Dependence of in vitro Demineralization of Apatite and Remineralization of Dental Enamel on Fluoride Concentration J.D.B. FEATIlERSTONE, R. GLENA, M. SHARIATI, and C.P. SHIELDS Department of Oral Biology, Eastman Dental Center, 625 Elmwood Avenue, Rochester, New York 14620 The anti-caries activity of fluoride is contributed to in several ways. Two major aspects of fluoride action are (i) the inhibition of demineralization at the crystal surfaces within the tooth, and (ii) the enhancement of subsurface remineralization resulting in arrestment or reversal of caries lesions. Fluoride present in the aqueous phase at the apatite crystal surface may play a determining role in the inhibition of enamel or dentin demineralization. In one part of the present study, the initial dissolution rate of synthetic carbonated-apatite in acetate buffers was measured with fluoride present in the buffer in the 0-2.6 mmolll. (0-50 ppm) range. Inhibition of demineralization was shown to be a logarithmic function of the fluoride concentration in solution. In the second part of the present study, an in vitro pHcycling model was used for determination of the effect on net del remineralization of enamel by treatment solutions containing fluoride in the 0-26 mmolll. (0-500 ppm) range. The net mineral loss was shown to be negatively related to the logarithm of the fluoride concentration. These studies have demonstrated an exponential quantitative relationship between fluoride' concentration and inhibition of apatite demineralization or enhancement of remineralization. The clinical implications are (i) that simply increasing fluoride concentration may not necessarily give increased cariostatic benefit, and (ii) that improving the means ofdelivery ofrelatively low fluoride concentrations for longer times should be more appropriate for enhancing clinical efficacy.
J Dent Res 69(Spec 155):620-625, February, 1990
Introduction. The anti-caries activityof fluoride has been clearly established by clinical and laboratory studies. It is currently considered (Guggenheim, 1984; Leach, 1986; Leach and Edgar, 1983; ten Cate, 1989) that fluoride works in several ways and that two major aspects of fluoride action are (i) the inhibition of demineralization at the crystal surfaces, and (ii) the enhancement of subsurface remineralization resulting in arrestment or reversal of caries lesions. Although it is well-established that fluoride is incorporated into dental apatitecrystalsduring tooth development, the importance of the role of this incorporated fluoride in caries prevention is debatable (Fejerskov et at., 1981; Larsen and Jensen, 1985). Evidence is accumulating which demonstrates the importance of the presence of fluoride during an acid challenge at the tooth surface and in the subsurface. Fluoride present in the aqueous phase at the apatite crystal surface may play a determining role in the inhibition of enamel demineralization. In vitro studies by ten Cate and Duijsters (1983a,b), Arends et at. (1984), and Larsen and Jensen (1985), in which fluoride was incorporated into acid buffers prior to enamel demineralization, have provided evidence for this conclusion. However, in studies that useddental enamel, it is difficult for aspects of diffusion and variation in apatite composition to be isolated in order to establish direct evidence Ptesented at a Joint IADR/ORCA International Symposium on Fluorides: Mechanisms of Action and Recommendations for Use, held March 21-24, 1989, Callaway Gardens Conference Center, Pine Mountain, Georgia These studies were partially supported by NIHlNIDR Grant DE0551O.
of fluoride/apatite crystal interactions. Although several investigators have shown that fluoride inhibits enamel subsurface demineralization, no comprehensive dose-response studies have been reported. One study by ten Cate and Duijsters (1983a,b) did suggest a possible quantitative relationship between fluoride concentration in the acid buffer and the degree of inhibition of demineralization as observed by microradiography. However, this effect was not described mathematically. Numerous other studies have been conducted which have demonstrated the solubility-reducing effect of fluoride on enamel, enamel powder, or hydroxyapatite, but it is not the aim of this paper to review those studies (refer to ten Cate, 1989). Although enamel mineral is comprised of highlysubstituted carbonated-apatite, few studies have utilized carbonated-apatite as a model substrate (see below) rather than hydroxyapatite. Further, a direct quantitative relationship between fluoride concentration and carbonated-apatite dissolution rate has not been established. Studies with synthetic apatites incorporating fluoride within the crystalline structure at concentrations up to 1000 ppm (ug/g) showed that the dissolution rates of these carbonated-apatites in acid were not measurably different from those of non-fluoride-containing carbonated-apatites, and that the carbonate was the principal driving force for the reactivity to acid (Nelson and Featherstone, 1982; Nelson et at., 1983; Featherstone et at., 1983a).These experiments were conducted with a large volume of acid relative to the surface area of the apatite. Therefore, any fluoride dissolving out of the crystal did not have an observable effect at the surface of the apatite. One of the aims of the present study was to utilize synthetic carbonated-apatites to determine quantitatively the effect of fluoride concentration on apatite dissolution rate in an acid buffer of a concentration that may be found in the pores of a caries lesion. These experiments were designed to investigate the effect of fluoride present in solution at or near the crystal surface, without the confusing influence of intercrystalline diffusion and variable composition of the substrate which are inevitable influences in enamel experiments. Many studies have demonstrated in recent years that the enhancement of remineralization by fluoride plays an important role in its caries-preventive activity (Leach, 1986; ten Cate, 1990). Early studies by Koulourides et at. (1961) and Koulourides and Reed (1964) indicated that there was some dependence of remineralization on fluoride concentration. Numerous studies such as those summarized by Leach and Edgar (1983), Guggenheim (1984), and Leach (1986) have reported in vitro studies repeatedly confirming the role of fluoride in remineralization of dental enamel, at least in vitro. Studies involving cycling models where enamel was cycled through demineralization and remineralization have been carried out by several authors (for example, ten Cate and Duijsters, 1982; Featherstone et al., 1986), and these are reviewed by ten Cate in the present issue (1990). The cycling model developed in our laboratories (Featherstone et al., 1986; White and Featherstone, 19$7)was designed so that it paralleled results found in vivo around orthodontic appliances (O'Reilly and Featherstone, 1987). Therefore, the advantage of that model is that by design the degree of net mineral change in vitro was in an
620 Downloaded from jdr.sagepub.com at UCSF LIBRARY & CKM on February 25, 2015 For personal use only. No other uses without permission. © 1990 International & American Associations for Dental Research
DE/REMINERALIZATION AND FLUORIDE CONCENTRATION
Vol. 69 Special Issue
amount similar to that found in vivo for the same fluoride treatments. The second aim of the present study was to use this in vitro cycling model to investigate the relationship of net mineral change to the concentration of fluoride in the treatment solution used between de- and remineralization. The overall aims of the present studies were to quantify further the effects of fluoride on de- and remineralization in model systems and to provide information that can subsequently be extrapolated to make predictions about improving caries prevention methods and/or delivery of fluoride in humans.
