Treatment of Necrotic Teeth by Apical Revascularization: Meta-analysis

Each year ~5.4 million children and adolescents in the United States suffer from dental infections, leading to pulp necrosis, arrested tooth-root development and tooth loss. Apical revascularization, adopted by the American Dental Association for its perceived ability to enable postoperative tooth-root growth, is being accepted worldwide. The objective of the present study is to perform a meta-analysis on apical revascularization. Literature search yielded 22 studies following PRISMA with pre-defined inclusion and exclusion criteria. Intraclass correlation coefficient was calculated to account for inter-examiner variation. Following apical revascularization with 6- to 66-month recalls, root apices remained open in 13.9% cases (types I), whereas apical calcification bridge formed in 47.2% (type II) and apical closure (type III) in 38.9% cases. Tooth-root lengths lacked significant postoperative gain among all subjects (p = 0.3472) or in subgroups. Root-dentin area showed significant increases in type III, but not in types I or II cases. Root apices narrowed significantly in types II and III, but not in type I patients. Thus, apical revascularization facilitates tooth-root development but lacks consistency in promoting root lengthening, widening or apical closure. Post-operative tooth-root development in immature permanent teeth represents a generalized challenge to regenerate diseased pediatric tissues that must grow to avoid organ defects.

tooth-root growth 13 . Apexification is the current treatment for pulp necrosis in children and adolescents by filling the disinfected root canal with inert materials, but leads to ~45.9% post-operative tooth fractures in nine years 14 . Following tooth loss in children and adolescents, dental implants are contraindicated because metallic implants are embedded in the growing alveolar bone in children whose alveolar bone undergoes active growth, for which metallic implants cannot adapt 15 . Thus, pulp necrosis at pediatric age is a pandemic without a clinically satisfactory therapy, either before or after tooth loss.
Regenerative therapies that enable the treated necrotic immature permanent teeth to complete root development have been tirelessly pursued. In 1961, Nygaard-Østby reported the first recognized study of evoked bleeding 16 . Histologic sections from the patient's extracted teeth showed ingrowth of vascularized connective tissue in the root canal 16 . Subsequently, the practice of evoked bleeding in clinically treated necrotic immature teeth proliferated 14,17 . In 2011, the American Dental Association (ADA) issued clinical codes (D3351, D3352 and D3354) for the practice of evoked bleeding, also known as apical revascularization (AR) 13 . Presently, teaching of AR is incorporated in postgraduate endodontics training programs in the United States 8,9 . The European Society of Endodontology recently adopted AR 18 . Dental and/or endodontic societies in China and several other Asian countries are in the process of adopting AR.
To date, no meta-analysis has been reported on AR's efficacy. Bose et al. (2009) performed a retrospective evaluation of radiographic outcomes in immature teeth with necrotic dental pulp treated with apical revascularization 5 , with a collection of 54 published and unpublished clinical cases but without performing PRISMA or meta-analysis. The primary goal of Bose et al. (2009) was to compare two root-canal disinfection protocols. The primary objective of the present study is to perform a patient-level meta-analysis of tooth-root development among all qualified AR cases in the literature using PRISMA with strict inclusion and exclusion criteria 19,20 . As opposed to the majority of tissue-engineering work with a focus on regenerating adult tissues, AR aims to regenerate pediatric tissues, and is not considered successful unless the treated immature teeth complete root development 12 . In general, whether and how injured or damaged pediatric tissues regenerate to heal defects is largely elusive 21,22 . Necrotic immature teeth offer powerful models in both experimental animals and human patients for devising effective therapies that regenerate growing, pediatric tissues to complete organ development.

