Predicting infiltration of the surface layer of natural enamel caries

Predicting infiltration of the surface layer of naturalenamel caries

Kassia Regina Simoes Meira a, Camila Santos de Mattos Brito a,Frederico Barbosa de Sousa a,b,c,*

aMaster Program in Dentistry, Health Sciences Center, Federal University of Paraiba, Cidade Universitaria, 58051-900

Joao Pessoa, Paraiba, Brazilb Laboratory of Microscopy and Biological Image, Health Sciences Center, Federal University of Paraiba,

Cidade Universitaria, 58051-900 Joao Pessoa, Paraiba, BrazilcDepartment of Morphology, Health Science Center, Federal University of Paraiba, Cidade Universitaria, 58051-900

Joao Pessoa, Paraiba, Brazil

a r c h i v e s o f o r a l b i o l o g y 6 0 ( 2 0 1 5 ) 8 8 3 – 8 9 3

a r t i c l e i n f o

Article history:

Accepted 2 March 2015

Keywords:

Enamel caries

Histopathology

Permeability

Capillarity

Infiltration

a b s t r a c t

Aim: To test the hypothesis that the water volume more easily available for diffusion (ad) is

the best predictor among all major components of the proportion of pore volume infiltrated

by a liquid in the surface layer of dry natural enamel caries (NEC).

Materials and method: Two aqueous solutions of mercuric and potassium iodide (Thoulet’s

solutions) with different refractive indexes (1.4 and 1.47) and penetration coefficients

(3212 cm/s and 2297 cm/s) were tested at histological points (n = 63) of ground sections of

NEC lesions. Component volumes were measured with microradiography and interpreta-

tion of birefringence. Real-time 2D mapping of capillary flow was performed with orienta-

tion-independent polarizing microscopy.

Results: ad was a good predictor for both liquids (T1.40: R2 = 0.413; T1.47: R2 = 0.505), but was

similar to the water and air volumes for Thoulet’s 1.47, and to the mineral and organic

volumes for Thoulet’s 1.40. From real-time 2D mapping, infiltration in ground sections

occurred in two propagation directions, perpendicularly to the prism paths (at the centre of

the lesion bodies) and axially to the prism paths (at all parts of the lesions), with two

penetration rates, the faster related to prisms sheaths and the slower related to intrapris-

matic pores, affecting penetration length and air displacement.

Conclusions: ad was a good predictor for both liquids, but was similar to the water and air

volumes for T1.47 and to the mineral and organic volumes for T1.40. Both flow mechanics

and component volumes are required to interpret infiltration of liquids into NEC.

# 2015 Elsevier Ltd. All rights reserved.

* Corresponding author at: Departamento de Morfologia, Centro de Ciencias da Sau de, Universidade Federal da Paraıba, Cidade Uni-versitaria, S/N, CEP 58051-900 Joao Pessoa, Paraıba, Brazil, Tel.: +55 83 3216 7254; fax: +55 83 3216 7094.

E-mail addresses: [email protected] (K.R.S. Meira), [email protected] (C.S. de Mattos Brito), [email protected](F.B. de Sousa).

Available online at www.sciencedirect.com

ScienceDirect

journal homepage: http://www.elsevier.com/locate/aob

http://dx.doi.org/10.1016/j.archoralbio.2015.03.0010003–9969/# 2015 Elsevier Ltd. All rights reserved.

a r c h i v e s o f o r a l b i o l o g y 6 0 ( 2 0 1 5 ) 8 8 3 – 8 9 3884

1. Introduction

The early stages of the carious process result in a complex

network of nano-scaled larger-than-normal pores in dental

enamel. Options for treating natural enamel caries (NEC)

lesions have long been investigated by arresting further

development and reducing pore size. Reduction of pore size

can be achieved by remineralization1 or infiltration of the

pores with fluid resins (known as infiltrants).2 The outcomes of

such treatments depend, among other factors, on the ease

with which remineralizing agents and infiltrants penetrate

into NEC. Data indicated that the surface layer (SL) of NEC

reduces the penetration of materials into NEC. Its removal has

been shown to be important for proper penetration of resin

adhesives and infiltrants into NEC,3–6 and increased penetra-

tion of remineralizing materials has been achieved by

procedures involving demineralization (for more information

see review by Cochrane et al.)1 and removal of organic matter.7

As was recently highlighted in a study, development of

procedures to increase penetration of materials through the SL

of NEC will represent significant advance in clinical reminer-

alization1 as well as infiltration. A recent report showed that

infiltrants penetrate deeper into NEC when they are active

(probably because their SL is more porous)8 or when inactive

lesions are pretreated with acid (to enhance pore sizes).9

Compared to sound enamel and inactive NEC lesions, active

NEC lesions more frequently present pH � 5.5 in the pores of

their lesion bodies, which might favour a more porous SL.10 To

preserve more tissue over time, it was emphasized that

preserving the SL of NEC is preferable to acid pretreatment,9

and a more conservative pre-treatment of the SL of active NEC

was recently reported.11

Resin infiltrants are transported into enamel pores by

capillary forces when carious enamel is dry,12,13 filling the air

volume of enamel pores. Such air volume is the volume of air

that replaces the water pore volume removed at room

temperature (known as the loosely bound water volume),

which, in turn, is a fraction of the water pore volume,14 and the

total pore volume of carious enamel is the sum of the organic

and water volumes.15 The volume of certain foreign liquids

infiltrated into the pores of carious enamel has been

previously quantified: quinoline and alcohols in the dark

zone of NEC, using interpretation of enamel birefringence16;

and a resorcinol–formaldehyde resin in the whole enamel

caries lesion (including both natural and artificial lesions, but

with no data at any particular histological layer) by measuring

the mass of chloronaphthalene imbibed into the air pore

volume created by drying in vacuum at room temperature.17

Those measurements neglected the organic volume and the

volume of water that remains after drying at room tempera-

ture (known as the firmly bound water volume). There are no

quantitative data on the volume filled by liquids (including

infiltrants) in carious enamel taking into account all consti-

tuents of the pore volume, and such data are important for

studying the permeability of the SL of NEC. Recent methodo-

logical developments with ground sections of NEC have

enabled spatially resolved measurement of the water volume

more easily available for diffusion,15 which is expected to be

directly proportional to the air volume in dry carious enamel.

