A Systematic Review and Meta-Analysis on the Efficacy of Locally Delivered Adjunctive Curcumin (Curcuma longa L.) in the Treatment of Periodontitis

This meta-analysis intended to assess evidence on the efficacy of locally delivered curcumin/turmeric as an adjunctive to scaling and root planing (SRP), on clinical attachment level (CAL) and probing pocket depth (PPD), compared to SRP alone or in combination with chlorhexidine (CHX). RCTs were identified from PubMed, Cochrane Library, BASE, LIVIVO, Dentistry Oral Sciences Source, MEDLINE Complete, Scopus, ClinicalTrials.gov, and eLibrary, until August 2022. The risk of bias (RoB) was assessed with the Cochrane Risk of Bias tool 2.0. A random-effects meta-analysis was performed by pooling mean differences with 95% confidence intervals. Out of 827 references yielded by the search, 23 trials meeting the eligibility criteria were included. The meta-analysis revealed that SRP and curcumin/turmeric application were statistically significantly different compared to SRP alone for CAL (−0.33 mm; p = 0.03; 95% CI −0.54 to −0.11; I2 = 62.3%), and for PPD (−0.47 mm; p = 0.024; 95% CI −0.88 to −0.06; I2 = 95.5%); however, this difference was considered clinically meaningless. No significant differences were obtained between patients treated with SRP and CHX, compared to SRP and curcumin/turmeric. The RoB assessment revealed numerous inaccuracies, thus raising concerns about previous overestimates of potential treatment effects.


Introduction
Severe periodontal disease affected about 1.1 billion people globally in 2019 [1], with an overall prevalence of 67.4% (probing pocket depths between 4 and 5 mm), with 9.1% accounting for adolescents, 27.7% for adults, and 30.6% for elderly people [2]. Consequently, periodontitis can be considered a cross-generational, global public health problem with widely ranging effects such as bleeding gums, periodontal pockets, bone loss, and functional as well as aesthetic issues [3]. Furthermore, it is considered a risk factor for several systematic diseases, including cardiovascular disorders, rheumatoid arthritis, and chronic obstructive pulmonary diseases, as well as non-alcoholic liver diseases [3,4].
The primary goal of periodontal treatment is the removal of highly organized microorganisms embedded in an extracellular, polysaccharide matrix attached to the tooth's surface [5]. Biofilm management is usually achieved by sustainable biofilm disintegration consisting of instructions on effective oral hygiene, mechanical debridement of tooth surfaces, and removal of co-factors favoring re-accumulation, together with a regular recall system in which the current periodontal situation is evaluated [6]. Local therapeutical outcome) and probing pocket depths, (PPD: secondary outcome)(O) in a randomized splitmouth design and/or parallel group design studies (S) in a defined period of time (T)? The protocol of the present review was registered at the Prospero Register of Systematic Reviews (registration number: CRD42022290324,10/01/2022; registration name: Effect of locally delivered adjunctive Curcumin in the Treatment of Periodontitis-a Systematic Review and Meta-analysis). The current review was conducted in accordance with the "Preferred Reporting Items for Systematic Reviews and Meta-analyses" (PRISMA) statement checklist [58].
The following databases were searched until August 2022: PubMed, Cochrane Library (Wiley), BASE (base-search.net), LIVIVO, Dentistry Oral Sciences Source (Ebsco), MEDLINE Complete (Ebsco), Scopus, ClinicalTrials.gov, and eLibrary (https://www. elibrary.ru/defaultx.asp; accessed on 3 August 2022.). This combination of information sources retrieved both published journal articles and gray literature (e.g., dissertations, or study register entries). The search strategies were designed by an experienced information specialist (IK). In addition to the search in electronic databases, reference lists of included studies were checked manually. Search results were imported and deduplicated in Endnote 20 (Version 2013; The Endnote Team; Clarivate Analytics, Philadelphia, PA, USA). Additional data for individual search strategies are presented in the Supplementary Materials. The study selection process was performed stepwise. First, two reviewers (LWT and MW) independently screened the titles and abstracts of references found with the literature search. Second, the full texts of the studies included during the previous step were assessed for eligibility. Randomized controlled trials (RCTs) were included that compared SRP alone or in combination with chlorhexidine to SRP and local curcumin/turmeric regarding clinical attachment level and probing pocket depth. Table 1 presents details of the study eligibility criteria.
The quality of the included trials was methodically assessed by two authors (LWT and MW) using the revised Cochrane risk of bias tool 2.0 for randomized trials (RoB2). Any possible dissensions were resolved by discussion and mutual agreement. RoB2 is arranged into five different disciplines (randomization process, deviations from intended interventions, missing outcome data, measurement of the outcome, and selection of the reported results) that aim to evaluate all aspects of the study that are related to the risk of bias [59]. The five different disciplines were judged as having either low risk, some concerns, or high risk, and according to this judgment, an overall assessment of the risk of bias level in each individual study was made.
One author (LWT) collected the relevant data from the included articles. This was crosschecked for accuracy and completeness by another author (MW). The data of interest were methodology, number of participants, participants baseline characteristics, concentration of curcumin/turmeric and chlorhexidine, evaluation period, and results for primary and secondary outcomes. The authors of the Dave et al. (2018) study were contacted to gather missing information concerning initial PPD and curcumin concentration.
A random-effects meta-analysis was performed using an inverse-variance model with the DerSimonian-Laird estimate of squared tau (τ 2 ) by pooling mean differences with 95% confidence intervals, if the number of identified investigations that were similar in population and outcome was sufficient. The statistical heterogeneity was assessed across trials by visually inspecting the forest plots and calculating the I 2 statistics. STATA release 17.0 was used for all analyses (StataCorp LLC; College Station, TX, USA). Additional calculations had to be performed for the data published by Raghava et al. (2019) and Farhood et al. (2020); CAL was presented in variance (PPD in standard error), and CAL and PPD were presented in standard error, respectively. Conversions were made to the standard deviation.