Materials and methods. Apatite .synthesis.-Synthetic carbonated-apatite containing approximately 3% (by weight) carbonate was synthesized by precipitation as previously reported (Featherstone et a!., 1983a). This is a modification of the method of LeGeros et a!. (1967). The apatite was precipitated by the slow, drop-wise, addition, at a constant rate over 2 h, of 200 mL of 0.18 mol/L sodium phosphate containing 10 mmol/L NaHC0 3 , to 200 mL of 0.30 mol/L solution of Ca(N03hAH20 that was maintained at 92 ± 1°C at a constant pH of 9.5 by use of ammonium hydroxide. The temperature was then raised to 100°C, and the mixture was stirred under reflux for 2 h. The precipitated apatite was filtered, washed with hot de-ionized water, and dried overnight at 120°C. The final powdered apatite was characterized by x-ray diffraction, infrared spectroscopy for carbonate content (Featherstone et a!., 1984), and chemical analysis for calcium, phosphate, sodium, and fluoride, as previously described (Featherstone et a!., 1983a). Carbonated-apatite was chosen as a model material for investigation of the dissolution rate, since it more closely approximates the composition of dental enamel than does pure hydroxyapatite. Studies by several investigators during the last 20 years have demonstrated that the carbonate component of synthetic apatites, enamel apatite, and other biological apatites dramatically alters the reactivity of these materials (for example, LeGeros, 1967; Nelson et a!., 1983). Recent studies by Budz et a!. (1987) have shown that carbonated-apatite more closely mimics the behavior of enamel apatite than does hydroxyapatite. Consequently, in the present study we synthesized the carbonated-apatite to contain calcium, phosphate, and sodium at levels similar to those found in dental enamel and also to contain 3% carbonate which likewise approximates that found in dental enamel (Weatherell et al., 1968). Acid reactivity-dissolution measurements. - A specially constructed dissolution apparatus (Featherstone et al., 1983a) was used to measure the acid reactivity of the apatite in acetate buffer (see below) and acetate buffers containing fluoride in a range of concentrations. The apparatus utilizes pressed pellets of apatite. The apatite was ground in an agate mortar and pestle, screened through a 60-mesh nylon screen, blended with 10% polyethylene powder (BDH polyethylene for spectroscopy), slurried in acetone, mixed until uniform, dried, pressed into pellets under vacuum by means of a Perkin-Elmer IR die, and cured at 120°C overnight. This formed an essentially nonporous pellet of uniform surface area and largely overcame the effect on dissolution rate of variation in apatite crystal size (Featherstone et al., 1983a; Nelson et a!., 1989). Synthetic carbonated-apatite containing 3% carbonate was used for all experiments as a standard background material for comparison of the effect of fluoride at different concentrations in the dissolution buffer. Each pellet was embedded in methyl methacrylate, with only the reacting surface left exposed for testing in
621
the dissolution apparatus. Before use and between dissolutions, each pellet was abraded by 600-grade silicon carbide, followed by a slurry of 0.05-J.Lm alumina (Buehler) in double-de-ionized water (DDW) and cleaning by sonication in DDW. This produced reproducible and uniform dissolution profiles (Featherstone et a!., 1983a). In the present study, we were aiming to approximate conditions which might be found inside the tooth during caries lesion formation, and therefore used 0.01 mol/L acetate buffer (200 mL per pellet) at pH 4.5 and 37°C as the acid dissolution medium. This is discussed in more detail in our previous publications (Featherstone et a!., 1983a; Nelson et a!., 1989). Constant stirring at 300 rpm was routinely used, and 2.5-mL aliquots were taken at 10-minute intervals for calcium analysis by atomic absorption spectroscopy. Calcium loss was found to be linear with time for at least the first 50 min, and initial dissolution rates were calculated by linear regression from the data over this period. All experiments were carried out at least in quadruplicate. Fluoride solutions. -Fluoride was added to the 0.01 mol/L acetate buffer at approximately 0, 1, 3, 5, 8, 20, and 50 mg! L for the above dissolution experiments. These concentrations were accurately determined by chemical analysis. The nonfluoride buffer was found to contain 0.02 mg!L fluoride from the reagents. Prior to the dissolution experiments, each acetate buffer was added to the reaction vessel and equilibrated to 37°C. Cycling demineralizationlremineralization test system. -The cycling enamel demineralization/remineralization model of ten Cate and Duijsters (1982) was modified as described previously (Featherstone et a!., 1986) to produce lesions approximately 75 J.Lm deep with a mineral loss of 10-15% in the presence of daily sodium fluoride toothpaste treatments over the course of the test-results similar to those found around orthodontic appliances in vivo after one month (O'Reilly and Featherstone, 1987). Each test cell consisted of ten teeth (human enamel crowns) with two exposed windows (one upper window and one lower window). The test regimen in each 24hour period proceeded as follows: (1) Teeth underwent 6 h of demineralization at 37°C in a buffer containing 2.0 mmol/L calcium, 2.0 mmol/L phosphate, and 0.075 mol/L acetate at pH 4.3. Each tooth was immersed individually in 40 mL of solution. (2) The teeth