Periodontal Therapy and Systemic Inflammation in Type 2 Diabetes Mellitus: A Meta-Analysis

Aim The aim of this systematic review was to assess the effect of periodontal therapy (PT) on serum levels of inflammatory markers in people with type 2 diabetes mellitus (T2DM). Methods of Study Selection A literature search was carried out using MEDLINE via Pubmed, EMBASE, LILACS and Cochrane Central Register of Controlled Trials (CENTRAL) databases. Randomized-controlled trials (RCTs) and controlled clinical trials (CCTs) evaluating the effect of PT on systemic inflammatory markers were deemed eligible. Case series (CS), reports and pilot trials were excluded. Study quality was assessed using the Cochrane Collaboration’s risk assessment tool. Meta-analysis was carried out using random effect methods. Results The search strategy identified 3,164 potential studies of which 61 were assessed for eligibility and 9 (6 RCTs and 3 CCTs) were included in this systematic review. Three RCTs were classified by the authors as being at low risk of bias and three were “unclear” and classified as uncertain risk of bias. All CCTs were considered to be at a high risk of bias. The meta-analysis showed a statistically significant mean difference (MD) for TNF- α (-1.33 pg/ml, 95% CI: -2.10; -0.56, p<0.001) and CRP (-1.28 mg/l, 95% CI: -2.07; - 0.48, p<0.001) favoring periodontal intervention versus control. Conclusion The results of this meta-analysis support the hypothesis that PT reduces serum levels of TNF- α and CRP in T2DM individuals. The decrease of inflammatory burden has important implications for metabolic control and can, in part, explain the mechanisms linking periodontitis and increased risk for complications in people with T2DM.


Results
The search strategy identified 3,164 potential studies of which 61 were assessed for eligibility and 9 (6 RCTs and 3 CCTs) were included in this systematic review. Three RCTs were classified by the authors as being at low risk of bias and three were "unclear" and classified as uncertain risk of bias. All CCTs were considered to be at a high risk of bias. The metaanalysis showed a statistically significant mean difference (MD) for TNF-α (-1.33 pg/ml, 95% CI: -2.10; -0.56, p<0.001) and CRP (-1.28 mg/l, 95% CI: -2.07; -0.48, p<0.001) favoring periodontal intervention versus control.

Conclusion
The results of this meta-analysis support the hypothesis that PT reduces serum levels of TNF-α and CRP in T2DM individuals. The decrease of inflammatory burden has important implications for metabolic control and can, in part, explain the mechanisms linking periodontitis and increased risk for complications in people with T2DM.

Introduction
Periodontitis is a chronic inflammatory/infectious disease, characterized by the loss of both soft and hard tooth supporting tissues [1]. Dental biofilm is the main etiological factor for oral production of inflammatory biomarkers [2,3], although several biological mechanisms may be involved. Local production of cytokines in response to periodontal bacteria and their products is related to higher serum concentrations of proinflammatory biomarkers [4]. Periodontal treatment (PT) is associated with a reduction in inflammatory burden in patients with periodontitis [5,6]. Type 2 diabetes mellitus (T2DM) is a disorder with both metabolic and vascular components. It is characterized by hyperglycemia due to defective insulin function or altered insulin cell receptors (insulin resistance) rather than its deficiency [7]. Many factors contribute to the onset and development of diabetes complications, such as genetics, diet, lifestyle, age and obesity [8,9]. Studies indicate that control of inflammatory processes may be related to novel approaches in treating this disorder [10][11][12]. T2DM is a recognized risk factor for periodontal diseases (PD) [13,14], affecting their prevalence, incidence and severity [15]. Nevertheless, the relationship between both disorders is bidirectional [14][15][16][17]. Altered systemic inflammatory response has been recognized in both PD and T2DM [18]. PD may therefore represent an additional factor contributing to the total inflammatory burden in people with T2DM [19][20][21]. Recent systematic reviews reported beneficial effects of PT on the glycemic control in individuals with T2DM [22][23][24], although the largest clinical trial published to date was inconclusive [25], having limited focus on other systemic outcomes (i.e., inflammatory serum markers). The aim of this systematic review (SR) was to answer the following question: can PT affect serum inflammatory biomarkers in people with PD and T2DM?

Materials and Methods
The protocol of this review was registered at the National Institute for Health Research PROS-PERO, International Prospective Register of Systematic Reviews (http://www.crd.york.ac.uk/ Prospero, registration 4 number CRD42012002988). We designed the protocol in accordance to Cochrane standards for analysis and reporting of methods. The search criteria met the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) guidelines [26].