Literature Search and Screening
A total of 827 records were identified through the literature search. After deduplication, 222 studies were screened by title and abstract. Consecutively, 33 full-text articles were assessed for eligibility, and, finally, 23 investigations were included . Details of the study selection process are presented in Figure 1.
CAL and PPD were presented in standard error, respectively. Conversions were made to the standard deviation.

Literature Search and Screening
A total of 827 records were identified through the literature search. After deduplication, 222 studies were screened by title and abstract. Consecutively, 33 full-text articles were assessed for eligibility, and, finally, 23 investigations were included . Details of the study selection process are presented in Figure 1.

Quality Assessment
The possibility of bias in design and analysis was evaluated by the Cochrane Risk of Bias tool 2.0 [59] (this tool was designed for randomized parallel group design investigations, and, consequently, this should be carefully considered when interpreting the following results). In total, 23 studies were rated with a moderate risk of bias. The most common source of potential bias was domain four ("measurement of the outcome"). In all included studies, investigators probed manually, which was judged to have a poor validity. The second most common source of bias was domain three ("missing outcome data"); many investigations [31][32][33][34][35]38,[41][42][43][44][45][46][47][48][49][50][51]53] failed to comment on the loss of followup, which led to the present judgment. The third most common source of bias was the "randomization process", which certain investigations could have described it in greater detail [31,32,35,36,[41][42][43][44]50,53]. The remaining domains, "effect of assigning to intervention", as well as "selection to reported results", performed acceptably throughout the included investigations. Details of the RoB Assessment are provided in Figure 2.

Quality Assessment
The possibility of bias in design and analysis was evaluated by the Cochrane Risk of Bias tool 2.0 [59] (this tool was designed for randomized parallel group design investigations, and, consequently, this should be carefully considered when interpreting the following results). In total, 23 studies were rated with a moderate risk of bias. The most common source of potential bias was domain four ("measurement of the outcome"). In all included studies, investigators probed manually, which was judged to have a poor validity. The second most common source of bias was domain three ("missing outcome data"); many investigations [31][32][33][34][35]38,[41][42][43][44][45][46][47][48][49][50][51]53] failed to comment on the loss of follow-up, which led to the present judgment. The third most common source of bias was the "randomization process", which certain investigations could have described it in greater detail [31,32,35,36,[41][42][43][44]50,53]. The remaining domains, "effect of assigning to intervention", as well as "selection to reported results", performed acceptably throughout the included investigations. Details of the RoB Assessment are provided in Figure 2.

SRP and Chlorhexidine Compared to SRP and Local Curcumin
The effect of SRP and chlorhexidine application, in comparison to SRP and local curcumin/turmeric application, on the loss of CAL was evaluated by six investigations [47][48][49][50][51][52]. Random effects meta-analysis showed a statistically non-significant mean difference of −0.42 mm (95% CI −1.15 to 0.31; p = 0.258, I 2 = 93.6%; 185 sites; see Figure 4) in favor of SRP and curcumin/turmeric.

