Early wound healing outcomes after regenerative periodontal surgery with enamel matrix derivatives or guided tissue regeneration: a systematic review

In periodontal defects, are the early wound healing outcomes after periodontal regenerative surgery better after the use of EMD as compared to GTR?

The studies were selected according to the following criteria:

A comprehensive and systematic electronic search of US National Library of Medicine (Pubmed) was performed. The search was conducted for trials in the period up to July 2018. The following key words were used: (guided tissue regeneration OR regenerative OR Emdogain OR enamel matrix derivatives OR amelogenin OR membrane) AND (periodontitis OR periodontal therapy OR periodontal surgery) AND (RCT OR clinical trial).

The literature research was performed without language restrictions.

Previous to the screening process, the first 50 titles and abstracts were used to calibrate the two reviewers (MR and LM) with a senior researcher (PS). Consequently, two reviewers (MR and LM) screened independently all titles and abstracts. Then, studies potentially complying with the inclusion criteria were selected for full text assessment. After independent assessment, any disagreement between both reviewers was resolved by discussion with a third reviewer (PS).

Relevant data, including population characteristics, intervention sites characteristics, description of the treatment prior to and at completion of the interventions, post-surgical indications and medications, time of the study, maintenance therapy characteristics and early wound healing parameters assessed were independently extracted by two reviewers (MR and LM).

Following the guidelines of the Cochrane Collaboration [34] a quality assessment of the included studies was performed independently by MR and LM.

Therefore, six domains were evaluated: 1) sequence generation, 2) allocation concealment, 3) blinding of participants and outcome assessors, 4) incomplete outcome data, 5) selective outcome reporting, 6) other sources of bias. In each assessment tool previously mentioned, a judgement of “Yes” or “No” indicated low and high risk of bias, respectively; whereas “Unclear” judgement indicated uncertain risk of bias.

A study was assigned as “Low risk of bias” when all the domains were of low risk of bias. However, when one or more key domains resulted with unclear or high risk of bias, the study was assigned as “Unclear or High risk of bias”.

The quality of non-randomized clinical trials was assessed using the Cochrane Collaboration tool -ROBINS-I tool- (“Risk Of Bias In Non-randomized Studies - of Interventions”) [35, 36]. Seven domains were evaluated: pre-intervention: 1) bias due to confounding, 2) bias in selection of participants into the study; at intervention: 3) bias in classification of interventions; post- intervention: 4) bias due to deviations from intended interventions, 5) bias due to missing data, 6) bias in measurement of outcomes, 7) bias in selection of the reported result. The bias of the studies was assigned as follows:

Any disagreement for data extraction and quality assessment was discussed and resolved by consensus. A third reviewer (PS) was consulted when necessary.

The search strategy generated 968 potentially fitting articles. After title and abstract screening, 26 articles were eligible for possible inclusion (Fig. 1). During full-text assessment, nineteen articles were excluded due to smoking (nine articles) [37‐45], patients younger than 30 years of age (three studies) [46‐48], diagnosis of aggressive periodontitis (two studies) [49, 50], missing outcomes for early wound healing evaluation (one study) [51] and not-fitting treatments group (one study) [52]. Moreover, three studies [53‐55] were found to report long-term results of already included publications and were likewise excluded. Finally, seven studies were included [56‐62]. Reviewer agreement for title and abstract screening was 92% and agreement for the full text screening before discussion was 88%(Table 1).

To assess and detect similarities and differences between the studies - and to determine if it was possible to perform a further synthesis or comparison methods - data were summarized into evidence tables and a summary was performed.

Considerable heterogeneity was found in the studies, regarding the whole range of assessed parameters, including different follow-up time period evaluation, study design and specific GTR treatment. Moreover, the morphologic baseline characteristics of the surgical sites showed strong heterogeneity between groups of different studies, or it was often not reported. For this reason, it was not possible to conduct a reasonable data synthesis for the included studies and a meta-analysis could not be performed.