were then removed from the solution, rinsed in de-ionized water, and shaken on an orbital shaker for 5 min while immersed in one of the test fluoride solutions (see below). One group with no fluoride in the test solution was the remineralization/demineralization-only control. (3) After the five-minute immersion period, the teeth were rinsed with de-ionized water, and immersed individually for 17 h at 37°C in 20 mL of a mineralizing solution overnight to simulate the remineralizing stage of the caries process. The mineralizing solution was supersaturated with calcium phosphate (calcium = 1.5 mrnol/L, phosphate = 0.9 mmol/ L), with potassium chloride at 150 mmol/L, and cacodylate buffer to pH 7.0 (20 mmoIlL). This solution approximates the saturation degree of apatitic minerals found in saliva and is similar to that utilized by ten Cate and Duijsters (1982). The test solutions used consisted of sodium fluoride in deionized water at concentrations of 0.05, 0.53, 2.63, 13.2, and 26.3 mmol/L (1, 10, 50, 250, and 500 ppm) of fluoride. This concentration range was used to bracket the commonly used 0.05% NaF (220 ppm F) mouthrinse concentration. The teeth were prepared as described in detail previously (Featherstone et a!., 1986, 1988). The upper windows were used as the principal test site. However, because cervically situated enamel often behaves differently, the lower windows were assessed in the same way, but as a separate group.
Downloaded from jdr.sagepub.com at UCSF LIBRARY & CKM on February 25, 2015 For personal use only. No other uses without permission. © 1990 International & American Associations for Dental Research
J Dent Res February 1990
FEATHERSTONE et al.
622
Carb-Ap In
Dissolution
Presence
of
Fluoride Dose Response Experiment
F
Normalized Data
100
6.0
.... ....
10
...
~
__ 50.0 ppm F ~
e 0
U
CIS
~
o 4.0 - - - - - -Non-Carb Apatite
i
10
A
rr'"' 40
tI:
/
"0
>
CIS
a:
__ CoI\traI 0.0 ppm F
/
1//
•e
G)
__ 250.0 ppm F __ 500.0 ppm F
e
~ 5.0
-
....
~
-s » ~/
s:
-&- 1.0 ppm F __ 10.0 ppm F
v
20
e 3.0 0
a
:::l
0
I/) I/)
h--..
2.0
c
CIS
a
100
150
200
250
Depth (mIcrons)
Fig. 2-Volume percent mineral vs. depth profiles for each group in the cycling de/remineralization study. Each point represents the mean of the values at each depth for each sample in the group.
1.0
'c
O+-------''''------'l---''--~
o
10
20
30
40
50
F conc in buffer • mgll Fig. I-Initial dissolution rate (mg/L/h calcium) vs. fluoride concentration (mg/L) in the acetate buffer for dissolution of 3% (by wt) carbonated-apatite. The horizontal dotted line is the initial dissolution rate for a non-carbonated hydroxyapatite disc (4.00 ± 0.2 SD). Error bars are standard deviations.
Assessment of demineralization/remineralization. -At the conclusion of the immersions (14 days of cycling), the teeth were rinsed in de-ionized water, sectioned longitudinally through the lesions, and embedded in epoxy resin with the cut face exposed, as previously described (Featherstone et al., 1983b, 1986; White and Featherstone, 1987). After serial polishing, each lesion was assessed by cross-sectional microhardness evaluation according to the techniques routinely used in our laboratory (Featherstone et al., 1986), starting at 25 urn from the anatomical tooth surface and progressing at 25-/-Lm intervals across the lesion and into the underlying enamel. The set of data representing each artificial caries lesion was curvefitted by means of a Simpson approximation (White and Featherstone, 1987) from 25 urn inward, and the area under the lesion tracing was calculated (in units of volume percent mineral x urn), and subtracted from the normal enamel value to give the parameter b"Z, being the relative mineral loss for each lesion, familiar from quantitative microradiography profiles. The 8.Z values for each lesion in each group were combined to give means and standard deviations for b"Z for each of the treatment solutions.
Results. Apatite studies. - The initial dissolution rates of the carbonated-apatites in the presence of the various acidic fluoride solutions are shown in Fig. 1. The horizontal dotted line is the value for the dissolution of a non-carbonated-apatite measured in a previous study (Glena et al., 1984). Fluoride at 0.1 mmol/
L (2 ppm) reduced the initial dissolution rate by approximately 40% compared with the non-fluoride buffer (1 urnol/L fluoride by chemical analysis), and higher concentrations reduced the dissolution rate further. Linear regression analysis of the initial dissolution rate data vs. the logarithm of the fluoride concentration (expressed as mg/L) produced a straight-line fit with the equation: Initial dissolution rate = -1.21 log [F]
+ 3.69
(1)
The correlation coefficient r = 0.997 showed an excellent fit of the data. Cycling demineralization/remineralization of enamel. - The az values for upper and lower windows, respectively, are given in the Table, and the mean profiles for the upper window are shown in Fig. 2. Linear regression analysis of the az values vs. the logarithm of the fluoride concentration (expressed as mgIL) produced straight-line fits as follows. The upper window equation was:
az = - 680 log [F] +
1843
(2)
The lower window equation was: az = - 989 log [F] + 2790 (3) The correlation coefficients for upper and lower windows were 0.984 and 0.950, respectively, indicating good linear fits of the data. The calculated values of F concentration for az = o were 513 mg/L (27 mmollL) and 662 mg/L (35 rnmol/L), respectively, for upper and lower windows.