By testing this parameter, the influence of the components of

enamel (mineral, water, and organic matter) on the perme-

ability of the SL can be further investigated and contribute to

improve infiltration of materials into NEC.

The aim of this study was to test the hypothesis that the

water volume more easily available (compared to the total

water, firmly bound water and loosely bound water volumes)

for diffusion is, among all enamel component volumes, the

best predictor of the proportion of total pore volume infiltrated

by a liquid in the SL of dry (at room temperature) NEC. The

liquids tested were two aqueous solutions with high refractive

indexes and high penetration coefficients (Thoulet’s solution),

which were previously used to define histological layers of

NEC under polarizing microscopy.18

2. Materials and methods

2.1. Samples

Sixty-three histological points were selected from 21 non-

cavitated approximal inactive (opaque enamel with shiny

surface) NEC lesions from human third molars and premolars,

which were extracted for oral health reasons. Lesion activity

was determined by a consensus of calibrated examiners

(intra-examiners kappa 0.739 and 0.856; inter-examiner kappa

0.812) with regard to the NYVAD system19 applied to a pool (30

lesions) of non-cavitated NEC lesions that were analyzed

under the stereomicroscope after removal of organic coating

(dental plaque) on the tooth surface (1% sodium hypochlorite

for 30 s), washing with water (30 s), and air drying with

compressed air (10 s). This study was approved by the Ethical

Committee on Research in Humans of the Lauro Wanderley

University Hospital (Federal University of Paraiba, Brazil;

protocol number 09285912.3.0000.5188), and adult volunteers

donated all teeth with signed consent. Teeth and ground

sections were kept in 0.02% aqueous sodium azide solution

before and after infiltration experiments with test solutions.

Longitudinal sections were obtained from the approximal

surfaces with NEC using a diamond disc under water

irrigation. Then, each section was ground to a thickness of

50–90 mm using a lapping jig as recently described.20 Sample

thickness at the selected histological points were measured

with the sections positioned edge-on in a transmitted light

polarizing microscope equipped with a reticle and a 20�objective (resolution of 0.7 mm).

2.2. Quantification of mineral volume

Glass plates sensitive to X-rays (resolution of 2000 lines/mm;

AGHD, Microchrome Tech., USA) were covered by all ground

sections and an aluminium step-wedge calibration standard

(10 sheets, each with a thickness of 20 mm and purity of

0.999%; ESPI Chemicals, USA); subsequently, they were

exposed to X-rays in a PCBA Inspector (General Electric,

Germany; Tungsten anode filtered by a 0.254 mm thick

beryllium window) operating at 40 kV and 0.25 mA for

25 min. The corresponding emission peak is 24 keV,21 which

was used to calculate linear attenuation coefficients of enamel

mineral (using density of 2.99 g cm�3)22 and aluminium (using

a r c h i v e s o f o r a l b i o l o g y 6 0 ( 2 0 1 5 ) 8 8 3 – 8 9 3 885

density of 2.7 g cm�3) as previously described.20 Developed

radiograph plates were photographed in a transmitted light

microscope using a digital camera (Nikon D80, Japan) in a

single session (with standardized illumination conditions); a

number of images (of NEC and aluminium step-wedge, raw

images R, of the field of view without the plate, flat field F, and

the field of view with the lens of the camera blocked, dark field

D) were obtained for use in an image analysis freeware

(ImageJ, NIH, USA) with the following image processing:

(R � D)/(F � D).23 This corrected the heterogeneous illumina-

tion of the field of view in order to avoid bias in measuring grey

levels. Using a calibration curve between grey levels and

aluminium thickness fitted with a fourth order polynomial

(R2 = 0.999), percent mineral volume (V1) was measured at the

selected histological points of the SL of NEC from24:

V1 ¼mAl � tAl

mm � ts� 100 (1)

Linear attenuation coefficients of aluminium (5.884) and

enamel hydroxyapatite (11.445) are mAl and mm, respectively. tAl

represents the aluminium thickness corresponding to the same

grey level measured at a histological location (15 � 15 mm),

obtained from the calibration curve described above. tsrepresents the sample thickness at the same histological point,

measured under light microscopy as described above.

The sample size was comprised of three histological points

in a row with a separation of 50 mm from one to the next at the

SL of each lesion (n = 63). All subsequent measurements of

component volumes were performed at the same locations.

2.3. Quantification of water and organic volumes andwater volume more easily available for diffusion(permeability)

Birefringence at the selected histological points under water

immersion (24 h of immersion) was measured (mean of five

measurements) with a Berek compensator in a polarizing

microscope (Axioskop 40, Carl Zeiss, Germany) equipped with

a green filter (546 nm, bandwidth of 10 nm; Edmund Optics,

USA). Observed birefringence (BRobs) in water was interpreted

using the approach described by Sousa et al.25 in order to

quantify water (a) and organic (b) volumes. This procedure was

described in detail in various previous reports.15,20,26 Current-

ly, these calculations are performed by customized software

(developed by our group) that is available for free use on the

internet (http://hotfile.com/dl/204330734/6099675/EBI-0.1.jar.

html). Then, we calculated the amount of the water volume

more easily available for diffusion (ad, which we consider a

permeability parameter) by15:

ad ¼a2

a þ b� 100 (2)

This equation is the product of a by a/(100 � V1), represent-

ing effective pore size and the inverse of pore viscosity,

respectively.

2.4. Thoulet’s solutions

Two aqueous solutions with refractive indexes higher than

that of water were prepared to infiltrate the ground sections

with carious enamel. Such solutions were prepared by

dissolving equal amounts of mercuric and potassium iodide

in deionized water until the refractive index (measured in an

Abbe refractometer) reached the desired values: 1.40 and 1.47.

The final pH of both solutions was 6.5. Hereafter, these

solutions are referred to as Thoulet’s 1.40 and Thoulet’s 1.47.

The surface tensions of the solutions, measured in a surface

tensiometer (model Q-6000, Sensadyne Instruments) at 25 8C,

were 71.24 mN/m (Thoulet’s 1.4) and 71.14 mN/m (Thoulet’s

1.47). Their densities (at 25 8C), calculated from the densities of

the solutes and their volume fractions in the solutions, were

1.367 g cm�3 (Thoulet’s 1.40) and 1.871 g cm�3 (Thoulet’s 1.47).

