Artigo submetido ao periódico Journal of Endodontics, visando publicação (Anexo 2). A estruturação do artigo segue as “Instruções aos autores”, preconizadas pela editora do periódico.
Thiago Oliveira Sousa, DDS, MSc1
Bassam Hassan, BDS, MSc, PhD2
Hesam Mirmohammadi, DDS, MSc, PhD3
Hagay Shemesh, DMD, PhD3
Francisco Haiter-Neto, DDS, MSc, PhD1
1Department of Oral Diagnosis, Piracicaba Dental School, University of Campinas,
Brazil
2Department of Oral Implantology and Prosthetic Dentistry, Academic Center for
Dentistry Amsterdam (ACTA), University of Amsterdam and VU University, Netherlands
3Department of Endodontology, Academic Center for Dentistry Amsterdam (ACTA),
University of Amsterdam and VU University, Netherlands
Corresponding author: Thiago Oliveira Sousa
Programa de Pós-Graduação em Radiologia Odontológica – Avenida Limeira, nº 901, Bairro Areião, CP 52. Piracicaba – SP – Brasil. CEP: 13414-903. Phone number: +55 19 2106-5361.
Abstract
Introduction: The study objective was to evaluate the effectiveness of cone-beam computed tomography (CBCT) for detection of lateral canals (LC) in endodontically treated premolars.
Methods: Two evaluators classified eighty extracted premolars into two groups based on the absence (n=40) or presence (n=40) of LC according to micro computed tomography (µCT) analysis. The extracted teeth were fixated in a human mandible and scanned with CBCT. Subsequently, each tooth was endodontically treated and CBCT scans were repeated. Three experienced examiners evaluated all images randomly. Receiver operating characteristic (ROC) curves were compared using McNemar test and sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were obtained.
Results: The area under the ROC curves values were 0.58 and 0.49 prior to and following RCT, respectively. These values were statistically significantly different (p<0.001). Prior to RCT sensitivity, specificity, PPV, NPV were 55%, 52%, 55% and 56% while following RCT the values were 33%, 61%, 46% and 48%, respectively.
Conclusions: LC detection in non-treated teeth presented low accuracy, while among treated teeth CBCT showed no efficacy. The results suggest that CBCT is not an effective diagnostic tool for LC detection.
Keywords
Introduction
Failure of root canal treatment (RCT) resulting from incomplete canal debridement and obturation due to aberrant anatomic composition is a clinical reality. Research demonstrates that at least 9% of endodontic failures could be attributed to complex canal morphology including the presence of apical ramifications and other morphologic aberrations (1,2). Lateral canals (LC) are ramifications connecting the main canal to the periodontal ligament that can exhibit different sizes and shapes (3,4). These canals, also dubbed secondary canals when located in the apical third of the root (5), are frequently found to be perpendicular to the main canal and represent a challenge for endodontic diagnosis and treatment since they are extremely difficult to access, clean, disinfect and fill (6). If remained untreated, they could result in lateral lesions (7), which could be misdiagnosed as periodontally induced bone loss or root fractures. Detecting LC prior to treatment or during the investigation of possible endodontic failure is crucial for determining the appropriate treatment protocol and for predicting treatment outcome (8). Additional diagnostic tools are required to detect LC since clinical observation is difficult to achieve, even with the use of operating microscope.
One important imaging method, which is currently accessible in the clinical practice is cone beam computed tomography (CBCT). It allows a volumetric visualization of hard tissues with relatively low radiation dose while eliminating superimposition artefacts of anatomical structures routinely encountered in conventional 2D imaging (9). CBCT was previously shown to reliably demonstrate anatomical variations regarding the main canal configuration (10–14). However, to the best of our knowledge, there is no investigation focusing on detecting LC. Therefore, the objective of this study was to evaluate CBCT reliability in detecting LC prior to and following RCT.