Discussion. In the present study, the carbonated-apatite dissolution experiments were aimed to model one part of the caries process, namely, the dissolution at the apatite crystal surface as a result of proton action, and the interference of F- in the acid buffer with this action. Another important distinction from almost all previously reported studies was that a synthetic carbonatedapatite was utilized with a carbonate content similar to that of dental enamel, rather than hydroxyapatite or powdered enamel. Nelson et al. (1983) previously reported studies with carbonated-apatite where 1 mg/L (53 umol/L) fluoride in the acid buffer reduced the dissolution rate by approximately 30%,
Downloaded from jdr.sagepub.com at UCSF LIBRARY & CKM on February 25, 2015 For personal use only. No other uses without permission. © 1990 International & American Associations for Dental Research
Vol. 69 Special Issue
623
DE/REMINERALIZATION AND FLUORIDE CONCENTRATION
whereas 1000 ppm (ug/g) F incorporated in the carbonatedapatite crystals had no measurable effect on dissolution rate. In the present study, we have shown that concentrations of 1 mg/L (53 urnol/L) of F or higher in the acid buffer reduced the initial dissolution rate of the carbonated-apatite below that of hydroxyapatite, and that this reduction in dissolution rate increased in proportion to the logarithm of the fluoride concentration in the buffer. One advantage of this model system is that the buffer volume/ apatite surface ratio is high (approximately 105 Llm2) , effectively creating an 'infinite sink'. The observed reductions in dissolution rates weretherefore a result of either fluoride in solution, fluoride adsorbed to the carbonated-apatite surface, or of fluoride effectively promoting the formation of a very thin coating of a low solubility phase suchas fluorapatite. Theycould nothave resulted from fluoride incorporated in the original apatite or from mass action effects due to rapid increases in calcium or phosphate concentrations in the buffer solution as dissolution proceeded. The maximum concentration of calcium or phosphate present in the acidic fluoride buffer bulk solution after 1 h of dissolution was0.1 mmol/L, Thismodel therefore provides a means of direct comparison between the effects of various fluoride concentrations in acidic buffers in contact withapatite crystal surfaces. Although diffusion to and from the crystal surface and even within the pellets will play a part in the empirically measured dissolution rates in these experiments, the stirring speed and geometry were the samefor each experiment, and consequently the direct comparison of initial dissolution rates from onefluoride concentration to the other gives a comparative measure of the relative effects of fluoride concentration in the acid buffer. A recent study by Wong et al. (1987) using pelletized hydroxyapatite, fluorapatite, or hydroxyapatite with surface-adsorbed fluoride showed that incorporated fluoride only had an effect when it was present at concentrations equivalent to those in pure fluorapatite (38,000 ppm). In other words, crystal surfaces which behaved as though they were essentially fluorapatite masked the reactivity of the underlying apatite crystals of differing compositions. Chen and Nancollas (1986) used powdered bovine enamel and powdered human enamel in a constant composition demineralization model with a partially saturated buffer at pH 4.5 and concluded that 'the dissolution reactions suggest a surface dislocation mechanism, and the presence of fluoride ion in the buffer markedly retarded the reaction'. In their experiments, a fluoride level of only 0.5 mg/L (26 umol/L) reduced the rate of dissolution of human enamel ten-fold. In contrast, they reported that the dissolution of pure hydroxyapatite was best interpreted in terms of a polynucleation process. In the present study, the inverse dependence of carbonated-apatite initial dissolution rate on the logarithm of the fluoride concentration in the surrounding aqueous acid suggests a direct effect of surface-adsorbed fluoride ions on dissolution sites at the crystal surface. It is expected thatthesame mechanism of inhibition willoccur at the surfaces of carbonatedapatite enamel crystals. The natural extension of the above information in relation to caries progression or inhibition is as follows. The plaque fluid and the fluid among the crystals will certainly not behave as an "infinite sink". The aqueous fluid among the crystals in a forming caries lesion will most likely have much higher concentrations of calcium and phosphate present, and be saturated with respect to the adjacent apatite crystal surface phases until acid diffuses in from the external plaque during a caries challenge. As the local pH falls, mineral components will dissolve-in particular calcium phosphate, carbonate, and fluoride-until saturation is again reached. Fluoride from within the crystal which appears in solution in this way will then inhibit further dissolution. More importantly, fluoride already present in this intercrystalline aqueous phase, or that diffused