Their kinematic viscosities were 0.812 � 10�6 m2/s (centiS-

tokes; Thoulet’s 1.40) and 0.828 centiStokes (Thoulet’s 1.47)

(mean of 30 measurements in a manual Ubbelohde viscosim-

eter at 25 8C), and the corresponding dynamic viscosities were

1.109 mPa s (Thoulet’s 1.40) and 1.549 mPa s (Thoulet’s 1.47).

The approximate penetration coefficient (PCprox2; we use

‘‘prox2’’ to distinguish this coefficient from approximate PC

reported by others13) for each solution, which was adapted to

the situation of infiltration into enamel nanochannels with

walls previously covered by water and organic matter. It was

taken into account that NEC dried at room temperature

present firmly bound water and organic matter in their pores,

as has been shown to be the case for both NEC and enamel

under maturation dried at room temperature,14,20 rendering

measurements of contact angles practically impossible with

current technology. PCprox2 was calculated by12:

PCprox2 ¼g

2h(3)

where g and h are the surface tension and the dynamic viscosi-

ty, respectively. The PCprox2 values were 3211.9 cm/s (Thoulet’s

1.40) and 2296.5 cm/s (Thoulet’s 1.47). The published approxi-

mate penetration coefficients (hereafter referred to as PCprox1)

of 65 resins,13 which were measured by taking into account the

cosine of the contact angle of the liquid with polished normal

enamel, surface tension, and viscosity, were compared with

their corresponding PCprox2 values (Eq. (3)). We tested both

correlation and agreement (using Bland & Altmant plots) in

order to evaluate the usefulness of PCprox2 relative to PCprox1

values of resin infiltrants currently used in dental research.

2.5. Quantification of the infiltration of Thoulet’ssolutions

After measuring BRobs in water, each sample was dried at

room temperature (258 C and 50% relative humidity) for 48 h,

immersed in Thoulet’s 1.47 for 24 h, and then birefringence

was measured using the same equipment (Berek compensa-

tor) at the same points previously analyzed.

Subsequently, all samples were immersed in water again

and the water was changed until the retardance returned to

values initially found under water immersion. The recovery of

retardance to the initial values was an indication that mineral

dissolution did not occur or was negligible during immersion

in Thoulet’s solution. Then, a new drying procedure and

immersion for 24 h in Thoulet’s 1.40 was performed in order to

measure birefringence in this medium; Thoulet’s 1.40 was

prepared in the same way as Thoulet’s 1.47.

a r c h i v e s o f o r a l b i o l o g y 6 0 ( 2 0 1 5 ) 8 8 3 – 8 9 3886

One of the main challenges in NEC is that the lesions

become opaque (no birefringence) after drying; hence, loosely

bound water (a2) cannot be measured directly. It is important

to quantify how much air, which replaces a2 after drying, was

replaced by the infiltrating material and whether or not it

replaced part of the firmly bound water (a1) found in enamel

pores after drying. The sum of a2 and a1 yields the total water

volume (a). A recent discovery regarding the relationship

between loosely bound water and organic volume facilitated

the development of a way to calculate an approximation of a2

from experimental mineral and organic volumes. Once those

data were available for all NEC lesions in this study,

approximate experimental a2 was determined by14:

aexp2 ¼ atheo

2 � 0:5759 � 0:7396Db (4)

atheo2 is the predicted a2 as a function of the experimental

mineral volume and its volume fraction is calculated by26:

atheo2 ¼ 0:6987 � 1:2487 � V1

100

� �þ 0:544 � V1

100

� �2

(5)

Additionally, Db is the difference between experimental

and predicted organic volumes. The fraction of the predicted

organic volume is determined by26:

btheo ¼ 0:08654 þ 0:46808 � V1

100

� �� 0:584 � V1

100

� �2

(6)

The values determined by Eqs. (5) and (6) have to be

multiplied by 100 in order to be inserted in Eq. (4). aexp2 is a

parameter related to the empty capillary volume.

By using these equations, we assumed that dried NEC had an

air volume equal to aexp2 . Using this value in the mathematical

approach for the interpretation of BRobs published by Sousa

et al.,25 it is possible to quantify the volumes infiltrated by

Thoulet’s solutions, by air (if any) and by a1 reasonably

approaching the real situation. The calculations revealed that

the refractive index of Thoulet’s solution first replaced aexp2 ,

either partially or completely (depending on the situation), in

order to equalize the value of theoretical BRobs with the value of

the experimental BRobs. The volume of air and a1 after infiltration

of Thoulet’s 1.47 were different from those of Thoulet’s 1.40.

2.6. Real-time 2D mapping of the infiltration of Thoulet’ssolutions

Some NEC lesions were selected to perform real-time mapping

of the infiltration of Thoulet’s 1.47 and 1.40 into NEC by

registering changes in enamel birefringence using orientation-

independent polarizing microscopy (the polscope system).

The polscope system measures phase retardance at all points

of the sample in the field of view independently of the sample

orientation by rotating the polarization states of light rather

than rotating the sample,27 allowing the exploration of new

information on the transport of materials in enamel. The

infiltration was mapped with two temporal resolutions: 0.5 s

(seven NEC lesions, using orientation-independent polarizing

microscopy with two liquid crystals; hereafter called as dual

polscope27), and time lapses of 1 and 5 min (10 NEC lesions,

using orientation-independent polarizing microscopy with

one liquid crystal, hereafter called as single polscope28). Both

systems used four polarization states, and the four-frame

algorithm27 was used to evaluate the final image, where the

grey levels of the sample represent the retardance values.

Seven lesions had infiltration of Thoulet’s 1.40 mapped

during 150 s (300 frames; 0.5 s interval/frame), as determined

with the dual polscope system, which was mounted in an

inverted polarized light microscope (Eclipse Ti, Nikon, Japan)

with a 4�/0.2 Plan Apo objective, two liquid crystals (model

SPR 0399, CRI, USA), a 546 nm bandpass filter (bandwidth of

12 nm), a quarter wave-plate, a voltage generator, and a

monochromatic digital camera (model C 11440, Hamamatsu,

Japan). In 10 lesions, infiltrations of both Thoulet’s solutions

were mapped during 1 h with the single polscope, which was

mounted in an upright polarizing microscope (model BA310

Pol, Motic, Canada) with a 5�/0.2 Plan Fluorite objective, one

liquid crystal covered with achromatic quarter wave plate

(Meadowlark, USA), a 546 nm interference filter (bandwidth of

10 nm), a quarter wave plate, a voltage generator, and a digital

camera in monochromatic mode (Nikon D7000, Nikon, Japan).

