Use of preoperative cone-beam computed tomography to aid in establishment of endodontic working length: A systematic review and meta-analysis.

PURPOSE: This study was performed to assess the accuracy of preoperative cone-beam computed tomography (CBCT), when justified for other reasons, in locating the apical foramen and establishing the working length.
MATERIALS AND METHODS: Six electronic databases were searched for studies on this subject. All studies, of any type, were included if they compared measurements of working length with preoperative CBCT to measurements using an electronic apex locator (EAL) or histological reference standard. Due to the high levels of heterogeneity, an inverse-variance random-effects model was chosen, and weighted mean differences were obtained with 95% confidence intervals and P values.
RESULTS: Nine studies were included. Compared to a histological reference standard, CBCT indicated that the apical foramen was on average 0.40 mm coronal of its histological position, with a mean absolute difference of 0.48 mm. Comparisons were also performed to an EAL reference standard, but the conclusions could not be considered robust due to high levels of heterogeneity in the results.
CONCLUSION: A low level of evidence is produced suggesting that preoperative CBCT shows the apical foramen to be on average 0.40 mm coronal to its histological position, with a mean absolute difference of 0.48 mm.

Introduction

Preoperative cone-beam computed tomography (CBCT) should not be used for the sole purpose of estimating the endodontic working length. However, with increasing numbers of tomograms being taken for other reasons, CBCT could still be helpful in the estimation of this measurement.

It is important that the positions of endodontic files, irrigation needles, and obturation are controlled in relation to specific topographical landmarks on the internal surface of the apical root canal wall. Several topographical landmarks have been described, notably the apical foramen (AF) (also known as the major diameter), the apical constriction (AC) (also known as the minor diameter), and the apical vertex (AV). The classical description is that the root canal is widest at the coronal orifice and tapers apically to its narrowest point, the AC, which is near the root apex. The canal is then described as widening for a short period in a funnel shape until it reaches the AF, which is the point where the canal meets the external surface of the root. Often, the AF is not located at the most apical portion of the root (the AV) but rather is slightly or markedly offset.1

Establishing the true positions of the AC and AF is important. Many endodontists believe that the working length should be terminated at the AC because this will create the smallest possible wound and promote healing.2 These clinicians believe that the extension of instruments, irrigant, endodontic core materials, or sealers to or beyond the AF affects healing due to damage to periapical tissues.2 Other endodontists argue that this approach could leave a portion of potentially infected canal between the AC and AF untouched by mechanical or chemical debridement. Currently, insufficient high-quality evidence is available to provide definitive answers to these questions.3

Several techniques exist for establishing the location of these landmarks and therefore the working length. Each has advantages and disadvantages, and accuracy varies. Periapical radiographs can only be used to identify the radiographic apex (RA), which correlates with the AV rather than the AC or AF, and often leads to overestimation of the working length.

Electronic apex locators (EALs) are believed to be usable to locate the AC and AF; however, their accuracy is subject to debate. The accuracy of EALs was first assessed in relation to the RA, but this has limited validity, as the RA and AF are often substantially offset.4 While subsequent studies involved the longitudinal sectioning of extracted teeth, micro-computed tomography (micro-CT) studies have since revealed the inaccuracies of this technique.56

In a recent study, micro-CT was used to measure the accuracy of EALs, since the former is highly valid for locating the true position of the AC.7 That study demonstrated that several apex locators were highly accurate in identifying the AC and AF. The only drawback of the in vitro study was the use of electroconductive materials to replace the periodontal ligament and extracted teeth.7 The validity of this technique is questionable, as it is unclear whether the electroconductive materials can replicate the exact functioning of the periodontal ligament.8

CBCT is a relatively new radiographic imaging system that can provide images in 3 dimensions. Recently, it has been established that the AF can be identified on preoperative CBCT, and proprietary measuring software can be used to measure its distance from a coronal reference point (CRP).91011

The main concern related to radiographic imaging is that it may cause a patient to develop cancer years after an exposure (a somatic stochastic effect).12 Any radiographic imaging must show a potential net benefit to the patient over the risks of malignant change from exposure, considering the efficacy, benefits, and risks of available alternative techniques. Unfortunately, even under optimized conditions, CBCT provides a dose that is 4 to 10 times higher, and therefore carries a higher risk of a somatic stochastic effect, than the alternative technique (periapical radiography).13

