Characterization of sound enamel and natural white spot lesions

Objective: To compare optical, morphological, chemical, and physical aspects of the sound enamel and white spot lesions (WSL) classified as ICDAS 2. Material and Methods: Seventeen human molars with one surface presenting WSL and a sound surface (2 x 2 mm window) were characterized by Quantitative light-induced fluorescence (QLF ® ), Optical coherence tomography (OCT), microhardness, and Raman spectroscopy. The ANOVA and Tukey’s test were used at 5% significance level. Results: The QLF comparison between distinct substrates yielded decreased ∆ Q (integrated fluorescence loss) of -15,37%mm 2 and -11,68% ∆ F (fluorescence loss) for WSL. The OCT detected mean lesion depth of 174,43 µm. ANOVA could not detect differences in the optical attenuation coefficient between the substrates (p>0.05). Lower microhardness measures were observed in WSL than on sound enamel (p<0.05). The Raman spectra showed four vibrational phosphate bands (v1, v2, v3, v4), where the highest peak was at 960.3 cm -1 (v1) for both substrates. However, a 40% decrease in phosphate (v1) was detected in WSL. The peak at 1071 cm -1 was higher for sound enamel, indicating the presence of a phosphate band instead of the B-type carbonate. The spectra showed higher intensity of the organic composition at 1295 cm -1 and 1450 cm -1 for WSL. Conclusion: Non-invasive QLF, OCT and Raman spectroscopy were able to distinguish differences in fluorescence, optical properties, and organic/inorganic components, respectively, between sound enamel and WSL, validated by the destructive microhardness analysis.


INTRODUCTION
The early diagnosis of enamel caries lesions in their subclinical stage is still a great challenge for clinicians and researchers. Therefore, the use of multiple caries diagnostic tools and their association are desirable to help dentists during their clinical routine. A precise early diagnose is essential to control and prevent the progression of the disease by implementing minimally invasive strategies.
It is well known that the deposition of dental biofilm on the enamel surface, the inefficiency of its removal, and the frequent intake of fermentable carbohydrates predispose the initial caries lesions development [1]. The organic acids penetrate the dental enamel, partially dissolve the hydroxyapatite crystals and lead to the loss of calcium and phosphate [1]. This phenomenon may be delayed or paralyzed by saliva, which acts towards mineral deposition and pH increase [1]. Hence, lesion formation is a dynamic process in which the remineralization by saliva makes the environment less critical for the balance of dental structure [1], and the demineralization can be paralyzed; thus, the surface will recrystallize [2][3][4].
In this context, the minimally invasive dentistry seeks for early diagnostic tools to implement management of lesions at their initial stage. Clinical treatment options for active WSLs have shown to be highly successful, promoting lesion arrest and avoiding cavitation and can include fluoride-containing toothpaste, varnishes, and restorative materials [5]. In some cases, esthetic micro-invasive or minimally invasive procedures can mask the white appearance of those lesions [6]. However, the development and improvement of different treatment options rely on understanding how caries initiation affects enamel to better predict clinical outcomes. In addition, predicting the performance of in vitro treatments in clinical lesions can be challenging since artificial lesions may not present the same optical, chemical, and mechanical characteristics as the sound enamel. Understanding the particularities of white spot lesions is essential to indicate the best treatment [7], and it will directly influence the reproducibility and reliability on in situ and in vivo studies.
Visual clinical examination is the conventional diagnostic tool used worldwide, routinely refined using magnification and proper illumination. But when doubts arise, the association of the visual exam with other caries diagnostic tools could improve diagnostic precision [8,9] of enamel lesions in its early stages. Noninvasive methods, such as Quantitative Lightinduced fluorescence (QLF) that analyzes enamel autofluorescence comparing sound and demineralized areas [10-12]; Optical Coherence Tomography, which scans the caries lesions, determines their depths, birefringence and the attenuation coefficient of the light inside the substrate that quantifies the interaction between light and the components inside the analyzed substrates [13][14][15][16][17]; and chemical analysis using Raman spectroscopy [9,[18][19][20][21][22] that provides information regarding changes in the inorganic and organic component of substrates could act as extra strategies in caries diagnosis. To validate these non-invasive above mentioned methods, microhardness was proposed as it is well known to easily distinguish between sound and carious enamel [23][24][25][26][27] especially subclinical lesions. With this rationale, this study aimed to evaluate optical, morphological, chemical, and physical features of enamel surface with WSL and compare them with sound enamel. The null hypothesis tested was: There would be no difference between the characteristics of the sound enamel and WSL analyzed.

