Contribution of Interleukin-10-592 (-590, -597) C>A Polymorphisms to Periodontitis Susceptibility: An Updated Meta-Analysis Based on 18 Case-Control Studies

Introduction The association between interleukin-10- (IL-10-) 592 (-590, -597) C>A polymorphisms and susceptibility to chronic or aggressive periodontitis (CP or AgP) is conflicting. This meta-analysis is aimed at quantitatively estimating the association. Materials and Methods PubMed, Embase, Web of Science, and WANFAN were searched for studies performed prior to January 31, 2018, to collect data for our research. Meta-analysis was performed using RevMan 5.3 or STATA 14.0. Results In total, 18 studies that met our criteria were included. Overall or HWE subgroup analysis of individuals with this polymorphism revealed that in terms of CP susceptibility, there was a significant difference between case groups and control groups in the A allele versus C allele model (OR = 1.38, 95% CI = 1.17–1.64 or OR = 1.38, 95% CI = 1.12–1.70), in the AA versus CC+CA model (OR = 1.49, 95% CI =1.06–2.10 or OR = 1.42, 95% CI = 1.13–1.78), and in the CC versus CA+AA model (OR = 0.69, 95% CI = 0.51–0.92 or OR = 0.68, 95% CI = 0.49–0.93); subgroup analysis based on a nonsmoking population also displayed significance in the A allele versus C allele model (OR = 1.43, 95% CI = 1.15–1.79) and CC versus CA+AA model (OR = 0.62, 95% CI = 0.44–0.87). For this polymorphisms and AgP susceptibility, our analyses revealed a significant association in both the A allele versus C allele model (OR = 1.29, 95% CI = 1.01–1.63) and the AA versus CC+CA model (OR = 1.93, 95% CI = 1.30–2.89); subgroup analysis based on Caucasian or nonsmoking populations showed significant differences in the AA versus CC+CA model (OR = 6.29, 95% CI = 1.78–22.21 or OR = 3.24, 95% CI = 1.59–6.61). Conclusions IL-10-592 (-590, -597) A allele and the associated AA genotype may be risk factors for the onset of CP or AgP—particularly for the AA genotype and the increased risk of AgP in Caucasian or nonsmoking populations. Conversely, the CC genotype may act as a protective factor against the onset of CP.


Introduction
It was previously reported that specific genetic factors can account for as much as 50% of the overall onset risks of periodontal disease [1,2]. Recently, gene polymorphisms in some cytokines such as interleukin-(IL-) 1, IL-6, TNF-α, IL-8, and IL-10 have been shown to play a vital role in the occurrence of both chronic and aggressive periodontitis (CP and AgP) immune pathogenesis [3][4][5][6][7].

Materials and Methods
The meta-analysis was performed in accordance with the PRISMA-P (preferred reporting items for systematic reviews and meta-analysis protocols) statement which was recommended for the establishment of a systematic review and meta-analysis [34].

Focused Question.
Is there an association between IL-10-592 (-590, -597) gene variations and CP or AgP? 2.2. Inclusion Criteria. The inclusion criteria that published studies need to meet to be included in the meta-analysis were as follows: (1) case-control studies; (2) the case groups consisted of patients diagnosed with CP or AgP, and the control groups consisted of periodontally healthy individuals; (3) IL-10-592 (-590, -597) C>A polymorphisms were detected, and sufficient data regarding genotype distributions were provided for the calculation of odds ratio (ORs) and corresponding 95% confidence intervals (95% CIs); and (4) studies with no repeated data. Studies that did not meet each of these criteria were excluded from the meta-analysis.

Search Strategy.
A systematic literature search for studies published up to January 31, 2018, was performed using the electronic databases, PubMed, Embase, Web of Science, and WANFAN. In addition, the reference lists of the selected manuscripts and related reviews were also manually screened for comprehensive results. The search strategies were presented as follows.