Results
Strict inclusion and exclusion criteria are shown in Table 1. The patients' age range was 8-18 years old, reflective of the population with immature permanent teeth (Table 2). Although the included 36 cases were treated by clinicians from multiple continents including Americas, Asia, Australia, and Europe, disinfection and intra-canal medicament protocols were substantially similar ( Table 2). Post-operative recalls ranged from 6 to 66 months (17.8 ± 11.6 months) ( Table 2). Figure 1 shows PRISMA flow diagram. A total of 320 reports resulted from 616 hits following removal of duplicates. Next, a total of 268 studies were excluded: 91 with titles and abstract not meeting the inclusion criteria; 76 review articles; 47 non-human animal studies; 51 for treatment without AR performed; and 3 in non-English language (Fig. 1). Among the resulting 52 articles, 30 were further excluded: 13 without adjacent reference teeth or anatomic reference points; 11 with immature reference teeth; 5 without dental radiographs or poor radiographic quality and one remaining study with <6-month recall (Fig. 1). Thus, 22 full-length articles that fit the pre-defined inclusion criteria and were immune from the pre-defined exclusion criteria were selected for meta-analysis, with a total of 36 cases ( Table 2). The Intraclass correlation coefficient (ICC) for the tooth length ratio, the optical width ratio, and the root area ratio were 0.85, 0.47, and 0.73, respectively,

Inclusion
• Human clinical studies assessing the treatment effect of apical revascularization on necrotic immature permanent teeth • Apical revascularization or evoked bleeding was the primary treatment protocol besides disinfection and antibiotics use  Table 1. Inclusion and exclusion criteria. *Unable to measure the cemento-enamel junction (CEJ), the most prominent incisal edge or cuspal edge or root apex, despite effort made by at least four clinically qualified coauthors and further verified by an oral and maxillofacial radiologist (S.J.Z.). **Elimination of one case with the concern of unfair disadvantage for apical revascularization due to insufficient postoperative time. The majority of the clinical cases had more than 6-month recalls.
SCientifiC RepoRTs | 7: 13941 | DOI:10.1038/s41598-017-14412-x indicating that strong inter-examiner agreement for tooth length ratio and root area ratio but moderate for apical width ratio.
Tooth-root lengths and root/crown ratios are of paramount importance to fracture rates and hence tooth loss 23 . Three cases were demonstrated in Fig. 2, which were selected from the included 36 patients to illustrate not only representative clinical outcome of tooth-root development and its diversity following AR, but also intrinsic radiographic image distortion. In Case 1 (PMID 15085044, corresponding to Patient #1 in Table 2), AR was performed in the necrotic second mandibular premolar of an 11-year-old male ( Table 2). The absolute tooth-root length (red line) indeed increased at 6-, 12-and 18-month recalls, so was root length of the reference tooth (blue line) also increased (Supplemental Clinical Data 1: Patient 1), with virtually no increases in root-length ratios: 0.88, 0.90, 0.87 and 0.86 ( Fig. 2A-D Table 2) represents a necrotic central incisor of a 16-year-old female (Fig. 2E). The linear tooth-root lengths of the treated tooth (red line) indeed increased at 6-and 12-month recalls, but so were the root lengths of the reference tooth (blue line) (Supplemental Clinical Data 1: Patient 21) (Fig. 2F,G). Accordingly, root-length ratios showed no substantial increases: 0.92, 0.93 and 0.94 for pre-treatment and 6-and 12-month recalls ( Fig. 2E-G), suggesting little gain in root lengths. Apical calcification bridge (ACB) was present at both 6-and 12-month recalls (Fig. 2F,G). Apical radiolucency was modest before treatment ( Fig. 2E), but became pronounced at 6-and 12-month recalls (Fig. 2F,G). Case 3 (PMID 24332005, corresponding to Patient #27 in Table 2) is a necrotic second mandibular premolar of an 11-year-old female (Fig. 2H). The root-length ratios were 0.77 pre-treatment and 0.88 by the 12-month recall (Fig. 2I), indicative