Microbial Association with Genus Actinomyces in Primary and Secondary Endodontic Lesions, Review

The main reason for root canal treatment failure is the persistence of microorganisms after therapy, or the recontamination of the root canal system due to an inadequate seal. In the mouth, Actinomyces spp. constitute a significant part of the normal flora, which is indicative of their ability to adhere to oral tissue and resist cleansing mechanisms, such as salivary flow. This review, performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA), aims to clarify the prevalence of microbial genera that are associated with the genus Actinomyces in primary and secondary endodontic infections (primary outcome), and to identify the most prevalent species of the Actinomyces genus in endodontic lesions (secondary outcome). A total of 11 studies were included in the qualitative and quantitative analysis, and a total of 331 samples were analyzed. Bacteria of the genus Actinomyces were found in 58 samples, and 46 bacterial genera were detected in association with bacteria of the genus Actinomyces. Bacteria of the genus Streptococcus and Propionibacterium were those most frequently associated with Actinomyces in the endodontic lesions considered, and Actinomyces israelii was the most frequently involved species.


Introduction
The main reason for the failure of root canal treatment is the persistence of microorganisms after therapy, or the subsequent contamination of the root system due to an inadequate seal (Nair, P.N., 2004). In endodontic failures, the presence of microorganisms has been reported in 35% to 100% of cases, with Cheung et al. reporting the presence of cultivable microorganisms in 66% of samples from teeth with endodontic failures [1,2]. some isolated oral bacteria-for example, members of Eubacterium-are difficult to identify using morphological and biochemical methods [23], leading to the requirement of a combination of methods in order to confirm their identification biochemically; i.e., by sequencing the 16S rRNA gene.
Molecular genetic methods-in particular, PCR-have been widely used for microbial identification purposes. PCR tests are very sensitive, and can allow the identification of microbial species that are difficult to cultivate [11,24].
More information on the different bacterial associations present within the same primary and secondary endodontic lesion can help in outlining an optimal treatment strategy for eradicating the microorganisms associated with endodontic lesions.
The purpose of this review is to investigate the possible microbial associations of actinomycetes in endodontic infections. Actinomyces is one of the perpetrators of persistent intra and extraradicular infections, and knowledge regarding its possible microbial associations may be important in applying a suitable therapy for eradication. In addition, the persistence of infections on the external surface of the root apex, with the formation of a biofilm, often leads to the failure of antibiotic and endodontic therapies.

Materials and Methods
The following review was performed on the basis of PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analysis) [25] indications. The methodology has already been adopted in other systematic reviews on the topic (Actinomyces and Propionibacterium) by the same authors [26][27][28].
The PICO question are the following: • Population-patients with teeth with primary and secondary endodontic infections; • Intervention-microbial associations with the genus Actinomyces; • Control-patients with teeth that have no Actinomyces infections; • Outcome-odds ratio of microbial genera that are found in association with the genus Actinomyces in primary and secondary endodontic infections.
The primary outcome of the review is to answer the following questions: Which genera of bacteria are found in association with the genus Actinomyces in primary and secondary endodontic infections? What is the odds ratio of microbial genera that are found in association with the genus Actinomyces in primary and secondary endodontic infections? Finally, which among the species of the genus Actinomyces has the greatest prevalence in endodontic lesions (secondary outcome)?
After an initial selection phase, in which records were identified in databases, the potentially eligible articles were qualitatively evaluated in order to investigate the role of bacteria in endodontic infections and in apical periodontitis, with particular attention being paid to the role of Actinomyces in endodontic infections.

Eligibility Criteria
Scientific studies concerning the role of bacteria in primary and secondary endodontic lesions were considered. In particular, all studies that investigated the presence of microorganisms within dental elements subject to endodontic treatment or retreatment, conducted in recent years (40 years) and published with abstracts in English, were considered potentially eligible.
We decided to choose articles published within the last 40 years because an increasing number of new bacterial species have been identified since 1980 (according to the approved lists of bacterial names in Med. J. Aust. 1980, 2, 3-4) [29].
The potentially eligible articles were finally subjected to a full-text analysis so as to verify their use for qualitative analysis and quantitative analysis.
The inclusion and exclusion criteria applied in the full-text analysis were the following: • Studies were included if they identified both bacteria of the genus Actinomyces and bacteria of other genera in dental elements subjected to endodontic treatment or retreatment, or in the teeth subjected to apicectomy or extraction following endodontic failure; • Studies were excluded if they did not report the prevalence data for bacteria of the genus Actinomyces in the primary and secondary lesions of the dental elements, did not consider the microbial composition of each analyzed sample, tested the presence of only a few species of bacteria, were not written in English or were published before 1980.

