Efficacy of Enamel Derivatives to Improve Keratinized Tissue as Adjunct to Coverage of Gingival Recessions: A Systematic Review and Meta-Analysis

Background: The systematic review was designed to answer the following focused question: Are enamel matrix derivatives able to improve the quantity of keratinized tissue (KT) around natural dentition in patients with recessions defects after their treatment with periodontal plastic procedures? Methods: Only Randomized Clinical Trials (RCT) in English language evaluating root coverage procedures in combination with enamel matrix derivatives (commercially known as Emdogain®—EMD), with at least 10 subjects and a minimum duration of six months, were included. The search was applied to PUBMED and SCOPUS and it consists of a combination of MeSH terms and free text words (from January 2000 to June 2019). Risk of bias in individual studies and across studies was also evaluated. Results: After the full text analysis and the exclusion of further 18 articles, 12 articles were finally included. In total 639 recessions were treated (334 tests and 305 control). The recessions defects were classified according to the classification of Miller (Class I, II, III, IV). Only one trial included Miller Class III recessions (7 in total). Enamel matrix derivatives were applied in conjunction with Coronally Advanced Flap (CAF), Coronally Advanced Flap + Sub Epithelial Connective Tissue Graft (CAF + CTG), Semilunar Flap (SF). For the group CAF vs CAF + EMD the mean difference between the keratinized tissue gain in the two procedures was 0.40 mm (95% Confindence Interval Lower/Upper: 0.014–0.81) (p < 0.058); for the comparison CAF + CTG + EMD vs. CAF + CTG the mean difference between the two groups resulted in −0.06 mm (95% Confindence Interval Lower Upper −0.45 to 0.33) (p = 0.7603). Discussion: Randomized clinical trials included medium-low quality evidence. The application of Enamel Matrix Derivatives to surgical procedures aimed to cover gingival recessions does not add robust clinical benefit to conventional plastic procedure alone.


Introduction
Isolated and multiple gingival recessions are a common finding among adult populations. In a recent cross sectional survey in a cohort of 349 young adults [1], every participant exhibited gingival recession affecting at least one tooth, with 42% having a maximum recession of 4-8 mm. There was a significant and linear association demonstrating an increase in maximum recession with age and prevalence increasing from incisors to molars in upper and lower arches, the premolars being the most affected.
Gingival recessions are treated to reduce dentin hypersensitivity, to treat radicular caries, and to improve aesthetics. Within the overall treatment plan of the periodontal patient, surgical treatment of

Electronic Search
We conducted a search on electronic databases from January 2000 until June 2019; the search was applied to PUBMED and SCOPUS. The strategy used was a combination of MeSH terms and free text words.
The search strategies were applied as follows: • PUBMED: ((((((EMD OR enamel matrix OR emdogain)) AND (recession OR recession coverage)) AND (coronally advanced OR coronally advanced flap OR surgical treatment OR crowned advanced)) AND "clinical study"[Publication Type])) NOT (intrabony OR intrabony defect OR intraosseous OR intraosseous defect OR infraosseous OR infra OR infrabony) • SCOPUS: (TITLE-ABS-KEY (emd OR enamel AND matrix OR emdogain) AND TITLE-ABS-KEY (recession OR recession AND coverage) AND TITLE-ABS-KEY (coronally AND advanced OR coronally AND advanced AND flap OR surgical AND treatment OR crowned AND advanced) AND NOT TITLE-ABS-KEY (intrabony OR intrabony AND defect OR intraosseous OR intraosseous AND defect OR infraosseous OR infra OR infrabony)) AND PUBYEAR > 1999 The criteria for considering studies for this review were organized by the P.I.C.O. method and were as follows: (P) Type of Participants: patients with a clinical diagnosis of localized or multiple gingival recessions. Studies involving only heavy smokers (≥10 cigarettes/day) were not enclosed. (I) Type of Interventions: any type of periodontal plastic procedure aimed to cover gingival recession with the adjunctive use of EMD. (C) Comparison between interventions: any type of periodontal plastic procedure for root coverage with and without enamel matrix derivatives with at least 6 months of follow-up.
(O) Type of Outcome measures: primary outcome was the improvement of keratinized tissue.

