Inflammatory mediator polymorphisms associate with initial periodontitis in adolescents

Abstract Several studies have addressed cytokine gene polymorphisms and their possible associations with periodontitis. We examined the association between salivary anti‐ and pro‐inflammatory mediator polymorphisms and initial periodontitis in Finnish adolescents, taking into account the effect of smoking. Salivary samples of 93 clinically examined adolescents from Eastern Finland were analyzed. Their oral health and smoking habits were recorded. Periodontal probing depth (PPD), and bleeding on probing (BOP) at four sites per tooth, root calculus (RC), and visible plaque index (VPI) were recorded from the index teeth. Salivary MMP‐8 median values were assessed. The sites with ≥4 mm PD were categorized as follows: PPD1 = one or more ≥4 mm pocket, PPD2 = two or more ≥4 mm pockets, and PPD3 = three or more ≥4 mm pockets. Genomic DNA was extracted from 300 μl of the saliva samples by genomic QIAamp® DNA Blood Mini Kit and genotyped for polymorphisms. Genetic variants for genotyping were selected from the following genes of interest: S100A8, FCGR2A, FCGR2B, IL10, MMP8, MMP3, MMP13, VDR, TLR4, MMP2, MPO, ELANE, IL1A, IL1B, IL1RN, CD28, MMP9, DDX39B, NFKBIL1, LTA, TNF, SOD2, IL6, TLR4, TIMP1, and SYN1. After false discovery rate control (FDR), polymorphisms in MMP3 (rs679620, rs520540, rs639752), CD28 (rs3116496), and VDR (rs2228570) associated (FDR q < 0.05) with deepened periodontal pockets. Smoking did not affect the results. Genetic polymorphisms of pro‐inflammatory mediators MMP3, CD28, and VDR seem to link to initial periodontitis.

The aim of the present study was to examine the association between the polymorphisms in salivary anti-and pro-inflammatory mediators and initial periodontitis. We hypothesized that the salivary anti-and pro-inflammatory mediator polymorphisms are associated with initial periodontitis.  Figure 1 for study flow of the participants).
Oral health of subjects was clinically examined in both cohorts.
No periodontal therapy has been performed in the last recent year among these participants before this study, and after diagnostic procedures, they were treated by specialist in clinical periodontics. In short, periodontal parameters were recorded according to the WHO recommendations (World Health Organization, International Statistical Classification of Diseases and Related Health Problems, ).
Periodontal probing depth (PPD), bleeding on probing (BOP), and visible plaque index (VPI) were recorded at four sites per tooth, and all sites with ≥4 mm PD were recorded and categorized as follows: PPD1 = one or more ≥4 mm pocket, PPD2 = two or more ≥4 mm pockets, and PPD3 = three or more ≥4 mm pockets . The cut-point of the BOP value was 20 percentage meaning as gingivitis (Heikkinen, 2011;Heikkinen et al., 2008). Root calculus (RC) and VPI were recorded from index teeth.
Smoking habits (cigarettes per week indicated regular smoking) and pack-years were recorded (Heikkinen et al., 2012). Stimulated salivary samples were collected and centrifuged, and the supernatants were used for the enzyme measurements, such as MMP-8 analysis.
FIGURE 1 Study flow of the participants MMP-8 levels were analyzed by immunofluorometric assay (Medix Biochemica, Kauniainen, Finland). The interassay coefficient of variation (CV)% for this study was 6.3%, and detection limit for the assay was 0.08 μg/l (Heikkinen et al., 2010).

| Candidate gene and SNP selection
Genetic variants for genotyping were selected from the following genes of interest: S100A8, Fc fragment of IgG, low affinity IIa, receptor Three SNPs (rs3795391, rs2125685, and rs2544480) were excluded during the optimization phase of genotyping. Table 1 gives all the SNPs successfully genotyped in this study. The data was collected using the MassARRAY Compact System (Agena Bioscience), and the genotypes were identified using Typer 4 software (Agena Bioscience). For quality control reasons, the genotype calls were also checked manually and corrected when necessary.

| DNA extraction and genotyping
Genotyping quality was examined by a detailed QC procedure consisting of success rate checks, duplicated samples, gender check for X-chromosomal markers, water controls, and Hardy-Weinberg Equilibrium (HWE) testing.
Genotyping success rates ranged from 90.3% to 98.9%. All duplicate samples gave concordant results. One subject was excluded because of discrepant gender check results. Table 1 gives the genotype and allele frequencies and HWE p-values. Five SNPs turned out to be non-polymorphic (rs16062, rs315935, rs4251998, rs17216656, and rs1062849). Three SNPs (rs5004021, rs4252022, and rs17223045) with MAF < 0.01 were excluded. Rs5906435 and rs6520279 are X-chromosomal explaining their deviation from the HWE. All other SNPs were in HWE (p > 0.01). Finally, 93 subjects and 63 SNPs were included in the statistical analysis.

| Statistical analysis
The association between dichotomic outcome variables (salivary MMP-8 > 163 μg/l, median value, BOP > 20%, RC > median, PPD1, PPD2, and PPD3) and SNPs were modeled using logistic regression model. For continuous outcome (PPD total), linear regression model was used. We assumed additive effect of SNPs. For all outcomes, two models were calculated: unadjusted with SNP as only explanatory variable, and VPI, regular smoking (i.e., those who reported weekly smoking), and two separate occasions for birth cohorts adjusted model.
All data analyses were carried out using R language (http://www.Rproject.org) with package "SNPassoc"(R Development Core Team. R: A Language and Environment for Statistical Computing. Vienna, Austria, 2011). P values for each variable separately were corrected using false discovery rate (FDR) correction (Benjamini & Hochberg, 1995;Glickman, Rao, & Schultz, 2014). FDR q-values less than 0.05 were considered significant. Population stratification or relatedness was not addressed.