Infiltrations were performed at 25 8C.

After drying, the sample was placed between the glass slide

and a cover glass, and Thoulet’s solution was injected with a

brush into the space between the cover glass and the glass

slide and diffused to the sample. With the single polscope (low

temporal resolution), a series of measurements of changes in

birefringence were performed at the following time intervals:

1 min intervals within the first 10 min and 5 min intervals

up to 1 h after injection. Technical problems did not allow

evaluation of longer periods of time.

2.7. Data analysis

We calculated the determination coefficients of each compo-

nent volume (mineral volume, a, b, ad, and aexp2 ) in relation to

the proportion of the pore volume infiltrated by each Thoulet’s

solution. The resulting statistical power for each determina-

tion coefficient (using two-tailed directionality) was deter-

mined, for both Thoulet 1.47 (n = 63) and Thoulet 1.40 (n = 48).

Paired comparisons between ad with the other component

volumes regarding R2 coefficients were performed using a

statistical Z test. A Z score for each determination coefficient

was calculated using Fisher’s transformation, and the Z score

for each paired difference of R2 coefficients was used to obtain

two-tailed p values from the Z table.29 Confidence intervals at

95% level were obtained. The magnitudes of the differences

between R2 parameters of predictors were tested using

Cohen’s effect size for correlation coefficients29:

q ¼ ðZ1 � Z2Þ2 (7)

and

z ¼ 12� ln 1 þ

ffiffiffiffiffiffiR2

p1 �

ffiffiffiffiffiffiR2

p !

(8)

where Z1 and Z2 are the Fisher’s Z scores of the R2 coefficients

of ad and another parameter (mineral volume, a, b, or aexp2 )

under comparison, respectively. For the statistically signifi-

cant differences, statistical power was computed.

To test if PCprox2 improved the proportion of pore volume

infiltrated by the tested solutions, the Wilcoxon test with

ranked data was performed. Using the same sample size of 48

a r c h i v e s o f o r a l b i o l o g y 6 0 ( 2 0 1 5 ) 8 8 3 – 8 9 3 887

histological points for each group (Thoulet’s 1.47 and Thoulet’s

1.4), type I error of 5%, power of 80%, and a standard deviation

of paired differences of 6.592% of infiltrated pore volume

(measured in this study), the minimum detectable difference

between groups was 2.72% of infiltrated pore volume.

Real-time 2D mapping of solution infiltration was qualita-

tively analyzed by visual examination.

3. Results

Typical histopathological features of NEC lesions are shown in

Fig. 1. Parts of the SL (negatively birefringent) and the

positively birefringent body of the lesion (the most deminer-

alized area) seen under water immersion (Fig. 1B) are shown as

opaque (no birefringence) under air immersion (before

Fig. 1 – Typical features of one NEC lesion. (a) Microradiography s

grey. (b) Under polarizing microscopy with Red I retardance filte

oriented at S458, water immersion shows the SL as negatively

the lesion. (c) Air immersion clearly shows opaque (no birefringe

positively birefringent body of the lesion. (d) Without Red I filte

parts, making it impossible (from this single image) to distingu

sample has the same colour of the background). (e) In Thoulet’s 1

positively birefringent body of the lesion. (f) Under immersion in

Scale bars = 500 mm. (For interpretation of the references to colou

the web version of this article.)

infiltration with Thoulet’s solution) (Fig. 1C–D). In Thoulet’s

1.40, the SL was negatively birefringent and the body of the

lesion presented a reduced positive birefringence (Fig. 1E).

Under immersion in Thoulet’s 1.47, the entire lesion was

negatively birefringent (Fig. 1F).

Fig. 2 shows the mineral (V1), organic (b), total water (a), and

ad volumes for all histological points. For the experiment with

Thoulet’s 1.40, five lesions could not be analyzed because of

technical problems during handling, yielding a total of 48

points (16 lesions � 3 points = 48 points), while all 63 points

were included in the experiment with Thoulet’s 1.47. The

volumes infiltrated by Thoulet’s 1.40 (a1:402 ) and Thoulet’s 1.47

(a1:472 ), and the firmly bound water volumes found after 24 h of

immersion in Thoulet’s 1.47 (a1:471 ) and 1.40 (a1:40

1 ) are shown in

Fig. 3a–d. For both solutions, some small portions remained

filled with air (data partially shown in Fig. 5c).

howing a surface layer (SL) with a relatively high amount of

r (which results in red background) and the sample prisms

birefringent and outlines a positively birefringent body of

nce) enamel at both the surface layer (some parts) and the

r, the background assumes the same colour of the opaque

ish between opaqueness and pseudo-isotropy (when the

.40, the lesion has a negatively birefringent SL and reduced

Thoulet’s 1.47, the entire lesion is negatively birefringent.

r in this figure legend and the text, the reader is referred to

(a)

(b) (d)

(c)

Fig. 2 – Mineral (V1), organic (b), total water (a), and ad (permeability) volumes for all histological points (n = 63).

a r c h i v e s o f o r a l b i o l o g y 6 0 ( 2 0 1 5 ) 8 8 3 – 8 9 3888

Analysis of all candidate predictors of the proportion of the

pore volume infiltrated by Thoulet’s 1.47 (a1:472 =V2) and

Thoulet’s 1.40 (a1:402 =V2) showed that the best predictor for

Thoulet’s 1.47 was ad (R2 = 0.505), followed (in decreasing

order) by loosely bound water (aexp2 ), a, V1, and b (Fig. 4). For

Thoulet’s 1.40, ad was also the best predictor, but with a lower

predictive value (R2 = 0.413) than that for Thoulet’s 1.47; in

addition, b had a very close predictive value (R2 = 0.370). The

(a)

(b)

Fig. 3 – Volumes infiltrated by Thoulet’s 1.47 (a1:472 , ‘‘a’’; n = 63)

bound water volumes after 24 h of infiltration of Thoulet’s 1.47

remaining predictors were (in decreasing order): V1, aexp2 , and a

(Fig. 4). A test of agreement between volumes filled with air

after infiltration revealed a heterogeneous distribution, with

some points presenting much more air after infiltration of

Thoulet’s 1.40 than after infiltration of Thoulet’s 1.47 (Fig. 5c).