SRP and Chlorhexidine Compared to SRP and Local Curcumin
The effect of SRP and chlorhexidine application, in comparison to SRP and local curcumin/turmeric application, on the loss of CAL was evaluated by six investigations [47][48][49][50][51][52]. Random effects meta-analysis showed a statistically non-significant mean difference of −0.42 mm (95% CI −1.15 to 0.31; p = 0.258, I 2 = 93.6%; 185 sites; see Figure 4) in favor of SRP and curcumin/turmeric.

SRP and Chlorhexidine Compared to SRP and Local Curcumin
The effect of SRP and chlorhexidine application, in comparison to SRP and local curcumin/turmeric application, on the loss of CAL was evaluated by six investigations [47][48][49][50][51][52]. Random effects meta-analysis showed a statistically non-significant mean difference of −0.42 mm (95% CI −1.15 to 0.31; p = 0.258, I 2 = 93.6%; 185 sites; see Figure 4) in favor of SRP and curcumin/turmeric.

Discussion
This systematic review and meta-analysis revealed a statistically significant difference between SRP alone, compared to SRP and curcumin/turmeric application for CAL and PPD. However, these significant decreases (CAL −0.33 mm; PPD −0.47 mm) are not considered clinically relevant. Included studies showed a notable degree of heterogeneity,