Hand Search
Hand searching (N.D., R.M.) was performed on relevant journals (Journal of Clinical Periodontology, Journal of Periodontology) from January 2000 up to June 2019, consisting of a manual page by page examination of the two journal's all issues and volumes. Bibliographies of all retrieved papers and review articles were searched as well.

Study Selection and Data Collection
Titles, abstract, and full-text analysis was performed to assess the eligibility. Titles and abstracts were screened for possible inclusion in the review by two reviewers (N.D., R.M.). Reviewers were calibrated for study screening with an unweighted k score of 0.90 [20]. Abstracts were to be excluded if they did not fulfil the inclusion criteria listed before. In order to avoid excluding potentially relevant articles, abstracts providing unclear results or absent were included in the full-text analysis. Full text of studies of possible relevance was then obtained for independent assessment by two reviewers (N.D., R.M.) against the stated inclusion criteria. Any disagreement was resolved by discussion between reviewers. The two reviewers conducted all data collection and quality assessments independently. If retrieved articles were unclear, authors were contacted directly. Data of the included articles were extrapolated through an "ad hoc" extraction sheet.

Data Items
Primary outcome measure considered was KT gain (KTg), both at site and patient/area level, obtained subtracting the width of KT at the baseline to the same measurement assessed at the follow-up visit. KTg at patient/area level was defined as the width of keratinized tissue of all recessions present in the subject or treated area for parallel or split-mouth studies, respectively. The mean difference and the standard deviation between baseline and the last follow up for test and control group were analyzed. If they were not calculated by the authors in the text, they were obtained applying the following formula: MD = X 1 − X 2 and the SE(MD) = √ (s 2 1 /n 1 + s 2 0 /n 0 ).

Risk of Bias in Individual Studies
Risk of bias was evaluated by two authors (N.D.; R.M.) independently using an individual component approach based on 5 domains (the tools acronyms is RoB 2) [20,21]. Disagreements were solved by discussion till a consensus was achieved. The assessment of risk of bias of each RCT was performed following the analysis of pertinent items suggested by the Cochrane reviewers' handbook [20,22] (RoB 2, Figure 1). The five domains assessed were (i) risk of bias arising from randomization process, (ii) risk of bias due to deviations from the intended intervention, (iii) missing outcome data, (iv) risk of bias in measuring of the outcome, and (v) risk of bias in selection of the reported result ( Figure 2). Studies have been categorized as being at low risk of bias (all domains were at low risk of bias), high risk of bias (one or more domains were at high risk of bias), or unclear risk of bias (if one or more domains were at unclear risk of bias). up visit. KTg at patient/area level was defined as the width of keratinized tissue of all recessions present in the subject or treated area for parallel or split-mouth studies, respectively. The mean difference and the standard deviation between baseline and the last follow up for test and control group were analyzed. If they were not calculated by the authors in the text, they were obtained applying the following formula: MD = X1 − X2 and the SE(MD) = √(s 2 1/n1 + s 2 0/n0).

Risk of Bias in Individual Studies
Risk of bias was evaluated by two authors (N.D.; R.M.) independently using an individual component approach based on 5 domains (the tools acronyms is RoB 2) [20,21]. Disagreements were solved by discussion till a consensus was achieved. The assessment of risk of bias of each RCT was performed following the analysis of pertinent items suggested by the Cochrane reviewers' handbook [20,22] (RoB 2, Figure 1). The five domains assessed were (i) risk of bias arising from randomization process, (ii) risk of bias due to deviations from the intended intervention, (iii) missing outcome data, (iv) risk of bias in measuring of the outcome, and (v) risk of bias in selection of the reported result ( Figure 2). Studies have been categorized as being at low risk of bias (all domains were at low risk of bias), high risk of bias (one or more domains were at high risk of bias), or unclear risk of bias (if one or more domains were at unclear risk of bias).