Criteria for considering studies for this review
Type of studies. Randomized controlled trials (RCTs) were considered the most appropriate study design to answer the research question for this review. Owing to the limited number of RCTs found, controlled clinical trials (CCTs) were also included.
Type of participants and inclusion/exclusion criteria. Participants were included if they were diagnosed with type 2 diabetes (according to the WHO criteria for diagnosis: fasting plasma glucose 126 mg/dl and/or 2h post-glucose challenge of 220 mg/dl). Further inclusion criteria were the following: (1) Participants having received PT with at least 3 months of followup; (2) assessment of serum inflammatory biomarkers related to insulin resistance; and (3) at least 30 individuals included in the type 2 diabetes group. All other study designs, such as case series reports and pilot studies, were excluded from this review. Trials in which inflammatory markers were not available for analysis (when original values could not be retrieved after contacting the original authors) were not eligible for inclusion.
Type of periodontal interventions. The periodontal interventions were based on professional oral hygiene instructions; full-mouth scaling and root planing (supra/subgingival biofilm and calculus removal) (SRP); surgical procedures (i.e., periodontal flap surgery) and SRP plus local or systemic antimicrobial. Periodontal interventions were compared with the passive option (no periodontal treatment).

Types of outcome measures
Primary outcome. Serum laboratory markers of inflammation following PT. Secondary outcome. Occurrence of adverse effects related to PT.

Search Strategy
Identification of studies for this SR was performed through detailed search strategies developed for each database searched (MEDLINE, EMBASE, LILACS and Cochrane central register of controlled trials). Articles in the English language were included by searching MEDLINE via Pubmed (1950 to November 2013), the Cochrane Central Register of controlled trials (CEN-TRAL), EMBASE via Ovid (1980 to November 2013), and LILACS (1982 to November 2013). Reference lists of previous reviews and identified studies were manually examined (hand search) in an attempt to identify any additional articles. The databases Clinical Trials Gov and OpenGray were also searched to identify unpublished trials. Authors of the included manuscripts were contacted for missing data as needed.

Quality of evidence
Quality assessment of selected studies was performed with the Cochrane Collaboration tool for assessing risk of bias [27].

Diagnostic assessment
Adequate diagnosis of type 2 diabetes was defined when recorded by medical staff using the WHO diagnostic criteria 2006 [28] defined as fasting plasma glucose 126 mg/dl and/or 2h post-glucose challenge of 220 mg/dl. Diagnosis was deemed inadequate if made through other methods, and unclear when the methods used were not clear or not reported in the studies. Due to the heterogeneity of criteria and case definitions of periodontitis, we accepted any periodontitis diagnosis as defined by the authors.

Data synthesis
Pooling of data was based on study design, outcomes, population characteristics, inflammatory mediators affected by periodontal treatment and types of PT. Specific inflammatory mediators that improved after therapy were also recorded. Descriptive summaries of included studies were entered into evidence tables and a narrative synthesis of evidence was performed. If outcome data could not be extracted from an article (e.g., data presented in a graph, use of median values) the corresponding author was asked to provide the data.
For meta-analysis, the mean difference (MD) and 95% confidence interval (CI) values among PT group and no-treatment group (control group) at both baseline and end were only calculated for two serum biomarkers: tumor necrosis factor alpha (TNF-α) and high sensitive C-reactive protein (hsCRP). Reviewers focused on these two markers based on the studies included in this review. Pooled estimates of the mean differences were calculated using random effects models in order to take potential inter-study heterogeneity into account and to adopt a more conservative approach. The statistical analysis was performed using the commercial statistical software Stata (Stata for Windows, v.11, Stata Corporation College Station, Texas, USA). The significant level established for analysis was 5% (p<0.05). The study by Chen et al. (2012) [29] had two treatment arms, and data from the control group was used in more than one comparison. Thus, the number of cases in this group was divided by number of comparisons (Chen et al., 2012a, Chen et al., 2012b) [29].

Search results
The initial search identified 3,164 potentially relevant articles, of which 2,301 were excluded based on their titles and abstracts. A second level, full-text search was performed on the 61 remaining studies. Nine studies were selected for full-text assessment with a total of 623 participants [5,29,[30][31][32][33][34][35][36]. Of these, 4 studies were included in CRP meta-analysis [31,33,29,35] and 2 in TNF-α meta-analysis [29,33] (Fig 1). The most common reasons for the exclusion of studies were study designs (e.g., reviews, commentaries, case series, or case reports) and missing data (e.g., data lacking exact serum biomarkers value description and/or lack of response to our requests by the corresponding author).