Study Selection and Data
Extraction. For study selection, duplicate studies or datasets were firstly removed from the included titles using EndNote software. Then titles and abstracts of the remaining results were screened, followed by full-text paper screening according to the inclusion criteria described above (Figure 1). The results were screened by two authors independently, and a third author (JL Song) was consulted if any discrepancies existed. The following characteristics were extracted from the included studies by two authors independently, and discrepancies were resolved through discussion as follows: (1) the name of the first author and year of publication, (2) country (or district) and ethnicity of study participants, (3) group size, (4) smoking status, (5) gender ratio comparability, (6) type of controls, (7) genotype distribution, and (8) the Hardy-Weinberg equilibrium (HWE) results for the controls.
2.5. Quality Assessment. The Newcastle-Ottawa Scale (NOS) was used to assess the quality of the included case-control studies, which was performed by two authors independently.
The composition of NOS includes three sections for consideration, which were "Selection" (0-4 points), "Comparability" (0-2 points), and "Exposure" (0-3 points). For the "Comparability" chapter, smoking status, age, and sex were selected as the main confounding factors to be matched in the present study. If two out of three factors were matched, one point (asterisk) was scored; if all three factors were matched, two points were assigned. The final scores were calculated ranging from 0 to 9. Studies with scores of 0-3, 4-6, and 7-9 points were considered of low, moderate, and high quality, respectively [6,35].
2.6. Data Analysis. The ORs and 95% CIs were calculated to evaluate the association between IL-10-592 (-590, -597) polymorphisms and susceptibility to chronic or aggressive periodontitis. Heterogeneity between studies was estimated by χ 2 and I 2 . An I 2 > 50% or P < 0 05 was considered to have significant heterogeneity. Next, the Mantel-Haenszel random effects model was used to assure the pooled efficiency. Otherwise, the Mantel-Haenszel fixed-effects model was used. The following three genetic models were applied for the meta-analyses of the IL-10-592 (-590, -597) polymorphisms: (1) allele comparison, (2) the dominant model, and

Characteristics and Quality Assessment of Included
Studies. A total of 18 articles were included in the current meta-analysis on the gene IL-10-592 (-590, -597) C>A (rs1800872) polymorphisms. The selection process for including publication articles is presented in the flowchart ( Figure 1). The characteristics of included studies are presented in Table 1. As shown in Table 1 [29], Toker et al. [30,33], and Moudi et al. [32], the genotype distributions in the control groups of the other studies were consistent with HWE. The scores of NOS ranged from 4 to 8. Nine studies were considered to be of high quality [16, 17, 22-24, 29, 30, 32, 33], and the other 9 studies were classed as moderate quality [18-21, 25-28, 31] (Table 2, e- Table 1).
When we considered only the Caucasian subgroup, we found no significant differences in the A allele versus C allele model, in the AA versus CC+CA model, or in the CC versus CA+AA model (Table 3). Conversely, in the nonsmoker subgroup, we did find significant differences in both the A allele versus C allele model (OR = 1.43, 95% CI = 1.15-1.79) and CC versus CA+AA model (OR = 0.62, 95% CI = 0.44-0.87) ( Table 3, Figures 3(a) and 3(c)).

Meta-Analysis
Results of Association between IL-10-592 (-590, -597) C>A Polymorphisms and the Risk of AgP. Six studies were involved, comprising 288 cases and 399 controls. The results are summarized in Table 3. The overall analyses of these studies yielded significant estimates in the A allele versus C allele model (OR = 1.29, 95% CI = 1.01-1.63) and in the AA versus CC+CA model (OR = 1.93, 95% CI = 1.30-2.89) ( Table 3, Figures 4(a) and 4(b)) but no significant estimates in the CC versus CA+AA model (Table 3). Subgroup analyses by ethnicity showed significant trends in the Caucasian population in the A allele versus C allele model (OR = 1.55, 95% CI = 0.97-2.48) and significant differences in the AA versus CC+CA (OR = 6.29, 95% CI = 1.78-22.21) ( Table 3, Figure 4(c)), but no significant estimates were found in the CC versus CA+AA model ( Table 3). The nonsmoker subgroup analysis did, however, reveal significant differences in the AA versus CC+CA model (OR = 3.24, 95% CI = 1.59-6.61) (Table 3, Figure 4(d)).

