1.

Introduction

Early detection of caries lesions provides for more efficient arrestment of the caries process, avoiding operative treatment.¹ Current clinical procedures for caries detection, such as visual and radiographic methods, are able to detect caries lesions only at an advanced stage.^{1, 2, 3, 4, 5} When dentists use visual inspection with extensive drying of the suspected site, an earlier and more accurate detection of initial demineralization can be achieved.⁶ Nevertheless, visual inspection is not a quantitative method and minor changes due to the caries process cannot be detected within short periods.²

New technologies are being developed to detect and quantify early caries lesions on smooth and occlusal surfaces. A quantitative laser/light-induced fluorescence (QLF) device measures the intrinsic fluorescence of mineralized tooth tissues. This method provides a good correlation between mineral loss^{3, 7, 8, 9} and early detection of enamel smooth surface caries lesions.^{3, 10} Optical coherence tomography (OCT) is another advanced method that has achieved good results in detecting and quantifying caries lesions on occlusal and smooth surfaces.^{11, 12} However, QLF is not suitable in occlusal surfaces or on dentin caries lesions¹³ and few clinical studies have been performed with OCT.

Another device, based on emission of a light $(\lambda =655\mathrm{nm})$ from a diode laser, is called DIAGNOdent (KaVo, Biberach, Germany). This laser fluorescence (LF) device captures the fluorescence emitted by dental tissues and translates it on a numerical scale from 0 to 99. The higher the number, the deeper the caries lesion.¹⁴ Previous studies with LF have presented good results in detecting occlusal caries lesions.¹⁵ However, the device does not provide a good correlation with mineral loss on smooth surface caries lesions in permanent^{7, 16} and primary teeth.¹⁷ Furthermore, despite the fact that some authors have claimed that the device is able to detect early enamel caries lesions,^{14, 18} studies do not confirm this assertion.^{15, 17} In fact, LF reflects changes in the organic material, rather than in the inorganic content of the teeth,^{7, 14, 16} probably because of the presence of porphyrins in caries lesions, mainly protoporphyrin IX (PPIX).^{14, 19}

Because of these facts, LF is not indicated to detect very early demineralization. Nevertheless, the LF device is portable, easy to use, and available for clinical practice. Thus, the aim of the present in vitro study is to propose the association of the LF device with two fluorescent dyes and to evaluate the performance of the methods in detecting and quantifying very early demineralization from smooth surface caries lesions in primary teeth.

2.

Material and Methods

3.

Results

With the sample destined for TMPyP experiment, there was a significant correlation between LF TMPyP values and calcium loss of the caries lesions demineralized for 24, 48, and 96 h. No significant correlation was found between LF TMPyP readings and calcium loss of lesions produced for 12 h, as well as the LF values and calcium loss with all teeth (demineralized for 12, 24, 48, and 96 h) (Table 1 ). The calcium losses [ $\text{mean}\pm \text{standard}$ deviation (SD)] in these samples were 12 h, 0.49±0.31 ppm/ ${\mathrm{mm}}^2$ ; 24 h; 2.15±1.03 ppm/ ${\mathrm{mm}}^2$ ; 48 h; 3.66±1.59 ppm/ $\mathrm{ppm}∕{\mathrm{mm}}^2$ ; and 96 h; 8.02±2.33 ppm/ ${\mathrm{mm}}^2$ .

Table 1

Pearson’s correlation coefficient between amount of calcium loss and LF values without dye (LF), and LF associated with tetrakis(N-methylpyridyl)porphyrin (LF TMPyP), and LF and LF with protoporphyrin IX (LF PPIX), and areas under ROC curve (Az) in the measurements performed in teeth subjected to different demineralization times and sound teeth.

Concerning the performance, we observed high values of the area under ROC curves with the LF TMPyP method to detect early caries lesions (higher than 0.94, for all periods of demineralization in contrast with measurements in sound surfaces). The areas under ROC curves obtained with the LF method were lower than with LF TMPyP, but there were significant differences only in lesions demineralized for 12 and 24 h (Table 1).

When the lesions were jointly assessed, considering all demineralized teeth (for 12, 24, 48, and 96 h) as carious teeth and the measurements performed in teeth not submitted to demineralization as sound teeth, higher specificity, accuracy, and area under ROC curve was achieved by LF TMPyP association. There was no significant difference in the sensitivity values between the methods, although the value was higher with LF TMPyP method (Table 2 ). Moreover, when considering all samples, there was significant correlation between LF TMPyP and calcium loss ( $R=0.707$ , $p<0.001$ ), while with LF without TMPyP, there was no significant correlation ( $R=0.310$ , $p⩾0.05$ ).

