INTRODUCTION

External root resorption (ERR) is a multifactorial biological process that can be defined as the physiologic and pathologic dissolution of the mineralized tissues, including dentin, cementum, and adjacent alveolar bone, primarily as a result of osteoclastic cell activity.¹ The typical course of ERR development is initiated by the degeneration of the protective layer such as precementum, followed by the damage to the denuded and thus more susceptible root surface,² which might account for the rapid advance of root resorption.³ Therefore, the diagnosis of incipient ERR could be helpful to prevent the occurrence of severe ERR and the resulting tooth loss.⁴

ERR is one of the common and undesirable complications in orthodontic treatment.⁵ Previous studies have suggested that either individual predisposition or mechanical factors could serve as etiological factors of ERR.⁶ Lupi et al.⁷ observed that the incidence of teeth with ERR increased from 15% to 73% after orthodontic treatment. Histologic studies reported that the occurrence of ERR in orthodontically treated teeth was higher than 90%.⁸ Because there seems to be no effective approach to prevent its occurrence completely, the accurate diagnosis of ERR and early intervention to minimize its adverse effects is critical to orthodontists.

Unfortunately, a diagnosis of root resorption is difficult for clinicians because of a lack of pathognomonic symptoms. Periapical radiograph (PR) has been the most prevalent diagnostic tool for ERR in past decades. However, recent studies suggested that the diagnosis of ERR using PR could be highly inaccurate.⁹ In recent years, cone-beam computed tomography (CBCT) has been widely applied in dental clinics because it enables dental practitioners to visualize the three-dimensional (3D) data and eliminates the structural superimposition.¹⁰ However, the expenses and radiation of CBCT are significantly higher than that of PR.¹¹ Thus investigations of the actual diagnostic efficacy of CBCT and PR for ERR could help dental clinicians make more appropriate diagnostic decisions and treatment plans. At present, several studies have compared the diagnostic efficacy of CBCT for ERR in vivo.^12,13 These studies are unconvincing because of the lack of an appropriate gold standard for ERR in vivo. As a result of the difficulties in performing such investigations among patients, an increasing number of studies have explored the diagnostic accuracy of CBCT and PR for ERR in vitro. The in vitro studies enable ERR to be precisely simulated, thus consummating the study design and making the results more reliable. Therefore, reviewing the in vitro studies would be a better alternative to clarify this issue than dealing with in vivo studies. In the present study, based on currently available in vitro evidence, we carried out a systematic review to comprehensively compare the diagnostic accuracy of CBCT and PR for detecting ERR.

MATERIALS AND METHODS

This systematic review and meta-analysis was conducted and reported using the Preferred Reporting Items for Systematic Reviews and Meta-Analysis checklist.¹⁴ Two reviewer authors carried out the study inclusion and data extraction and evaluated the risk of bias and eligibility of retrieved studies independently. Any disagreement was solved by discussion with a third author.

Search Methods

An electronic search was undertaken in databases including Cochrane Central Register of Controlled Trials, EMBASE, PubMed, China National Knowledge Infrastructure, and System for Information on Grey Literature in Europe (SIGLE). The search strategies in PubMed were the combinations of Medical Subject Headings terms with free text words optimized for each database, respectively. The electronic search was conducted on August 7, 2016 with no language restriction. The specific search strategies are presented in Table 1. In addition, a manual search of relevant journals and reference lists of the enrolled studies was carried out by reviewing titles and abstracts.

Table 1. Search Strategies for Each Database

^a CENTRAL indicates Cochrane Central Register of Controlled Trials; CNKI, China National Knowledge Infrastructure; SIGLE; MeSH, Medical Subject Headings.