MATERIAL AND METHODS
Seventeen human molars were obtained from the Biobank of Human Teeth at the University of Sao Paulo, School of Dentistry, following a protocol approved by the Local Ethics Committee (#3.742.709). As inclusion criteria, the teeth should present a surface (smooth or proximal) with a natural WSL classified as ICDAS 2 and diferenciar a fluorescência, propriedades ópticas e conteúdo orgânico/inorgânico do esmalte sadio comparado com esmalte com lesões de mancha branca, sendo validado pela análise de microdureza.

Araújo et al.
Characterization of sound enamel and natural white spot lesions

Araújo et al.
Characterization of sound enamel and natural white spot lesions sound enamel on its homologous surface. The teeth were cleaned and then stored without any solution in a refrigerator at 4ºC until use.
Before using the teeth for the different methods described, the lesions' size was evaluated, and a reading window (2 × 2 mm) on both the WSL and sound enamel surfaces was standardized using adhesive tape.

Quantitative light-induced fluorescence (QLF)
The teeth were dried for 10 s using a triple syringe immediately before the readings and positioned perpendicularly to the QLF handpiece, and focused. A single examiner performed all the QLF readings, and the images were analyzed using the QLF software (Inspektor Research System BV, Amsterdam, The Netherlands). Differences in fluorescence (∆F) and integrated fluorescence loss (∆Q = ∆F by lesion area) values were recorded at a threshold level of 5%, so 95% of fluorescence loss between sound enamel and the white spot lesion were detected. Data collected included average fluorescence loss expressed in percentage (∆F), size of the WSL area (mm 2 ), and difference in fluorescence integrated by lesion area (∆Q), expressed -%mm 2 .

Optical Coherence Tomography (OCT) -Lesion depth and Optical attenuation coefficient
The teeth were examined by OCT using a superluminescent LED operated at 930 nm with 2mW of power (OCP930RS Thorlabs Inc.), with an axial resolution of 4 µm, the lateral resolution of 6 µm and considering a refractive index of 1.6, for human enamel. Images were acquired and used to determine the lesion depth and the optical attenuation coefficient. Lesion depth was evaluated using the software Image J (NIH) to calculate lesion depth. The lesions were analyzed by scanning them along five horizontal lines, separated by 300µm, and analyzing the image with the software. The mean of the lesion depth (in µm) on the five lines was determined for each surface.
The total optical attenuation coefficient obtained from 5 images per sample was analyzed using homemade software developed in LabView 8. The coefficient was calculated assuming an exponential decay of light intensity inside enamel and lesion (backscattered), according to the Equation 1: Where: I(z) represents detected in-depth intensity z; I 0 is the source intensity value at the sample surface; α is the whole optical attenuation coefficient; z is the evaluated depth; and C is a constant to reduce background noise signal [16]. The arithmetic mean for five values of the attenuation coefficient was obtained for each sample.

Cross-sectional microhardness analysis
After the non-invasive QLF and OCT measurements, dental roots were removed 2 mm below the root crown junction in each tooth, and the crown was vertically sectioned in the buccal-lingual or mesio-distal direction. After that, samples were sequentially polished with silicon carbide papers (P1200 for 30 s, P2500 for 30 s, and P4000 for 4 min) and ultra-sonicated between each abrasive paper for 5 min [28]. The opposite surface was also polished to obtain a flat surface, an indispensable requirement for microhardness measurements using a microhardness Knoop indenter (HMV-G21DT, Shimadzu Co., Tokyo, Japan). A 25 g load was used for 10 s to produce a diamond-shaped indentation under 40 x magnification [23]. Measurements were performed as shown in Figure 1. On each cross-section from the area near the enamel surface (Surface layer-SL) (between 30-50 µm from outer enamel) until a sound enamel (SE), for WSL. For sound enamel, the same WSL reference was used, both following reference values for Knoop hardness. KHN (Knoop hardness number) results from all samples in each condition (sound or WSL) were averaged.