in from the plaque at the same time as the acid, will have an inhibitory effect on dissolution in a manner similar to that observed in the experiments reported here. Further, fluoride in this aqueous phase between the crystals will contribute, together with calcium and phosphate, to the degree of saturation with respect to FAP, thereby inhibiting dissolution by thermodynamic considerations. This is best illustrated in the phase diagram of Fig. 3. Each line represents the boundary between solid and solution as calculated from published solubility product values for each mineral. Enamel apatite does not have a definite single solubility product (Ksp) because of its variablecomposition. The moresoluble mineral, however, can be represented by a line drawn based on pKsp = 104 reported by Patel and Brown (1975). Lines for FAP equilibria are shown for solution F concentrations of 0.1 and 1.0 ppm (5.3 and 53 umol/L). It can be seen that even in the presence of these low fluoride concentrations a crystal surface behaving like FAP would have an effective solubility several orders of magnitude lower than the original enamel apatite crystalat the same pH-for example, at pH = 5.0. Therefore, if enamel apatitecrystalsurfaces come in contactwith fluoride from topical sources and the fluoride adsorbs together with calcium and phosphate, producing a FAP-like surface, such as thatformed and tested in the experiments of Wong et al. (1987), this would at least partly explain the role of fluoride as an inhibitor of enamel demineralization when present during a caries challenge. In the present study, the pH-cycling de/remineralization model that we used was designed to simulate results found in vivo at highly susceptible sites around orthodontic brackets (Featherstone et al., 1986; O'Reilly and Featherstone, 1987). This model combines periods of acidchallenge and periods of lesion repair. It purposely eliminates salivary components exceptcalcium, phosphate, and fluoride and allows for daily, or twicedaily, fluoride product treatments. The model can also be used to study the effects of salivary organic components (Cooper and Featherstone, 1986), the effects of fluoride at low concentrations in the mineralizing solution (Featherstone et al., 1986), and root caries progression (Featherstone et al., 1989). Although care must be taken in the interpretation of results, this model has been used successfully to screen fluoride products as part of the American DentalAssociation (ADA) profile system (ADA Council on Dental Therapeutics, 1988), to predict the lack of interference of pyrophosphate with remineralization in calculus-control dentifrices (Featherstone et al., 1988). This latter conclusion has been supported by clinical trials (Lu et al., 1985; Triol et al., 1988). In the presentstudy, the net effect of daily fluoride solution treatments on the de/remineralization process was measured to establish whether the net mineral change was dependent on fluoride concentration in the treatment solutions. The inverse TABLE RELATIVE MINERAL LOSS, IlZ, IN pH CYCLING EXPERIMENT Fluoride Concentration (mg/L) in Treatment Solution
o 1 10 50 250 500
Upper window (SE")
IlZ" 3132 1845 836 578 470 51
(387) (366) (339) (293) (247) (234)
Lower window mean IlZ (SP)
4257 2616 980 1060 895 208
(605) (466) (296) (399) (123) (75)
""IlZ = mean relative mineral loss, volume percent x urn, for each test group. "SE = standard error of the mean.
Downloaded from jdr.sagepub.com at UCSF LIBRARY & CKM on February 25, 2015 For personal use only. No other uses without permission. © 1990 International & American Associations for Dental Research
FEATHERSTONE et ai.
624
J Dent Res February 1990 ,DO
2
\ \ \
3
80
\
\
SUPERSATURATED
c: 0
't
\
4
~
\
60
e,
C
5 """ ~
..•• u
X
4D
a,
DCPD
J Dent Res 69(Spec 155):620-625, February, 1990
Introduction. The anti-caries activityof fluoride has been clearly established by clinical and laboratory studies. It is currently considered (Guggenheim, 1984; Leach, 1986; Leach and Edgar, 1983; ten Cate, 1989) that fluoride works in several ways and that two major aspects of fluoride action are (i) the inhibition of demineralization at the crystal surfaces, and (ii) the enhancement of subsurface remineralization resulting in arrestment or reversal of caries lesions. Although it is well-established that fluoride is incorporated into dental apatitecrystalsduring tooth development, the importance of the role of this incorporated fluoride in caries prevention is debatable (Fejerskov et at., 1981; Larsen and Jensen, 1985). Evidence is accumulating which demonstrates the importance of the presence of fluoride during an acid challenge at the tooth surface and in the subsurface. Fluoride present in the aqueous phase at the apatite crystal surface may play a determining role in the inhibition of enamel demineralization. In vitro studies by ten Cate and Duijsters (1983a,b), Arends et at. (1984), and Larsen and Jensen (1985), in which fluoride was incorporated into acid buffers prior to enamel demineralization, have provided evidence for this conclusion. However, in studies that useddental enamel, it is difficult for aspects of diffusion and variation in apatite composition to be isolated in order to establish direct evidence Ptesented at a Joint IADR/ORCA International Symposium on Fluorides: Mechanisms of Action and Recommendations for Use, held March 21-24, 1989, Callaway Gardens Conference Center, Pine Mountain, Georgia These studies were partially supported by NIHlNIDR Grant DE0551O.