View Fullscreen

RESULTS

Description of Included Studies

A total of 720 records were retrieved from the literature search, among which 689 irrelevant articles were removed after screening. After that, the full texts of the reserved 31 citations were obtained and evaluated per the inclusion criteria. Finally, 15 studies were included in the systematic review^17–31(Figure 1). Among the enrolled studies, nine studies reported the diagnostic efficacy of CBCT with a sample size of 1097 suspected ERR sites, and 11 reported information of PR with a sample size of 1072 sites (five studies focused on both modalities). All studies adopted the artificially created ERR as the reference text. The detailed information of each study is presented in Table 2.

View Figure

• Download as Powerpoint
• Download Figure

Table 2. General Information of Included Studies^a

^a ERR indicates external root resorption; CBCT, cone-beam computed tomography; PR, periapical radiograph.

View Fullscreen

Quality of Included Studies

Among the 15 included studies, 11 were assessed with a medium risk of bias and the other four with a high risk of bias. In terms of applicability, 11 studies were assessed with low applicability concerns and the other four with medium applicability concerns. The detailed results of the quality assessment using QUADAS-2 are summarized in Table 3.

Table 3. Quality Assessment of the Included Studies According to the Quality Assessment of Studies of Diagnostic Accuracy-2^a

^a High indicates high risk; medium, medium risk; low, low risk.

View Fullscreen

Synthesized Results of Diagnostic Accuracy

The feasible data regarding diagnostic accuracy was statistically pooled using a bivariate model. Overall, the meta-analysis showed that CBCT had a mean sensitivity of 0.89 (95% confidence interval [CI]: 0.77–0.96), a mean specificity of 0.92 (95% CI: 0.84–0.96), a +LR of 11.24 (95% CI: 5.26–24.02), a −LR of 0.11 (95% CI: 0.05–2.27), and an AUC of 0.96 (95% CI: 0.94–0.98; Figure 2. For the PR, the pooled results showed a sensitivity of 0.68 (95% CI: 0.56–0.78), a specificity of 0.89 (95% CI: 0.82-0.94), a +LR of 6.39 (95% CI: 3.82–10.69), a −LR of 0.36 (95% CI: 0.26–0.50), and an AUC of 0.88 (95% CI: 0.85–0.90; Figures 3 and 4).

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

Comparisons between CBCT and PR of the diagnostic efficacy to detect ERR were performed. The results showed that CBCT had a significantly higher sensitivity (P = .005) and a significantly lower −LR (P = .006) than PR. Moreover, the AUC of CBCT was significantly higher than that of PR (P = .0002). The overall specificity and +LR were similar between the two imaging modalities (P > .05). The detailed statistic is summarized in Table 4.

Table 4. Comparison of Meta-Analysis Results on Diagnostic Efficacy Between CBCT and PR^a

^a CBCT indicates cone-beam computed tomography; PR, periapical radiograph; CI, confidence interval; Sen, sensitivity; Spe, specificity; +LR, positive likelihood ratio; −LR, negative likelihood ratio; AUC, area under the curve.

View Fullscreen

Subgroup Analysis and Meta-Regression

The sources of heterogeneity among the enrolled studies were determined using meta-regression and subgroup analysis. Risk of bias (0 = medium risk of bias; 1 = high risk of bias), publication year (0 = published in or before 2011; 1 = published after 2011), tooth type 1 (0 = single-root tooth; 1 = single- and multiroot tooth), and tooth type 2 (0 = incisors, 1 = premolars, 2 = others) were investigated in single-covariate meta-regression. The results suggested that the risk of bias significantly affected the assessment of diagnostic efficacy of CBCT (P = .0017; Table 5). Thus we performed a sensitivity analysis by omitting the studies with a high risk of bias (one study using CBCT, three studies using PR) and observed no significant changes in both modalities (Table 6). None of the other characteristics (publication year, tooth type 1, tooth type 2) introduced any heterogeneity to the overall estimates (Table 5).

Table 5. Results of Meta-Regression

^a SE indicates standard error; CI, confidence interval; RDOR, relative diagnostic odds ratio; Cte, constant term in the equation; S, indicator of threshold; CBCT, cone-beam computed tomography.

