77, 57–74. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12227737 Kennedy, G. C., Matsuzaki, H., Dong, S., Liu, W., Huang, J., Liu, G., … Jones, K. W. (2003). Large-scale genotyping of complex DNA. Nature Biotechnology, 21(10), 1233–1237. https://doi.org/10.1038/nbt869 Kinane, D. F., & Chestnutt, I. G. (2000). Smoking and periodontal disease. Critical Reviews in Oral Biology and Medicine: An Official Publication of the American Association of Oral Biologists, 11(3), 356–65. Retrieved from TABLES Table 1. Demographic characteristics of the study populations analysed in the articles included in the review. Reference Cases Controls Ethnic Group (Country) Smoking Diabetes Diagnosis n Origin Age (years) n Origin Age (years) (Suzuki et al., 2004) AgP 134 U and Col 40 ± 11 125 U and Col 40 ± 15 Asian (Japan) NS NS CP 117 U and Col 54 ± 9 (Brett et al., 2005) AgP* 51 H NS 100 Pop NS Caucasian (United Kingdom) NS NS CP* 57 H NS (Fiebig et al., 2008) AgP 415 U 30 ± 2 874 Pop 48 ± 4 Caucasian (Germany and The Netherlands) S S* (Gunji et al., 2007) AgP* 25 U 33 ± 6 349 U NS Asian (Japan) NS NS (Kobayashi et al., 2009a) PD* 117 DC, H and U 52 ± 1 108 NS 51 ± 1 Asian (Japan) S NS (Kobayashiet al., 2009b) AgP* 172 H 32 ± 6 303 H 38 ± 13 Asian (Japan) S S* CP* 147 H 53 ± 7 (Schaefer et al., 2010a) AgP 283 Pop 30 ± 5 972 H 51 ± NS Caucasian (Germany) S S (Schaefer et al., 2010b) AgP* 532 Pop 33 ± NS 1472 Pop 42 ± NS Caucasian (Germany and The Netherlands) S S CP* 805 Pop 54 ± NS 319 Pop 43 ± NS (Scapoli et al., 2010) AgP† 95 U and RC 43 ± 8 121 Col 30 ± 5 Caucasian (Italy) NS NS (Chai et al., 2010a) PD* 229 DC, H and U 44 ± 7 207 DC, H and U 42 ± 8 Asian (China) S S* (Chai et al., 2010b) CP‡ 190 DC, H and U 46 ± 5 169 DC, H and U 45 ± 6 Asian (China) S S* (Scapoli et al., 2011) AgP† 122 U and RC 43 ± 7 246 U 30 ± 8 Caucasian (Italy) NS NS (Schaefer et al., 2011) AgP* 159 Pop 34 ± 5 421 Pop 32 ± 9 Caucasian (The Netherlands) S S CP* 154 Pop 45 ± 10 Pop (Letra et al., 2012) CP 169 U and Col 50 ± NS 506 U and Col 46 ± NS Caucasian (Brazil and USA) NS NS (AtanasovskaStojanovska et al., 2013) CP* 114 U 39 ± 10 301 U 35 ± 10 NS (Macedonia) NS NS (Schaefer et al., 2013) AgP 600 Pop NS 1448 Pop NS Caucasian (Germany) S NS Table(s) Abbreviations: PD, periodontal disease; AgP, aggressive periodontitis; CP, chronic periodontitis; S, specified; NS, not specified. S*, specified as exclusion criterion; U, university; H, hospital; DC, dental clinic; RC, research centre; Col, College of Dentists; Pop, general population. *Armitage, 1999 (Armitage, 1999). †Lang et al., 1999 (Lang et al., 1999). ‡ Fleming, 1999 (Flemmig, 1999). Age is expressed as mean ± SD. (de Jong et al., 2014) AgP 283 Pop 30 ± 5 979 Pop 51 ± 13 Caucasian (Germany) S S (Feng et al., 2014) CP 712 U 57 ± NS 1973 U 55 ± NS Caucasian, African and other (Brazil and USA) S S (Schaefer et al., 2014) AgP 828 Pop NS 2119 Pop and U NS Caucasian (Germany and The Netherlands) S NS (Zhang et al., 2014) CP* 400 Pop 50 ± 9 750 Pop 50 ± 8 Asian (China) S* S (Schaefer et al., 2015) AgP 1042 U NS 3094 Pop NS Caucasian (Germany and