of root lengthening. Remarkably, there was no apical closure, suggesting that apical closure and root lengthening are not necessarily associated with each other. Substantial pre-treatment apical radiolucency was resolved at 12 months (Fig. 2I). Table 3 provides root-length ratios of all 36 included cases. Statistical analysis of root-length ratios of all 36 cases lacked post-operative gain (p = 0.3472) (Fig. 3A), contrasting to self-reported 52.8% root lengthening in 19 of the 36 cases (Supplemental Table 1). Plots of root-length ratios of 36 cases are shown in Supplemental Fig. 1A. Radiographic images of all 36 included clinical cases with and without current tooth-length measurements are provided in Supplemental Clinical Data 1 and 2 for any interest in verifying data validity and reproducibility. Another important indication of tooth-root development is apical closure, or lack thereof, because immature permanent teeth without apical closure are 2.75 times more likely to fracture than fully mature teeth 24 . Apical closure status of post-operative immature permanent teeth was divided into three categories in reference to Moorrees stages of normal root development 25 Table 2). Apical closure was self-reported in 25 cases at 69.4%, along with no self-reporting on apical closure status in 10 cases at 27.8% (Supplemental Table 2). To resolve this discrepancy between self-reported apical closure status (Supplemental Table 2) and the present evaluation (Fig. 4), apical-width ratios were measured by dividing the linear apical-opening width against tooth-crown width at CEJ. Relatively high fidelity emerged in apical-width ratios (Supplemental Fig. 2A) with author-graded apical closure status (Fig. 4). Type I cases showed no significant apical narrowing (p = 0.1568), but apical narrowing was present in both types II cases (p = 0.0002; slope: −0.00261) and III cases (p < 0.0001; slope: −0.00655) (Supplemental Fig. 2B).  19,20 . Out of 616 records emerged, duplicates removed to yield 320 full-length publications. Cases were excluded using pre-defined inclusion and exclusion criteria in Table 1. Titles and abstracts of the 320 full-length studies without reviewing any data in each report's Result section, and excluded 268 studies to yield the resulting 52 studies. The full text of the 52 fulllength reports were carefully reviewed to further exclude 30 studies, with specific reasons as stated, to yield the final 22 studies included in meta-analysis with a total of 36 clinical cases.
SCientifiC RepoRTs | 7: 13941 | DOI:10.1038/s41598-017-14412-x Tooth-root lengthening was further analyzed among types I, II and III subgroups and again found no significant post-operative root-length gain in any subgroup: p = 0.1546 for type I, p = 0.4981 for type II, and p = 0.2439 for type III (Supplemental Fig. 1B). The average root-dentin area ratio showed significant post-operative gain (p = 0.0003), but this gain was restricted to type III apical closure cases (p < 0.0001), with no significant gain in root-dentin area ratio among type I or II cases (Supplemental Fig. 3B). Type II was associated with a significant decrease in root-dentin area ratio (p = 0.0229) ( Supplementary Fig. 3B), likely due to enlarged root-canal space following curved ACB formation. Root-area ratios showed high fidelity in differentiating type III from types I and II cases. Per receiver operating characteristic (ROC) analysis, type I was associated with root-area ratios < 0.58, type II between 0.58 and 0.79, and type III > 0.79.
Apical radiolucency in patients with necrotic dental pulp suggests metastasis of microbial infections to the peri-apical space. Peri-apical infections are severe clinical conditions, and if left untreated, may cause systemic infections 8,9 . With a uniform standard and by stratifying apical radiolucency into four categories, apical radiolucency indeed was resolved in 26 out of the total 36 cases at 72.2% (Supplemental Table 3). However, apical radiolucency remained present in 8 cases at 22.2% (Supplemental Table 3). Two cases were free from pre-treatment and post-operative apical radiolucency, accounting for 5.6% (Supplemental Table 3).