View Fullscreen

Table 6. Sensitivity Analysis and Subgroup Analysis^a

^a LR indicates likelihood ratio; AUC, area under the curve; CBCT, cone-beam computed tomography; ERR, external root resorption; PR, periapical radiograph.

View Fullscreen

Because ERR during orthodontic treatment was usually less than 2 mm,⁶ a subgroup analysis was performed to analyze the studies focusing on ERR cavities with diameters less than 2 mm, but no significant changes were found (Table 6). Both conventional and digital PR were used in the included studies. Thus we conducted a subgroup analysis and found that the diagnostic accuracy of conventional PR was similar to digital PR (Table 6).

Publication Bias

The funnel plots for publication bias are shown in Figure 5. Although slight asymmetry existed in the funnel plots, no statistical significance of the publication bias was observed for CBCT (P = .579) and PR (P = .919).

View Figure

• Download as Powerpoint
• Download Figure

DISCUSSION

Although ERR has long been considered as an iatrogenic complication in orthodontic treatment, the accurate detection and subsequent intervention still remain a challenge to orthodontists.⁶ Because of the lack of pathognomonic symptoms, ERR in orthodontic treatment is often noticed by routine examinations such as panoramic radiograph. Further radiographic techniques are normally required for diagnosis because of the weak diagnostic efficacy of a panoramic radiograph.³² Currently, CBCT and PR are the main tools to evaluate the presence of ERR. However, in most clinical situations, only one imaging technique will be performed. Thus we conducted a systematic review and meta-analysis comparing the efficacy of CBCT and PR in diagnosing ERR. Unfortunately, the current clinical trials seem to be unconvincing because there is no gold standard for ERR in vivo. Therefore, this study focused on well-designed in vitro studies.

To the best of our knowledge, the present study is the first systematic review and meta-analysis comparing the diagnostic efficacy of CBCT and PR for the detection of ERR. In this systematic review and meta-analysis, 15 studies were included for quantitative analysis. The meta-analysis indicated that CBCT is superior to PR in the accuracy of diagnosing ERR (Figures 2–4, Table 4). The subgroup analysis and sensitivity analysis showed that the overall estimates were consistent (Tables 5 and 6). Moreover, no significant publication bias was detected in the enrolled studies for both modalities (Figure 5). Thus in general, the pooled results in this meta-analysis were robust.

We found that CBCT was associated with a higher sensitivity (0.89; 95%CI: 0.77–0.96) when compared with PR (0.68; 95%CI: 0.56–0.78), suggesting that CBCT is a more reliable tool to detect the true ERR among suspected patients. The advantage in sensitivity is caused by the basic mechanism of CBCT, which enables examiners to observe three-dimensional images of suspected teeth, whereas limited information of ERR in the buccal or lingual root surface could be provided by PR. No significant differences in the pooled specificity were observed in the present study, indicating that the ability to rule out the false ERR among patients were similar between the two imaging techniques (Figures 2 and 3, Table 4). Because the trade-off effect always exists between sensitivity and specificity, either of the two measurements could not act as an independent indicator of diagnostic efficacy.³³ Thus the summary reviewer operator characteristic curve was been adopted to reflect the diagnostic accuracy of tested modalities in this systematic review.³⁴ The AUC of CBCT is 0.96 (95%CI: 0.94–0.98; Figure 4). This result could strongly demonstrate the high potency of CBCT in diagnosing ERR.³⁴ Moreover, the comparison of AUC indicates that CBCT is superior to PR in diagnosing ERR (Figure 4, Table 4).