Table 2

Best cutoff points to distinguish between teeth with early caries lesions and sound teeth, and performance of laser fluorescence (LF) and LF associated with tetrakis(N-methylpyridyl)porphyrin (LF TMPyP), and LF and LF associated with protoporphyrin IX (LF PPIX) methods in detecting early artificial demineralization expressed in sensitivity, specificity, accuracy and area under ROC curve (Az)

In the LF PPIX experiments, one tooth demineralized for 48 h was excluded, because it presented a much higher LF value (20.3) than other samples. There was no significant correlation of LF and LF PPIX values with the amount of calcium loss in any period of demineralization (Table 1). Furthermore, there was no significant difference between areas under ROC curves with LF and LF PPIX methods, except in teeth demineralized for 48 h, where LF without PPIX presented better performance than with the dye (Table 1). The calcium losses in the samples destined for PPIX experiment were $(\text{mean}\pm \mathrm{SD})$ 12 h; $0.74\pm 0.40\mathrm{ppm}∕{\mathrm{mm}}^2$ ; 24 h, $2.44\pm 1.37\mathrm{ppm}∕{\mathrm{mm}}^2$ ; 48 h; $4.03\pm 2.01\mathrm{ppm}∕{\mathrm{mm}}^2$ ; and 96 h; $7.95\pm 4.01\mathrm{ppm}∕{\mathrm{mm}}^2$ .

When the samples were assessed in combined groups, there was no significant correlation between LF readings and amount of calcium loss ( $R=0.091$ , $p⩾0.05$ ), but this correlation was significant between LF PPIX values and calcium loss ( $R=0.597$ , $p<0.001$ ). With regard the performance, the LF method showed significantly higher values of sensitivity, accuracy, and area under ROC curve, and similar specificity compared with the LF PPIX method (Table 2).

4.

Discussion

The LF device is portable and available for clinical practice, but it is not able to detect very early mineral loss. Thus, we tested an association of the LF device with two porphyrins to improve the detection of the LF device in very early caries lesions. This is a new approach for the utilization of the LF device.

The dyes were selected based on literature and properties of the products. In the absorption spectrum of TMPyP and PPIX, there are^{20, 23} two strong features at around $400\mathrm{nm}$ (Soret band) and a set of bands around $520\mathrm{nm}$ ( $Q$ bands). Excitation at $655\mathrm{nm}$ is resonant with the $Q$ bands for both porphyrins,^{20, 23} thus, originating emission at longer wavelengths used in the dye-assisted caries lesion detection presented in our study.

For this purpose, we used an artificial caries-induction method for shorter periods than other studies. While some studies used periods of demineralization of about $14\text{to}21\text{days}$ ,^{24, 25, 26, 27} other authors have used shorter periods to evaluate methods to detect early demineralization,^{3, 8, 9, 10, 11, 12} similar to our study.

Our results showed good performance in detecting early demineralization with the LF TMPyP method, even with lesions produced in $12\mathrm{h}$ . With different periods of demineralization, we accomplished good performance of LF TMPyP in detecting the caries lesions, since the areas under ROC curves were higher than 0.94. This association (LF TMPyP) presented significant correlation with calcium loss in caries lesions produced for 24, 48, and $96\mathrm{h}$ . Without dye the LF device did not achieve significant correlation with any lesion.

We observed that the earlier the caries lesions, the lower the areas under ROC curves with the LF TMPyP method. The performance of the LF device on natural smooth surface caries lesions in primary teeth was also better in more advanced caries lesions.¹⁷ These lesions could retain a higher quantity of TMPyP. Thus, the performance in more advanced lesions could improve, as observed in our study.

When the analyses were performed with all samples (independent of demineralization periods), the LF TMPyP also provided good results in detecting (area under ROC $\text{curve}=0.978$ ) and quantifying early caries lesions (Pearson’s correlation $\text{coefficient}=0.707$ ). The other methods (LF and LF PPIX) did not present good results. In a previous study, the performance of LF in detecting artificial caries lesions was better than that obtained in this study.²⁶ Those differences could be related to different methods of caries lesions induction, since the artificial caries lesions in the earlier study were created in $14\text{days}$ using a pH cycling method. The disparity in methodology could explain these differences.²⁶ Inequalities related to the sample drying before the measurements could also explain the differences.²⁸ Furthermore, in the previous study,²⁶ measurements were taken on the same sample before and after the demineralization, while in this study, we carried out the measurements on different samples. Other authors have claimed that LF is not adequate in detecting artificial caries lesions.¹⁴

