Abstract
Objectives
To assess the accuracy of near-infrared-light transillumination (DIAGNO) compared to visual-tactile (VT) and radiographic (RA) evaluation of proximal carious lesions adjacent to composite restorations in vitro.
Methods
Two hundred extracted posterior permanent human teeth with occluso-proximal composite restorations were allocated to 50 groups of four posterior teeth, and mounted in a pilot-tested diagnostic model in a dummy head. The teeth were independently assessed by two examiners. Transverse microradiography and visual assessment served as reference tests to detect any lesions (prevalence 24%) and cavitated lesions (18%), respectively, adjacent to restorations. Sensitivity, specificity, positive and negative predictive values and the area under the receiver-operating-characteristics curve (AUC) were calculated.
Results
To detect any proximal carious lesions adjacent to composite, the mean sensitivity/specificity were 0.63/0.95 for DIAGNO, 0.70/0.88 for RA when lesions radiographically extending into enamel and dentin were considered, 0.26/0.98 for RA when only lesions extending into dentin were considered, and 0.31/0.96 for VT. For cavitated lesions adjacent to proximal composite restorations, these values were RA (enamel and dentin) 0.84/0.88, RA (dentin) 0.34/0.99, DIAGNO 0.69/0.94 and VT 0.40/0.97. AUC did not differ significantly between RA and DIAGNO, while VT showed significantly lower values (p < 0.05).
Conclusion
Within the limitations of this study, DIAGNO seems useful for detecting proximal carious lesions adjacent to restorations.
Clinical relevance
Near-infrared-light transillumination could be used as a radiation-free adjunct or alternative to RA for detecting carious lesions adjacent to composite restorations.
1
Introduction
The replacement of dental restorations is one of the most frequently provided dental treatments [ ]. Carious lesions adjacent to existing restorations, that are secondary/recurrent or residual carious lesions, are one of the main reasons for such replacement. These lesions occur years after placement [ ], and are oftentimes associated with faulty restorations allowing acid and bacterial penetration into the tooth-restoration interface [ ].
The detection of carious lesions adjacent to restorations is challenging [ ], especially on proximal surfaces. A range of detection methods are available, each having different sensitivity, specificity and reliability. Visual-tactile examination (VT) and bitewing radiographs (RA) used alone or in combination are widely used for detecting lesions adjacent to restorations [ ]. The sensitivity of VT is limited, especially on proximal surfaces of posterior teeth; RA has a higher sensitivity, but requires ionizing radiation [ ]. Both tend to underestimate the lesion extension [ ]. Methods involving laser-induced fluorescence have been found to yield high sensitivity, but low specificity for lesions adjacent to restorations [ ]. Low specificities lead to over-detection, and unnecessary repair or replacement of the restorations, generating costs and limiting the lifespan of the tooth [ ]. Low sensitivities lead to lesions remaining undetected, allowing them to progress.
An alternative to RA, which does not require any ionizing radiation, is near-infrared-light transillumination. A diagnostic camera system based on this concept has been introduced (DIAGNOcam, Kavo, Biberach, Germany). It consists of a near-infrared light source, two flexible arms, each containing an optical fiber to transmit the light into the tooth and a charge-coupled device sensor to capture the signal, with the resulting images or videos being monitored or stored digitally. A number of in vitro and in vivo studies addressed the accuracy of near-infrared-light transillumination at wave lengths ranging from 700 to 1700 nm to detect occlusal and proximal primary lesions, suggesting that near-infrared-light transillumination is suitable for detecting early carious lesions [ ]. Simon et al. [ ] also concluded that near-infrared-light transillumination at a wavelength of 1310 nm is appropriate for detecting secondary caries lesions around composite restorations.
The aim of the present study was to assess the diagnostic accuracy of a near-infrared-light transillumination device (DIAGNOcam) at a wave length of 780 nm compared to RA and VT for detecting proximal carious lesions adjacent to composite restorations in vitro .
2
Materials and methods
2.1
Study design
Two-hundred extracted posterior teeth with occluso-proximal composite restorations were allocated to 50 groups of four teeth. These 50 sets of teeth were independently examined by two examiners (KE, HA) using all three diagnostic tests (RA, VT, DIAGNO). As therapeutic decisions might differ for different lesions (e.g. all lesions regardless of their surface status, or only lesions with surface cavitation), we assessed the accuracy of the tests at two cut-offs; detecting all lesions and detecting only cavitated lesions. The presence of surface cavitation was assessed visually-tactile, while transverse microradiography (TMR) was used to detect the presence of any lesions (reference tests). The setup and the resulting images are shown in Figs. 1 and 2 .