Because high heterogeneity was detected when pooling data from all included studies (Figures 2 and 3), meta-regression and subgroup analysis were used to explore the potential sources of heterogeneity. Among the variables investigated in the meta-regression, only the risk of bias was found to affect the overall estimates of CBCT (Table 5). Thus the sensitivity analysis by omitting the studies with high risk of bias was performed, and no obvious changes were observed in the results of these two techniques (Table 6). As most ERR occurring in orthodontic treatment is less than 2 mm,⁶ we further analyzed the studies reporting data of simulated ERR with diameters ≤ 2 mm in the subgroup analysis. The outcome is consistent with the meta-analysis (Table 6). Two studies used simulated ERR cavities with diameters more than 2 mm (Table 2), and both of them reported the overall data of the ERR with various diameters rather than focused on the ERR with a diameter > 2 mm. Thus the subgroup with ERR diameters > 2 mm was not conducted. Both conventional and digital PR were involved in the included studies, and another subgroup analysis was performed based on these studies. The finding suggests no significant difference of the diagnostic accuracy between the two types of PR. The consistent outcomes in the sensitivity analysis and subgroup analysis could be the indicative of the robustness of the results in the meta-analysis.

The QUADAS-2 tool was tailored to produce review-specific guidance for the present study according to the QUADAS-2 group's suggestion.¹⁵ The signaling question “Was a consecutive or random sample of patients enrolled?” was omitted because no patients were involved in the recruited in vitro studies. The question “If a threshold was used, was it prespecified?” was disused because it is not applicable to the diagnosis of ERR. As the most commonly used tool for quality assessment in the diagnostic systematic review, QUADAS-2 assessed the studies based on the following two categories: risk of bias and applicability concerns.¹⁵ Because the simulated ERR was first created on the root surface as the reference test and then examined by CBCT or PR in all included studies, the case-control study design could hardly be avoided, and the domain concerning patient selection was assessed with a high risk of bias in all recruited studies according to the instruction of QUADAS-2.¹⁵ When omitting this inherent deficiency of in vitro studies and the resulting case-control design, 11 studies would be assessed as a low risk of bias. Moreover, when concerning the clinical applicability, this is more similar to the real clinical setting. The ERR in orthodontic treatment is often noticed in routine examinations using panoramic radiographs, and the suspected teeth would be further identified using CBCT or PR. In terms of applicability concerns, 11 of the 15 included studies were assessed as low applicability concerns (high clinical applicability; Table 3). Because the radiographic diagnosis for ERR is not a theoretical issue but a real clinical application, the category of applicability concerns should be more emphasized in the quality assessment. Thus we would assess the quality of the recruited studies in this systematic review to be with a medium or low risk of bias (Table 3).

The present review suggests CBCT to be a more precise technique for the diagnosis of ERR. However, current evidence also indicates that CBCT could bring about more cost and radiation exposure.³⁵ Because the diagnostic efficacy of PR (AUC = 0.88) is acceptable, we suggest patients with clinically suspected ERR be first evaluated using PR. Subsequent interventions should be performed when positive results are observed. When PRs display negative results, CBCT should be adopted for further examinations. Moreover, the cost-effective studies of this issue are needed for more reliable clinical guidance.

Although the present systematic review and meta-analysis was conducted carefully referring to the normalized process, some limitations still persisted that need to be addressed. First, the present systematic review only focused on in vitro studies, which might be different from the real in vivo situations. Second, the heterogeneity among included studies could influence the overall results. Although meta-regression and subgroup analysis were performed to detect the heterogeneity sources and sensitivity analysis by omitting confounding studies indicated the robustness of results. Some characteristics such as the diagnostic ability of examiners could not be assessed quantitatively. Third, the radiographic devices and exposure parameters vary among studies of both modalities, which might affect the overall estimates. Fourth, no cost-effective or radiation data were available in the enrolled studies, thus no comparison of the cost-effective ratio and radiation exposure could be provided. The results of this study should be interpreted in light of the aforementioned factors. More well-designed studies, especially clinical trials, are critical to obtain more reliable conclusions.

CONCLUSION

• This study suggests that CBCT could be reliable in detecting the presence of ERR in clinical practice, which has a higher diagnostic efficacy than PR.