Random Sequence generation
Deviation of intended intervention Missing outcome data Measuring outcome Selection reported results

Study Design and Study Population
Characteristics of included studies are described in Table 1. Follow-up varied, from 6 months (three trials [29][30][31]) to 12 months in six trials [15,[32][33][34][35][36], and 18 and 24 months in Pilloni et al. and Spahr et al., respectively [37,38]. Smokers were excluded in all the trials except in Cueva et al. in which 2 smokers were included, and they were asked to quit smoking 2 weeks before the surgical treatment until 2 weeks after the surgery. One trial [36] provided a 25% of the sample who were former smokers.
Previous periodontal treatment, consisting in oral hygiene instruction and non-surgical periodontal therapy (supragingival debridement), was reported in all the trials.
In total 639 recessions were treated (334 tests and 305 control), consisting in 632 Miller class I/II. Only one trial included Miller Class III recessions (7 in total). One trial [35]reported to treat only Class I. The treated teeth were incisors, canines, and premolars. Molar teeth were included and treated in two trials [29,31]. Maxillary recessions were selectively included in four trials [29,30,33,35] while 6 trials included both maxilla and mandible. One trial did not report the anatomical location [37].

Summary Measures and Synthesis of the Results
Tables were created for the review question to summarize an overview of the included studies, characteristics of the intervention, characteristics of primary outcome reporting (measurement, methods, timing), and risk of bias in individual studies. KT gain (KTg) between baseline and the last follow up was reported (calculated), and the results were expressed as mean difference (MD) and standard deviation (sd).
The statistical heterogeneity among studies was assessed using the Q test according to Der Simonian and Laird [23]. To overcome the intrinsic limitation of the Q test (power dependent on the number of included studies), two additional parameter will be calculated [24]: the H value and the I 2 index [25]. The latter was calculated in order to quantify the percentage of variation in the total estimate that was associated to heterogeneity. The study specific estimates were pooled together with the random effect model for meta-analysis [23].
Subgroup analysis was carried out when different surgical procedures were applied. In each subgroup were analyzed only the studies which used exactly the same procedures.
Forrest plots were created to illustrate the effects in the meta-analysis [26]. All the statistical analysis was formulated with STATA 15 software (StataCorp LP, Lakeway Drive, College Station, TX, USA) with the grid used to develop the analysis available in Appendix B. Statistical significance was defined as a P value < 0.05.

Risk of Bias across Studies
The method used to assess the presence of a publication bias was the Egger test [27]. The presence of bias is valued by the significance of the ordinate at the origin for a value p < 0.10.

Assessment of the Quality of Evidence Using GRADE
We evaluated the body of evidence grading the quality of the evidence for each outcome across studies, using the Grading of Recommendations: Assessment, Development, and Evaluation (GRADE) tool [28]. Then we developed a meta-analysis based on the strength of evidence for each outcome. This approach allows to classify the results in four levels of evidence quality: high, moderate, low, very low.
The first step of the GRADE approach is to define the study design (randomized clinical trials or observational trials); the second step is to rate the quality of evidence using 5 tools which may decrease the rating (risk of bias, inconsistency, indirectness, imprecision, publication bias) and 3 which could raise it (large magnitude effect, dose-response gradient, effect of plausible residual confounding).
To classify the quality of each outcome as explained above, at each one of the tools was addressed a value among: (1) no limitation, (2) serious limitation, (3) very serious limitation. When the rank is high, it suggests a high confidence that the true effect is close to the estimate of the effect, whereas a very low quality suggests that the estimate reported can differ significantly from the measure evaluated.