2.2
Study samples
Two-hundred extracted posterior permanent human teeth (106 premolars and 94 molars) with occluso-proximal composite restorations were obtained with informed consent under an ethics-approved protocol (ethics committee of the Charité – Universitätsmedizin Berlin, EA4/102/14). The teeth were rinsed with water for 15 s, calculus and debris were removed using a scaler (S1296, Hu-Friedy, Chicago, IL, USA), and all surfaces cleaned using polishing paste (Proxyt, Ivoclar Vivadent, Schaan, Lichtenstein) for 10 s. The teeth were then stored individually in plastic containers containing 0.1% thymol solution at 4 °C until starting the evaluations.
Teeth were allocated to 50 groups of four teeth. Each study surface was paired with an adjacent sound or restored surface; surfaces with no adjacent tooth (first and last surfaces) were excluded from the examination. The roots of all teeth were embedded in transparent epoxy resin (Epo-Thin 2, Buehler, Lake Bluff, IL, USA) ensuring tight contacts. In a pilot study, such resin models ( Fig. 2 ) had been found to produce DIAGNO imaging similar to that yielded by clinical application. In this pilot study, three examiners assessed 50 randomly shown images taken in a clinical setting as well as 50 randomly shown images taken under the described in vitro conditions (see Fig. 2 ). They were asked to assign an imaging status (clinical or in vitro ) to the images. We found the examiners to not being able to discriminate images taking under clinical conditions from those taken under in vitro conditions. Also, note that another recent study also showed the in vitro evaluation of DIAGNO to be possible if using a suitable model [ ]. For tactile assessment, pink wax was additionally placed to simulate the gingiva and prevent visual evaluation of the proximal surfaces (see Fig. 2 ).
2.3
Index tests
The study models were mounted in a dummy head (Phantomkopf P-6, Frasaco, Tettnang, Germany) to allow examination under a simulated clinical situation. As described, three tests were used; RA, DIAGNO and VT. A two-week period was left between each test method. Firstly, bitewing radiographs were taken of all 50 models, then both examiners used the DIAGNO diagnostic camera to evaluate the occluso-proximal fillings, at last the wax gingiva was added to the models and VT took place.
For RA, digital bitewing radiographs were taken using a radiation source (Heliodent Plus, Sirona Dental Systems, Bensheim, Germany) and sensor (XIOS XG, Sirona Dental Systems, Bensheim, Germany), operating at 65 kV and 7 mA with an exposure time of 0.06 s. The source-to-film distance was 250 mm. A 15-mm thick Plexiglas mold was placed between the source and the model to stimulate soft tissue scattering [ ]. All radiographs were standardized via mounting the study models in a custom-made device, allowing for orthograde radiographs being taken [ ]. The digital radiographs were assessed (Sidexis, Sirona Dental Systems, Bensheim, Germany) under the same lighting conditions on a 27-inch diagnostic screen. No modification of the original image or screen adjustments was allowed and no image enhancement software was used. The radiographs were scored as follows: 0 = no carious lesion adjacent to the composite restoration, 1 = carious lesion extending into the enamel adjacent to the composite restoration, 2 = carious lesion extending into the dentin adjacent to the composite restoration.
For DIAGNO, the DIAGNOcam diagnostic camera was moved perpendicularly to the occlusal surface over each restoration/tooth of interest. To prevent any light interference, the dental unit light was switched off during examination. To simulate the clinical application of the device, we did not capture pictures but allowed the examiners to evaluate the teeth live and in motion using the KID software (KaVo Integrated Desktop/version 2.4.1.6821, KaVo), which allows the examiners to imitate the clinical situation. Carious lesions were defined as a less translucent shadow under or adjacent to the composite restoration ( Fig. 1 ), no further grading system was used.
For VT, a dental light (Heliodent Plus, Sirona Dental Systems, Bensheim, Germany), a dental mirror (M4C, Hu-Friedy, Chicago, IL, USA), a pointed, curved dental probe (EDX3CH, Hu-Friedy, Chicago, IL, USA), dental floss (Oral B, Procter & Gamble, Schwalbach am Taunus, Germany) and a 3-way air syringe (Sirona Dental Systems) were used. The probe was carefully entered into the proximal space in contact with the tooth. It was then moved up and down along the gingivo-cervical margin, carefully probing it without force. Carious lesions were suspected if any irregularities, positive or negative steps or gaps between the tooth and the restoration were detected, no further grading system was used.
2.4
Reference test
As described, two reference tests were applied: (1) Visual-tactile assessment to detect cavitated lesions; this was done before teeth were embedded for application of the index tests (see above), allowing full access and view onto the proximal area. (2) TMR to detect any lesions; this was done after application of the reference tests (as TMR is a destructive method).