Because of this higher risk, it is not currently recommended that preoperative or intraoperative CBCT be taken for the purpose of estimating working length.14 However, evidence is growing in other areas, particularly within the fields of implantology, endodontics, and oral surgery, that in certain other situations the benefits of preoperative CBCT may outweigh the risks.15 It is therefore increasingly likely that a patient receiving root canal treatment either will need preoperative CBCT before treatment for another reason16 or may have recently had CBCT for other treatment, such as placement of an implant.15

The objective and rationale of this meta-analysis was to use a state-of-the-art evidence-based approach, utilizing statistical methods and aligning with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines, to summarize the results of all available independent studies on the subject. This will provide a more precise estimate of the accuracy of preoperative CBCT in locating the AF and establish its usefulness in estimating the working length in teeth undergoing root canal treatment.17

Materials and Methods

A protocol was finalized after a brief pilot search, and the review was submitted for registration on the PROSPERO register in April 2019.

Six databases were searched by 2 authors in April 2019 and February 2020: MEDLINE, Embase, PubMed, Web of Science, Scientific Electronic Library Online, and the Cochrane Library. The MeSH terms “cone beam computed tomography,” “tooth apex,” “tooth root,” “root canal preparation,” “endodontics,” and “root canal therapy” were utilized. The keywords “CBCT,” “working length,” “apical foramen,” “tooth apex” “root canal length,” “root canal length determination,” “root canal length measurement,” and “root canal treatment” were also used.

Pre-specified study and report eligibility criteria were stated in the PICOS format.18 The population was teeth requiring root canal treatment, and the intervention was preoperative CBCT. The comparison was a suitable reference standard such as an in vivo EAL or in vitro assessment under magnification after extraction. The outcome was the determination of working length, and the included studies contained basic, observational, and interventional primary research, as well as secondary research. No restrictions were placed on language, publication date, or publication status. Textbooks and leading journals were manually searched for additional articles, and gray literature was identified on the OpenGrey Database and Google Scholar.

Data were collected from articles that met the inclusion criteria, and authors were contacted to request any additional information required.

The risk of bias was assessed for randomized controlled trials with the Cochrane risk-of-bias tool19 and for basic research studies with the Office of Health Assessment and Translation risk-of-bias questions.20

Heterogeneity was found in the outcome parameters utilized in the studies, with some using mean difference and others mean absolute difference. The 2 parameters could not be statistically combined and were therefore recorded separately. Attempts were made to contact all authors to obtain the raw data so that the outcome parameters could be combined; however, a response was only received from a single author.10

In addition to the 2 outcome parameters identified (mean difference and mean absolute difference), 2 reference standards such as histology and EAL were also identified: histology and EAL. Due to the 2 outcome parameters and 2 reference standards included, 4 meta-analyses were planned to cover each combination. Because of the high levels of statistical heterogeneity, an inverse-variance random-effects model was chosen.19 Weighted mean differences were subsequently obtained with 95% confidence intervals and P values. The results were plotted as forest plots. The level of significance was set at 5%, and R version 3.0.2 software (R Foundation for Statistical Computing, Vienna, Austria; Metafor package) was used to perform the meta-analyses.

Results

Characteristics of included and excluded studies

Figure 1 shows the results of the search. Five studies were excluded because they used intraoperative rather than preoperative CBCT.212223242526 Two studies were excluded because they measured to the apical vertex rather than the apical foramen, resulting in a different measurement from the working length.2728 One study was excluded due to the lack of a suitable reference standard.29 Two studies included insufficient data for inclusion in the meta-analysis, and the authors could not be contacted to provide the missing data.3031 One study involved the use of multidetector CT rather than CBCT,32 and 1 study was excluded since it was a systematic review of studies that had already been included.33

Fig. 1

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow chart.

The characteristics of the included studies are presented in Tables 1 and 2. In the 3 randomized controlled crossover-design trials, EALs were used as the reference standard (Table 1). Regarding outcomes, the mean difference was calculated in 2 studies, whereas the mean absolute difference could be calculated in all 3 studies (Table 2).

Table 1

General characteristics of the included studies

EAL: electronic apex locator

Table 2

Summary of parameters and results for included studies

EAL: electronic apex locator

The remaining 6 studies were basic research studies with histological assessment as the reference standard (Table 1). Three were undertaken on extracted teeth,343536 2 on dry mandibles,3738 and 1 on cadavers.39 Regarding outcomes, the mean difference was calculated in 3 studies, whereas the mean absolute difference was calculated in the other 3 studies (Table 2).