Raman spectroscopy
The chemical characterization of the sound enamel and WSLs (n=3) was performed using the LabRAM HR Evolution Raman spectrometer (HORIBA, Scientific France SAS, France). Samples were analyzed without any previous treatment (sectioning or polishing) using a wavelength of 633 nm, the laser power was 120mW, the pinhole was 200 µm, and 90 acquisitions of 1 s each were made using 0.7 cm -1 of spectral resolution (600 lines/mm grating), and a 10x objective [20]. The spectral range was fixed at 400-1500 cm -1 to analyze the inorganic and organic composition, namely phosphates, carbonates, and organic content. Additionally, photobleaching was performed using the maximum laser power for 5 min before all the readings. Three points were analyzed on sound enamel and three points on natural white spot lesion for each sample. After this, the spectra were processed with a baseline correction, normalized by mean, and smoothened by an 8-point third order polynomial, using LabSpec 6 software (HORIBA, Scientific France SAS, France). The chemical composition analysis was performed on the surface of the substrates, and the collected data were processed and normalized. The data normalization was done using the final average value of the entire spectra specified for each tooth.

Statistical analysis
QLF and Raman spectroscopy results were qualitatively described. One-way ANOVA or twoway ANOVA followed by Tukey's post hoc tests (α=0.05) were conducted to compare sound enamel and WSLs for OCT and microhardness results, using the Minitab 9 software (Minitab Inc., State College, PA, USA).

RESULTS
The QLF data are described in Table I. The average ∆Q (-% mm 2 ) indicates loss of integrated fluorescence, when compared sound and demineralized areas, on the lesion size of -15%mm 2 (on average) of the total area analyzed. Therefore, the values of ∆Q described above have a high coefficient of variation. These values were calculated by multiplying the ∆F (fluorescence loss value) by the demineralized area (mm 2 ) WSL mean size was 1.4 mm 2 with a 71% variation in size within the different lesions. The difference in fluorescence between the sound enamel and the WSL was -11.68%.
Regarding lesion depth, OCT was able to detect lesions between the range 240.70 to 115.74 µm, with an average lesion depth of 174.43 ± 7.52 µm, with a coefficient of variation of 17.25%.
O n e -w a y A N O V A d e m o n s t r a t e d n o significant difference between the values of the optical attenuation coefficient when comparing the sound enamel and the WSL region (p > 0.05) (Table II).
For the microhardness data (Table III), two-way ANOVA demonstrated a statistically significant difference between the substrates (p < 0.001), but similar results for location of the readings in the same sample (surface vs. crosssectional) (p > 0.05) and no interaction between these two factors (p > 0.05). Additionally, comparisons were made between the surface layer (up to 50µm) and the other depths evaluated (100µm-500µm), both for the sound enamel and the white spot lesion. For sound enamel, the comparison of the surface layer (50µm) vs. all the depths described above showed no significant difference (p>0.01). When the white spot lesion was evaluated, there was no significant difference between the microhardness of the surface layer The resultant spectra averages for all teeth in each condition (sound enamel and WSL) obtained with the Raman spectroscopy are shown in Figure 2. Regarding enamel's inorganic content, the highest peak is at 960.3 cm -1 (v1 PO 4 ) for both sound enamel and the WSL, with a 40% decrease in WSL compared to sound enamel. Peaks at 1052 cm -1 (v3 PO 4 ), 579 cm -1 (v3 PO 4 ), 431 cm -1 and 446 cm -1 (v2 PO 4 ) were also clearly seen, with higher values for the sound enamel when compared to the WSL. Higher peak intensity in the 1071 cm -1 (v3 PO 4 ) band was detected for the sound enamel than WSL. In addition, two other peaks were verified, 1295 cm -1 (amide III) and 1450 cm -1 (CH 2 ), which showed greater intensity on the WSLs than sound enamel.

DISCUSSION
This study compared sound enamel and WSLs in several aspects, including chemical composition, mechanical and optical properties. Overall, all properties are altered in WSLs compared to sound enamel, which led to the rejection of the null hypothesis. In addition, lesion depth and changes in mineral content at the lesion body were also evaluated.
As a method to evaluate the fluorescence, QLF ® was used to analyze variations in the mineral content of the surface. The QLF is *Different superscript letters (column) represent statistically significant difference (p < 0.05).