For visual-tactile assessment, study surfaces were first evaluated visually wet and then dry, using 2.5× magnification, a dental light and a 3-way air syringe and a pointed, curved dental probe (EDX3CH). Loss of marginal integrity (cavitation) was recorded.
For TMR, the crowns were separated from the roots (Band Saw Exakt 300 CL, Exakt Apparatebau, Norderstedt, Germany) and embedded in acrylic resin carrier (Technovit 4071, Heraeus Kulzer, Hanau, Germany). Each tooth was cut perpendicularly into thin sections (100 μm; Band Saw 300 cl; Mikroschleifsystem 400 CS, abrasive paper 1200, 2400, and 4000). A nickel-filtered copper X-ray source (PW3830, Pananalytical, Kassel, Germany) operated at 20 kV and 20 mA was used to obtain radiographs. Films (35 mm B/W positive, Fujifilm, Tokyo, Japan) were exposed for 10 s and developed (Kodak Professional D-19) under standardized conditions according to the manufacturer’s recommendations. Microradiographs were analyzed using a digital-image analyzing system (CFW 1312 M, Scion, Frederick, USA) interfaced with a universal microscope (Axioplan 60318, Zeiss, Oberkochen, Germany) and a personal computer (TMR 2008; UMCG, Groningen, The Netherlands). Mineral Loss (ΔZ) was measured in the most affected area beneath the restoration. A carious lesion was defined as any mineral loss exceeding 400 vol.% × μm) to exclude line-like radiolucent areas resulting from possible bonding material.
2.5
Calibration and reliability
Examiners had been calibrated for all diagnostic methods prior to the study with a theoretical lecture followed by supervised practical examination training on 20 independent teeth not used for the main study. Repeated examination of this subset of teeth, separated by a one-month interval, allowed to estimate intra-rater reliability. Given that we had two examiners in our main study, inter-rater reliability was estimated from the assessment of all teeth. Reliability was graded as good (κ = 0.61–0.80) or excellent (κ > 0.80) [ ].
2.6
Statistical analysis
Accuracy was expressed as sensitivity, specificity and positive (PPV) and negative (NPV) predictive values. Receiver-operating-characteristics (ROC) curves and the resulting area under the curve (AUC) were used to express the diagnostic value of each test. Differences in mean accuracy between tests were assumed if p < 0.05 (Mann-Whitney-u-Test). Statistical evaluation was performed using SPSS 22 (IBM, Armonk, USA).
2
Materials and methods
2.1
Study design
Two-hundred extracted posterior teeth with occluso-proximal composite restorations were allocated to 50 groups of four teeth. These 50 sets of teeth were independently examined by two examiners (KE, HA) using all three diagnostic tests (RA, VT, DIAGNO). As therapeutic decisions might differ for different lesions (e.g. all lesions regardless of their surface status, or only lesions with surface cavitation), we assessed the accuracy of the tests at two cut-offs; detecting all lesions and detecting only cavitated lesions. The presence of surface cavitation was assessed visually-tactile, while transverse microradiography (TMR) was used to detect the presence of any lesions (reference tests). The setup and the resulting images are shown in Figs. 1 and 2 .
2.2
Study samples
Two-hundred extracted posterior permanent human teeth (106 premolars and 94 molars) with occluso-proximal composite restorations were obtained with informed consent under an ethics-approved protocol (ethics committee of the Charité – Universitätsmedizin Berlin, EA4/102/14). The teeth were rinsed with water for 15 s, calculus and debris were removed using a scaler (S1296, Hu-Friedy, Chicago, IL, USA), and all surfaces cleaned using polishing paste (Proxyt, Ivoclar Vivadent, Schaan, Lichtenstein) for 10 s. The teeth were then stored individually in plastic containers containing 0.1% thymol solution at 4 °C until starting the evaluations.
Teeth were allocated to 50 groups of four teeth. Each study surface was paired with an adjacent sound or restored surface; surfaces with no adjacent tooth (first and last surfaces) were excluded from the examination. The roots of all teeth were embedded in transparent epoxy resin (Epo-Thin 2, Buehler, Lake Bluff, IL, USA) ensuring tight contacts. In a pilot study, such resin models ( Fig. 2 ) had been found to produce DIAGNO imaging similar to that yielded by clinical application. In this pilot study, three examiners assessed 50 randomly shown images taken in a clinical setting as well as 50 randomly shown images taken under the described in vitro conditions (see Fig. 2 ). They were asked to assign an imaging status (clinical or in vitro ) to the images. We found the examiners to not being able to discriminate images taking under clinical conditions from those taken under in vitro conditions. Also, note that another recent study also showed the in vitro evaluation of DIAGNO to be possible if using a suitable model [ ]. For tactile assessment, pink wax was additionally placed to simulate the gingiva and prevent visual evaluation of the proximal surfaces (see Fig. 2 ).