Discussion
This systematic review and meta-analysis revealed a statistically significant difference between SRP alone, compared to SRP and curcumin/turmeric application for CAL and PPD. However, these significant decreases (CAL −0.33 mm; PPD −0.47 mm) are not considered clinically relevant. Included studies showed a notable degree of heterogeneity, which is probably due to different application methods, varying concentrations of curcumin/CHX, and the unreliability of outcome measurements.
All periodontal pockets were probed manually by all investigators, even though one investigation used a pressure-sensitive manual probe [53] and two investigations measured six sites around each tooth [40,43]. It is well accepted that probing manually might be unreliable, and can result in certain inaccuracies [61,62], at least to some extent. A standardized measuring method using electronic periodontal probes, thus controlling probing force and measuring to the closest tenth of a millimeter and, therefore, generating reproducible results, even with different examiners, would seem generally preferable [63,64]. Furthermore, factors like the design of the probe, probing position, visual observational error, and tissue inflammation could influence the reproducibility of readings, thus leading to detection bias [65,66]. Nevertheless, the assessors of the present investigation decided to override the suggested overall judgment for the risk of bias assessment. This is justified by the fact that all treatment groups measured the outcomes manually, in most cases uniformly using a UNC 15 periodontal probe [33,37,39,40,45,[50][51][52][53]. Additionally, studies have shown that there is a tendency to have similar reliability between manual and electronic probes [67,68]. Furthermore, many investigations [31][32][33][34][35]38,[41][42][43][44][45][46][47][48][49][50][51]53] failed to comment on the loss of follow-up, and this sort of attrition bias led to the present judgment.
A further plausible explanation for the observed heterogeneity might be the varying application methods. To a certain extent, gels have several advantages over other application methods. They are biocompatible and bioadhesive, allowing them to attach to periodontal pockets and, furthermore, enabling a controlled drug release and minimum dose frequency [69]. While most studies applied a conventional gel [32][33][34][35][36][37][39][40][41][42][43][44][45]48,50,51], two used a nanoparticle gel system, which is expected to have several advantages. Due to their nanoparticle size, these gel systems are able to penetrate into the most apical regions of periodontal pockets, thus ensuring a homogenous disposal of the drug over a long interval, along with a reduction in drug quantity and high bioavailability [69]. Matrix delivery systems such as chips and strips, as used in four trials [46,47,49,52], have the benefit of sustained drug release patterns. In contrast, solutions, as used in one investigation [54], provide high concentrations initially and will be diluted promptly by liquids, for example, by gingival crevicular fluid [69].
Although the studies under investigation all revealed varying curcumin concentrations, it can be assumed that the latter was effective, since an inhibitory effect on 61.01% of the MMP-9 activity has been determined at a curcumin concentration of 1500 µg/mL [73]. Additionally, a few investigations repeated the curcumin application. Patients were either instructed to apply the gel daily [37], or the application was repeated once weekly (over a three-week period) [36], with a second application of gel after one week [41], repeated application at day 15 [48], or repeated applications after 7, 14, and 21 days [53]. Furthermore, the use of a COE pack after drug application, aiming to ensure the persistence of the applied drug and prevent site contamination, could be a positive influencing factor for the concentration of the drug applied; however, this has not been proven clinically up to now.
Previous trials have commented on curcumin's poor bioavailability, and this was probably due to low absorption, rapid metabolism, and quick systemic elimination [74,75]. To date, a number of investigations have begun to examine the use of synthetic, structural analogues and various nanoforms of curcumin, as they have improved plasma and tissue levels [74][75][76]. The effect of oral application of curcumin and modified curcumin on bone resorption, inflammation, and apoptosis in rats, was compared by a previous study. It was concluded that administration of chemically modified curcumin significantly reduced the inflammatory infiltrate in comparison to natural curcumin [77]. This should encourage interest in planning and conducting trials to explore the qualities of chemically modified curcumin.
Limitations of this work are that the risk of bias assessment used was based on a study design used in general medicine. There seems to be a lack of risk of bias in assessments designed for split-mouth trials, which is a common design in oral health. It is worth noting that, even though a split-mouth design can be considered powerful, this latter methodology may result in considerable variability, which is a result of characteristic differences between examiners. No doubt, and this cannot be ruled out, there was a potential risk of contamination of the control site, as it could be possible for the drugs to diffuse to the other site (carry-across effects) [78]. In comparison to the protocol registration, a few amendments would seem worth mentioning. No subgroup analysis was conducted; more data bases were searched; the search was updated in April 2022, and a slight alteration was made to the title; the STATA release 17.0 was used for all analyses (StataCorp LLC; College Station, TX, USA).
The search for new anti-microbial and anti-inflammatory substances as potential agents in the treatment of oral diseases, especially those that cannot develop antibiotic resistance, has gained much importance in recent years. Curcumin appears to possess these valuable properties and has reached the clinical testing phase. At first glance, the results of these clinical studies appear very promising. On a closer inspection, however, it must be stated that several issues in relation to the risk of bias, as discussed in detail, must be elucidated. Uniform study designs and methods with accurate and reproducible measurements of endpoints and homogenous concentrations would be desirable. Interestingly, a recently published meta-analysis of the effect of adjuvant curcumin in the treatment of periodontitis, came to the conclusion that curcumin can be successfully used in periodontal therapy [57]. Another meta-analysis concerning this topic, evaluating gingival index, sulcus bleeding index, and bleeding on probing as primary outcomes, concluded that curcumin is a "good candidate as an adjunct treatment for periodontal disease" [56]. Another metaanalysis concludes that locally applied curcumins "were found to be equally effective compared to the routinely used agents for reduction of plaque and gingival inflammation" [55]. Undoubtedly, such statements should be carefully but critically weighed; when reflecting on several aspects, like the lack of sufficient details concerning study quality, this would seem more than justified. Moreover, to recommend curcumin for periodontal therapy would call for clear endpoints assessing the effectiveness of curcumin in periodontal treatment, and any possible effects of the oral hygiene of participants must be clearly distinguished.
When pondering on the treatment of periodontal disease with adjuvant curcumin, the data available from this present meta-analysis would suggest that there is obviously no reason for any further investigations, and this would refer both to large-scale and high-quality studies. At the end of the day, the available data base could reveal that the proven (and noteworthy, no doubt) biochemical properties of curcumin would justify the previous research projects in the first instance, but, notwithstanding, no clinically significant improvements could be proven with the current systematic review. Consequently, a clinical implementation of adjuvant curcumin for periodontal treatment is not recommended.

Conclusions
In conclusion, with reference to clinical attachment level and probing pocket depth, the present results cannot indicate that curcumin's/turmeric's anti-bacterial and anti-inflammatory properties result in an additionally beneficial clinical outcome, when combining this adjunct to scaling and root planing. Therefore, our findings do not support the application of curcumin-/turmeric-based products in non-surgical periodontal treatment scenarios.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/biomedicines11020481/s1, Table S1: Search strategies. Funding: This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Patient consent was waived as specific patient details were not mentioned or anonymized. Data Availability Statement: Datasets generated in this study can be found in the Supplementary Materials.

Conflicts of Interest:
The authors declare no conflict of interest.