Six studies [56, 58‐62] were randomized trials. According to the guidelines of the Cochrane Collaboration [34], three studies showed a high risk of bias [59, 60, 62] and three studies an unclear risk of bias [56, 58, 61] (Table 2).

One study [57] was reported as a non-randomized clinical trial (a case-cohort study) and according to Cochrane Collaboration tool -ROBINS-I tool [35, 36], resulted to be a low risk of bias (Table 3).

Table 7 illustrates the early wound healing outcomes of the seven included studies.

Flap dehiscence was evaluated in two studies [57, 61] and 79 sites. Dehiscences were observed in 6/50 (12%) of the GTR treated sites and in 3/29 (10.3%) of the EMD treated sites. Membrane exposure was evaluated in five studies [56, 58‐60, 62] and was registered in 21/73 (28.8%) of the defects. In one of these studies [56], in which 7 of the 16 (43.7%) of the GTR treated defects showed exposition of the membrane at 3 weeks, swelling also was observed in 8 of the 16 sites at the first post-surgical week. Flap dehiscence, however, was not registered in sites treated with EMD while swelling were found in the same number of cases.

When flap dehiscence and membrane exposure were evaluated together it was observed that flap dehiscence was registered only 3.1% in the EMD treated sites whereas flap dehiscence/ membrane exposure was observed in the 22% of defects treated with GTR.

In all the remaining studies none of the others parameters evaluated (suppuration, abscess formation and/or allergic reaction) were observed. In only one study [60] the early wound healing was reported to be uneventful.

It has been reported that the use of inter-dental tissue preservation surgical techniques provides a better flap stabilization [66]. It is important to note that in the three studies [56, 60, 62] where intracrevicular incisions were made without any interdental surgical preservation technique, membrane exposure was observed in almost half of the GTR of the treated defects (42.4%); conversely, when SPPF or MPPT were used [57‐59, 61] flap dehiscence/membrane exposure was registered in only 10.3% of the EMD treated defects [57, 61] and in the 14.4% of the GTR treated sites (Table 6).

Suturing is one of the most important factors related to wound stability [67], especially during the first post-surgical weeks when adherence of the flap to the underlying hard tissues is only guaranteed by a thin blood clot that is converting to fibrous and osseous tissue [68, 69]. In fact, in regenerative procedures the suture is normally removed after 10 to 14 days post-surgery [68, 70]. It has been demonstrated that the use of thick sutures (4–0) and/or early suture removal can result in dehiscences of the formerly adapted flaps [71]. In the studies included in the present review this was not explicitly reported, even in three studies [56, 60, 62] in which the suture was removed at 14 days the percentage of sites with membrane exposure was higher (42.4%) as compared to studies where this time was shorter (7–10 days), with only 13.4% of the treated sites with dehiscence/membrane exposure [57‐59, 61] (Table 6).

The post-surgical controls and maintenance therapy have also been evaluated in this article. In general, the first follow-up visit was scheduled 1 week after surgery [72] and, in regenerative therapies with membranes, the recall visits turned out to be more frequent during the first 2–3 weeks, when professional tooth cleaning was performed [67, 73]. In all evaluated studies the first post-surgical control was performed at one post-surgical week and the recall visits were indicated every 1 or 2 weeks for the first 6–8 weeks. In one study [61] the first evaluation was made at 5 days and, at this time, a wound dehiscence could be observed in the 14% (5/39) of the treated sites. This aspect is of paramount importance because early controls might help to detect early complications as can be a small flap dehiscence without a membrane exposure. Moreover, the time to resume hygiene oral procedures must be considered since it has been reported that only after 4–5 weeks the flap is completely reattached to teeth and bone [65, 68]. This period was considered in all the studies and the oral hygiene procedures were resumed between 4 to 6 post-surgical weeks (Table 6).