Research Methodology
The articles were identified using electronic databases-namely PubMed and Scopus-and their bibliographies were examined and consulted in order to further identify articles.
The search for sources was conducted between 13.03.2020 and 25.03.2020.

Screening Methodology
Before the identification phase of records, the keywords to be searched and their combinations were first agreed upon by the two reviewers (with the task of selecting potentially eligible articles). The records obtained were subsequently examined by two independent reviewers (M.D. and C.Q.), and a third reviewer (G.T.) acted as a decision-maker in situations of doubt.
The screening included the analysis of the title and the abstract and, in cases of doubt, a text analysis to eliminate records that were not related to the topics of the review. The articles obtained were subjected to full-text analysis by the two reviewers (81 articles), from which those eligible for qualitative analysis and inclusion in the meta-analysis for the two outcomes were identified.
The results sought by the two reviewers were the following: (1) Primary outcome-which genera of bacteria are found in association with the genus Actinomyces in primary and secondary endodontic infections? What is the odds ratio of microbial genera that are found in association with the genus Actinomyces in primary and secondary endodontic infections? (2) Secondary outcome-the determination of the prevalence of the species of the genus Actinomyces that has the greatest prevalence in endodontic lesions.

Statistical Analysis Protocol
A meta-analysis was conducted on five sub-groups identified among the genus bacteria that had the highest number of positive samples together with the genus Actinomyces (primary outcome). The analyzed sub-groups were the following: Streptococci, Propionibacterium, Peptostreptococci, Staphylococci and Eubacterium. With the meta-analysis of the sub-groups, odds ratios (OR) were calculated to establish whether the bacteria of the respective sub-groups were more likely to present themselves in the samples with Actinomyces than in those without Actinomyces.
The protocol with which the meta-analysis was performed is based on the indications of the Cochrane Handbook for Systematic Reviews of Interventions. It was decided to use Reviewer Manager 5.4 (Cochrane collaboration, Copenhagen, Denmark) as a software for metanalysis [30]. In particular, pooled odds ratios (OR) and their 95% confidence intervals were calculated, and the inverse of variance test was applied to test for differences in overall effects between groups. The presence of heterogeneity was assessed by calculating the Higgins Index (I 2 ); if the measure proved to be higher than 50%, the rate of heterogeneity was considered to be high. The pooled results of meta-analysis were represented via forest plots for each of the analyzed sub-groups.
The risk of bias in the studies was calculated following the guidelines reported in the Newcastle-Ottawa Scale (NOS) for assessing the quality of studies in meta-analyses [31].
The risk of bias between studies was assessed graphically through the use of funnel plots and the calculation of heterogeneity determined through the Rev-manager 5.4 software.
A meta-regression was conducted with the use of Open Meta-Analyst version 10 (Tufts University, Medford, MA, USA) for those sub-groups that had high heterogeneity, reporting the risk of bias as a covariant.

Results
From searches in the PubMed, Scopus, EBSCO and Web of Science databases, 883 records were identified; furthermore, 51 articles included in the references of the identified full-text publications were selected. With the use of EndNote software, the overlaps were removed, resulting in 475 records. After the elimination of articles prior to 1980, 462 records remained. With the application of the eligibility criteria (all studies that studied the presence of bacteria in endodontic infection), we retained 81 articles.
Applying the inclusion and exclusion criteria, we retained 11 articles in the meta-analysis. All articles were analyzed according to the primary and secondary outcomes as defined above. All selection and screening procedures are described in the flowchart shown in Figure 1.