Materials and Methods
Sample Preparation
The study protocol was approved by the institutional review board. Eighty extracted maxillary and mandibular single and multi-rooted premolars were cleaned, disinfected and kept in physiologic solution. Roots with open apex, excessive restorations or resorptive lesions were excluded. All teeth underwent micro computed tomographic (μCT) examination using Skyscan 1174 (Bruker, Kontich, Belgium) with the following settings: 50 kV, 800 µA, voxel dimension of 15.91 µm, 1.0 mm-thick aluminum filter, rotation step 0.4° and 4 frames. Two experienced maxillofacial radiologists analysed the images using CTAn software (v.1.14.4.1, Bruker, Kontich, Belgium) in order to classify the teeth according to the presence/absence and number of LC. If a consensus could not be reached between the two examiners, a third examiner would assist in making the decision. The included teeth were separated into two groups: LC absent (n=40), and LC present (n=40) (Figure 1). The widest diameter of the lateral canal was also measured in order to assess the possible correlation between LC detection and their size.
CBCT Data Collection
Each tooth was placed in a dry mandible (socket position: 29) and scanned using 3D AccuiTomo 170 (Morita Inc, Tarumi-cho Suita City, Osaka, Japan), with the following protocol: FOV 4x4, voxel size of 0.08 mm, 90 kVp and 5 mA. Water was used to simulate soft tissues as previously described in the literature (15).
Three experienced examiners evaluated the CBCT volumes independently and scored the LC detection on a 5-point rank scale: 1 – LC definitely present; 2 – LC probably
present; 3 – unsure whether present or absent; 4 – LC probably absent; 5 – LC definitely absent (16).
Subsequently, root canals were prepared with the Mtwo rotary system (VDW Silver; VDW GmbH, Munich, Germany) until file size 40. Sodium hypochlorite 2.5% was used for canal irrigation followed by administering 17% ethylenediaminetetraacetic acid (EDTA ). Teeth were filled with a single-cone technique with Mtwo gutta-percha and AH 26 sealer (Dentsply DeTrey GmbH, Konstanz, Germany) per the manufacturer's instructions. CBCT acquisitions were repeated using identical settings. The images were randomized and analysed 30 days after the first evaluation. For intra-agreement analysis, 25% of the images were randomly reviewed 30 days later. Finally, μCT acquisitions of the treated teeth were repeated, aiming to investigate whether LC were filled, partially filled or not filled.
Statistical Analysis
Data was analysed using SPSS for Windows (v. 22, SPSS Corp., Chicago, IL). Receiver operating characteristic (ROC) curve analysis was plotted to evaluate CBCT diagnostic accuracy before and after RCT. The areas under the curves were compared by the McNemar test and the level of significance for rejecting the null hypothesis was set at 5% (α=0.05). To calculate sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV), the evaluators’ responses were dichotomized into the presence or absence of LC. Scores of 1 and 2 were considered as LC present while scores 3, 4 and 5 as LC absent (16). Spearman correlation coefficient was performed to verify possible positive correlation between maximum diameter of LC and evaluators’ responses. Kappa test accessed the reliability intra- and inter-evaluators.
Results
Accuracies were obtained from the areas under ROC curves, considering the 5 previously defined scores. The CBCT ROC curves prior to and following RCT are shown in Figure 2. Prior to RCT an area value of 0.58 (p<0.05) was obtained and following RCT an area value of 0.49 (p>0.05) was calculated. The differences between both areas were statistically significantly different (p<0.001) (Figure 3). Sensitivity, specificity, PPV and NPV for CBCT evaluations prior to and following RCT are shown in Table 1. The largest diameter of lateral canal as assessed using micro CT was (0.14±0.04). Spearman coefficient showed no correlation between LC diameter and evaluators’ responses (p>0.05). µCT analysis after RCT showed that out of 40 treated teeth, 13 were fully filled with sealer, nine were partially filled and 18 were totally empty. Intra and inter evaluators agreement ranged from moderate (0.59) to substantial (0.67) and fair (0.26) to moderate (0.56), respectively (17).
Discussion
This ex-vivo study investigated the detection accuracy of LC with CBCT prior to and following non-surgical root canal treatment. The area under the ROC curve calculated for non-treated teeth was low, while the area for treated teeth was not statistically significant. When evaluating areas under the ROC curve, an ideal diagnostic test should be close to 1.0 (with a curve distancing from the reference line), while values below 0.5 indicate no effectiveness (18). The value of 0.58, obtained in this study, before RCT, indicates that CBCT was more effective than random chance but not strong enough (18) to justify the indication of CBCT for LC detection. The comparison between both areas under the curve, with difference statistically significant (p<0.001),
leads to the assumption that presence of a root canal filling negatively influences the detection of LC. However, it is worth emphasizing that the curve for non-treated teeth demonstrates unacceptable reliability for diagnosis (p>0.05).