Quality assessment

Figure 2 shows the risk of bias of the selected studies. The 3 studies using an EAL reference standard were crossover randomized controlled trials subjectively assessed to be at low risk of bias.91011 In all 3 trials, only a few minor sources of potential bias were present. First, we noted a failure to randomize the order in which CBCT and EAL readings were taken. However, this randomization would have been inappropriate, as in clinical settings, preoperative CBCT would always be taken before use of an EAL. Second, it was unclear how blinded the EAL and CBCT operators were to the study conditions. Third, 1 study reported 10% attrition bias,9 and an unknown delay separated the date of the CBCT and EAL reading, during which time the position of the AF could have changed due to resorption.40 Fifth, the potential for bias from the spill-over effect was present, as several subjects offered multiple teeth for the studies.

Fig. 2

Quality assessment of the selected studies. A. Randomized controlled crossover trials. B. Basic research studies.

The studies using extracted teeth, dry mandibles, or cadavers and a histological reference standard were considered to have a high risk of bias due to the basic nature of the research.3435363839

Meta-analysis 1: CBCT identification of the AF position compared to histological measurements

Meta-analysis 1 estimated that CBCT measurements approximate the position of the AF to be 0.40 mm coronal of the histological assessment, and this difference was statistically significant (95% confidence interval, - 0.71 mm to - 0.10 mm) (P=0.009) (standard error, 0.15 mm). Both the I2 (68.9%) and the Cochran Q values (P<0.05) suggested the presence of heterogeneity in the results; however, the tests lacked power and therefore could be misleading. With regard to publication bias, the Egger test indicated a lack of bias at P=0.917 (Fig. 3).

Fig. 3

A. Meta-analysis forest plots for histology and mean difference. B. Histology and mean absolute difference. C. Electronic apex locator (EAL) and mean difference. D. EAL and mean absolute difference.

Meta-analysis 2: Difference between CBCT and histological measurements

Meta-analysis 2 estimated that CBCT measurements of the location of the AF were statistically significantly different from the histological assessments, with a weighted mean difference of 0.48 mm (95% confidence interval, 0.39 mm to 0.57 mm) (P<0.001) (standard error, 0.05 mm). Both the I2 (77.5%) and the Cochran Q values (P<0.05) suggested the presence of heterogeneity in the results; however, the tests lacked power and therefore could be misleading. With regard to publication bias, the Egger test indicated a lack of bias at P=0.615 (Fig. 3).

Meta-analysis 3: CBCT identification of the AF compared to EAL

Meta-analysis 3 estimated that there was no statistically significant difference in the location of the apical foramen as obtained with CBCT or EAL measuring techniques. The weighted mean difference placed the CBCT estimate slightly coronal of the EAL assessment at - 0.11 mm; however, the confidence interval included zero (- 0.75 mm to 0.53 mm) (P=0.737) (standard error, 0.32 mm). Both the I2 (84.7%) and the Cochran Q values (P<0.05) suggested the presence of heterogeneity in the results; however, the tests lacked power and therefore could be misleading. The Egger test could not be conducted to evaluate publication bias due to an insufficient number of studies (Fig. 3).

Meta-analysis 4: Difference between CBCT and EAL measurements

Meta-analysis 4 estimated that CBCT measurements of the location of the AF were statistically significantly different from EAL assessments, with a weighted mean difference of 0.64 mm (95% confidence interval, 0.29 mm to 0.99 mm) (P<0.001) (standard error, 0.18 mm). Both the I2 (87.5%) and the Cochran Q values (P<0.05) suggested the presence of heterogeneity in the results; however, the tests lacked power and therefore could be misleading. With regard to publication bias, the Egger test indicated a lack of bias at P=0.788 (Fig. 3).

Discussion

Following this systematic review and meta-analysis, it was possible to assess the accuracy of preoperative CBCT for locating the AF and assisting in establishing the endodontic working length.

The conclusions of this meta-analysis could be accepted as robust. Despite the evidence of heterogeneity obtained via the I2 and Cochran Q test, all 3 studies were consistent in their assessment that CBCT estimated the AF to be coronal of the histological assessment by approximately 0.40 mm. The confidence intervals in 2 studies excluded zero, indicating a statistically significant difference.3739 The third had a very wide standard deviation, so the confidence interval did not exclude zero.36

Possible reasons for the larger standard deviation in the study by Aktan et al. include that the CBCT was taken at a slightly larger voxel size (0.2 mm compared to 0.133 mm and 0.125 mm), that slightly lower magnification was used for the histological assessments (×2.5 compared to ×10), and that the apical reference for CBCT measurements was not explicitly stated in the study and thus may have been inconsistent (Tables 1 and 2, Fig. 3A).