However, the proportion of the pore volume infiltrated by the

two liquids did not differ significantly statistically ( p = 0.63;

Wilcoxon test).

(d)

(c)

and Thoulet’s 1.40 (a1:402 , ‘‘c’’; n = 48) as well as the firmly

(a1:471 , ‘‘b’’; n = 63) and Thoulet’s 1.40 (a1:40

1 , ‘‘d’’; n = 48).

Fig. 4 – Linear correlations between some predictors and the proportion of the pore volume infiltrated by Thoulet’s 1.47

(a1:472 =V2; n = 63) and Thoulet’s 1.40 (a1:40

2 =V2; n = 48). (a)–(d) Data on Thoulet’s 1.47 showing ad as the best predicted

(R2 = 0.505). (e)–(h) data on Thoulet’s 1.40 showing ad as the best predictor (R2 = 0.413), followed by b (R2 = 0.370). aexp2 was a

poor predictor for Thoulet’s 1.40 (i), but the second best for Thoulet’s 1.47 (j).

a r c h i v e s o f o r a l b i o l o g y 6 0 ( 2 0 1 5 ) 8 8 3 – 8 9 3 889

Fig. 5 also shows the results of the analysis of correlation

and agreement of PCprox1 [Eq. (3)] and PCprox2 (both measured

at 25 8C) of 65 resins reported elsewhere.10 Strong positive

correlation and good agreement are shown (Fig. 5).

Results of the statistical analysis of paired comparisons of

R2 coefficients are shown in Table 1. For Thoulet’s 1.47, ad was

statistically significantly higher than V1 ( p < 0.0001, a large

effect size of 0.62, and power of 99%) and b ( p = 0.0012, a

medium effect size of 0.36, and power of 91%), but similar to a

and aexp2 . For Thoulet’s 1.40, ad was statistically significantly

higher than a ( p = 0.0002, medium effect size of 0.46, and

power of 88%) and aexp2 ( p = 0.0014, a medium effect size of 0.34,

and power of 80%), but similar to V1 and b.

Real-time 2D mapping of infiltration of Thoulet’s 1.40

with high temporal resolution (time interval of 0.5 s

between frames) showed that penetration started in the

centre of the body of the lesion and simultaneously

propagated fast in two directions: laterally and axially to

the prism paths. However, the extent of lateral propagation

decreased from the centre of the body of lesions outward in

all directions and was negligible at the most peripherical

parts of the body of the lesion after 2 min (video 1). At low

temporal resolution (at a lower time interval, 1 min),

penetration of Thoulet’s 1.40 occurred only axially to prism

paths, propagating from the centre of the body of the lesion

towards its periphery. After reaching the innermost part of

the body of the lesion, penetration proceeded (at a lower

rate than the previous one) axially towards the SL,

displacing air and resulting in an air bubble emerging from

the SL (video 2). The total length filled by the liquid was the

sum of the distance from the centre of the body of the lesion

to the innermost point of the lesion plus the distance from

that latter point to the SL.

Video 3 shows the same lesion during infiltration of

Thoulet’s 1.47 at low temporal resolution. This video confirms

that penetration propagates axially to the prism paths. The

penetration rate was slower and air bubbles were not seen.

The solution took longer to reach the bottom of the body of the

lesion, and the late outward flow was less intense compared

with that observed with Thoulet’s 1.40.

(a) (b)

(c)

Fig. 5 – Linear correlation (a) and agreement ((b) Bland and Altman plot) between PCprox1 and PCprox2 for the 65 resins

reported by Paris et al.10 (c) Agreement (Bland & Altman plot) between air volumes remaining after infiltration of Thoulet’s

1.40 and Thoulet’s 1.47.

a r c h i v e s o f o r a l b i o l o g y 6 0 ( 2 0 1 5 ) 8 8 3 – 8 9 3890

4. Discussion

The penetration power of a liquid in a dry capillary, as

originally described by Washburn,9 was determined based on

the product of Eq. (3) by the cosine of the angle of contact of the

liquid with the material composing the capillary walls. We

showed that air, firmly bound water, and organic matter fill

Table 1 – Results of statistical analysis of paired comparisonsthe determination coefficients R2.

Liquid

V1 b

Thoulet’s 1.47

ad p < 0.0001

CI = 0.606; 0.293

q = 0.6212

Power = 99%

p = 0.0012

CI = 0.477; 0.10

q = 0.3549

Power = 91%

Thoulet’s 1.40

ad p = 0.2892

CI = 0.203; 0.010

q = 0.037

p = 0.8572

CI = 0.112; 0.05

q = 0.003

p = two-tailed statistical significance; CI = upper and lower limits of 95% c

and two-tailed directionality.

Categories of effect sizes q for related groups24: <0.02 is negligible; 0.02

these nanochannels (outlined by hydroxyapatite walls) in dry

NEC. Evidence from transmission electron microscopy of dry

human enamel with low mineral content and high organic

content (similar to NEC) indicated that the hydroxyapatite

crystallites are coated by the organic content30 so that liquids

most likely contact the organic content and not the surface of

the hydroxyapatite crystals when infiltrating the nanochan-

nels of dry NEC lesions. Such organic content is composed of

between ad and other component volumes with regard to

Parameters

a aexp2

9

p = 0.0588

CI = 0.288; 0.009

q = 0.1196

p = 0.1416

CI = 0.230; 0.000

q = 0.0726

s

p = 0.0002

CI = 0.557; 0.132

q = 0.4592

Power = 88%

p = 0.0014

CI = 0.502; 0.084

q = 0.3497

Power = 80%

onfidence interval; q = effect size; power is related to type I error of 5%

is small; 0.18 is medium; and 0.50 is large.

a r c h i v e s o f o r a l b i o l o g y 6 0 ( 2 0 1 5 ) 8 8 3 – 8 9 3 891

both proteins and lipids.31 Contact angles vary depending on

the nature of the organic material that covers the wall

surface,32 and drying alters the conformation of adsorbed

proteins,33 thus affecting contact angle. It is not known which

material actually contacts the liquids that infiltrate the

nanochannels in dry carious enamel, and this uncertainty

leads us to propose PCprox2 as a measure of the penetration

power of the liquids tested here. PCprox2 is strongly correlated

and strongly agrees with PCprox1, which is calculated by

assuming that hydroxyapatite contacts the liquids (Fig. 5a–b)

and is the most common PC currently used in dental research

on caries infiltrants.13 Thus, the liquids tested here can be

compared to resin infiltrants with regard to penetration

power.