of fluoride/apatite crystal interactions. Although several investigators have shown that fluoride inhibits enamel subsurface demineralization, no comprehensive dose-response studies have been reported. One study by ten Cate and Duijsters (1983a,b) did suggest a possible quantitative relationship between fluoride concentration in the acid buffer and the degree of inhibition of demineralization as observed by microradiography. However, this effect was not described mathematically. Numerous other studies have been conducted which have demonstrated the solubility-reducing effect of fluoride on enamel, enamel powder, or hydroxyapatite, but it is not the aim of this paper to review those studies (refer to ten Cate, 1989). Although enamel mineral is comprised of highlysubstituted carbonated-apatite, few studies have utilized carbonated-apatite as a model substrate (see below) rather than hydroxyapatite. Further, a direct quantitative relationship between fluoride concentration and carbonated-apatite dissolution rate has not been established. Studies with synthetic apatites incorporating fluoride within the crystalline structure at concentrations up to 1000 ppm (ug/g) showed that the dissolution rates of these carbonated-apatites in acid were not measurably different from those of non-fluoride-containing carbonated-apatites, and that the carbonate was the principal driving force for the reactivity to acid (Nelson and Featherstone, 1982; Nelson et at., 1983; Featherstone et at., 1983a).These experiments were conducted with a large volume of acid relative to the surface area of the apatite. Therefore, any fluoride dissolving out of the crystal did not have an observable effect at the surface of the apatite. One of the aims of the present study was to utilize synthetic carbonated-apatites to determine quantitatively the effect of fluoride concentration on apatite dissolution rate in an acid buffer of a concentration that may be found in the pores of a caries lesion. These experiments were designed to investigate the effect of fluoride present in solution at or near the crystal surface, without the confusing influence of intercrystalline diffusion and variable composition of the substrate which are inevitable influences in enamel experiments. Many studies have demonstrated in recent years that the enhancement of remineralization by fluoride plays an important role in its caries-preventive activity (Leach, 1986; ten Cate, 1990). Early studies by Koulourides et at. (1961) and Koulourides and Reed (1964) indicated that there was some dependence of remineralization on fluoride concentration. Numerous studies such as those summarized by Leach and Edgar (1983), Guggenheim (1984), and Leach (1986) have reported in vitro studies repeatedly confirming the role of fluoride in remineralization of dental enamel, at least in vitro. Studies involving cycling models where enamel was cycled through demineralization and remineralization have been carried out by several authors (for example, ten Cate and Duijsters, 1982; Featherstone et al., 1986), and these are reviewed by ten Cate in the present issue (1990). The cycling model developed in our laboratories (Featherstone et al., 1986; White and Featherstone, 19$7)was designed so that it paralleled results found in vivo around orthodontic appliances (O'Reilly and Featherstone, 1987). Therefore, the advantage of that model is that by design the degree of net mineral change in vitro was in an
620 Downloaded from jdr.sagepub.com at UCSF LIBRARY & CKM on February 25, 2015 For personal use only. No other uses without permission. © 1990 International & American Associations for Dental Research
DE/REMINERALIZATION AND FLUORIDE CONCENTRATION
Vol. 69 Special Issue
amount similar to that found in vivo for the same fluoride treatments. The second aim of the present study was to use this in vitro cycling model to investigate the relationship of net mineral change to the concentration of fluoride in the treatment solution used between de- and remineralization. The overall aims of the present studies were to quantify further the effects of fluoride on de- and remineralization in model systems and to provide information that can subsequently be extrapolated to make predictions about improving caries prevention methods and/or delivery of fluoride in humans.
Materials and methods. Apatite .synthesis.-Synthetic carbonated-apatite containing approximately 3% (by weight) carbonate was synthesized by precipitation as previously reported (Featherstone et a!., 1983a). This is a modification of the method of LeGeros et a!. (1967). The apatite was precipitated by the slow, drop-wise, addition, at a constant rate over 2 h, of 200 mL of 0.18 mol/L sodium phosphate containing 10 mmol/L NaHC0 3 , to 200 mL of 0.30 mol/L solution of Ca(N03hAH20 that was maintained at 92 ± 1°C at a constant pH of 9.5 by use of ammonium hydroxide. The temperature was then raised to 100°C, and the mixture was stirred under reflux for 2 h. The precipitated apatite was filtered, washed with hot de-ionized water, and dried overnight at 120°C. The final powdered apatite was characterized by x-ray diffraction, infrared spectroscopy for carbonate content (Featherstone et a!., 1984), and chemical analysis for calcium, phosphate, sodium, and fluoride, as previously described (Featherstone et a!., 1983a). Carbonated-apatite was chosen as a model material for investigation of the dissolution rate, since it more closely approximates the composition of dental enamel than does pure hydroxyapatite. Studies by several investigators during the last 20 years have demonstrated that the carbonate component of synthetic apatites, enamel apatite, and other biological apatites dramatically alters the reactivity of these materials (for example, LeGeros, 1967; Nelson et a!., 1983). Recent studies by Budz et a!. (1987) have shown that carbonated-apatite more closely mimics the behavior of enamel apatite than does hydroxyapatite. Consequently, in the present study we synthesized the carbonated-apatite to contain calcium, phosphate, and sodium at levels similar to those found in dental enamel and also to contain 3% carbonate which likewise approximates that found in dental enamel (Weatherell et al., 1968). Acid reactivity-dissolution measurements. - A specially constructed dissolution apparatus (Featherstone et al., 1983a) was used to measure the acid reactivity of the apatite in acetate buffer (see below) and acetate buffers containing fluoride in a range of concentrations. The apparatus utilizes pressed pellets of apatite. The apatite was ground in an agate mortar and pestle, screened through a 60-mesh nylon screen, blended with 10% polyethylene powder (BDH polyethylene for spectroscopy), slurried in acetone, mixed until uniform, dried, pressed into pellets under vacuum by means of a Perkin-Elmer IR die, and cured at 120°C overnight. This formed an essentially nonporous pellet of uniform surface area and largely overcame the effect on dissolution rate of variation in apatite crystal size (Featherstone et al., 1983a; Nelson et a!., 1989). Synthetic carbonated-apatite containing 3% carbonate was used for all experiments as a standard background material for comparison of the effect of fluoride at different concentrations in the dissolution buffer. Each pellet was embedded in methyl methacrylate, with only the reacting surface left exposed for testing in