Study Characteristics and Data Extraction
The studies included for the quantitative analysis were those of Sunde [40].
The extraction of the data and the methods by which they have been reported follow the indications of the Cochrane Handbook for Systematic Reviews of Interventions, chapter 7 (selection of studies and data collection); specifically, from pages 156 to 182.
The extracted data included the bacterium species in the infection along with the bacterial species of the genus Actinomyces investigated (genus and species), the article information (data, author and journal), the number of samples examined, the types of samples (tooth in pulpitis or apical periodontitis, necrotic or vital tooth, tooth previously treated endodontically, endodontic canal, and tooth with failure subject to extraction or endodontic surgery), the number of samples for pathology in the presence of Actinomyces, and the bacterium identification method (culture or PCR).

Study Characteristics and Data Extraction
The studies included for the quantitative analysis were those of Sunde [40].
The extraction of the data and the methods by which they have been reported follow the indications of the Cochrane Handbook for Systematic Reviews of Interventions, chapter 7 (selection of studies and data collection); specifically, from pages 156 to 182.
The extracted data included the bacterium species in the infection along with the bacterial species of the genus Actinomyces investigated (genus and species), the article information (data, author and journal), the number of samples examined, the types of samples (tooth in pulpitis or apical periodontitis, necrotic or vital tooth, tooth previously treated endodontically, endodontic canal, and tooth with failure subject to extraction or endodontic surgery), the number of samples for pathology in the presence of Actinomyces, and the bacterium identification method (culture or PCR).
The data extracted for the two outcomes are shown in Table 2 and Table 3. The data extracted for the two outcomes are shown in Tables 2 and 3. Table 2 reports the number of samples of a particular bacterial genus found in association with Actinomyces, compared with the number of samples of Actinomyces. Then, the number of samples with that particular genus, compared with all the samples analyzed for each article, is reported. Table 3 reports the number of samples in which each Actinomyces species is present in each article.
For the studies selected for qualitative and quantitative analysis, a total of 331 samples were analyzed, and bacteria of the genus Actinomyces were found in 58 samples. For each sample, the microbial composition was available.
For the primary outcome, the bacterial genera present in the infections were considered together with species of the genus Actinomyces, and the prevalence relative to the infected samples together with Actinomyces was calculated in addition to the absolute prevalence relative to all the samples analyzed in each study (Table 4).      In some studies, only a cultural search of bacterial species was carried out; thus, an analysis by sub-group (cultures and PCR) was also carried out to remedy an evident limit of the review, as shown in Table 5. For the secondary outcome, the prevalence of each individual species of Actinomyces was calculated and compared with the total number of analyzed samples (Table 6). Table 6. Prevalence of the individual Actinomyces species, given the total number of samples for all articles selected for this review. We have the greatest number of positive samples with Actinomyces israelii and Actinomyces naeslundii.

Risk of Bias