Discussion
Coagulated blood is perceived to harbor factors that heal tissue defects 26,27 . As a clinician initiated "regenerative" procedure, AR is analogous to other blood-clotting therapies such as microfracture in orthopedics where blood from bone marrow is induced to coagulate in focal articular cartilage defects for cartilage regeneration 28 . Can the same circulating blood induce the regeneration of different tissues such as articular cartilage and mineralized dentin? The intuitive answer might be that coagulated blood only serves as a scaffold, and the local environment determines what tissue is regenerated. However, local regenerative cues likely are insufficient to heal the defect in the first place, or otherwise there is no defect. What blood components, if any, activate local cues to regenerate specific tissues warrants experimental studies. Bioactive cues, local and/or blood-derived, must form  a chemotactic gradient to recruit cells for tissue regeneration. Whether coagulated blood yields a chemotactic gradient lacks experimental evidence, and should be investigated. For AR or any other regenerative therapies to succeed in enabling post-operative tooth-root development in children and adolescents, tooth roots must undergo axial growth (lengthening), transverse growth (dentin-wall thickening) and convergence growth (apical closure) (Supplemental Movie 2). A lack of significant tooth-root lengthening in the present study is to our surprise and at variation with 52.8% self-reported root lengthening among the included 36 cases. The present finding of a lack of root-lengthening following AR is directly supported by absence of tooth-length increase among 34 trauma cases of immature permanent teeth by comparing AR with Apexification by cone-beam CT measurements 37 . Intuitively, root lengthening is necessary for apical closure. However, the present data only support root-dentin wall thickening, but not root lengthening, even in type III apical-closure cases. The average absolute tooth-root length increased by 1.2 millimeters in 18 months following AR 29 , perhaps too small to be clinically meaningful and may be susceptible to intrinsic measurement errors. A lack of significant tooth-root lengthening following AR is probably attributed to the observation that apical calcification bridge 30 , present in 47.2% of all included AR cases as the largest subgroup, may limit axial root lengthening. We speculate that apical calcification bridge (ACB) formation, perceived as a bridge in 2D radiographs, is actually a band of transverse mineralization in 3D, and may not develop into a pointy and closed apex. ACB differs from apical closure as a distinctive subgroup, as confirmed by a lack of root-dentin widening in the ACB subgroup but increased root-dentin area in the apical-closure subgroup. Post-operative root lengths in individual  Table 3. Ratios of tooth-root length, apical width and root-dentin area before and after apical revascularization.
patients may indeed increase as shown in Case 3. However, a substantial root-length gain of AR-treated necrotic immature permanent teeth in children and adolescents as a population may not be a realistic expectation. Ratios of root lengths, apical widths and root-dentin areas were used in the present study to account for radiographic image distortion that is unavoidable among radiographs taken over time and by multiple practitioners 31 . The present study is limited to the teaching of clinical cases included, in absence of any prospective and randomized clinical trial on AR's efficacy. In experimental animal studies and several clinical case reports in which  the teeth were extracted following AR, ingrowth of vascularized soft connective tissue was present in the treated root canal 32,33 . However, there is remarkable variation regarding the frequency, amount and nature of AR-induced tissue ingrowth 32,34 , potentially accounting for the diversity of types I, II and III outcome of the included clinical cases. Together, post-operative tooth-root development represents a general challenge to regenerate pediatric tissues that must grow to avoid organ defects 21,22 . Whether circulating blood activates local molecular cues to regenerate different and specific tissues warrant experimental and clinical investigations.

Methods
Search strategy. PubMed, Scopus and Google-Scholar searches were performed using ((revascularization OR revitalization OR (regenerative endodontics) OR (regeneration)) AND (immature teeth) from inception of each database to September 15, 2016. Study selection: PRISMA, inclusion and exclusion criteria. Transparent Reporting of Systematic Reviews and Meta-analyse (PRISMA) was adhered in study selection (Fig. 1). Inclusion and exclusion criteria were pre-defined ( Table 1). As in Fig. 1, cases were excluded, for example, because the cemento-enamel junction (CEJ) or root apex was unrecognizable on radiographs despite independent effort made by at least four clinically qualified coauthors, and further verification by an oral and maxillofacial radiologist (S.J.Z.). Three coauthors independently screened the titles and abstracts of the 320 full-length studies without reviewing any data in each report's Result section, and excluded 268 studies (Fig. 1) strictly based on the pre-defined exclusion criteria (Table 1) to yield the resulting 52 studies. Subsequently, three coauthors carefully read the full texts of the 52 full-length reports and further excluded 30 studies, with specific reasons stated in Results below (also c.f. Fig. 1).
