**1. Introduction**

Bacterial infection plays an important role in establishing pulp tissue inflammation, which may lead to subsequent pulp necrosis and the formation of periapical lesions [1]. The complete removal of, or at least significant reduction in, the bacterial load during nonsurgical endodontic treatment is an important factor determining the final prognosis of root canal treatment. However, the development of apical periodontitis was reported in 44.9% of studied cases [2], mainly related to persistent or secondary endodontic infections [3].

Endodontic surgery is recommended after unsuccessful retreatment, when retreatment is impossible, or when there is an unfavorable prognosis [4]. Surgical endodontic procedures include removing necrotic and infected periapical tissues, resecting the apical part of the tooth (apicoectomy), and preparing the root-end cavity for the insertion of retrograde filling material [5]. Conventional endodontic surgery has been reported to result in a complete periapical tissue healing rate of 90% [6].

Recently, guided tissue regeneration (GTR) techniques have been widely used in medicine, including in dentistry, to improve tissue healing. Furthermore, GTR techniques have been recommended as an adjunct to endodontic surgery to promote periapical tissue healing and improve the treatment outcome [7].

Complete periapical healing involves the regeneration of alveolar bone, periodontal ligament cells, and cementum [8]. However, the surrounding connective tissues may grow into the osseous defect, preventing bone healing [6]. GTR techniques have been proposed as an adjunct to endodontic surgery approaches to promote bone healing and prevent the collapse of connective tissues [9].

Numerous studies reported the clinical effectiveness of GTR techniques to promote healing and improve the outcome of surgical endodontic treatments [10,11]. However, the wide range of available biomaterials, the treatment protocols, and the lack of standardization in assessment criteria lead to inconsistent and confusing results. Therefore, an evidence-based review and meta-analysis of the available literature regarding the influence of GTR techniques on the outcome of surgical endodontic treatment is necessary to help clinicians select the most predictable tissue regeneration technique for surgical endodontic treatment success.

Network meta-analysis (NMA) extends the principles of meta-analysis to the evaluation of several treatments in a single analysis, comparing multiple treatments simultaneously by combining direct and indirect evidence within an array of randomized controlled trials [12]. It is the best tool to examine the success rates of different procedures, such as GTR techniques in endodontic surgery.

The aim of the present study is to conduct a systematic review and NMA to analyze the influence of GTR techniques on the success rate of surgical endodontic treatment. The null hypothesis (H0) was that GTR techniques do not influence the success rate of surgical endodontic treatment.

#### **2. Materials and Methods**

#### *2.1. Study Design and Registration*

This systematic review and NMA was conducted following the Preferred Reporting Items for Systemic Reviews and Meta-Analyses (PRISMA, http://www.prisma-statement.org, accessed on 30 July 2020) guidelines. The review also fulfilled the PRISMA 2009 Checklist [13]. The registration number is CRD42020203447 (PROSPERO).

#### *2.2. Literature Search Process*

The search strategy was based on the following population, intervention, comparison, outcome (PICO) question: in adult patients undergoing endodontic surgery (P), does the use of regeneration techniques (I) compared to not applying regeneration techniques (C) influence the success rate (O)? An electronic search was conducted in the PubMed, Scopus, EMBASE, and Web of Science databases. The search covered all of the literature published internationally up to December 2021. The search included the following medical subject heading (MeSH) terms: "apicoectomy", "periapical surgery", "endodontic surgery", "periapical lesion", "surgical endodontic treatment", "root-end surgery", "root-end resection", "periradicular surgery", "guided tissue regeneration", "bone graft", "bone regeneration", and "membrane". The Boolean operators applied were OR and AND. The search terms were structured as follows: (("apicoectomy") OR ("periapical surgery") OR ("endodontic surgery") OR ("periapical lesion") OR ("surgical endodontic treatment") OR ("root-end surgery") OR ("root-end resection") OR ("periradicular surgery")) AND (("guided tissue regeneration") OR ("bone graft") OR ("bone regeneration")) AND (("membrane")). Two researchers (R.T. and A.Z.M.) independently conducted the database searches in duplicate. Titles and abstracts were selected by applying the inclusion and exclusion criteria. One researcher (R.T.) extracted data for the relevant variables. The systematic review was carried out by R.T., and two researchers not involved in the selection process (A.Z.M. and J.M.C.) performed the subsequent meta-analysis.