Study Selection
The electronic search found a total of 55 articles ( Figure 2). Hand searching identified 19 additional articles for the full text analysis. Thus, a total of 74 studies were identified ( Figure 2). Screening of titles and abstracts led to rejection of 39 articles (Appendix A), and the full text PDFs of the remaining 30 articles were obtained. After the full text analysis and the exclusion of a further 18 articles, 12 articles were finally included ( Figure 2).

Study Design and Study Population
Characteristics of included studies are described in Table 1. Follow-up varied, from 6 months (three trials [29][30][31]) to 12 months in six trials [15,[32][33][34][35][36], and 18 and 24 months in Pilloni et al. and Spahr et al., respectively [37,38]. Smokers were excluded in all the trials except in Cueva et al. in which 2 smokers were included, and they were asked to quit smoking 2 weeks before the surgical treatment until 2 weeks after the surgery. One trial [36] provided a 25% of the sample who were former smokers.
Previous periodontal treatment, consisting in oral hygiene instruction and non-surgical periodontal therapy (supragingival debridement), was reported in all the trials.
In total 639 recessions were treated (334 tests and 305 control), consisting in 632 Miller class I/II. Only one trial included Miller Class III recessions (7 in total). One trial [35] reported to treat only Class I. The treated teeth were incisors, canines, and premolars. Molar teeth were included and treated in two trials [29,31]. Maxillary recessions were selectively included in four trials [29,30,33,35] while 6 trials included both maxilla and mandible. One trial did not report the anatomical location [37].

Type of Interventions
Adjunctive effect of Emdogain ® was coupled in a quite heterogeneous group of original treatments ( Table 2). Recessions were treated and Emdogain ® was applied more frequently with coronally advanced flap (CAF) that served as a control in four trials [31,34,37,38]. In the latter, test treatment consisted in coronally advanced flap plus Emdogain ® ( Table 2).
Recessions were also treated with different combination of treatments: tunnel technique plus EMD vs CAF plus EMD [33] and the semilunar flap technique vs Semilunar flap plus EMD [35] (Table 2).

Synthesis of the Results
Comprehensively, the width of keratinized tissue in test and control group was 2.6 mm (sd 1.01) and 2.49 (sd 1.03) respectively. CAF vs. CAF + EMD. This group of studies treated both Miller class I/II and III. They obtained loss of keratinized tissue in two studies after CAF technique [31,37] and negligible to moderate gain in the other two [34,38]. The application of EMD in this group (CAF+EMD) led to a gain of 0.65 mm (sd 0.99) and 0.82 mm (sd 0.2) in Spahr et al. [38] and Cueva et al. [31].

Risk of Bias in Individual Studies
Adequate methods of sequence generation were reported in all articles included. In 5 trials [15,29,31,34,37] allocation concealment was not specified. Blinding of personnel was not specified in any of the articles included. Incomplete reporting outcome was identified in two papers [15,37] and in one paper was defined as unclear. No information on masking of statisticians was reported. Unclear information consisted mainly on lack of clear definition of primary outcome, oral hygiene levels, and periodontal status at baseline. A summary, according to a specific graphic tool, was presented in Figure 1.

Assessment of the Quality of Evidence
In the summary of findings (Table 3), the Quality of Evidence was evaluated for the outcome KTg in each one of the subgroups characterized by the different interventions. Using the GRADE approach [28], the evidence was downgraded mostly due to high risk of bias in some studies and because of the imprecision due to the small sample size and large confidence intervals.