Figure 2 -
The spectra average for all surfaces in each condition representing all the bands for sounds enamel and WSL. Arrows indicate peaks at 431cm -1 (v2 PO 4 ), 579cm -1 (v3 PO 4 ), 960cm -1 (v1 PO 4 ), 1071cm -1 (v3 PO 4 ) related to phosphate vibrational modes and 1295cm -1 (amide III), 1450cm -1 (CH 2 ), related to organic content. In this study, the QLF was efficient in comparing the fluorescence of sound and WSL enamel. It showed a reduction of -15%mm 2 of integrated fluorescence loss (∆Q) and -11.68% of fluorescence loss (∆F) for WSLs. These values are related to mineral loss and to lesion depth, which presents a good correlation with changes in mineral content compared to methodologies such as microradiography [33]. It is noteworthy to mention that the loss of minerals was also evidenced in the Raman readings, which demonstrated a decrease of 40% in the phosphate peak (v1) intensity in WSL. This peak of the spectra is the symmetrical stretching mode of phosphate (v1) at 960.3cm-1, which is the highest peak related to mineralized tissues, such as enamel, dentin, and bone [9,18,19,21]. These results follow some studies demonstrating that decreased phosphate content indicates lower mineralization [9,19]. The enamel comprises hydroxyapatite Ca10(PO4)6(OH)2 crystals and other ions, such as carbonate. This study showed that the phosphate vibration modes (v1, v2, v3, and v4), presented higher intensity in sound enamel due to a more significant amount of phosphate on this surface [19] when compared to WSL.
The QLF® is a clinical diagnostic tool that has been used in laboratorial studies [10] to monitor the effect of different dental materials and treatments in incipient artificial lesions. Therefore, by understanding QLF parameters (fluorescence loss and the loss of integrated fluorescence) of natural lesions could help standardize the development of artificial lesions for future studies.
As a limitation of this method, we observed a high coefficient of variation of ∆Q; it is likely due to the variation of the demineralized areas captured. The natural WSL inherently vary in size and depth, thus this variation was expected and confirmed by the OCT results that detected lesion depths from 116 µm to 241 µm.
Several studies have been carried out using the OCT to diagnose incipient caries due to the optical properties that are altered after the demineralization process [34][35][36][37]. The OCT measures the optical attenuation coefficient, a parameter that quantifies the interaction between light and the components inside the analyzed substrates (A-scan). For this purpose, graphics obtained from OCT readings were analyzed based on the first peak referring to the change in light reflectivity when it contacts the dental enamel surface and the subsequent extinction of this signal resulting in a value described in the results section [15,16,32,38]. In this study, the optical attenuation coefficient observed in sound enamel and WSL were similar. The literature presents controversial results for demineralized enamel compared to sound enamel. While some studies [16,32] detected higher attenuation coefficient values and a high degree of sensitivity and specificity using artificial caries lesions, others reported a lower optical attenuation coefficient for natural WSLs than sound enamel [15,38]. The higher value presented by WSL is due to prism disorganization and enlargement of pores in dental enamel [32]. Tissue disorganization and pores inside the substrate increase the number of interfaces with which the light will interact, increasing the scattering of light and, consequently, the value of the coefficient [16,32]. However, other studies [15,38] suggest that the more significant demineralization presented by natural lesions leads to a greater loss of minerals and larger porosities that can increase the probability that the light does not interact with any structure (light scattering centers) inside the enamel, so light scattering on the tissue will decrease [38]. In this study, the intrinsic size variability of the WSL demonstrated above may be responsible for not detecting differences in the attenuation coefficient between sound enamel and WSL.
Moreover, our Raman results detected a peak in the 1295 cm-1 band which is related to the presence of proteins, such as amelogenin, in the pores of enamel crystallites [9]. Alike Sadyrin [9], we observed a peak in this region, which corresponds with amide III (= CH) [19]. Although the amount of organic content of enamel is not comparable to inorganic content, the enlargement of enamel pores after the demineralization of this substrate can account for protein and lipid content in dental enamel [39].
The OCT was used to measure the lesion depth that yielded an average depth of 174.42 µm, within the estimated value for white spot lesions without a clinically visible appearance. Studies [29,40] indicated that initial caries lesions in enamel are visible to the naked eye when they reach a depth between 300-500µm. Interestingly, even though the deepest lesion detected with OCT in this study was 241 µm, when the microhardness analysis was performed, it was possible to verify changes in the KHN until around 500 µm from the lesion surface. The OCT device has a depth range of up to 1 mm, and this value may decrease after years of use [15]. In this way, the amount of light can be insufficient to penetrate the evaluated lesions in-depth, resulting in scattering light by the surface. In addition, factors such as the absorption coefficient and the interaction between the tissue and the wavelength may have decreased light penetration depth [15,38].