#### *2.3. Inclusion and Exclusion Criteria*

The inclusion criteria for the selected studies were as follows: randomized clinical trials (RCTs) that had a minimum follow-up period of at least 6 months; studies that analyzed GTR techniques (bone graft, membrane, membrane plus bone graft, platelet-rich plasma, or membrane plus platelet-rich plasma) or compared GTR techniques with a control treatment; patients that were 18 years old or older; and endodontic surgery procedures that were used to treat apical and/or apical-marginal lesions. No restrictions were placed on the year of publication or language.

The exclusion criteria for the selected studies were as follows: systematic or bibliographic reviews, clinical cases, case series, retrospective studies, and editorials and studies, including patients younger than 18 years.

#### *2.4. Data Extraction*

The following data were extracted from each study by independent reviewers (S.H.M. and J.M.A.): author and year of publication, title, journal in which the article was published, sample size (n), follow-up time, measurement procedure, type of GTR technique, success rate, periapical reduction, and bone density. The success of healing was analyzed according to the radiographic criteria established by Rud et al. [14] and Molven et al. [15], with complete healing defined as the reformation of periodontal space (intact lamina dura) with one cavity filled with bone (which can be of different radiopacity) and complete bone repair, but no discernable PDL around the apex. A third reviewer (P.V.B.) was consulted if the independent reviewers did not agree.

#### *2.5. Risk of Bias*

The risk of bias in the selected studies was assessed using the Cochrane Collaboration tool for methodological quality assessment of clinical trials [16]. This tool consists of 7 items that evaluate sequence generation, allocation concealment, participant blinding, assessment blinding, incomplete data, free selective reporting, and other sources of bias (Table 1 and Figure 1).

**Figure 1.** Risk of bias. Green color means "low risk of bias", and yellow color means "unclair risk of bias".


**Table 1.** Cochrane Collaboration tool for assessing risk of bias.

#### *2.6. Data Synthesis and Statistical Analysis*

The meta-analysis was carried out using a random effects model to estimate the success rate of endodontic surgery with and without GTR techniques, along with the confidence intervals. Heterogeneity among the combined studies for each treatment group was assessed using the I2 statistical index [17], which describes the percentage of total variation of studies due to heterogeneity and is not random. The effect of heterogeneity was quantified as being between 0 and 100% (low 0–25%, mild 25–50%, moderate 50–75%, high > 75%) [17]. The results of the meta-analysis are represented by forest plots.

Direct treatment comparisons were combined with a fixed effects model in a frequentist NMA to estimate indirect comparisons. The estimated effect size of the comparisons was analyzed by the OR. The inconsistency of studies included in the NMA was assessed with the Q test [18], with a significance level of *p* < 0.01, and a net heat plot [19].

Direct comparisons were performed using a NETWORK graph, and treatments were ranked on a scale of 0 to 1 using a P-score measuring the degree of certainty and indicating whether one treatment was superior to another [20].

Publication bias was analyzed using the trim and fill adjustment method for funnel plot asymmetry. In this analysis, each study was represented by a point, and the effect size and standard error were represented on the X-axis (logit transformed proportion). If there were no significant differences between the initial and adjusted estimates, the publication bias was considered to be low. R software was used with the Metaprop and Netmetaprop statistical packages to perform the meta-analysis.

## **3. Results**

#### *3.1. Results of the Search Process*

The systematic electronic search identified 159 articles in PubMed, 40 in Web of Science, 64 in EMBASE, 12 in Scopus, and 1 in the gray literature, which was found in the bibliography of a previous review [21]. Of the 276 articles, 56 were discarded as duplicates using RefWorks (https://refworks.proquest.com/reference/upload/recent/, accessed on 14 August 2020). After reading the titles and abstracts, an additional 130 articles were eliminated, leaving 90 articles; a further 55 articles were rejected because they did not fulfil the inclusion criteria: they did not include complete success rate data, did not use in vivo patient data, or presented a minimum follow-up time of less than 6 months. Finally, 11 articles were included in the qualitative and quantitative synthesis because they included all of the required data and variables (Figure 2).

#### *3.2. Qualitative Analysis*

All 11 articles that were included were randomized clinical trials [22–32]. Among them, 7 studies analyzed both clinical and radiographic parameters [23,24,26–28,30,31], and 4 studies analyzed radiographic parameters, such as bone density and periapical defect volume [22,25,27,32]. Most of the studies presented a sample size of approximately

25–30 patients, although the sample size ranged from 25 [30] to 101 [23], with subject ages ranging from 18 to 70 years and a follow-up time from 12 to 24 months. The results are presented in Table 2.