Apical revascularization. The following generic AR treatments were representative among the included studies ( Table 2). Following rubber-dam isolation, local anesthesia and tooth access preparation, necrotic root canals were disinfected using irrigants such as sodium hypochlorite and chlorhexidine at the first visit. Calcium hydroxide or antibiotic pastes were applied into the canals as inter-appointment medicaments ( Table 2). At the second visit, root canals were cleaned with irrigants to remove the intracanal medicaments. Apical bleeding was induced by passing a hand instrument beyond the apical foramen to allow blood filling in the canal. Mineral trioxide aggregate was placed directly over the coagulated blood clot. The coronal access was sealed with permanent restorations.
Quantitative measurements of tooth-root development. To minimize potential bias, radiographs of all 36 cases were assigned to two clinically qualified coauthors with randomized and blinded radiograph sequence. Tooth-root development was quantified digitally on pre-and post-operative radiographs: root lengths as ratios of the treated and adjacent reference teeth, root-dentin geometry by subtracting root-canal area from tooth-root area, and apical closure as the ratio of crown and apical opening widths. The linear tooth-root length was measured independently by the two examiners from the CEJ to root apex digitally on all pre-and post-operative radiographs per existing methods 31,35 . To compensate for potential radiographic distortion with images taken over time and by multiple practitioners in the literature, the linear root length of not only the treated teeth (TT), but also the adjacent reference teeth (RT) were measured as long as the reference teeth had already completed root development 31 . TT/RT ratios were calculated not only to minimize radiographic image distortion, but also for cross-patient comparisons 31 . For immature permanent teeth with open apices, a linear line was drawn to connect apical tips with the center of the transverse line used as the apical end for root-length measurements 31 . Apical calcification bridge 30 was measured as a transverse mineralization band that joined apical root tips 30,31 blindly and independently by the two examiners. Root-dentin area was measured by subtracting root-canal area from the total root area that was defined from the apex to a transverse line connecting two CEJs 36 . Root-dentin area ratios were calculated by dividing root-dentin area against the total root area. Apical-width ratios were calculated by dividing apical-opening width against the width of a transverse line connecting the two CEJs per tooth. Apical radiolucency was stratified into four categories: −/−: no apical radiolucency before or after treatment; −/+: no apical radiolucency before treatment but detected apical radiolucency following treatment; +/+: apical radiolucency both before and after treatment; +/−: apical radiolucency before treatment but resolved following treatment.

Statistical analysis.
Random effects approach was adopted in absence of consistent measure of tooth-root development among the included 22 studies. The meta-analysis investigated whether tooth-root length ratios, apical width ratios, and root area ratios had any significant change following AR using a linear mixed effects model where recall time was the main predictor, controlling for the value at baseline, and the type of the treated tooth (e.g. incisors vs. premolars) if it proved significant. Random effects were included in the model to account for variations among the included 22 studies. Nested random effects within each study were included to account for variations among subjects within each study, with additional random effects included to account for variations between the two independent examiners. Intraclass correlation coefficient (ICC) was used to evaluate the inter-examiner reliability of the three outcomes. We further investigated whether tooth-root length ratios, apical width ratios, and root area ratios differed significantly among the three subgroups of apical closure status by assessing the moderating effect of apical closure type. For any significant post-operative change, cutoff values were determined by the receiver operating characteristic (ROC) method for apical closure prediction. All statistical analysis was conducted using SAS version 9.4, and plots were generated using R version 3.0.1. A p-value < 0.05 was considered as statistically significant.