621
the dissolution apparatus. Before use and between dissolutions, each pellet was abraded by 600-grade silicon carbide, followed by a slurry of 0.05-J.Lm alumina (Buehler) in double-de-ionized water (DDW) and cleaning by sonication in DDW. This produced reproducible and uniform dissolution profiles (Featherstone et a!., 1983a). In the present study, we were aiming to approximate conditions which might be found inside the tooth during caries lesion formation, and therefore used 0.01 mol/L acetate buffer (200 mL per pellet) at pH 4.5 and 37°C as the acid dissolution medium. This is discussed in more detail in our previous publications (Featherstone et a!., 1983a; Nelson et a!., 1989). Constant stirring at 300 rpm was routinely used, and 2.5-mL aliquots were taken at 10-minute intervals for calcium analysis by atomic absorption spectroscopy. Calcium loss was found to be linear with time for at least the first 50 min, and initial dissolution rates were calculated by linear regression from the data over this period. All experiments were carried out at least in quadruplicate. Fluoride solutions. -Fluoride was added to the 0.01 mol/L acetate buffer at approximately 0, 1, 3, 5, 8, 20, and 50 mg! L for the above dissolution experiments. These concentrations were accurately determined by chemical analysis. The nonfluoride buffer was found to contain 0.02 mg!L fluoride from the reagents. Prior to the dissolution experiments, each acetate buffer was added to the reaction vessel and equilibrated to 37°C. Cycling demineralizationlremineralization test system. -The cycling enamel demineralization/remineralization model of ten Cate and Duijsters (1982) was modified as described previously (Featherstone et a!., 1986) to produce lesions approximately 75 J.Lm deep with a mineral loss of 10-15% in the presence of daily sodium fluoride toothpaste treatments over the course of the test-results similar to those found around orthodontic appliances in vivo after one month (O'Reilly and Featherstone, 1987). Each test cell consisted of ten teeth (human enamel crowns) with two exposed windows (one upper window and one lower window). The test regimen in each 24hour period proceeded as follows: (1) Teeth underwent 6 h of demineralization at 37°C in a buffer containing 2.0 mmol/L calcium, 2.0 mmol/L phosphate, and 0.075 mol/L acetate at pH 4.3. Each tooth was immersed individually in 40 mL of solution. (2) The teeth were then removed from the solution, rinsed in de-ionized water, and shaken on an orbital shaker for 5 min while immersed in one of the test fluoride solutions (see below). One group with no fluoride in the test solution was the remineralization/demineralization-only control. (3) After the five-minute immersion period, the teeth were rinsed with de-ionized water, and immersed individually for 17 h at 37°C in 20 mL of a mineralizing solution overnight to simulate the remineralizing stage of the caries process. The mineralizing solution was supersaturated with calcium phosphate (calcium = 1.5 mrnol/L, phosphate = 0.9 mmol/ L), with potassium chloride at 150 mmol/L, and cacodylate buffer to pH 7.0 (20 mmoIlL). This solution approximates the saturation degree of apatitic minerals found in saliva and is similar to that utilized by ten Cate and Duijsters (1982). The test solutions used consisted of sodium fluoride in deionized water at concentrations of 0.05, 0.53, 2.63, 13.2, and 26.3 mmol/L (1, 10, 50, 250, and 500 ppm) of fluoride. This concentration range was used to bracket the commonly used 0.05% NaF (220 ppm F) mouthrinse concentration. The teeth were prepared as described in detail previously (Featherstone et a!., 1986, 1988). The upper windows were used as the principal test site. However, because cervically situated enamel often behaves differently, the lower windows were assessed in the same way, but as a separate group.
Downloaded from jdr.sagepub.com at UCSF LIBRARY & CKM on February 25, 2015 For personal use only. No other uses without permission. © 1990 International & American Associations for Dental Research
J Dent Res February 1990
FEATHERSTONE et al.
622
Carb-Ap In
Dissolution
Presence
of
Fluoride Dose Response Experiment
F
Normalized Data
100
6.0
.... ....
10
...
~
__ 50.0 ppm F ~
e 0
U
CIS
~
o 4.0 - - - - - -Non-Carb Apatite
i
10
A
rr'"' 40
tI:
/
"0
>
CIS
a:
__ CoI\traI 0.0 ppm F
/
1//
•e
G)
__ 250.0 ppm F __ 500.0 ppm F
e
~ 5.0
-
....
~
-s » ~/
s:
-&- 1.0 ppm F __ 10.0 ppm F
v
20
e 3.0 0
a
:::l
0
I/) I/)
h--..
2.0
c
CIS
a
100
150
200
250
Depth (mIcrons)
Fig. 2-Volume percent mineral vs. depth profiles for each group in the cycling de/remineralization study. Each point represents the mean of the values at each depth for each sample in the group.
1.0
'c
O+-------''''------'l---''--~
o
10
20
30
40
50
F conc in buffer • mgll Fig. I-Initial dissolution rate (mg/L/h calcium) vs. fluoride concentration (mg/L) in the acetate buffer for dissolution of 3% (by wt) carbonated-apatite. The horizontal dotted line is the initial dissolution rate for a non-carbonated hydroxyapatite disc (4.00 ± 0.2 SD). Error bars are standard deviations.
Assessment of demineralization/remineralization. -At the conclusion of the immersions (14 days of cycling), the teeth were rinsed in de-ionized water, sectioned longitudinally through the lesions, and embedded in epoxy resin with the cut face exposed, as previously described (Featherstone et al., 1983b, 1986; White and Featherstone, 1987). After serial polishing, each lesion was assessed by cross-sectional microhardness evaluation according to the techniques routinely used in our laboratory (Featherstone et al., 1986), starting at 25 urn from the anatomical tooth surface and progressing at 25-/-Lm intervals across the lesion and into the underlying enamel. The set of data representing each artificial caries lesion was curvefitted by means of a Simpson approximation (White and Featherstone, 1987) from 25 urn inward, and the area under the lesion tracing was calculated (in units of volume percent mineral x urn), and subtracted from the normal enamel value to give the parameter b"Z, being the relative mineral loss for each lesion, familiar from quantitative microradiography profiles. The 8.Z values for each lesion in each group were combined to give means and standard deviations for b"Z for each of the treatment solutions.