It has been demonstrated that non-resorbable membranes have a higher risk of exposure than resorbable membranes in GTR procedures [74]. In the included studies, no such effect was shown. Of 123 sites treated by GTR, in 71 resorbable membranes were used and in 52 e-PTFE membranes. Surprisingly, flap dehiscence/membrane exposure was present in 30% of the sites treated with resorbable membranes, whereas only in 11.5% sites with non resorbable e-PTFE membranes (Table 6). Regarding this point, it is important to underline that in the selected studies 20 of the 52 treated sites were 3-wall contained intrabony defects. Studies [11, 75, 76] show a good prognosis when these defect types are treated with others surgical approaches or with resorbable membranes.

It is important to highlight the fact that, when all the previous mentioned factors related with the early wound healing were evaluated, the group treated with EMD presented a lower percentage of sites with post-surgical complications respect the GTR group. This could be related to the aforementioned properties of the EMD in the wound healing [15, 31].

In fact, if both evaluated parameters (dehiscence and membrane exposure) are considered as one kind of post-surgical complications, complications were observed less often after EMD procedures than after GTR procedures. This is in agreement with the results observed in a systematic review [15] and clinical studies [41] in which more post-surgical complications following GTR than after EMD application have been reported. In fact, in a multicentre clinical trial [41]in which 75 patients were treated, it was observed that all cases treated with GTR presented a post-surgical complication, mostly membrane exposure, while only 6% of EMD treated defects showed complications. This study was not included in the present revision since smokers were also evaluated.

Another important parameter is the administration of systemic medications and especially antibiotics after or during the surgical procedures. While a few studies concluded that better healing and less discomfort is observed when antibiotics were given [77, 78], in many other studies [79‐82] the use of antibiotics was considered not necessary. Although there is currently no consensus regarding this aspect, this parameter was assessed in the present review (Table 6) in order to avoid a possible bias but, given that in all studies drugs and posology were vastly different, establishing any conclusion from the given data seems inappropriate. However, it was observed that in two of the studies [58, 59] in which antibiotics were not administered but only anti-inflammatory drugs (Ibuprofen or Acetaminophen), the early post-surgical complications (17.5% GTR group and 0% EMD group) were not more frequent than in the studies with antibiotic administration. In fact, the complications registered were even more frequent in the “antibiotic group” [56, 57, 60‐62], in which 25% of the GTR and 17.8% of the EMD group presented membrane exposure/dehiscence and/or swelling. This is coincident with a previous clinical study [82] that evaluated the role of antibiotics in preventing early post-operative complications after periodontal surgical procedures. The evaluation was performed 1, 2, 4, 7 days and 3 months after surgery and 3 groups were evaluated (amoxicillin, doxycycline and no antibiotics). The authors reported no differences in terms of early complications between the three groups. They concluded that performing the surgical procedures following strict asepsis the prevalence of complications is low. Accordingly, prophylactic antibiotic to prevent post-operative complications was considered unnecessary.

Two important aspects related to the included studies should be especially considered: first, the already mentioned heterogeneity observed among all the studies and especially for defect morphology and second the studies ‘quality.

With respect to the first point, one of the most notable differences between the studies was the morphology of the defects (Table 5). Although generally in all studies intrabony defects were included, the spectrum ranged from “advanced intrabony defects” (scheduled for extraction) over 3-wall defects, partially non-containing defects and GIII furcation defects. However, a descriptive analysis distinguishes the different types of defects with respect to the individual treatment that was performed.

In fact, the defect morphology –related also with the surgical approach and the biomaterials selected - strongly influences the results of the surgical procedures [11, 57, 75, 76]. Clinical success was reported when contained intrabony defects were treated with EMD. In non-contained intrabony defects, GTR procedures are more indicated [11, 75, 76] although it has been observed in a recent study successful clinical results when non-contained intrabony defects were treated with EMD [83].

In the present systematic review, when the 2–3 wall contained intrabony defects were evaluated it was observed flap dehiscence/membrane exposure in 26% of the sites treated by GTR whereas flap dehiscence was observed in only 6.5% of the sites treated with EMD [56, 57, 61]. Instead, in non-contained defects [57‐59] 14% of the GTR treated sites showed membrane exposure while no post-surgical complications were observed in the EMD group.

