Exploring the boundaries of caries detection : two advanced methods evaluated

Sammanfattning: Background: Caries detection methods require thorough validation. This should include studies which clarify what characteristics of the caries lesions are being measured, the limitations of the method and comparison of performance with conventional caries detection methods. The outcome of validation tests has important clinical implications, such as interpretation of the data at the cut-off points used by the clinician to differentiate between lesions requiring invasive and non-invasive intervention. Aim: The overall aim of the thesis was to evaluate the characteristics, technical efficacy and diagnostic accuracy/efficacy of two advanced methods for caries detection, laser-induced fluorescence (LF) and digital imaging fiber-optic transillumination (DIFOTI), with special reference to implications for clinical treatment decisionmaking at the threshold between non-invasive or invasive management of the lesion. Material and methods: In Study I the LF method was evaluated for occlusal caries detection and quantification in vivo. The method was validated by comparing LF readings to actual lesion depth, determined by opening the fissures. Reproducibility was tested by comparing readings from four different LF instruments for each site and repeatability by comparing two readings from the same instrument. The LF readings were also correlated with the microbial flora cultured from the measured site. Study II, in vitro, tested the ability of the LF instrument to measure fluorescence from the two major components of the caries lesion, demineralized enamel and cultivated caries-associated bacterial flora. Sound enamel was demineralized and the change in mineral content verified by quantitative light-induced fluorescence (QLF). LF readings were taken before and after demineralization, to determine the influence of the demineralization on LF readings. LF readings were obtained from cultures of caries-associated bacterial species, with and without surrounding culture media: the influence of the age of the culture on LF readings was investigated. In Studies III and IV the DIFOTI method was validated for approximal and occlusal caries detection and quantification. Diagnostic accuracy was validated by comparison with actual lesion depth determined by microradiography and histology. Reliability was tested in terms of inter- and intra-observer agreement, by comparing the results of scoring recorded by eight different observers. Results Validation: For three out of the four LF instruments tested there was a weak but significant correlation between LF reading and actual lesion depth (Study I). A different cut-off level was however obtained for different instruments, meaning that no general cut-off threshold could be used to differentiate between enamel and dentin caries. The DIFOTI method showed generally good correlation with actual lesion depth, especially for approximal surfaces, with good diagnostic accuracy and efficacy at both the diagnostic thresholds tested (Study III). Compared to film and digital radiography, DIFOTI showed better performance at diagnostic threshold D1, and similar performance at diagnostic threshold D3. For occlusal caries detection (Study IV), DIFOTI showed moderate to good diagnostic accuracy and efficacy at both diagnostic thresholds. Performance was better than for both types of radiographs, but similar to visual inspection. Reliability: The LF method showed very good reliability in terms of both intraobserver and inter-instrument correlation (Study I). For the DIFOTI method, good to very good intra-observer agreement was achieved for approximal caries detection (Study III) but for occlusal caries detection there was greater variation, agreement ranging from low to very good (Study IV). For approximal caries detection, interobserver agreement was good (Study III) while for occlusal caries detection agreement was generally lower, ranging from low to good (Study IV). Characteristics of the LF method: No change in LF readings was observed with respect to demineralization of dental enamel. LF readings > 20, generally accepted as indicative of dentin caries, were recorded only for mutans streptococci cultured on mitis salivarius bacitracin agar (MSB) and young colonies of Prevotella spp. The readings increased as the colonies of mutans streptococci aged and decreased when the colonies were moved to glass slides (Study II). Conclusions: The LF method can be useful for occlusal caries detection and quantification, under certain conditions: a) no general cut-off threshold can be recommended due to inconsistency between instruments. b) the method does not give information on demineralization or specific information on bacterial content of lesions, but rather responds to the synergistic effect of the caries process. The results of the in vitro investigations suggest that the DIFOTI method may be of value for caries detection and quantification on both approximal and occlusal surfaces. The method shows superior performance to both film and digital radiography, especially for detection of early caries lesions on approximal surfaces. Thus the method shows promise as a means of monitoring early caries lesions and warrants further investigation.