In this study, we used Thoulet’s solutions, which played an

important role in previous analysis of the histological layers of

NEC,18 because they do not result in either demineralization or

remineralization nor undergo light curing (as resins do),

allowing subsequent analysis over a long duration to be

performed in the same undisturbed sample. The fact that

component volumes differed among all histological points

justifies treating each histological point as an independent

sample, yielding a sample size of 63.

Inactive NEC lesions were chosen because they are more

challenging with regard to infiltration. Compared with active

NEC lesions, inactive lesions have been shown to present a

lower resin-infiltrated area9 and a less porous34 and thicker

SL.8 Moreover, the reduction in opacity resulting from resin

infiltration is more difficult to predict in inactive NEC lesions

compared with active lesions.35,36 Removal of dental plaque is

required to allow incident light to reflect directly on the

surface of NEC, enabling the detection of the dull or shiny

surface appearance taken into account to assess the activity of

NEC. Here we preferred using 1% sodium hypochlorite for 30 s

to applying mechanical forces with brushing for removal of

dental plaque because it has been shown that mechanical

forces applied by brushing might cause wear/polishing of the

surface layer of NEC37,38 – and active NEC lesions present more

loosely packed crystals in their SL than inactive lesions,39,40 –

rendering it less irregular and, thus, more likely to present a

shiny appearance under reflected light. This could biases the

detection of inactive/active NEC lesions. Considering that

application of 3% sodium hypochlorite for 30 min without

ultrasonication does not remove the enamel pellicle from the

surface of sound enamel,41 most likely enamel pellicle in our

samples was not removed by 1% sodium hypochlorite for 30 s.

This and the basic pH of sodium hypochlorite indicate that

demineralization of the enamel surface was unlikely.

Despite the fact that ground sections with all surfaces

exposed to the liquid are expected to be infiltrated faster than

intact teeth, the quantitative volumetric data and the time

range of 2D mapping used here allow important information

to be derived with regard to the transport of materials in intact

carious teeth, including resin infiltrants. Considering compa-

rable PCproxy2 and temperature, once the liquid has penetrated

past the enamel surface of NEC in intact teeth, its transport up

to the maximum infiltration is expected to follow the

propagation directions described here.

In normal (non-carious) enamel, long hydroxyapatite

crystallites (the same length of the enamel layer) with a small

diameter (30 nm � 70 nm) are packed in cylinders (prisms;

5 mm in diameter) containing two types of nanochannels

according to their diameter: (i) 2–3 nm in intraprismatic, and (ii)

4–6 nm located in prism sheaths.42 These nanochannels

outline 3/4 of the prism circumference and do not laterally

connect with each other in most of the normal enamel layer.43

All nanochannels are enlarged in carious enamel, but struc-

tural evidence indicates that prism sheaths are the largest.44,45

Because the penetrating rate is directly proportional to the

radius of the capillaries,12 different rates are expected for

intraprismatic nanochannels and prism sheaths. This is why

we mapped infiltration with two temporal resolutions. The

faster penetration shown at the early stages of infiltration with

high temporal resolution (video 1) is probably related to the

prism sheaths. The lateral propagation of infiltration in the

centre of the body of the lesion can be explained by lateral

connection of adjacent prism sheaths. This lateral connection

has been previously reported with methacrylate replicas of

ground sections of NEC.44 The absence of lateral propagation

indicates that prism sheaths do not connect laterally at the

periphery of the body of the lesion.

The dominance of axial infiltration seen with low temporal

resolution during 1 h is likely related to the combination of

intraprismatic nanochannels and the absence of lateral

connection of prism sheaths. Thoulet’s 1.40, with its faster

penetration, resulted in more air displacement towards the SL

than Thoulet’s 1.47, resulting in a higher air volume retained

in some points infiltrated with Thoulet’s 1.40 after 24 h

(Fig. 5c). This can explain why infiltration of Thoulet’s 1.47

was more efficiently predicted than infiltration of Thoulet’s

1.40, contrary to what was expected from their PCs.

After 1 h, infiltration of both Thoulet’s solutions was not

complete, even in the SL. Considering resin infiltrants with

similar PCs, a greater amount of time than currently used in

clinical procedures (5 min)9 is required to achieve similar

infiltrated volumes. Our results show that the total length filled

by the liquid was the sum of the distance infiltrated by the

faster penetration rate (through prism sheaths) from the centre

of the body of the lesion to its periphery plus the distance

infiltrated by the slower penetration rate (through intrapris-

matic nanochannels) from the innermost point of the body of

the lesion to the SL (videos 1–3). Accurate determination of

penetration length is important, because the penetration rate

is inversely proportional to the length previously filled by the

liquid.12 Scientific research on resin infiltrants in carious

enamel is entirely based on the assumption that the total

length is the distance from the initial penetration to the bottom

of the body of the lesion.9,11,46,47 This distance is referred to as

penetration depth, and its corresponding area is the penetra-

tion area, which can be measured by fluorescence microscopy

and scanning electron microscopy. Polarizing microscopy,

used here, is able to provide infiltration volume and real-time

2D mapping of infiltration, which indicates that the actual

penetration length should be two times the penetration depth

when penetration starts at the enamel surface.

The quantitative volumetric data on the infiltration of

liquids in NEC, reported for the first time, shows that most of

the pore volume was not infiltrated (Fig. 3). This is consistent

with evidence from confocal scanning light microscopy of NEC

infiltrated with a fluorescence-labelled resin infiltrant (during

a r c h i v e s o f o r a l b i o l o g y 6 0 ( 2 0 1 5 ) 8 8 3 – 8 9 3892

a shorter infiltration time, 5 min), which showed incomplete

volume infiltration because of the lack of resin at some layers

along z stacks within infiltrated parts of the lesion.7 Our data

demonstrate that the infiltrated pore volume can be increased

after removal of organic matter.