**Figure 2.** Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram.

#### *3.3. Assessment of Risk of Bias*

The methodological quality results were assessed using the Cochrane Collaboration tool and are shown in Table 1. All selected studies showed a low risk of bias related to incomplete data outcome assessment and other sources of bias. Moreover, most studies showed a low risk of bias related to sequence generation assessment and blinding of outcome assessors; however, most studies also showed an unclear risk of bias related to allocation concealment and participant blinding, and all studies showed an unclear risk of bias related to free selective reporting.


**Table 2.** Qualitative analysis of articles included in systematic review.


#### *3.4. Quantitative Analysis Results*

Odds ratios among regeneration techniques for the success of healing after endodontic surgery (meta-analysis):

Six meta-analyses of direct comparisons between GRT techniques (PL vs. control, MB vs. PL, MB + PL vs. MB, MB vs. control, Os vs. control, MB + Os vs. control) were carried out with the data obtained from the eleven selected RCTs. The meta-analysis of combined studies comparing MB-Os versus control (fixed effects model with the absence of heterogeneity; I2 = 0%) estimated a significant OR of 3.53 with a 95% confidence interval between 1.33 and 9.33. The remaining comparisons do not produce a significant OR (Figure 3).


**Figure 3.** Forest plot of ORs among guided tissue regeneration techniques for healing success after endodontic surgery. Column 1 lists the articles included in the meta-analysis. Columns 2 and 3 show us the results of the articles in the form of a proportion. Column 3 is the forest plot itself, the graphic part of the representation. It plots the effect measures for each study on both sides of the null effect line, which is the one for the odds ratio. In the lower part of the graph, the global result of the meta-analysis is represented. Column 4 describes the estimated weight of each study in percentage, and column 5 presents the estimates of the weighted effect of each one. Diamonds indicate the mean and confidence interval of combined effect, and squares indicate the mean and confidence interval of each study. Red lines represent the prediction interval.

Odds ratios were among the regeneration techniques for the success of healing after endodontic surgery (net meta-analysis).

Eleven RCTs (sixteen pairs of comparisons) were included in a frequentist NMA examining six GRT techniques (control, Os, PL, MB, MB + Os, and MB + PL) to analyze their influence on the success of healing after endodontic surgery. The data were combined with a fixed effects model (Mantel–Haenszel method). The nodes represent treatments, and the lines connecting the nodes are the six direct comparisons included in the NMA (Figure 4).

**Figure 4.** NETWORK plot of GTR techniques. Node size is proportional to the number of participants randomized to that technique, and the edge width is proportional to number of trials comparing two techniques.

The outcome of GTR techniques was estimated in terms of OR and 95% confidence interval. OR > 1 indicated that the treatment in the first column on the left was superior to the comparator, while OR < 1 indicated the opposite. Statistically significant ORs are shown in bold (*p* < 0.05). Direct comparisons (6/15) are highlighted in gray, and indirect comparisons are uncolored. Only two statistically significant ORs were found (in bold) (*p* < 0.05). The probability of obtaining a successful result was 3.67 times greater in the MB + Os group than in the control group (*p* < 0.05). The success of healing was 3.47 times greater in the MB + Os group than in the MB group (*p* < 0.05). The remaining comparisons among the groups do not show significance (*p* > 0.05) (Table 3 and Figure 5).

The ranking of the GTR techniques was performed according to the P-score, which measures the degree of certainty and indicates whether one alternative is superior to the others. The P-score is measured on a scale of 0 to 1. The MB + Os group presents the highest P-score (0.93), followed by MB + PL (0.60) and PL (0.53) (Figure 6).


**Table 3.** Comparison between GTR techniques using OR and 95% confidence intervals estimated in Netmeta. \* *p* < 0.05.

**Figure 5.** Forest plot of healing success using GTR techniques (odds ratio) compared to control group.

**Figure 6.** Ranking of GTR techniques by P-score.

No heterogeneity or inconsistency was found in the NMA (test of heterogeneity/ inconsistency Q = 0.29; *p* = 0.589). The net heat plot (Figure 7), which provides a detailed assessment of inconsistency, detected a very slight inconsistency between direct and indirect estimations, which was not significant.

#### *3.5. Publication Bias*

Six new studies were incorporated using the trim and fill method to adjust for funnel plot asymmetry, and a new OR for the six direct comparisons analyzed was estimated. No statistically significant differences were found with respect to the initially estimated OR (Figure 8).

**Figure 7.** Net heat plot. Gray boxes signify the importance of one treatment comparison to the estimation of another treatment comparison. Larger boxes indicate more important comparisons. Color background, ranging from blue to red, signifies the inconsistency of comparison (row) attributable to design (column).

**Figure 8.** Initial funnel plot after trim and fill adjustment of OR of periapical healing among guided tissue regeneration techniques.
