The role of three interleukin 10 gene polymorphisms (− 1082 A > G, − 819 C > T, − 592 A > C) in the risk of chronic and aggressive periodontitis: a meta-analysis and trial sequential analysis

Relevant studies were searched in the health-related electronic databases of PubMed, EMBASE, web of science and google scholar, with the combination of the keywords with Boolean operators: “periodontitis” OR “chronic periodontitis” OR “aggressive periodontitis” AND “interleukin-10” OR “IL10” OR “-1082 A/G”. OR “-819 C > T” OR “-592 A > C”.

The search was limited to the studies published in English and Chinese (abstract in English) from 1968 until March 2018. The references of retrieved articles and relevant reviews were checked for any additional studies.

Human studies of any sample size that assessed periodontitis were included, if:

CP and AP in the current analysis were as defined in the primary studies. Studies, which did not meet the inclusion criteria were excluded. Studies based on family or sibling-pairs were also excluded.

Three investigators (HCW, YO, CN) individually checked the titles and abstracts, and then selected the relevant full-text articles, according to the inclusion criteria. The three investigators independently extracted data, using a piloted data extraction form. The following information were collected from each study: author, publication year, country, design (e.g. hospital-based study, population-based study), the number of cases and controls, the racial descent (Asian or non-Asian), clinical types of periodontitis, genotyping method and genotype/allele distribution in both cases and controls. Any discrepancy was resolved by discussion and consensus. If allele frequency was zero, then a value of 1 was added to all allele, following Laplace approximation [15].

The two investigators (HCW, YO) independently assessed the methodological quality of studies, using the Newcastle-Ottawa Scale (NOS) for quality assessment [16], with necessary modifications. The scores were based on three main aspects: the selection of the study groups (maximum 4 points: whether the case definition was adequate with independent validation; whether the patients were consecutive or obviously representative series of cases; whether the controls were population-based; whether the controls were persons without history of periodontitis); the comparability of the groups (maximum 2 points: whether cases and controls were matched for age and gender; whether cases and controls were matched for region or ethnicity); and the ascertainment of exposure (maximum 3 points: whether the ascertainment of exposure was secure record; whether the same method of ascertainment was applied for cases and controls; whether the same non-response rate was reported for both groups). A total score for each study can vary from 0 (worst) to 9 (best). Studies scoring ≤4 were regarded as low quality, while a score of ≥7 was considered as a high-quality study. Any discrepancy between the two investigators was resolved by discussion and consensus.

Hardy-Weinberg Equilibrium (HWE) in the controls was tested, using the exact test for goodness-of-fit and p > 0.05 was regarded as consistent with HWE [17]. The strength of the association between IL10 (− 1082 A > G,-819 C > T, − 592 A/C) and the risk of periodontitis (CP or AP) was estimated using odds ratio (OR) and its 95% confidence interval (CIs). Heterogeneity was statistically assessed using the I² test. The I² test values describes the percentage of total variation across studies that is attributable to the heterogeneity rather than chance. I² values greater than 50% is considered as substantial heterogeneity [18]. For pooling of the estimates across studies included, the summary ORs and corresponding 95% CIs were calculated with random-effects model (The Der Simonian and Laird method) in the presence of between–study heterogeneity of the studies. Otherwise, fixed-effect model (the Mental-Haenszel method) was used. We estimated the pooled ORs and its 95% CIs in four genetic models: − 592 A > C the allelic contrast model (A vs C), the recessive model (AA vs CA + CC), the dominant model (AA+CA vs CC) and the additive model (AA vs CC). As for − 1082 A > G the allelic contrast model (A vs G), the recessive model (AA vs GA + GG), the dominant model (AA+GA vs GG) and the additive model (AA vs GG). As for − 819 C > T the allelic contrast model (T vs C), the recessive model (TT vs CT + CC), the dominant model (TT + CT vs CC) and the additive model (TT vs CC) was used. Analysis was further stratified by the clinical forms of periodontitis (CP, AP) and by ethnic decent. To assess the stability of results, the sensitivity analysis was done with leave-one-out meta-analysis by sequential omission of individual studies [19]. We assessed the publication bias by visualizing funnel plots [20, 21].

Trial sequential analysis (TSA), an approach that adjust for random error risk, was done for estimation of the required information size [22]. It is classified as ‘potentially spurious evidence of effect’, if the cumulative Z-curve did not cross the monitoring boundaries or as ‘firm evidence of effect’, if the cumulative Z-curve crossed the monitoring boundaries [23].