Determination coefficients above 0.25 are considered

large.24 Thus, ad was a strong predictor of the proportion of

pore volume infiltrated by both Thoulet’s solutions (Fig. 4),

which indicates that ad closely measures the heterogeneity of

the pore volume available for infiltration. The facts that ad was

a predictor similar to the water and air volumes for Thoulet’s

1.47, and similar to the mineral and organic volumes for

Thoulet’s 1.40, suggest that the relative predictive strengths of

the component volumes depend on the PC of the liquid.

Mineral volume was a very poor predictor of infiltration of

Thoulet’s 1.47, but was a strong predictor of Thoulet’s 1.40

(Fig. 4f). This indicates that demineralization of the SL does not

contribute to increase the infiltration of NEC with liquids with

a PC close to that of Thoulet’s 1.47. Removal of organic matter,

which increases ad [Eq. (3)], is a procedure that can more

efficiently increase infiltration, supporting the improved

infiltration of calcium ions into NEC after deproteinization.5

5. Conclusion

We partially confirmed our hypothesis that ad is the best

predictor of the proportion of pore volume infiltrated by

Thoulet’s solutions in dry NEC. ad was a strong predictor for

both liquids, but was similar to the water and air volumes for

Thoulet’s 1.47, and similar to the mineral and organic volumes

for Thoulet’s 1.40. Removal of organic matter is likely the best

procedure to improve infiltration. Real-time 2D mapping

revealed two types of infiltration regarding both penetration

rate and direction of propagation relative to prism paths;

moreover, real-time 2D mapping showed that the length of

penetration is much higher than previously thought. Quantita-

tive volumetric data on component volumes combined with

information on fluid flow mechanics provided unprecedented

information on the nature of the infiltration of liquids in the SL of

NEC, which is important for the transport of materials in carious

enamel, including the improvement of infiltration of NEC with

resins in order to optimize arrestment of lesion progression.

Funding

Meira KRS and Brito CM received master scholarships from

CAPES (Brazilian Ministry of Education).

Competing interests

The authors declare that there are no conflicts of interest.

Ethical approval

Any aspect of the work covered in this manuscript that has

involved either experimental animals or human patients has

been conducted with the ethical approval of all relevant

bodies.

Acknowledgements

The authors greatly appreciate the help provided by Dr

Michael Shribak (Marine Biological Laboratory, USA) regarding

the selection of the appropriate parts of the single polscope

and alignment of those parts for the single polscope setup. To

Dr. Rudolf Oldenbourg (Marine Biological Laboratory, USA) for

providing access to his dual polscope, and to Miss Mai Tran

(Marine Biological Laboratory, USA) for her technical assis-

tance with the dual polscope. The invaluable help of Mr. Yuri

Gonzaga (Master in Computer Science, Federal University of

Paraiba, Brazil), who developed the software used to control

the single polscope and obtain retardance images, is greatly

acknowledged. The first two authors of this study were

financially supported by scholarships from CNPq (Brazilian

Ministry of Science, Technology and Innovation).

Appendix A. Supplementary data

Supplementary data associated with this article can be

found, in the online version, at http://dx.doi.org/10.1016/j.

archoralbio.2015.03.001.

r e f e r e n c e s

1. Cochrane NJ, Cai F, Huq NL, Burrow MF, Reynolds EC. Newapproaches to enhanced remineralization of tooth enamel. JDent Res 2010;89(11):1187–97.

2. Paris S, Hopfenmuller W, Meyer-Lueckel H. Resin infiltrationof caries lesions: an efficacy randomized trial. J Dent Res2010;89(8):823–6.

3. Davila JM, Buonocore MG, Greeley CB, Provenza DV.Adhesive penetration in human artificial and natural whitespots. J Dent Res 1975;54(5):999–1008.

4. Robinson C, Hallsworth AS, Weatherell JA, Kunzel W. Arrestand control of carious lesions: a study based on preliminaryexperiments with resorcinol–formaldehyde resin. J Dent Res1976;55(5):812–8.

5. Meyer-Lueckel H, Paris S, Kielbassa AM. Surface layererosion of natural caries lesions with phosphoric andhydrochloric acid gels in preparation for resin infiltration.Caries Res 2007;41(3):223–30.

6. Meyer-Lueckel H, Paris S. Infiltration of natural caries lesionswith experimental resins differing in penetration coefficientsand ethanol addition. Caries Res 2010;44(4):408–14.

7. Robinson C, Hallsworth AS, Shore RC, Kirkham J. Effect ofsurface zone deproteinisation on the access of mineral ionsinto subsurface carious lesions of human enamel. Caries Res1990;24(4):226–30.

8. Cochrane NJ, Anderson P, Davis GR, Adams GG, Stacey MA,Reynolds EC. An X-ray microtomographic study of naturalwhite-spot enamel lesions. J Dent Res 2012;91(2):185–91.

9. Neuhaus KW, Schlafer S, Lussi A, Nyvad B. Infiltration ofnatural caries lesions in relation to their activity status andacid pretreatment in vitro. Caries Res 2013;47(3):203–10.

a r c h i v e s o f o r a l b i o l o g y 6 0 ( 2 0 1 5 ) 8 8 3 – 8 9 3 893

10. Braga MM, Ekstrand KR, Martignon S, Imparato JCP, RickettsDNJ, Mendes FM. Clinical performance of two visual scoringsystems in detecting and assessing activity status ofocclusal caries in primary teeth. Caries Res 2010;44(3):300–8.

11. Yim HK, Kwon HK, Kim BI. Modification of surface pre-treatment for resin infiltration to mask natural white spotlesions. J Dent 2014;42(5):588–94.

12. Washburn EW. The dynamics of capillary flow. Phys Rev1921;17(3):273–83.

13. Paris S, Meyer-Lueckel H, Colfen H, Kielbassa AM.Penetration coefficients of commercially available andexperimental composites intended to infiltrate enamelcarious lesions. Dent Mater 2007;23(6):742–8.

14. Macena MS, Leite MLA, Gouveia CL, De Lima TAS, AthaydePA, Sousa FB. A comparative study on component volumesfrom outer to inner dental enamel in relation to enameltufts. Arch Oral Biol 2014;59(6):568–77.

15. Barbosa de Sousa F, Dias Soares J, Sampaio Vianna S.Natural enamel caries: a comparative histological study onbiochemical volumes. Caries Res 2013;47(3):183–92.