| Comparison of allele frequencies between populations
Allele frequencies of the SNPs with statistically significant findings (rs679620, rs520540, rs639752, rs3116496, and rs2228570) were compared with allele frequencies given in the ExAC database (http:// exac.broadinstitute.org/). Frequencies for SNP rs639752 were not found in the ExAC database, thus they were searched for HapMap-CEU from dbSNP database (http://www.ncbi.nlm.nih.gov/SNP/). In addition, the frequencies were compared with American population  (Table 2).

| RESULTS
Saliva MMP-8 median values were calculated and appeared to be 163.1 μg/l. However, no statistical significant findings were observed between salivary MMP-8 median values and SPNs, which were analyzed in this study (Table 1). Of the subjects (n = 94), 18 were healthy (BOP < 20% and no ≥ 4 mm pockets), 15 subjects had gingivitis (BOP ≥ 20% and no ≥ 4 mm pockets), and 61 had one or more ≥4 mm pocket (PPD1).
Fifty-one subjects had two or more ≥4 mm pockets (PPD2) and 37 had three or more ≥4 mm pockets (PPD3). Table 3 gives the basic characteristics of the study population.
Smoking or VPI did not affect these results. The significant results for the adjusted model are given in Table 4, and all association results (all phenotypes and both models) are given in the Supporting Information.
Allele frequencies of the five SNPs with statistically significant findings (rs679620, rs520540, rs639752, rs3116496, and rs2228570) were compared with allele frequencies in Finnish, European, and American populations (ExAC, HAPMAP-CEU, 1000 genomes AMR and HAPMAP-MEX). No major population specific allele differences were detected.

| DISCUSSION
The main finding of this study was that three MMP3 SNPs (rs67620, rs520540, and rs639752) were associated as exposure agents with three or more at least 4-mm deep periodontal pockets by the strict statistical analysis (FDR). Interestingly, the same SNPs have previously been significantly associated to chronic adult periodontitis (rs679620 in US Caucasians), rs520540 on a trend level in US Caucasians, and rs639752 in US Caucasians and Brazilians (Letra et al., 2012). Our results also showed that those with clinical signs of initial periodontitis were separated from the healthy ones using the same definitions as in our previous publications of the whole cohort (Heikkinen et al., 2010;Heikkinen, 2011;Heikkinen et al., 2008;Heikkinen et al., 2016).
Another finding of this study was that CD28 SNP rs3116496 [also known as +17(T/C)] major allele T associated with PPD3 in the adolescents. Previously, the polymorphism did not show any association with periodontitis in adults (e Silva et al., 2013). However, among non-smokers, they reported a higher frequency of the T − (CC) genotype in aggressive periodontitis compared with chronic periodontitis.
Furthermore, we observed that Vitamin D receptor SNP rs2228570 (also known as FokI) major allele C (also known as F allele) was protective to PPD3. In line, Naito et al. (2007) reported that heterozygous Ff individuals had a lower risk of severe chronic periodontitis than individuals without the F allele (Naito et al., 2007). On the contrary, Li et al. (2008) reported that F allele increased the susceptibility of aggressive periodontitis in Chinese (Li et al., 2008) and Park, Nam, and Choi (2006) reported that CC genotype associated with increased risk for generalized aggressive periodontitis in Koreans (Park et al., 2006). However, several studies have also reported no association between rs2228570 and periodontis (Wang, Zhang, & Chen, 2015;Wang et al., 2009;El Jilani et al., 2015;Tachi et al., 2003). Laine et al. (2012) pointed in their review article that an association could be found between periodontitis and haplotypes in theIL4 and IL6 and VDR genes (Laine et al., 2012). Earlier, Kornman et al.
(1997) observed that "severe periodontitis patients were accounted  for by either smoking or the IL-1 genotype" (Kornman et al., 1997). In our study, no such effects could be seen; in our study only 17% were regular smokers, and thus this potential confounding factor could not be detected. In this respect, cytokine gene polymorphisms are observed to be quite ambivalent.
However, as a limitation of this study of adolescents, the sample size was quite small. Majority of the participants were lost because in Finland it is difficult to obtain permission from adolescents and their parents for a genetic study. This study has several strengths such as comprehensive oral health examination, unique age group, and adolescents. Despite the strengths, the limited sample size needs to be acknowledged.
To conclude, our study was the first to investigate in an adolescent population the genetic background of pathogenesis of initial or early periodontitis. We found that the genetic polymorphisms in MMP3, CD28, or VDR gene seem to be important in this respect, otherwise than MMP8 gene, as might be assumed (Holla, Hrdlickova, Vokurka, & Fassmann, 2012). However, more investigations are needed in larger materials for final conclusion. Further, new point-of-care chair-side diagnostic tools have been developed to conveniently and consistently identify those adolescents with elevated risk for ongoing active gingivitis and periodontitis to be guided to regular examination and treatments . Nevertheless, our study hypothesis was confirmed by showing an association between genetic polymorphism and periodontal disease parameters and markers.