All diagnostic values decreased after RCT, except specificity, which increased from 33% to 61%. One of the disadvantages inherent to CBCT is the production of beam hardening artefacts, which masks the visibility of anatomical structures juxtaposed to highly radiopaque materials, such as gutta-percha (19,20). This could explain the low diagnostic values obtained here following obturation. The findings in the present study agree with other studies that investigated the influence of gutta-percha in root fracture detection (21-23). The higher specificity value may be explained by the evaluators’ difficulty in detecting LC, tending to answer “probably absent” or “absent”, which resulted in a high number of “negative” answers, consequently increasing the number of “true negative” cases (24). The relatively low intra and inter evaluators agreement, ranging from fair to substantial, also support the difficulty in detecting LC on CBCT images.
Several clinical studies evaluated and classified endodontic anatomy with CBCT. The evaluations included the main canal configuration (25–27) and even fine structures such as apical foramina (27–29) but with no gold standard. Our results emphasize the importance of a more critical view regarding the limitations of CBCT for the detection of intricate anatomical variations within the root canal system. Prior to radiographic examination, the efficacy of the complementary imaging method should be demonstrated following the ALADA principle (As Low As Diagnostically Acceptable) (30). The radiographic exposure to the patient cannot be ethically justified if the information obtained from complimentary diagnostic instrument does not aid in diagnosis.
The subjective image quality of CBCT-based images and their ability to display various features are influenced by a myriad of variables including the CBCT machine type, the FOV, voxel size, the tube voltage and current among other technical factors (31,32). In the present study, images were acquired using a high-resolution voxel size of 0.08 mm considering that the higher the spatial resolution, the higher the diagnostic accuracy (19). No statistically significant correlation was found between LC diameter and their overall detection. These findings are in agreement with Thönissen et al (33), that investigated thin bony structures by CBCT imaging using different voxel size and concluded that CBCT did not allow exact representation of the actual clinical status when used to evaluate fine anatomical structures. Likewise, we also concluded that contrast-to-noise ratio (CNR), which is considered crucial to image quality (34), may have contributed to the poor LC detection. Although voxel sizes smaller than 0.2 mm represent a desirable high spatial resolution, this choice results in increased noise level on the final images and consequently a decrease of CNR (34).
Although we used a filling technique that did not include lateral or vertical compaction, aiming to result in LC as empty as possible (35) (simulating a clinical situation of failure/LC not filled), we were not able to control whether the LC was filled or not. To assess filling quality, post-operative μCT acquisition and analysis was performed. It was found that 67.5% of the LC were either not or only partially filled, while 32.5% were fully obturated. Due to the hyper dense characteristic of the endodontic cement, fully filled LC are expected to be more distinguishable on the images; however this fact did not yield better diagnostic detection with CBCT. LC detection in non-treated teeth presented low accuracy, while among treated teeth CBCT showed no efficacy. The results suggest that CBCT is not an effective tool to detect LC.
Accuracy of LC detection in non-treated teeth was low, while following RCT it was not accurate at all. Although CBCT is a helpful tool in Endodontics, ALADA principles should never be neglected. In vitro/ex vivo results should not be extrapolated to clinical situations, however this study suggests additional caution when prescribing and interpreting CBCT regarding the presence of LC, as they might be often missed, especially in the cases of treated teeth.
Acknowledgements
The authors deny any conflicts of interest related to this study.
References
1. Song M, Kim HC, Lee W, Kim E. Analysis of the cause of failure in nonsurgical endodontic treatment by microscopic inspection during endodontic microsurgery. J Endod 2011;37:1516–9.
2. Wu MK, Shemesh H, Wesselink PR. Limitations of previously published systematic reviews evaluating the outcome of endodontic treatment. Int Endod J 2009;42:656–66.
3. De Deus QD. Frequency, location, and direction of the lateral, secondary, and accessory canals. J Endod 1975;1:361–6.
4. Carvalho-Sousa B, Almeida-Gomes F, Carvalho PRB, et al. Filling lateral canals: evaluation of different filling techniques. Eur J Dent 2010;4:251–6.
6. Ricucci D, Siqueira JF. Fate of the Tissue in Lateral Canals and Apical Ramifications in Response to Pathologic Conditions and Treatment Procedures. J Endod 2010;36:1–15.