16. Darling AI, Mortimer KV, Poole DF, Ollis WD. Molecular sievebehaviour of normal and carious human dental enamel.Arch Oral Biol 1961;5(3–4):251–73.

17. Robinson C, Brooks SJ, Kirkham J, Wood SR, Shore RC. Invitro studies of the penetration of adhesive resins intoartificial caries-like lesions. Caries Res 2001;36(2):136–41.

18. Darling AI. Studies of the early lesion of enamel caries. BrDent J 1958;105:119–35.

19. Nyvad B, Machiulskiene V, Baelum V. Reliability of a newcaries diagnostic system differentiating between active andinactive caries lesions. Caries Res 1999;33:252–60.

20. De Medeiros RC, Soares JD, De Sousa FB. Natural enamelcaries in polarized light microscopy: differences inhistopathological features derived from a qualitative versusa quantitative approach to interpret enamel birefringence. JMicrosc 2012;246(2):177–89.

21. Boone JM, Fewell TR, Jennings RJ. Molybdenum, rhodium,and tungsten anode spectral models using interpolatingpolynomials with application to mammography. Med Phys1997;24(12):1863–74.

22. Elliott JC. Structure, crystal chemistry and density of enamelapatites. Ciba Found Symp 1997;205:54–67. [discussion 67–72].

23. Murphy DB. Fundamentals of light microscopy and digitalimaging. New York: Wiley-Liss; 2001.

24. Angmar B, Carlstrom D, Glas JE. Studies on theultrastructure of dental enamel. IV. The mineralization ofnormal human enamel. J Ultrastruct Res 1963;8(1–2):12–23.

25. Sousa FB, Vianna SS, Santos-Magalhaes NS. A new approachfor improving the birefringence analysis of dental enamelmineral content from polarizing microscopy. J Microsc2006;221(2):79–83.

26. Sousa FB, Vianna SS, Santos-Magalhaes NS. Dental enamelbirefringence for a wide mineral content range and fordifferent immersion media’s refractive indexes: animproved mathematical interpretation. J Microsc2009;23(1):69–75.

27. Oldenbourg R. Polarization microscopy with the LC-PolScope. In: Goldman RD, Spector DL, editors. Live cellimaging: a laboratory manual. Cold Spring Harbor: Cold SpringHarbor Laboratory Press; 2005. p. 1–42.

28. Shribak M. Complete polarization state generator with onevariable retarder and its application for fast and sensitive

measuring of two dimensional birefringence distribution. JOpt Soc Am A Opt Image Sci Vis 2011;28(3):410–9.

29. Cohen J. Statistical power analysis for the behavioral sciences.Hillsdale: Lawrence Erlbaum Associates; 1988.

30. Wright JT, Lord V, Robinson C, Shore R. Enamelultrastructure in hypomaturation amelogenesis imperfecta.J Oral Pathol Med 1992;21(9):390–4.

31. Shellis RP, Hallsworth AS, Kirkham J, Robinson C. Organicmaterial and the optical properties of the dark zone in carieslesions of enamel. Eur J Oral Sci 2002;110(5):392–5.

32. Van Der Scheer AT, Smolders CA. Dynamic aspects ofcontact angle measurements on adsorbed protein layers. JColloid Interface Sci 1978;63(1):7–15.

33. Mcmillin CR, Walton AG. A circular dichroism technique forthe study of adsorbed protein structure. J Colloid Interface Sci1974;48(2):345–9.

34. Fejerskov O, Nyvad B, Kidd EAM. Pathology of dental caries. In:Fejerskov O, Kidd EAM, editors. Dental caries: the disease and itsclinical management.. Oxford: Wiley-Blackwell; 2008. p. 19–48.

35. Neuhaus KW, Graf M, Lussi A, Katsaros C. Late infiltration ofpost-orthodontic white spot lesions. J Orofac Orthop2010;71(6):442–7.

36. Rocha Gomes Torres C, Borges AB, Torres LM, Gomes IS, DeOliveira RS. Effect of caries infiltration technique andfluoride therapy on the color masking of white spot lesions.J Dent 2011;39(3):202–7.

37. Holmen L, Thylstrup A, Artun J. Surface changes during thearrest of active enamel carious lesions in vivo. Acta OdontolScand 1987;45(6):383–90.

38. Holmen L, Mejare B, Malmgren B, Thylstrup A. The effect ofregular professional plaque removal on dental cariesin vivo. A polarized light and scanning electron microscopestudy. Caries Res 1988;22(4):250–6.

39. Holmen L, Thylstrup A, Ogaard B, Kragh F. A scanningelectron microscopic study of progressive stages of enamelcaries in vivo. Caries Res 1985;19(4):355–67.

40. Ogaard B, Rolla G, Arends J. Orthodontic appliances andenamel demineralization. Part 1. Lesion development. Am JOrthod Dentofac Orthop 1988;94(1):68–73.

41. Hannig M, Khanafer AK, Hoth-Hannig W, Al-Marrawi F, AcilY. Transmission electron microscopy comparison ofmethods for collecting in situ formed enamel pellicle. ClinOral Invest 2005;9(1):30–7.

42. Zahradnik RT, Moreno EC. Structural features of humandental enamel as revealed by isothermal water vapoursorption. Arch Oral Biol 1975;20(5–6):317–25.

43. Shellis RP, Dibdin GH. Enamel microporosity and itsfunctional implications. In: Teaford MF, Smith MM,Fergunson MWJ, editors. Development, function and evolution ofteeth. New York: Cambridge University Press; 2000. p. 242–51.

44. Shellis RP, Hallsworth AS. The use of scanning electronmicroscopy in studying enamel caries. Scanning Microsc1987;1(3):1109–23.

45. Shellis RP. Scanning electron-microscopic study of solubilityvariations in human enamel and dentine. Arch Oral Biol1996;41(5):473–84.

46. Paris S, Meyer-Lueckel H, Colfen H, Kielbassa AM. Resininfiltration of artificial enamel caries lesions withexperimental resins. Dent Mater J 2007;26(4):582–8.

47. Paris S, Lausch J, Selje T, Dorfer CE, Meyer-Lueckel H.Comparison of sealant and infiltrant penetration into pitand fissure caries lesions in vitro. J Dent 2014;42(4):432–8.