Sensitivity Analysis.
To assess the effect of an individual dataset on pooled ORs, a sensitivity analysis was performed through the sequential omission of each study. The results suggested that no single study greatly influenced the pooled estimations under any of the three genetic models for CP (e-Tables 2-4).

Publication Bias.
Egger's test proved that there was no significant publication bias except HWE fulfillment or Caucasian population subgroup analysis in the CP versus controls allele comparison (Table 3). There was also not any obvious evidence of publication bias by Egger's test in overall and subgroup analysis in the AgP versus controls alleles and genotypes comparison (Table 3). Investigations of the correlations between IL-10-592 (-590, -597) C>A polymorphisms and CP risk suggested that IL-10-592 (-590, -597) A allele and AA genotype may increase the risk of CP, while CC genotype provides increased protection against the risk of the disease. These results were expected because 11 out of 16 individual studies included in our meta-analysis presented these trends in their populations. Furthermore, sensitivity analysis revealed no quantitative changes for the interstudy heterogeneity, suggesting that these results were stable and trustworthy.

Discussion
It was demonstrated that the associations between genetic polymorphisms and certain diseases varied with different geographical regions and ethnic groups [6]. Therefore, we also investigated the contributions of ethnicity to the risk of disease. Our results suggested that the IL-10-592 (-590, -597) A allele and AA or CC genotypes were not associated with any increased risk that Caucasian individuals may have toward CP. These results varied from the meta-analysis results reported by Zhong et al. [14] and Albuquerque et al. [15] in which they found out that the A allele might increase the risk for CP, and Albuquerque et al. [15] found out the CC genotype might resist the risk among Caucasians. We hypothesize that part of the reason for this variation is that our meta-analysis included four additional studies that yielded inconsistent results [22,25,26,30].
For IL-10-592 (-590, -597) C>A polymorphisms and AgP, our meta-analysis results indicated that the A allele may confer a relative increase in the risk for developing AgP, especially in Caucasian individuals, as described in the meta-analysis results reported by Albuquerque et al. [15]. The same conclusions can be drawn in relation to the comparison AA versus AC/CC genotypes, which is in contrast to the results reported by Albuquerque et al. [15].
Smoking may increase the risk of periodontitis onset [36][37][38], so we performed nonsmoker subgroup analyses  [16], and Silveira et al. [27] did not show these trends. Our present meta-analysis also did not show significantly positive results, suggesting that the AA genotype might not be the risk for CP in a nonsmoking population. However, the AA genotype might increase the risk for AgP onset in nonsmoker population because our metaanalysis results showed a threefold greater difference between the case and control groups under the AA versus AC+CC model for the AgP risk. These results were partially expected, as three related studies performed by Silveira et al. [27], Gorgun et al. [29], and Toker et al. [30] all showed the risk trend.