Characteristics of the included studies
All studies included were RCTs or CCTs published in English. Duration of the follow-up period was at least 1 month, with 2 studies [31,35] including a 3-month and a 6-month follow-up, respectively, although one study was carried out for 12 months [34]. Inclusion criteria would have included studies with other inflammatory serum biomarkers; however, none were available that met the inclusion criteria, and hence all studies included were of subjects with serum hs-CRP or CRP, IL-6 and TNF-α data. The largest studies in terms of sample size of the above data were Sun et al. [33], with 157 participants and Chen et al. [29], with 134. These two studies reported that PT can reduce serum inflammatory cytokines in T2DM individuals. Of the nine studies included, four reported CRP data, three reported hs-CRP, four reported TNF-α and four reported IL-6. Table 1 describes details of the included studies, such as number of participants, methods, definitions for periodontitis, type of periodontal interventions, outcomes, and conclusions. Only three studies described sample size calculation [29,35,36].

Types of outcome measures
Included studies observed hs-CRP or CRP, IL-6 and TNF-α (inflammatory serum biomarkers) as primary outcomes.

Quality assessment of included studies
Randomization was performed in all RCTs included, but only three of them appropriately described the sequence of generation and allocation concealment [29,35,36]. In fact, these three studies also described adequate methods of examiner masking and were therefore considered at low risk of bias. Three of the studies included [30,32,34] were not described as RCT and were hence deemed as being at high risk of bias due to lack of description about randomization scheme, allocation of therapy group concealment, masking of examiners, withdrawals and missed follow-up as described by Cochrane group [27]. Follow-up period and dropouts were clearly described by all studies included.

Effects of periodontal therapy (individual outcomes)
All treatment group interventions consisted of non-surgical periodontal therapy with or without adjunctive topical or systemic antibiotics, while only one study included surgical therapy.  Eight of nine studies were parallel arm two group studies, while one study [29] included three therapy groups. Of these three group studies, group 1 received SRP at baseline and additional subgingival debridement at the 3-month follow-up, while group 2 only received prophylaxis at 3-months. The third group received no intervention throughout the study period. Of the two group studies, three included SRP only [30,32,34], one study described minimal treatment group as supragingival cleaning [35]. -Analysis of variance, Kruskal-Wallis test and Mann-Whitney U test with Bonferroni correction were used to compare and correlate biochemical analysis. CRP exhibited higher levels for poorly controlled group (P<0.05), whereas well-controlled group and systemically healthy exhibited similar CRP values (P>0.05). Serum TNF-α decreased slightly after periodontal therapy without significant differences (P>0.05) and IL-6 serum levels showed significantly lower values after 3 months in both well-controlled and systemically healthy groups (p<0.05). Sun et al. (2011) [33]-One way ANOVA was performed, followed by LSD multiple comparison or Student's t-test (independent samples) to estimate statistical significance between means. All inflammatory mediators evaluated (hsCRP, TNF-α, and IL-6) decreased significantly after 3 months of periodontal intervention in treated group (T2DM-T) when compared to the group without periodontal treatment (T2DM-NT) (p<0.01). Auyeung et al. (2012) [34]-Comparisons of inflammatory parameters between baseline and 12 months post-therapy were analyzed by paired t-test. The results of this CCT showed no difference between CRP values at baseline and 12 month post-therapy for the group with moderate-to-severe periodontitis (p = 0.62).
Chen et al. (2012) [29]-ANOVA was used to analyze immunologic and metabolic variables. Both groups (group 1-non surgical therapy + additional subgingival debridement after 3 months; group 2-non surgical therapy + supragingival prophylaxis at 3 months) showed lower hs-CRP at 6 months (p<0.05), whereas hs-CRP showed no significant difference when compared with the control group after 6 months (group 3-without intervention). The results for TNF-α showed no statistical difference in any group after 6 months (P>0.05). Koromantzos et al. (2012) [35]-Authors performed Wilcoxon Mann-Whitney U test to analyze differences between baseline and the 6-month follow-up. With regard to log-transformed hs-CRP values at the 6-month follow-up, there was no statistical difference between baseline and groups (IG-intervention and CG-control group)(p = 0.06). Lin et al. (2012) [36]-Authors performed latent growth curve modeling (LGCM) to test differences in changes from baseline. IL-6 values increased in SRP group and decreased in SRP + minocycline group, but there was no statistical difference after the 6-month follow-up (p = 0.17). The LGCM for log-transformed CRP showed no statistical significance for both SRP and SRP + minocycline groups after 6 months period.