Finally, in the study treating furcation III defects [62], membrane exposure was observed in seven of the ten treated sites (70%). This study was the only one that compared EMD with either GTR or GTR + EMD. When the membrane exposure was assessed in the GTR groups 2 (67%, with GTR) of 3 and 5 (71%, with GTR + EMD) of 7 sites were find to show that. Although no meta-analysis was performed due to the small power of the published data it seems that EMD as an adjunct to GTR did not provide an additional benefit in the treatment of furcation GIII defects. At the final follow-up, the results also demonstrated that only a partial closure of the furcation entrance was achieved. This finding was in accordance with a previous clinical study [84] which could not be considered in this review since smokers were included.

Regarding the quality of the included studies [34‐36], the present review included six RCTs and one non-randomized clinical trial (a case-cohort study) comparing EMD and GTR surgical procedures. The quality of the  RCTs were found to be moderate to low, considering that three studies showed a high risk of bias [59, 60, 62] and three studies an unclear risk of bias [56, 58, 61] (Table 2). The only study reported as non-RCT [57] resulted to have a low risk of bias (i.e., the study is comparable to a well-performed randomized trial with regard to all the domains; Table 3). Furthermore, a potential bias regarding the funding sources has to be considered. In concerning this matter there was only one study included that was supported by external companies that sponsored the different biomaterials used for both study groups (EMD and GTR) [62], what rendered the risk for bias rather low.

In the present systematic review, we decided to exclude smokers to avoid possible bias considering that smoking affects the wound healing process [85]. Indeed, in a recent clinical study [86] in which the impact of smoking status on the clinical outcomes after regenerative surgical procedures were evaluated, the authors concluded that in smoker patients wound healing quality was significantly hampered when compared to non-smokers. A dose-dependent effect of smoking was observed with respect to the values of PD reduction and CAL gain at 6 months with a tendency to lower values in patients consuming 11–20 cigarettes/day than in smokers from 1 to 10 cigarettes/day. Accordingly, even light smokers were excluded from the present analysis.

Specifically assessing early wound healing outcomes, there are no RCTs comparing EMD and GTR for the treatment of intrabony defects. Most of the studies focus however on the long-term clinical outcomes after 12 months [76]. In addition, in none of the studies included in the present revision, early wound healing was evaluated with any of the indices/systems already proposed in the literature [23, 24, 86].

Within the present systematic review, no relevant differences in the early wound healing results between EMD and GTR surgical treatment in periodontal intrabony defects can be found, although - when a deeper and detailed evaluation of the studies was performed - a tendency for better early healing in the group treated with EMD seems evident. Particularly, when the analysis was performed considering the different defect types it was observed that both contained and non-contained intrabony defects presented a higher percentage of dehiscence/membrane exposure when GTR treatment was performed. These findings however should be interpreted with care given the heterogeneity and the quality of the studies included. The higher risk for dehiscence and membrane exposure in GTR procedures, however, cannot be interpreted as general superiority of EMD in the early wound healing of the treatment of intrabony defects. In fact, when only flap dehiscence was analysed the results observed were similar for both treatment groups (12% GTR versus 10.3% EMD treated sites).

Therefore, future RCTs comparing EMD and GTR surgical procedures in terms of early wound healing are necessary to understand if EMD presents an additional benefit in this regard. In order to render a quantitative meta-analysis possible study designs should be standardized to reduce heterogeneity and possible biases. Moreover, long-term studies that compare early wound healing outcomes to the final results would allow to deepen the insight into the effect of uneventfully early healing.

Finally, it is important to mention that the purpose of the present systematic review is not to suggest the one or the other treatment type. Clinically, the decision for/against EMD/GTR is multifactorial [76] and depends especially on the defect morphology. Furthermore, the number of sites to be treated and their localization might be of relevance, since multiple defects or defects difficult to reach might easier and less expensively be treated with EMD due to a quicker and easier application as compared to the GTR protocol.