Meta-analysis was done with RevMan 5.3 (The Cochrane collaboration, Copenhagen) and Leave-one-out meta-analysis was with meta package (metabin command) in R 3.4.3 software (The R Foundation). TSA was done with TSA version 0.9 beta (Copenhagen Trial Unit, Centre for Clinical Intervention Research, Copenhagen).

Overall, there was statistical significant associations between IL10–1082 A > G polymorphism and the risk of AP in a subgroup of the non-Asian population under the recessive model (OR,1.42;95% CI,1.11–1.8,I²:43%) (Fig. 2). For IL10–592 A > C, only the non-Asian populations under the allele contrast model (C vs A) (OR,4.34; 95%CI:1.87–10.07, I²:65%) and the recessive model (OR,2.1;95% CI,1.16–3.82,I²:0%) showed significant relationship between this SNP and the risk of AP (Fig. 3). For IL10–819C > T, there was no significant associations between this SNP and the risk of CP/AP in any population groups under any of four genetic models (Fig. 4). Summary of the evaluation of the three candidate SNPs stratified by population groups and clinical types under four genetic models are provided in Additional file 5.

To assess the influence of individual data on the pooled ORs, leave-one-out meta-analysis was performed. The overall estimate of recessive model of IL10–1082 A > G remained statistical significance, while omitting any single study (Fig. 5). This indicates that the results are stable. This was true for all other models with the 3 candidate SNPs. As an example, a subgroup analysis stratified by diagnostic criteria was done on CP patients with − 1082 A > G under recessive model: CP patients diagnosed only by clinical criteria was significantly associated with this SNP (Additional file 6). This implied that the diagnostic criteria for CP had impact on the direction of association. Begg’s and Egger’s test showed no evidence of publication bias in all analyses concerning IL10 (− 1082 A > G, − 819 C > T, 592C > A) polymorphisms. An example illustration with IL10–1082 A > G under dominant model is shown in Additional file 7.

We performed TSA with studies on IL10–1082 A > G in CP that followed HWE. We used an overall 5% risk of a type I error and 20% risk of a type II error (power of 80%). The cumulative z-curve crossed the traditional boundary and the trial sequential monitoring boundary and reached the required information size, suggesting there is no need for more evidence to establish additional study of − 1082 A > G in CP among the non-Asian population (Additional file 8).

There are some limitations in the present study. Only two studies included in this meta-analysis (11.7%) were from the Asia population. A small sample size in the Asian population subgroup analyses may not be the representative of the population. Hence, a selection bias with geographical imbalance is a concern. Our analysis was done with the unadjusted raw data provided in the primary studies, in which patients might have some common factors (i.e. age, gender, diets, smoking habits, diabetes). An earlier meta-analysis had highlighted that the inclusion of both smoking and non-smoking subjects in the primary studies can be an additional source of variability [54]. Stratified analysis by age group, smoking habits, oral hygiene habits or presence of comorbidity were not possible due to limited data. Hence, the findings could be confounded with these common factors. A stratification analysis on − 1082 A > G in the CP patients documented that the estimates could vary with the type of diagnosis criteria. If cases and controls have been genotyped in separate batches in the primary studies, differential misclassification of exposure is a concern. The sample sizes in most of the included studies were relatively small and were under power to detect statistically significant differences between the cases and controls. Meta-analysis, however, is a retrospective synthesis of published studies and power analysis is not applicable and type II errors are expected to be less common in a meta-analysis than in single studies [18, 19]. The manifestation of many systemic diseases which may lead to compromised host function should also be considered. For example, diabetes and rheumatoid arthritis are examples of diseases which may have a genetic component and may have enhanced periodontal breakdown as a secondary feature. In the presence of publication bias in this analysis, non-published studies and/or studies in other language might have been missed in the current review. Moreover, there might be some extent of interaction with IL10 (− 1082 A > G, − 819 C > T, IL10 592 A > C) and other genes (gen-gen interaction/synergism). Hence, findings in the current meta-analysis should be interpreted with caution.

Interventions with the manipulation of anti-inflammatory cytokines have been suggested as adjuvants for treating periodontal disease. The efficacy of anti-cytokine biotherapies in patients with inflammatory diseases is a proof that blocking the effects of a cytokine can slow down the disease process [59]. Investigating the association of the IL10 gene polymorphisms with periodontitis susceptibility may promote our understanding of its pathogenesis and explain individual differences in the risk.