Additional Analysis
Meta-analysis was performed evaluating the mean difference between test and control group in terms of KTg (in mm) (mean difference, 95% interval confidence) for the comparison (a) CAF vs. CAF + EMD (Figure 3, Tables 4 and 6) (b) CAF + CTG vs. CAF + CTG + EMD (Figure 4, Tables 5 and 7) Due to the elevated heterogeneity (I 2 = 70.12%; Tau 2 = 0.15) between the two studies [29,36], meta-analysis was not performed for the group CAF + EMD vs. CAF + CTG.
Forrest plot for random effect was presented. For the group CAF vs. CAF + EMD the mean difference was 0.40 mm (95% Conf. Interval Lower Upper: 0.014-0.81) (p < 0.058).   Publication bias was not present for KTg for the experimental procedures analyzed. It was not statistically significant (P > |t| = 0.555, 95% Conf. Interval −10.20 to 15.56) A meta-analysis was performed for the comparison between CAF + CTG + EMD vs. CAF +CTG (Figure 4)   The mean difference between the two groups resulted in −0.06 mm (95% Conf. Interval Lower Upper −0.45 to 0.33) (p = 0.7603), so the difference was in favor of the control group. This difference was not statistically significant. Publication bias (Egger Method) [27] was not statistically significant (t = −11.18, p < 0.057).  A meta-analysis was performed for the comparison between CAF + CTG + EMD vs. CAF + CTG (Figure 4).

Discussion
The mean difference between the two groups resulted in −0.06 mm (95% Conf. Interval Lower Upper −0.45 to 0.33) (p = 0.7603), so the difference was in favor of the control group. This difference was not statistically significant. Publication bias (Egger Method) [27] was not statistically significant (t = −11.18, p < 0.057).