Meta-analysis 2. Difference between CBCT and histological measurements

The conclusions of this meta-analysis could also be accepted as robust. Despite the evidence of heterogeneity obtained via the I2 and Cochran Q tests, all 3 studies were consistent in their assessment that a statistically significant difference was present in the measurements made using CBCT and histological assessment. All 3 studies had narrow confidence intervals that excluded zero (Fig. 3B).

This meta-analysis did not yield valid conclusions. The 2 included studies demonstrated a high level of heterogeneity in their I2 and Cochran Q values, as well as disagreement regarding whether CBCT estimated the AF to be apical or coronal compared to an EAL reference standard. It is also important to note that the confidence intervals barely overlapped (Fig. 3C).

The heterogeneity in the results may have been due to the different EALs used in each of the studies, leading to differing capacity to detect the AF. Both studies were also conducted on teeth with large radiolucencies; this may have caused the resorption of apical landmarks, which is believed to affect EAL accuracy.42

A recent systematic review on this subject also found no significant differences in measurements made by EAL and CBCT. However, those researchers also could not draw valid conclusions due to an insufficient sample size and high levels of heterogeneity in equipment brands and outcome measurements.33

This meta-analysis did not yield valid conclusions. The 3 studies demonstrated a high level of heterogeneity in their I2 and Cochran Q values. It is also clear that the confidence intervals of Ustun et al. do not overlap with those of the other studies. This heterogeneity could have been due to the use of different EALs in the studies (Fig. 3D).10

Completeness and applicability of evidence

The conclusions of meta-analyses 1 and 2 could be considered robust from a statistical and methodological perspective, which made it relevant to answer the objectives of this meta-analysis. In terms of the completeness of the evidence, it would be beneficial to increase the number of studies in each of the 2 meta-analyses by obtaining raw data from research excluded due to heterogeneity in outcome parameters.

Regarding the applicability (or external validity) of the evidence, unidentified effect modifiers may be present and should be investigated. A particular issue was the time period between the CBCT and treatment, as the positions of the CRP and AF may change over time due to cementum deposition,143 apical inflammatory root resorption,40 dental caries, tooth wear, or restorative treatment.

Another important potential effect modifier was the variation of working length during canal preparation. The distance between the CRP and AF tends to shorten by an average of 0.22 mm with the use of NiTi rotary files due to the straightening of any curvature.43 The magnitude of this decrease has also been shown to be larger with stainless steel hand files than with NiTi rotary instruments due to the capacity of the latter to stay centered in the canal and reduce straightening. It has also been shown to be larger with aggressive coronal flaring - for example, with Gates-Glidden drills that remove cervical dentin - and in more curved canals.43

Other unidentified effect modifiers may include tooth type, tooth age, presence of apical periodontitis and apical inflammatory root resorption, type of CBCT machine, tube voltage, tube current, image detector type, detector pixel size, number of basis projections, extent of rotational arc, size of field of view, voxel size, computed tomography reconstruction algorithm, image processing filters, viewing conditions, and viewing software.

The results from meta-analyses 3 and 4 were incomplete and lacked external validity.

Quality of evidence

Although the conclusions of meta-analyses 1 and 2 can be considered methodologically robust, the level of evidence they provide is undermined by the basic research upon which they are based. For this reason, the evidence level must be graded as very low according to the Grading of Recommendations, Assessment, Development, and Evaluation approach.19

The conclusions of meta-analyses 3 and 4 must also be triple-downgraded to very low due to indirectness through reference standard accuracy issues, heterogeneity and inconsistency of results, and imprecision of results as evidenced by wide confidence intervals.19

The main limitation of this review in terms of limiting bias is that not all of the available data could be used in each of the 4 meta-analyses. The use of the mean absolute difference as an outcome parameter alongside the mean difference was necessary due to the failure of many studies to report the latter.

In this study, a low level of evidence was produced to suggest that preoperative CBCT measures the AF to be on average 0.40 mm coronal to its histological position. Although this evidence is low-quality, it is the best currently available, and therefore it should be introduced as an additional tool to provide evidence of the distance between a CRP and the AF.