The Netherlands) S S* (Shang et al., 2016) CP 471 H NS 1312 Pop NS Asian (China) S S* (Offenbacher et al., 2016) AgP 717 U NS 4210 Pop NS Caucasian (Germany) S S (Zupin et al., 2017) CP* 399 Pop NS 148 Pop NS Caucasian (Italy) S NS (Munz et al., 2017) AgP 1116 H and U NS 7179 Pop NS Caucasian (Germany, The Netherlands and Turkey) S NS Table 2. Clinical parameters recorded in publications included in the review. Abbreviations: R, recorded; NR, not recorded; PPD, probing pocket depth; CAL, clinical attachment loss; BOP, bleeding on probing. Reference Clinical parameters PPD CAL BOP Alveolar bone loss Others (Suzuki et al., 2004) R NR R R: On x-rays Degree of tooth mobility (Brett et al., 2005) NR NR NR NR X-rays and 6-point periodontal probing (Fiebig et al., 2008) NR NR NR R: On x-rays Number of teeth present (Gunji et al., 2007) R NR NR NR Number of natural teeth present (Kobayashi et al., 2009a) R R R NR Number of missing teeth, accumulation of supragingival plaque, mean distance from the free gingival margin to bottom of pocket and mean distance from the cement-enamel junction to bottom of pocket (Kobayashiet al., 2009b) R R R R: On x-rays (2 sites per tooth) Number of natural teeth, presence/absence of supragingival plaque (4 sites per tooth) (Schaefer et al., 2010a) NR NR NR R: On x-rays NR (Schaefer et al., 2010b) R R R R: On x-rays and orthopantomography Number of natural teeth and of exposed furcations (Scapoli et al., 2010) NR NR NR NR Loss of interproximal insertion > 2 mm on probing of any tooth (Chai et al., 2010a) R R R R: On orthopantomography NR (Chai et al., 2010b) R R R NR Number of natural teeth (Scapoli et al., 2011) NR NR NR NR Interproximal insertion loss > 2 mm on probing of any tooth (Schaefer et al., 2011) R R R R: On x-rays and orthopantomography Number of natural teeth and of exposed furcations (Letra et al., 2012) R R NR NR NR (Atanasovska-Stojanovska et al., 2013) R R R NR Löe and Silness gingival index (Schaefer et al., 2013) NR NR NR R: On x-rays NR (de Jong et al., 2014) NR NR NR R NR (Feng et al., 2014) NR R NR NR NR (Schaefer et al., 2014) NR NR NR R: On x-rays Number of natural teeth and of exposed furcations (Zhang et al., 2014) R R R R: Vertical or horizontal loss of alveolar bone Degree of gingival recession, tooth mobility and chronic gingival bleeding (Schaefer et al., 2015) NR NR NR R: On x-rays and orthopantomography NR (Shang et al., 2016) R R NR NR NR (Offenbacher et al., 2016) R R R NR Number of missing teeth, gingival index and plaque index (Zupin et al., 2017) R R R NR X-rays, number of teeth present, O´Leary plaque index and gingival recession (Munz et al., 2017) NR NR NR NR NR Table 3. Genome-wide association studies (GWASs) not included in the review and reasons for their exclusion. Reference Reasons for Exclusion (Divaris et al., 2012) Investigation of the susceptibility loci for colonisation with periodontal microbiota. (Divaris et al., 2013) The control group was formed of individuals with mild periodontitis. (Teumer et al., 2013) The control group was formed of individuals with mild periodontitis. (Rhodin et