7. Schilder H, Hargreaves KM. Filling root canals in three dimensions. J Endod 2006;32:281–90.
8. Vertucci FJ. Root canal morphology and its relationship to endodontic procedures. Endod Top 2005;10:3–29.
9. Scarfe WC, Levin MD, Gane D, et al. Use of cone beam computed tomography in endodontics. Int J Dent 2009;2009:634567.
10. Neelakantan P, Subbarao C, Subbarao C V. Comparative evaluation of modified canal staining and clearing technique, cone-beam computed tomography, peripheral quantitative computed tomography, spiral computed tomography, and plain and contrast medium-enhanced digital radiography in studying root canal morphology. J Endod 2010;36:1547–51.
11. Baratto Filho F, Zaitter S, Haragushiku GA, et al. Analysis of the Internal Anatomy of Maxillary First Molars by Using Different Methods. J Endod 2009;35:337–42.
12. Matherne RP, Angelopoulos C, Kulild JC et al. Use of Cone-Beam Computed Tomography to Identify Root Canal Systems In Vitro. J Endod 2008;34:87–9.
13. Soares De Toubes KMP, Côrtes MIDS, Valadares MADA, et al. Comparative analysis of accessory mesial canal identification in mandibular first molars by using four different diagnostic methods. J Endod 2012;38:436–41.
14. Mirmohammadi H, Mahdi L, Partovi P, et al. Accuracy of cone-beam computed tomography in the detection of a second mesiobuccal root canal in endodontically treated teeth: An ex vivo study. J Endod 2015;41:1678–81.
15. Neves FS, Freitas DQ, Campos PSF, et al. Evaluation of cone-beam computed tomography in the diagnosis of vertical root fractures: The influence of imaging modes and root canal materials. J Endod 2014;40:1530–6.
16. Junqueira RB, Verner FS, Campos CN, et al. Detection of vertical root fractures in the presence of intracanal metallic post: A comparison between periapical radiography and cone-beam computed tomography. J Endod 2013;39:1620–4.
17. Landis JR, Koch GG. The Measurement of Observer Agreement for Categorical Data. Biometrics 2008;33:159–174.
18. Mandrekar JN. Receiver operating characteristic curve in diagnostic test assessment. J Thorac Oncol 2010;5:1315–6.
19. Helvacioglu-Yigit D, Demirturk Kocasarac H, Bechara B, et al. Evaluation and Reduction of Artifacts Generated by 4 Different Root-end Filling Materials by Using Multiple Cone-beam Computed Tomography Imaging Settings. J Endod 2016;42:307–14.
20. Schulze R, Heil U, Groß D, et al. Artefacts in CBCT: A review. Dentomaxillofacial Radiol 2011;40:265–73.
21. Hassan B, Metska ME, Ozok AR, et al. Detection of Vertical Root Fractures in Endodontically Treated Teeth by a Cone Beam Computed Tomography Scan. J Endod 2009;35:719–22.
22. Bechara B, McMahan CA, Noujeim M, et al. Comparison of cone beam CT scans with enhanced photostimulated phosphor plate images in the detection of root fracture of endodontically treated teeth. Dentomaxillofacial Radiol 2013;42:20120404.
23. Neves FS, Freitas DQ, Campos PSF, et al. Evaluation of cone-beam computed tomography in the diagnosis of vertical root fractures: The influence of imaging modes and root canal materials. J Endod 2014;40:1530–6.
24. Liedke GS, Spin-Neto R, Vizzotto MB, et al. Diagnostic accuracy of cone beam computed tomography sections with various thicknesses for detecting misfit between the tooth and restoration in metal-restored teeth. Oral Surg Oral Med Oral Pathol Oral Radiol 2015;120:131–7.
25. Park JB, Kim N, Park S, et al. Evaluation of root anatomy of permanent mandibular premolars and molars in a korean population with cone-beam computed tomography. Eur J Dent 2013;7:94–101.
26. Ok E, Altunsoy M, Nur BG, et al. A cone-beam computed tomography study of root canal morphology of maxillary and mandibular premolars in a Turkish population. Acta Odontol Scand 2014;8:1–6.