Meta-analysis
The effect of PT on TNF-α and hs-CRP serum levels was analyzed in two meta-analysis. The mean difference (MD) and confidence interval (CI) at baseline and between PT and no periodontal treatment groups were calculated (Figs 2 and 3). A significant MD was found for both TNF-α (-1.33 pg/ml, CI: -2.10; -0.56, p<0.0001) and hs-CRP (-1.28 mg/l, CI: -2.07;-0.48, p<0.0001). A random effects model was used to pool the data. Heterogeneity among studies was assessed using I2 statistic with 95% confidence (uncertainty) intervals [37,38]. I2 values of 25%, 50% and 75% imply small, moderate and high heterogeneity, respectively [38]. The pooled results for meta-analysis with 'high heterogeneity' are not presented; instead, individual study results are presented for informal comparison. We could not assess the possible presence of publication bias with funnel plots because there were too few studies included for analysis (at least 10 studies would be necessary).

Occurrence of adverse effects/complications associated with periodontal therapy
None of included studies reported the occurrence of adverse effects/complications.

Summary of the main results
This is the first systematic review investigating the effect of PT on serum inflammatory mediators in people with T2DM. The main finding of the review was that hs-CRP and TNF-α were statistically significantly lowered by PT in people with type 2 diabetes when compared to controls. Only nine studies met the proposed inclusion criteria [29,5,[30][31][32][33][34][35][36]. Six of them were RCTs and three were CCTs. Each study was assessed for a risk of bias. In all studies included in this systematic review, at least one serum inflammatory biomarker was available to evaluate the effect of PT on systemic levels of inflammation. The included studies analyzed TNF-α, hs-CRP and/or IL-6. It was possible to perform two meta-analysis with regards to TNF-α and hs-CRP. Data from CCTs and IL-6 studies were few and showed high heterogeneity to perform metaanalysis.

Agreements and disagreements with previous studies
Relevant inflammatory biomarkers considered to play an important role in the pathogenesis of type 2 diabetes and insulin resistance were selected as primary outcomes [39]. The studies included in this review described 3 markers: TNF-α, hs-CRP and/or IL-6. Poor glycemic control and elevated levels of HbA1c in patients with type 2 diabetes have been associated with oxidative stress and high risk of cardiovascular diseases. Oxidative stress is associated with elevated serum levels of advanced glycation end products, which leads to release of serum IL-6, TNF-α and CRP [39]. A recent meta-analysis reported a cumulative improvement on HbA1c levels in periodontitis patients with diabetes [40]. A great debate aroused from this recent trial and put into question whether the periodontal treatment delivered was sufficient. No other biomarkers were measured in this study population.
Another recent meta-analysis showed that periodontal treatment improves endothelial function and reduces biomarkers of atherosclerotic disease [39,40]. This review reported a statistically significant improvement in CRP and TNF-α serum levels after periodontal therapy in people with a number of comorbidities but not limited to type 2 diabetes. Both Teeuw et al. (2010) [23] and our study tend to suggest a consistent effect of periodontal therapy in reducing systemic inflammation in people suffering from periodontitis with or without other co-morbidities. These results could have a great impact in explaining the association of PD with increased risk of vascular events and complications in people with T2DM and it is worthy of further investigation.

Quality of evidence and potential biases in the review process
An important reason for study exclusion was related with design (cases series reports or pilot trial) or lack of clarity about the diabetes type of participants. Such papers were thus excluded  [34] reported sample size calculation. Quality analysis for these studies revealed a high risk of bias. The most frequent criteria were reporting of sample size calculation and randomization. Absence of sample size calculation is particularly troubling because it means we cannot be sure that the study has been designed to avoid random error. Furthermore, without a sample size calculation, we cannot be sure that enough individuals were recruited to detect a difference in outcome between treatment groups if one exists (a Type 2 error). This is of special concern if the studies that have not reported a sample size calculation have found no difference in outcome between treatment groups. Randomization is a means of assigning participants to groups, such that the groups are balanced for known and unknown risk factors to minimize bias. In the absence of randomization the groups of subjects may differ in terms of severity of their periodontitis, and we cannot have a balance between known and unknown prognostic factors in the assignment of treatment. Long-term randomized clinical trials evaluating the effects of PT on serum markers of diabetic individuals are scarce. This could be related to some important factors, such as difficulty in recruiting individuals that fulfill inclusion criteria; need to maintain the same medications during the follow-up period, and need of a controlled body mass index (BMI) for diabetic individuals. Another limitation is the unclear report about glycemic control during the studies' follow-up. Despite the limitations of this study, we feel confident in ascribing a causal role to PT in improving serum inflammatory markers, but it is difficult to determine the magnitude of the likely benefit.

Authors' Conclusions
This study supports the hypothesis that PT reduces systemic inflammation in people with T2DM. Our findings emphasize the importance of periodontal health in the management of individuals with type 2 diabetes mellitus to reduce their long term risk of complications.