The risk of bias was assessed using the Newcastle-Ottawa case-control scale, modified by the authors to adapt it to microbiological studies, as already done in previous systematic reviews with meta-analyses [26,27]. The results are reported in detail in Table 7. For each category, a value of one to three was assigned (where one = low and three = high). Studies presenting a high risk of bias were not included in the meta-analysis. Articles with a high bias risk were excluded from the scale and eliminated during the inclusion phase. Other articles were excluded because for the outcomes investigated; they presented the same data and samples.
The bias risk assessment of the 11 articles included was conducted by the first reviewer (M.D.). The risk of bias between the studies was considered low for five sub-groups of the primary research outcome; in fact, the heterogeneity that emerges from the meta-analysis shows an I 2 equal to 54% for sub-group 2 (Propionibacterium), 44% for sub-group 3 (Peptostreptococcus), 30% for sub-group 4 (Staphylococcus) and 0% for sub-groups 1 and 2 (Streptococcus, Eubacterium). The low heterogeneity is also confirmed by the funnel plot (Figures 2-6). three was assigned (where one = low and three = high).
Studies presenting a high risk of bias were not included in the meta-analysis. Articles with a high bias risk were excluded from the scale and eliminated during the inclusion phase. Other articles were excluded because for the outcomes investigated; they presented the same data and samples.
The bias risk assessment of the 11 articles included was conducted by the first reviewer (M.D.). The risk of bias between the studies was considered low for five sub-groups of the primary research outcome; in fact, the heterogeneity that emerges from the meta-analysis shows an I 2 equal to 54% for sub-group 2 (Propionibacterium), 44% for sub-group 3 (Peptostreptococcus), 30% for sub-group 4 (Staphylococcus) and 0% for sub-groups 1 and 2 (Streptococcus, Eubacterium). The low heterogeneity is also confirmed by the funnel plot (Figures 2-6).
For the second sub-group, graphical analysis of the funnel plot indicates the studies of Fukushima et al. 1990 [39] and Sunde et al. 2002 [32] as possible sources of heterogeneity and bias.
Graphic evaluation of the confidence intervals for the individual studies (forest plot) shows a good overlap for the Streptococci and Eubacterium sub-groups, and poor overlap for the Propionibacterium group, confirming the lack of heterogeneity in the Streptococci and Eubacterium sub-groups, and the high heterogeneity for Propionibacterium (Cochrane Handbook for Systematic Reviews of Interventions, chapter 9.5.2, identifying and measuring heterogeneity). Since heterogeneity is a sign of a possible risk of bias between the studies, it was decided to investigate meta-regression as a function of the risk of bias determined for each individual studies.            For the second sub-group, graphical analysis of the funnel plot indicates the studies of Fukushima et al. 1990 [39] and Sunde et al. 2002 [32] as possible sources of heterogeneity and bias.
Graphic evaluation of the confidence intervals for the individual studies (forest plot) shows a good overlap for the Streptococci and Eubacterium sub-groups, and poor overlap for the Propionibacterium group, confirming the lack of heterogeneity in the Streptococci and Eubacterium sub-groups, and the high heterogeneity for Propionibacterium (Cochrane Handbook for Systematic Reviews of Interventions, chapter 9.5.2, identifying and measuring heterogeneity). Since heterogeneity is a sign of a possible risk of bias between the studies, it was decided to investigate meta-regression as a function of the risk of bias determined for each individual studies.