27. Estrela C, Bueno MR, Couto GS, et al. Study of root canal anatomy in human permanent teeth in a subpopulation of Brazil’s center region using cone-beam computed tomography - Part 1. Braz Dent J 2015;26:530–6.
28. da Silva EJNL, de Castro RWQ, Nejaim Y, et al. Evaluation of root canal configuration of maxillary and mandibular anterior teeth using cone beam computed tomography: An in-vivo study. Quintessence Int 2016;47:19–24.
29. Yang H, Tian C, Li G, et al. A cone-beam computed tomography study of the root canal morphology of mandibular first premolars and the location of root canal orifices and apical foramina in a Chinese subpopulation. J Endod 2013;39:435– 8.
30. Jaju PP, Jaju SP. Cone-beam computed tomography: Time to move from ALARA to ALADA. Imaging Sci Dent 2015;45:263–5.
31. Spin-Neto R, Gotfredsen E, Wenzel A. Impact of voxel size variation on CBCT- based diagnostic outcome in dentistry: A systematic review. J Digit Imaging 2013;26:813–20.
32. Kamburoğlu K, Murat S, Kolsuz E, et al. Comparative assessment of subjective image quality of cross-sectional cone-beam computed tomography scans. J Oral Sci 2011;53:501–8.
33. Thönissen P, Ermer MA, Schmelzeisen R, et al. Sensitivity and specificity of cone beam computed tomography in thin bony structures in maxillofacial surgery e A clinical trial. J Cranio-Maxillofacial Surg 2015;43:1284–8.
34. Bechara B, A. McMahan C, S. Moore W, et al. Contrast-to-noise ratio difference in small field of view cone beam computed tomography machines. J Oral Sci 2012;54:227–32.
35. Celikten B, Uzuntas CF, Orhan AI, et al. Micro-CT assessment of the sealing ability of three root canal filling techniques. J Oral Sci 2015;57:361–6.
Table
Table 1. Diagnostic tests of sensitivity, specificity, PPV and NPV for CBCT prior to and following RCT.
Sensitivity Specificity PPV NPV
CBCT before treatment 0.55 0.52 0.55 0.56
Figure Captions
Figure 1. μCT three-dimensional reconstruction: (A) An example of a tooth with absence of lateral canal; and (B) example of a tooth with a lateral canal.
Figure 2. ROC curves for CBCT accuracies prior to and following RCT.
Figure 3. CBCT sagittal images of the same example tooth with no LC, prior to (1A) and following (1B) RCT. Images (2A) and (2B) illustrate an example tooth with one LC, prior to and following RCT. The arrows indicate the regions where LC is expected and the influence of gutta-percha artifacts on its visibility.
Figure 1
3 DISCUSSÃO
O objetivo no presente estudo foi avaliar a eficácia da TCFC e da radiografia periapical na identificação da configuração de canais radiculares; e da TCFC na detecção de canais laterais, antes e após tratamento endodôntico em pré-molares. Na avaliação da configuração do canal radicular, a TCFC apresentou alta acurácia (89%), não havendo diferença significativa com a microtomografia, enquanto as radiografias periapicais apresentaram uma pobre acurácia de 55%, com diferença estatística tanto em relação à TCFC quanto ao padrão-ouro. Na detecção de canais laterais, a TCFC apresentou baixa acurácia, com área abaixo da curva ROC de 0.58, que decaiu após o tratamento endodôntico (0.49).
A maioria dos canais de Tipo I de Vertucci (Vertucci, 1984) foi corretamente identificada nas radiografias periapicais, enquanto todos os outros tipos foram diagnosticados como Tipo I em praticamente todos os casos. Isto é explicado pela característica bidimensional das radiografias, sobrepondo as variações anatômicas na imagem final, impedindo a sua correta identificação. Nossos resultados estão em consonância com Matherne et al. (2007), que relataram que os observadores falharam em identificar pelo menos um canal radicular em aproximadamente 40% dos dentes avaliados. Entretanto, eles utilizaram duas incidências radiográficas e a TCFC foi utilizada como padrão-ouro. Portanto, outros estudos deveriam ser realizados, usando a microtomografia como padrão-ouro em vez da TCFC, para investigar se a dupla incidência, comparada a uma única incidência ortorradial, melhoraria a acurácia na identificação da configuração dos canais radiculares de pré-molares.
No presente estudo, a detecção de canais acessórios não foi avaliada.