Histologic and radiographic assessment of caries-like lesions localized at the crown margin.

STATEMENT OF PROBLEMS Little is known about the impact of crown margins on the demineralization and remineralization process or on the interpretation of radiographs for the detection of early secondary caries. PURPOSE The aim of this study was to correlate the findings from 2 different radiographic techniques with the results of a histologic assessment of carieslike lesions at the crown margin. MATERIAL AND METHODS Ten fully impacted third molars were extracted and restored with complete metal crowns with chamfer preparations. For each tooth, the prepared margin ended in enamel on one side and extended into cementum/dentin on the other. After being thoroughly cleaned, the teeth were covered with wax except for a 2- x 3-mm window at the crown margin in each of 2 test areas localized at the interproximal enamel and cementum/dentin junction. During demineralization with 6% hydroxyethyl cellulose gel at pH 4.9, direct digital radiographs or conventional radiographs were made at baseline and after 7, 14, 21, and 28 days. Each radiograph was blinded and examined for the presence/absence of lesions by 3 calibrated observers who used predetermined criteria required for receiver operating characteristic (ROC) analysis. After 28 days, the teeth were imbibed with methyl methacrylate and serially ground sectioned for assessment of the lesions in polarized light. Morphometric analysis of lesion extension included the calculation of mean values and standard deviations. Kappa and ROC analysis were applied to assess radiographs made at different demineralization time periods, localization of the crown margins in enamel or cementum/dentin, and direct digital and conventional radiographs. Data were tested for normal distribution (QQ plot), and statistical differences were calculated with Tukey's honestly significant difference post hoc test. RESULTS Radiographic evaluation revealed only slight interexaminer agreement (Kappa) for enamel lesions (direct digital 0.19, conventional radiographs 0.2) and a fair interexaminer agreement for cementum/dentin lesions (direct digital 0.35, conventional radiographs 0.33). ROC analysis revealed no differences between the radiographic techniques. Lesions localized in cementum/dentin (direct digital 0.86 +/- 0.14, conventional radiographs 0.86 +/- 0.09) were diagnosed more reliably than those in enamel (direct digital 0.79 +/- 0.05, conventional radiographs 0.72 +/- 0.10). As microscopically assessed, the depth of lesions in enamel was less than that of lesions in cementum/dentin (49.3 +/- 7.3 microm vs 89.5 +/- 13.2 microm). Erosions were found in cementum/dentin (depth 25.2 +/- 5.3 microm). CONCLUSION Within the limitations of this study, small artificial lesions were detected equally well by conventional and direct digital radiography. However, because of low interexaminer agreement, the radiographic assessment did not reproduce true histopathologic lesion characteristics documented by morphometry of serial ground sections.