Meta-Analysis
The statistical analysis of the data was performed using the Rev-manager 5.4 software (Copenhagen, 153 Denmark, The Nordic Cochrane Centre, The Nordic Cochrane Collaboration, 2014).
The meta-analysis of the first sub-group (Streptococci) showed an absence of heterogeneity with I 2 equal to 0%, and a fixed effects model was applied. The results shown in Figure 7 show that Streptococci are more likely to occur in samples with Actinomyces (OR = 2.49; 95% confidence interval (CI): [1.27, 4.86]).
The meta-analysis of the second sub-group (Propionibacterium) showed high heterogeneity, with I 2 equal to 56%, and a random effects model was applied. The results shown in Figure 8 show that Propionibacterium are not more likely to occur in samples with Actinomyces (OR = 1.26, 95% CI: [0.31, 5.13]).

Meta-Analysis
The statistical analysis of the data was performed using the Rev-manager 5.4 software (Copenhagen, 153 Denmark, The Nordic Cochrane Centre, The Nordic Cochrane Collaboration, 2014).
The meta-analysis of the first sub-group (Streptococci) showed an absence of heterogeneity with I 2 equal to 0%, and a fixed effects model was applied. The results shown in Figure 7 show that Streptococci are more likely to occur in samples with Actinomyces (OR = 2.49; 95% confidence interval (CI): [1.27, 4.86]). The meta-analysis of the second sub-group (Propionibacterium) showed high heterogeneity, with I 2 equal to 56%, and a random effects model was applied. The results shown in Figure 8 show that Propionibacterium are not more likely to occur in samples with Actinomyces (OR = 1.26, 95% CI: [0. 31, 5.13]).
With the identification and elimination of the two sources of heterogeneity, it is evident that I 2 drops to values equal to 0%; despite the elimination of the two studies, the forest plot does not report data with statistically insignificant odds ratios in favor of the samples with Actinomyces.
Furthermore, a meta-regression was conducted as a function of the risk of bias evaluation within the studies, in order to investigate whether the risk of bias within the studies could be a source of heterogeneity and bias between the studies. From the statistical analysis, we find a regression coefficient equal to −0.140, with a p value 0.639 (Table 8). The meta-regression data are not statistically significant, and the high heterogeneity index does not depend on the bias in the studies (Figure 9).   With the identification and elimination of the two sources of heterogeneity, it is evident that I 2 drops to values equal to 0%; despite the elimination of the two studies, the forest plot does not report data with statistically insignificant odds ratios in favor of the samples with Actinomyces.
Furthermore, a meta-regression was conducted as a function of the risk of bias evaluation within the studies, in order to investigate whether the risk of bias within the studies could be a source of heterogeneity and bias between the studies. From the statistical analysis, we find a regression coefficient equal to −0.140, with a p value 0.639 (Table 8). The meta-regression data are not statistically significant, and the high heterogeneity index does not depend on the bias in the studies (Figure 9).    The meta-analysis of the third sub-group (Peptostreptococci) showed average heterogeneity, with I 2 equal to 44%, and a fixed effects model was applied. The results shown in Figure 10 show that Peptostreptococci are more likely to occur in samples with Actinomyces (OR = 2.14, 95% CI: [1.1, 4.11]). The meta-analysis of the third sub-group (Peptostreptococci) showed average heterogeneity, with I 2 equal to 44%, and a fixed effects model was applied. The results shown in Figure 10 show that Peptostreptococci are more likely to occur in samples with Actinomyces (OR = 2.14, 95% CI: [1.1, 4.11]). The meta-analysis of the fourth sub-group (Staphylococci) showed average heterogeneity, with I 2 equal to 30%, and a fixed effects model was applied. The results shown in Figure 11 show that Staphylococci are not more likely to occur in samples with Actinomyces (OR = 1.54, 95% CI: [0.54, 4.37]). The meta-analysis of the fourth sub-group (Staphylococci) showed average heterogeneity, with I 2 equal to 30%, and a fixed effects model was applied. The results shown in Figure 11 show that Staphylococci are not more likely to occur in samples with Actinomyces (OR = 1.54, 95% CI: [0. 54, 4.37]). The meta-analysis of the fourth sub-group (Staphylococci) showed average heterogeneity, with I 2 equal to 30%, and a fixed effects model was applied. The results shown in Figure 11 show that Staphylococci are not more likely to occur in samples with Actinomyces (OR = 1.54, 95% CI: [0. 54, 4.37]). The meta-analysis of the fifth sub-group Eubacterium showed an absence of heterogeneity, with I 2 equal to 0%, and a fixed effects model was applied. The results shown in Figure 12 show that Eubacterium are more likely to occur in samples with Actinomyces (OR = 2.68, 95% CI: [1.10, 6.51]). The meta-analysis of the fifth sub-group Eubacterium showed an absence of heterogeneity, with I 2 equal to 0%, and a fixed effects model was applied. The results shown in Figure 12 show that Eubacterium are more likely to occur in samples with Actinomyces (OR = 2.68, 95% CI: [1.10, 6.51]).

Discussion
Follow-up studies report success rates of around 80-90% when canals are treated endodontically in aseptic conditions [5,41,42]. Endodontic failures mainly manifest when procedures are used that have not fulfilled the standard conditions for the elimination of microorganisms inside the endodontic lesion. Long-term follow-ups have demonstrated the presence of endodontic failures, with the presence of apical radiolucent lesions, even on teeth apparently treated adequately with procedures that meet high standards, demonstrating the persistence of infections that affect the apical portion of the dental roots.
Factors that may contribute to the perpetuation of periapical radio transparencies after root canal treatment include the following: an intraradicular infection that persists in the apical part of the root canal [42]; an extraradicular infection, generally in the form of periapical actinomycosis [42]; the filling of the extruded root canal, or other materials that cause reactions to foreign bodies [43][44][45]; and cysts, especially those with a significant accumulation of cholesterol crystals [46,47].
The samples considered in this review are primary and secondary endodontic lesions; of the 331 analyzed samples, 46 bacterial genera were detected in association with bacteria of the genus