The better performance obtained by the LF TMPyP method than that achieved with the LF PPIX method could be explained by the chemical properties of the dyes. While the PPIX is a lipophilic substance,¹⁹ the TMPyP is a hydrophilic one, and it has a high affinity for surfaces.²⁰ As the initial demineralization occurs in surface,²⁵ TMPyP should have an advantage over PPIX in the early caries lesions. For longer periods of demineralization, a characteristic subsurface enamel caries lesion is formed, with an apparently intact surface.²⁵ In this way, PPIX could interact better with more advanced caries lesions since it presents anionic properties.¹⁹ In fact, better performance was obtained in quantifying mineral loss with LF PPIX, and a similarity in detecting artificial caries lesions created from 8– to 16–day lesions with two dyes (unpublished data).

Previous research has reported methods of early caries lesions detection. A QLF device (using an argon-ion laser) has shown a significant correlation with artificial caries lesions depth created from $2\text{to}24\mathrm{h}$ . Another study proposed the association of a fluorescent dye with the QLF method.³ The dye-enhanced laser fluorescence method (DELF) showed a very early detection of demineralization in artificial caries lesions formed in $2\mathrm{h}$ . However, the DELF method has not achieved significant correlation with mineral loss. A QLF device (without dye) was able to detect caries lesions with $8\mathrm{h}$ of demineralization and showed better correlation with mineral loss.³ A comparison between these studies and our results is complicated, because the earlier studies have used bovine teeth and a different method of caries induction, while our study uses human primary teeth.

Another study, using primary teeth, observed that a QLF device was able to detect 60% of the teeth demineralized for $24\mathrm{h}$ , and 100% of the samples demineralized for $48\mathrm{h}$ . This study did not present performance values, such as ROC curves, sensitivity, specificity, or accuracy values.¹⁰ This study used primary teeth and an artificial caries method using a demineralizing solution $(\mathrm{pH}=4.5)$ , with shaking, and at $37°\mathrm{C}$ . In our study, we achieved a good performance in detecting tooth demineralized for $12\mathrm{h}$ . We also used primary teeth, the pH of demineralizing solution was higher $(\mathrm{pH}=4.8)$ , and the experiment was carried out at room temperature and with no shaking. Thus, it is possible that the method of artificial caries induction used in the earlier study¹⁰ was more aggressive than the method used in our research.²⁴ If this is the case, LF TMPyP could detect earlier caries lesions than the QLF device

Regarding OCT, it seems that the method is adequate to detect minor mineral changes due to demineralization and remineralization.^{11, 12} Nevertheless, the comparison with a LF device associated with fluorescent dyes is complicated on account of the differences in the methodology used. In fact, further studies aiming to compare these methods, using the same sample and procedures, are necessary to corroborate these findings. Moreover, different concentrations of the dyes, different periods of immersion into dyes, and other kinds of dyes should be tested to improve the performance of association between LF and fluorescent dyes.

Some authors have suggested that methods of early detection of caries lesions could also provide more false positive diagnosis.^{4, 5} Furthermore, a great number of non-cavitated early caries lesions could be arrested without professional intervention. Therefore, these methods should not be cost-effective.⁴ Nevertheless, in primary teeth, caries lesions progress faster than in permanent teeth.²⁷ Besides, we obtained high specificity and sensitivity values with LF TMPyP method, and therefore, a low number of false positive results. A quantitative method that detects minor changes in mineral content due to the caries process could be suitable for evaluation of a patient’s caries activity and for use in short-term clinical studies.²⁶ Moreover, better remineralization has been achieved in small enamel caries lesions than in larger ones.¹

Despite the good results provided by our study, we used an artificial caries method. There are significant differences between artificial and natural caries lesions. While natural caries lesions take a long time to form, the artificial ones are formed in a few days. Thus, the surface zone of artificial caries is more softened than that of natural lesions.²⁵ The method used in our study, however, simulates early demineralization, which occurs in the enamel surface. Nevertheless, other in vitro studies with natural enamel caries must be carried out. After that, in situ and in vivo studies must be designed to confirm the good performance of the LF TMPyP method in early detection of the enamel demineralization and to prove the efficacy in monitoring caries development.

5.

Conclusion

The LF TMPyP association provides excellent performance in detecting early enamel caries lesions (even with $12\mathrm{h}$ of demineralization) and a good correlation with mineral loss of these lesions in primary teeth.