Discussion
The current systematic review was designed to evaluate the adjunctive clinical benefit of Enamel Matrix Derivatives (Emdogain ® ) applied with mucogingival plastic procedures. The main outcome variable was elicited to be the gain of keratinized tissue (KTg) between EMD application versus periodontal plastic procedures applied to cover gingival recessions. No secondary outcome variables were analyzed.
The size of the adjunctive benefit of Enamel Matrix Derivatives resulted highly heterogeneous, depending on the specific plastic procedure with which it was applied, but clinically negligible.
Recessions that were treated with CAF + Emdogain ® , versus CAF alone, received an additional gain of keratinized tissue of less than 0.5 mm. This difference was almost statistically significant.
In three trials subepitelial connective tissue graft was added (CAF + CTG + EMD vs. CAF + CTG) in both experimental groups. The adjunctive effect of EMD in terms of KTg was almost null.
Albeit the most common flap design with which EMD was applied was the CAF, in one recent publication [35], it was applied with a semilunar flap: Even in this application, KTg was not superior in one experimental group in comparison to the other.
In the author's best knowledge, the current investigation represents the first attempt to focus and consequentially review, with a systematic approach, the impact of Enamel derivatives on the KTg after recessions coverage procedures.
The amount of keratinized tissue around the teeth has long been a debated and controversial topic. At first, clinical studies had recommended a specific limit amount in order to guarantee periodontal health [39]. Subsequently, prospective clinical and pre-clinical studies have shown that periodontal health was also compatible with the absence of keratinized gingiva [40].
It seems reasonable to think that teeth treated for one or more adjacent recessions can benefit from a greater band of keratinized gingiva. This in turn can act as a beneficial local factor in order to prevent future recurrences (secondary prevention).
Our results show that only if the surgical technique chosen is CAF, the addition of EMD appears to be slightly beneficial for the gain of keratinized gingiva.
Graziani and co-workers [6], applying a Bayesian Network meta-analysis, have tried to verify which was the most effective treatment in terms of keratinized gingival gain. The results obtained show a minimal effect of EMD in addition to CAF, both in statistical and in clinical significance (−0.05, 90% C.I.: [−0.68; 0.57]). The difference in magnitude obtained, compared to our review, can be partially explained by the use of a different data analysis system, which is not always comparable to the indirect analysis (pair-wise vs Bayesian Review) [41], and by the different PICO question of the review (Complete root coverage vs. Keratinized tissue gain).
An aspect that is worth to mention is the healing time of the experimental studies included. The information available in the literature shows the amount of KT gain over time after the execution of both CAF and CAF + CTG [42,43]. The studies included in this review in only two cases exceed 12 months of observation [37,38], and the study by Spahr and colleagues obtained the best value in terms of KT gain when CAF alone was considered (0.33 mm after 24 months of healing). Therefore, the results presented in the review should be interpreted with caution also due to the relatively short follow-up, as the process of "creeping attachment" is still ongoing [44].
There are several limitations for the present protocol that are worth discussing. In primis, the research protocol was not registered, before being applied, in an independent register (i.e., PROSPERO). Supposedly, the latter could lead to an increased risk of selective report of outcomes. In this perspective, one of the tools that are used to verify the methodological quality of a systematic review, AMSTAR 2 [45] foresees among its items (Item 2) to check whether the methods of the revision have been established before conducting the review itself. Interestingly, a recent systematic review [46] evaluates the association between registration of orthodontic systematic reviews in PROSPERO and review quality, assessed by the Amstar 2 tool. The results obtained, after proper adjustment, confirm that verifiable "a priori" protocol registration significantly improves the overall quality of the reviews. Notwithstanding, only a small percentage of reviews was registered so far.
The research strategy applied for the current review was deliberately limited to English language. Potentially, this can represent a source of bias [47]. However, recent reviews on the analysis of the aftermath of language restriction do not appear to bias the estimates of the intervention's effectiveness [48,49].
Another interesting issue that deserved to be mentioned is that the current systematic review is aimed to pool evidence coming from randomized clinical trials about the adjunctive effect that enamel matrix derivatives can produce in terms of keratinized tissue gain, when applied with periodontal plastic procedures. None of the included studies have been designed, and consequentially their sample size calculated, considering the gain of keratinized tissue as a primary outcome; complete root coverage (CRC) or recession reduction (RD) were, indeed, the primary outcome more frequently established. This aspect should be borne in mind when considering the results of the current meta-analysis. A pool estimate derived from potentially underpowered studies could be considered less precise. Original investigations should be designed in the future (with keratinized tissue gain as a primary outcome) to specifically answer to the question that the current review advance.
Some observations on the applicability of the results obtained can be formulated. The application of enamel matrix derivatives, indeed, does not seem to develop any additional benefit in terms of keratinized gingiva gain when applied together with bilaminar techniques. Therefore, the therapeutic indication of enamel matrix derivatives, with the aim of increasing keratinized gingiva, does not take place if the chosen technique involves the use of a connective tissue graft.
The design of the included studies did not allow answering a specific clinical question: Was there a difference in the result depending on the type of recession treated (miller class I or II)? In other words, from a biological standpoint, we still cannot argue which is the role of attached gingiva around the recession defect (Miller class I) in the terms of future gain.
In conclusion, applying Enamel Matrix Derivatives during procedure aiming to treat gingival recessions does not seem to add clinical benefit in terms of keratinized tissue gain, irrespective of the surgical technique applied. More specifically designed randomized clinical trials are needed to overhaul our hypothesis.

Conflicts of Interest:
The authors declare no conflicts of interest.      Four-year results of a prospective-controlled clinical study evaluating healing of intra-bony defects following treatment with an enamel matrix protein derivative alone or combined with a bioactive glass 0 3 Sculean 2014 The modified coronally advanced tunnel combined with an enamel matrix derivative and subepithelial connective tissue graft for the treatment of isolated

Shirakata 2019
Healing of localized gingival recessions treated with a coronally advanced flap alone or combined with an enamel matrix derivative and a porcine acellular dermal matrix: a preclinical study 0 9 Sipos 2005 The combined use of enamel matrix proteins and a tetracycline-coated expanded polytetrafluoroethylene barrier membrane in the treatment of intra-osseous defects Tunnel technique with connective tissue graft versus coronally advanced flap with enamel matrix derivative for root coverage: a RCT using 3D digital measuring methods. Part I. Clinical and patient-centred outcomes 1