al., 2014) The control group was formed of individuals with mild periodontitis. (Shaffer et al., 2014) Not a case-control study. Analysis of two distinct phenotypes related to CP. (Freitag-Wolf et al., 2014) Selected only sex-specific SNPs. (Hong et al., 2015) The control group was formed of individuals with a healthy periodontium or mild periodontitis. (Shimizu et al., 2015) The control group was formed of individuals with systemic diseases (cerebral aneurysm, oesophageal cancer, endometrial cancer, chronic obstructive pulmonary disease or glaucoma). (Sanders et al., 2017) Not a case-control study. Table 4. Characteristics of the genetic analysis in the articles included in the review. Reference SNPs Studied (n) Genotyping Method Results Associated gene (polymorphism) OR 95%CI Association (Suzuki et al., 2004) 310 TaqMan COL1A1 (NA) 2.5 1.4-4.6 AgP COL4A1 (NA) 2.1 1.3-3.5 AgP PTGDS (NA) 5.1 1.5-18.2 AgP KRT23 (NA) 6.3 1.4-29.2 AgP IL6ST (NA) NA NA AgP CTSD (NA) 3.2 1.6-6.5 CP EGF (NA) 4.8 1.5-14.7 CP CTSD (NA) NA NA CP HSPG2(NA) NA NA CP COL17A1(NA) NA NA CP (Brett et al., 2005) 10 PCR IL1B(NA) 3.6 1.4-9.7 AgP IL6(NA) 3.1 1.5-6.3 AgP and CP VDR (NA) 2.7 1.4-5.4 CP TLR4(NA) 3.0 1.2-7.6 PD (Fiebig et al., 2008) 10 TaqMan ND ND ND ND (Gunji et al., 2007) 30 PCR-RFLP FPR1 (rs11666254) 2.0 NA AgP FPR1 (rs12460836) 2.0 NA AgP (Kobayashi et al., 2009a) 16 Novel nano-Invader DNA chip system ND ND ND ND (Kobayashi et al., 2009b) 35 Novel nano-Invader DNA chip system, (PCR)-Invader and TaqMan VDR (NA) 2.5 1.4-4.3 CP (Schaeferet al., 2010a) 500.568 Affymetrix Human Mapping 500k Array set and Affymetrix Genome-Wide Human SNP Array 5.0(controls of GWAS 2) GLT6D1 (rs1537415) 1.67 (GWAS 1) 1.65 (GWAS 2) 1.27-2.18 1.26-2.17 AgP (Schaefer et al., 2010b) 12 SNPlex and TaqMan DEFB1 (rs1047031) 1.3 1.1-1.6 AgP and CP (Scapoli et al., 2010) 70 Sequenom MassARRAY ND ND ND ND (Chai et al., 2010a) 11 iPLEX Sequenom MassARRAY C5 (rs17611) 6.1 1.3-28.2 PD (Chai et al., 2010b) 102 iPLEX Sequenom MassARRAY FCGR3A (rs445509) 1.0 0.5-0.8 CP (Scapoli et al., 2011) 28 Sequenom MassARRAY and PCR IL6(NA) 2.5 1.5-4.4 AgP (Schaefer et al., 2011) 51/37* SNPlex and TaqMan CDKN2B (rs3217992) 2.5 1.3-5.1 AgP CDKN2B-AS1 (rs518394) 2.3 1.1-4.9 AgP CDKN2B-AS1 (rs1360590) 2.6 1.3-5.2 AgP CDKN2B-AS1 (rs11790231) NA NA AgP CDKN2B-AS1 (rs1360590) 2.1 1.1-3.9 CP CDKN2B-AS1 (rs10965224 NA NA CP (Letra et al., 2012) 34 TaqMan TIMP1 (rs5906435) NA NA CP MMP3 (rs679620) NA NA CP MMP3 (rs650108) NA NA CP (AtanasovskaStojanovska et al., 2013) 22 PCR-SSP IL1B (NA) 2.1 1.4-3.3 CP (Schaefer et al., 2013) 268 Illumina Inmunochip Upstream of IL10 (rs61815643) 0.8 0.7-0.9 AgP Upstream of IL10 (rs6667202) 0.8 0.7-0.9 AgP Upstream of IL10 (rs4072226) 0.8 0.7-1.0 AgP (de Jong et al., 2014) 27 Affymetrix Human Mapping 500K Array set SLC23A1 (rs6596473) 1.3 1.0-1.5 AgP SLC23A1 (rs10063949) 1.4 1.0-1.9 AgP SLC23A2 (rs6990309) 1.5 1.1-2.1 AgP SLC23A2 (rs1519860) 1.6 1.1-2.3 AgP SLC23A2 (rs1715395) 1.4 1.0-2.1 AgP (Feng et al., 2014) 620.901/20† Illumina Human610-Quadv1_B BeadChip Intergenic (rs1477403) NA NA CP Intergenic (rs2833017) NA NA CP Intergenic (rs9305434) NA NA CP Intergenic (rs2048126) NA NA CP (Schaefer et al., 2014) ~3.000 Illumina Inmunochip IRF5 (rs62481981) 3.1 1.6-6.1 AgP PRDM1 (rs6923419) 0.7 0.6-0.9 AgP (Zhang et al., 2014) 23 iPLEX Sequenom MassARRAY IL8 (rs4073) 1.2 1.0-1.4 CP (Schaefer et al., 2015) 46 Illumina Inmunochip and Affymetrix Human Mapping 500K Array set PLG (rs4252120) 1.3 1.3-1.4 AgP (Shang et al., 2016) 65 Sequenom MassARRAY FBXO38 (rs10043775) 1.2 NA CP (Offenbacher et al., 2016) ~2.1 million/37ʹ Illumina Human BeadChips ND ND ND ND (Zupin et al., 2017) 203 Illumina Exome 12v1 NLRC5 (rs289723) 2.0 1.3-3.1 CP (Munz et al., 2017) 502.332/15˄ iPLEX Sequenom MassARRAY, Illumina Human DEFA1A3 (rs297851) 1.3 1.2-1.4 AgP Abbreviations: PD, periodontal disease; AgP, aggressive periodontitis; CP, chronic periodontitis; OR, odds ratio; CI, confidence interval; ND, no data; NA, data not available. * In the association study for AgP, 51 SNPs were analysed. In the association study for CP, 37 SNPs were analysed. † In the association study for CP, 620 901 SNPs were analysed. An independent test for CP was performed in two Caucasians populations (Porto Alegro and Río de Janeiro), analysing 20 SNPs. ʹ In the association study for CP, 2.135.235SNPs were analysed. An independent test for CP was performed in German sample, analysing 47 SNPs. ˄ In the discovery stage 502.332 SNPs were analysed. A replication was performed in a sample of Turkish descent, analysing 15 SNPs. OmniBeadChip and TaqMan DEFA1A3 (rs2738058) 1.3 1.1-1.4 AgP SIGLEC5 (rs4284742) 1.3 1.2-1.5 AgP SLC1A3 (rs1122900) 1.3 1.1-1.4 AgP Articles valid for full-text reading (n = 87) Full-text articles excluded, with reasons (n = 62) Eligibility Articles identified in the PubMed search (n = 2392) Articles identified in the Cochrane search (n = 47) Articles after duplicates removed (n =2412) Identification Included Articles identified after combining the two databases (n = 2439) Screening Duplicate articles (n = 27) Articles excluded during evaluation of the title/abstract, with reasons (n = 2325) Articles suitable for data extraction and inclusion in the review (n = 25) - Only abstract in English (n=126) - Reviews and/or meta-analysis (n=148) - Focusing on treatment (n=72) - Focusing on microbial data (n=680) - Expression studies (n=163) - Focusing in other pathologies (n=489) - Focusing on pregnancy or preterm birth (n=9) - Focusing on epigenetics (n=8) - Focusing on tools, functional studies, studies at cellular level, microsatellites and others (n=205) - Not case-control study (n=82) - <10 SNPs studied (n=159) - Group of cases and controls composed for < 100 individuals (n=168) - Groups studied with potential biases (n=16) - Reviews and/or meta-analysis (n=7) - Not case-control study (n=3) - Group of cases and controls composed for < 100 individuals (n=13) - <10 SNPs studied (n=33) - Groups studied with potential biases (n=6) Additional records identified through other resources (n = 0) Figure(s) Supplementary Material Click here to download Supplementary Material: Supplemental Material.docx