After the use of non-invasive methods use (QLF, OCT and Raman spectroscopy), a crosssectional microhardness analysis was performed to validate differences between sound enamel and WSLs. As mentioned in the methodology, literature [41] previously defined that the reference value used for sound enamel is between 272 and 404 KHN. This study showed that the mean value of the sound enamel was inside this range, and the value for WSLs was significantly lower, demonstrating the demineralization of the substrate, which was also observed by the decrease of 40% in the phosphate peak (v1) intensity in WSL using Raman. When comparing the KHN near the enamel surface (first reading at 30 to 50µm) for the two substrates, a 52% difference was observed. However, no difference was observed when superficial and bottom readings were compared in each substrate. It is worth mentioning that, as in the study by Craig and Peyton [42], our study carried out indentations on the longitudinal surface apart from studies such as Caldwell [23], which found greater hardness for the enamel surface layer.
The surface hardness analysis of the enamel is indicated when it evaluates artificial caries lesions, which commonly present a smaller demineralization depth. However, longitudinal hardness measurements are more indicated to evaluate the natural lesions with a higher depth [43]. While studies [44,45] reported that the surface/outer layer of enamel has a greater hardness than the inner enamel (body), it was not observed in our study that observed similar hardness values for surface layer and enamel body lesion, following Craig and Peyton [42]. This discrepancy might be related to the variation in the direction of indentations within the same tooth [46]. However, one study [45] stated that orientation does not directly influence hardness values. Therefore, in this study, all indentations were carried out parallel to the enamel surface. Moreover, enamel hardness varies from the enamel surface to the adjacent amelo-dentinal junction [47] due to a variation of the organic and inorganic content in dental enamel. In general, the enamel surface presents greater mineral content related to dynamic ionic exchanges with the oral cavity, while deeper regions are composed of greater organic content [48][49][50]. Interestingly, these differences were not detected for sound enamel or white spot lesions in this study when performed the microhardness analysis. We believe that the small size of the enamel lesions classified as ICDAS 2 included in this study could account for the absence of differences in microhardness values.
Regarding the Raman spectroscopy readings, in addition to a decrease in the phosphate peak (v1) and the detection of the presence of organic content (amide III) in WSLs, it was possible to observe the peak of 1071 cm-1, referred to as phosphate v3 vibration mode and B-type carbonate [20,21]. To clarify, if this band represents carbonate or phosphate, observation of other bands is required. If other phosphate vibrations are expressed, then it is phosphate. On the other hand, if A-type carbonate is expressed in 1104 cm-1, the vibration in 1071 cm-1 can represent B-type carbonate. As other phosphate vibrations were expressed in this study, we consider that it represents phosphate (v3), with a higher vibration for sound enamel than for WSL [9,19].
It is known that the analyzed surface, although not intact, still has a high amount of phosphate because the demineralization process of WSL is initial and does not occur homogeneously, which could be observed in Raman results. Also, in deeper lesions, the carbonate substitution of phosphate in the apatite lattice can occur [20,21,28,51], creating carbonated hydroxyapatite that is more susceptible to acid demineralization and responsible for the crystallographic changes of the prismatic structure. A previous study that analyzed mineralized substrates [52] found a band overlap of carbonate under the phosphate band, which did not occur in this study.
One of the limitations of this study relies on sample size and specific knowledge of the origin of the tooth. However, most of the non-invasive methods were able to distinguish sound enamel from WSL. This characterization is crucial for revisiting those substrates' essential aspects and characteristics and further developing enamel caries-like lesions, constructing a substrate that mimics natural lesions, and enabling innovative materials and therapeutic strategies.

CONCLUSION
Non-invasive QLF, OCT and Raman spectroscopy were able to distinguish differences in fluorescence, optical properties, and organic/ inorganic components, respectively, between sound enamel and WSL. These results were validated by the destructive microhardness analysis. Future investigations should continue to explore the tissue properties, mainly from the subclinical incipient enamel lesion to the white spot lesion.