Discussion
Follow-up studies report success rates of around 80-90% when canals are treated endodontically in aseptic conditions [5,41,42]. Endodontic failures mainly manifest when procedures are used that have not fulfilled the standard conditions for the elimination of microorganisms inside the endodontic lesion. Long-term follow-ups have demonstrated the presence of endodontic failures, with the presence of apical radiolucent lesions, even on teeth apparently treated adequately with procedures that meet high standards, demonstrating the persistence of infections that affect the apical portion of the dental roots.
Factors that may contribute to the perpetuation of periapical radio transparencies after root canal treatment include the following: an intraradicular infection that persists in the apical part of the root canal [42]; an extraradicular infection, generally in the form of periapical actinomycosis [42]; the filling of the extruded root canal, or other materials that cause reactions to foreign bodies [43][44][45]; and cysts, especially those with a significant accumulation of cholesterol crystals [46,47].
Other studies in the literature have examined biofilm formation in the canal space or on the outer surface of the apical portion of the root [13,48].
However, information on extraradicular infections resulting in a persistent lesion is limited, and mainly references Actinomyces or Propionibacterium species [32,33,49,50]. Bacteria that are difficult to grow are often only cultured through non-traditional methods, which leads to an underestimation of the bacterial diversity associated with persistent disease [51].
Most bacteria isolated from infected root canals are oxygen-sensitive, and cannot be grown using conventional bacteriological methods [52].
In previous studies that assessed the influence of infection on the treatment outcome, bacteriological techniques that were unfavorable for use in the recovery of anaerobic bacteria were used [53,54]. Therefore, the presence of bacteria that may have been important for the outcome of the treatment may have been precluded, and cases that apparently did not contain bacteria could, in fact, have hosted persistent microorganisms.
As PCR can overcome some of the intrinsic limitations of the culture process, it has contributed significantly to our understanding of the endodontic microbiota associated with primary infections [11].
The lesions analyzed in this review were both primary and secondary endodontic lesions; the bacteria were identified in the lesions by culture and PCR.
A study that identified multiple bacterial genera associated with Actinomyces was conducted by Niazi et al. (2010), who identified 35 bacterial genera in refractory endodontic lesions (9 with abscesses and 11 without abscesses) and 11 species of Actinomyces [37]. A considerable number of bacterial genera were also identified by Pinheiro et al. (2003) and Debelian et al. (1995), who identified 12 bacterial genera [14,40].
The meta-analysis of the sub-groups of the five bacteria most commonly found in the samples with Actinomyces highlighted how the bacteria of the genera Streptococci, Peptostreptococci and Eubacterium are more likely to be found in samples positive for Actinomyces, compared to the negative samples, with odds ratios of 2.49, 2.14 and 2.68, respectively. The meta-analysis of the Propionibacterium and Staphylococci sub-groups indicated how testing positive for these two genera of bacteria is found with the same propensity in samples that are positive, as those which are negative, for "Actinomyces".
For all the other bacteria, there are no indications from this meta-analysis suggesting their greater frequency in primary and secondary lesions with Actinomyces; in fact, the literature review shows us that many bacteria are more frequently found in endodontic lesions than in Actinomyces.
Bacteria of the genus Actinomyces are constantly being reclassified with the identification of new bacterial species. Bacteria that are present in the oral cavity, and that can potentially cause secondary or persistent infections, are commonly found in the gastroenteric system or in the mucous membranes of the urogenital tract [56,57].
A group of these bacteria has been shown to cause actinomycosis, progressing chronically as diseases manifesting abscesses associated with tissue fibrosis and draining sinuses, and sometimes mimicking malignant tumors [58][59][60].
Actinomyces israelii is the most commonly isolated species in human actinomycosis [61,62]. The loss of the integrity of the mucous membrane of the oral cavity, caused by extractions, bone and dental fractures, anesthesia, periodontal disease and the endodontic treatment of pulp exposures, can give rise to infection by these microorganisms, which, upon the interruption of the continuity of oral tissue, infect and invade the underlying tissues, thanks also to the selective conditions of anaerobiosis [63][64][65].
The limits of the study are the heterogeneity of the outcomes sought from the clinical studies included in the meta-analysis, sometimes with different microbiological identification methods, and the continuously updated taxonomy of bacterial species. The data are therefore to be considered as an indication (with an analytical basis) of which might be main bacteria associated with Actinomyces, which we consider to be one of the main culprits of perpetrators intraradicular and extraradicular infections.

Conclusions
Bacteria of the genera Streptococcus and Propionibacterium are those that are most frequently associated with Actinomyces in the considered endodontic lesions, and Actinomyces israelii is the most frequently involved species.
The microorganisms found in endodontic failures remain in the root canal after previous treatment, or enter during or after treatment through a leak. For all the other bacteria, the literature review shows us that many bacteria are more frequently found in endodontic lesions than in Actinomyces. therefore, thorough knowledge and a deep understanding of these endodontic microbes can assist in making decisions for further surgical treatment or reprocessing.