Control
Odds ratio Odds ratio

Case Study or subgroup
Reichert et al. [19] Toker et al. [30] Total (95% CI)  distributions from the HWE [6]. There was obvious heterogeneity in the overall analysis involving CP individuals. We removed the studies from our analysis that deviated from HWE in the controls and performed the analysis again. Using these refined parameters, we still found that the results displayed higher heterogeneity in the A allele versus C allele model, as well as in the CC versus AC+AA model. Subgroup analysis based on Caucasian race or nonsmoking population also displayed higher heterogeneity. Further analysis found out that in the A allele versus C allele model, there was observed heterogeneity in the Atanasovska-Stojanovska et al. study [22], whose control sample sizes was larger, and from the Lopes et al. study [28], in which the pooled OR values were more than two, indicating over two times increased susceptibility to CP. When we concurrently removed the above two studies, the I 2 value decreased to less than 50%. In the CC versus AC+AA model, the main heterogeneity may have resulted from inclusion of the Atanasovska-Stojanovska et al. study [22], which had larger sample sizes in the control group, the Lopes et al. study [28] with zero CC individual in the case group, the Hu et al. study [20] with differential result, and the Scapoli et al. study [26], which had larger sample sizes in its case group. When these studies were removed, the I 2 value also decreased to less than 50%.
Another factor that we considered in our analysis was publication bias. Publication bias stems from the fact that positive results are much more readily published by journals, whereas negative results tend to be poorly received by journals and are collectively known as "grey literature" [6,14,15]. In the present study, we used Egger's test to probe the occurrence of publication bias, and the results indicated apparent publication bias in HWE fulfillment, as well as in the Caucasian population subgroup analysis in the CP versus controls allele comparison, which may have distorted our present results. We observed the funnel plot asymmetry of the two contrasts and found Atanasovska-Stojanovska et al. study [22], which contained a larger sample size, and Reichert et al. study [19], which comprised a smaller sample size, caused the publication bias. After excluding these two studies, the P value of Egger's test increased to over 0.05. However, the corresponding pooled OR values were not substantially altered without the publication bias.
Although a broad search in four different databases was used to find studies for inclusion in our meta-analysis, it is impossible to confirm that all available studies addressing the relationship between IL-10-592 (-590, -597) C>A polymorphisms and periodontitis were included, presenting another major limitation of the meta-analysis. haplotype has been associated with decreased synthesis of IL-10 and is frequently associated with periodontitis [18]. Owing to the linkage disequilibrium, the presence of those haplotypes can be fully determined by the analysis of the IL-10-592 (-590, -597) C>A polymorphism, in which the occurrence of the A allele indicates the presence of the [ATA] haplotype [18]. Our findings showed that IL-10-592 (-590, -597) A allele or AA genotype existed extensively in CP and AgP populations, especially in Caucasian AgP populations. Therefore, the IL-10-592 (-590, -597) A allele or AA genotype may be a putative biomarker for the diagnosis of CP and AgP. We suggest that when patients are initially diagnosed with periodontitis clinically, testing for IL-10-592 (-590, -597) polymorphisms may be helpful in confirming diagnosis. Doctors and dentists may also routinely consider monitoring the IL-10-592 (-590, -597) A allele or AA genotype in healthy population to prevent the occurrence of CP and AgP by recommending prophylactic measures. Such prophylactic measures include no smoking, regularly seeing dentists for professional examination, removal of microbial biofilm, and so forth, brushing teeth twice daily for 2 minutes with a soft toothbrush, brushing the tongue, cleaning the interdental spaces with interdental aids (such as floss or interproximal brushes), using a fluoride toothpaste, and having a balanced diet, among others [39].

Conclusion
Even considering the limitations of this study, the present meta-analysis supported the hypothesis that IL-10-592 (-590, -597) C>A polymorphisms may be associated with CP and AgP susceptibility. We not only identified that the IL-10-592 (-590, -597) A alleles and AA genotypes may be a risk factor for the development of CP and AgP but also found out that the IL-10-592 (-590, -597) CC genotype may play a protective role in preventing CP. It is noteworthy that the AA genotype was found to be more closely tied to the risk of AgP in Caucasian and nonsmoker population. Thus, IL-10-592 (-590, -597) A alleles or AA genotypes may be a putative biomarker for diagnosing CP and AgP. Large-scale studies to further validate our findings should be performed in the future.

Supplementary Materials
e- Table 1: Newcastle-Ottawa quality assessment scores for the studies included in the meta-analysis. e- Table 2: sensitivity analysis * for the A allele versus C allele model and CP risk. e- Table 3: sensitivity analysis * for the AA versus AC+CC model and CP risk. e- Table 4: sensitivity analysis * for the CC versus AA+AC model and CP risk. (Supplementary  Materials)