Results. Apatite studies. - The initial dissolution rates of the carbonated-apatites in the presence of the various acidic fluoride solutions are shown in Fig. 1. The horizontal dotted line is the value for the dissolution of a non-carbonated-apatite measured in a previous study (Glena et al., 1984). Fluoride at 0.1 mmol/
L (2 ppm) reduced the initial dissolution rate by approximately 40% compared with the non-fluoride buffer (1 urnol/L fluoride by chemical analysis), and higher concentrations reduced the dissolution rate further. Linear regression analysis of the initial dissolution rate data vs. the logarithm of the fluoride concentration (expressed as mg/L) produced a straight-line fit with the equation: Initial dissolution rate = -1.21 log [F]
+ 3.69
(1)
The correlation coefficient r = 0.997 showed an excellent fit of the data. Cycling demineralization/remineralization of enamel. - The az values for upper and lower windows, respectively, are given in the Table, and the mean profiles for the upper window are shown in Fig. 2. Linear regression analysis of the az values vs. the logarithm of the fluoride concentration (expressed as mgIL) produced straight-line fits as follows. The upper window equation was:
az = - 680 log [F] +
1843
(2)
The lower window equation was: az = - 989 log [F] + 2790 (3) The correlation coefficients for upper and lower windows were 0.984 and 0.950, respectively, indicating good linear fits of the data. The calculated values of F concentration for az = o were 513 mg/L (27 mmollL) and 662 mg/L (35 rnmol/L), respectively, for upper and lower windows.
Discussion. In the present study, the carbonated-apatite dissolution experiments were aimed to model one part of the caries process, namely, the dissolution at the apatite crystal surface as a result of proton action, and the interference of F- in the acid buffer with this action. Another important distinction from almost all previously reported studies was that a synthetic carbonatedapatite was utilized with a carbonate content similar to that of dental enamel, rather than hydroxyapatite or powdered enamel. Nelson et al. (1983) previously reported studies with carbonated-apatite where 1 mg/L (53 umol/L) fluoride in the acid buffer reduced the dissolution rate by approximately 30%,
Downloaded from jdr.sagepub.com at UCSF LIBRARY & CKM on February 25, 2015 For personal use only. No other uses without permission. © 1990 International & American Associations for Dental Research
Vol. 69 Special Issue
623
DE/REMINERALIZATION AND FLUORIDE CONCENTRATION
whereas 1000 ppm (ug/g) F incorporated in the carbonatedapatite crystals had no measurable effect on dissolution rate. In the present study, we have shown that concentrations of 1 mg/L (53 urnol/L) of F or higher in the acid buffer reduced the initial dissolution rate of the carbonated-apatite below that of hydroxyapatite, and that this reduction in dissolution rate increased in proportion to the logarithm of the fluoride concentration in the buffer. One advantage of this model system is that the buffer volume/ apatite surface ratio is high (approximately 105 Llm2) , effectively creating an 'infinite sink'. The observed reductions in dissolution rates weretherefore a result of either fluoride in solution, fluoride adsorbed to the carbonated-apatite surface, or of fluoride effectively promoting the formation of a very thin coating of a low solubility phase suchas fluorapatite. Theycould nothave resulted from fluoride incorporated in the original apatite or from mass action effects due to rapid increases in calcium or phosphate concentrations in the buffer solution as dissolution proceeded. The maximum concentration of calcium or phosphate present in the acidic fluoride buffer bulk solution after 1 h of dissolution was0.1 mmol/L, Thismodel therefore provides a means of direct comparison between the effects of various fluoride concentrations in acidic buffers in contact withapatite crystal surfaces. Although diffusion to and from the crystal surface and even within the pellets will play a part in the empirically measured dissolution rates in these experiments, the stirring speed and geometry were the samefor each experiment, and consequently the direct comparison of initial dissolution rates from onefluoride concentration to the other gives a comparative measure of the relative effects of fluoride concentration in the acid buffer. A recent study by Wong et al. (1987) using pelletized hydroxyapatite, fluorapatite, or hydroxyapatite with surface-adsorbed fluoride showed that incorporated fluoride only had an effect when it was present at concentrations equivalent to those in pure fluorapatite (38,000 ppm). In other words, crystal surfaces which behaved as though they were essentially fluorapatite masked the reactivity of the underlying apatite crystals of differing compositions. Chen and Nancollas (1986) used powdered bovine enamel and powdered human enamel in a constant composition demineralization model with a partially saturated buffer at pH 4.5 and concluded that 'the dissolution reactions suggest a surface dislocation mechanism, and the presence of fluoride ion in the buffer markedly retarded the reaction'. In their experiments, a fluoride level of only 0.5 mg/L (26 umol/L) reduced the rate of dissolution of human enamel ten-fold. In contrast, they reported that the dissolution of pure hydroxyapatite was best interpreted in terms of a polynucleation process. In the present study, the inverse dependence of carbonated-apatite initial dissolution rate on the logarithm of the fluoride concentration in the surrounding aqueous acid suggests a direct effect of surface-adsorbed fluoride ions on dissolution sites at the crystal surface. It is expected thatthesame mechanism of inhibition willoccur at the surfaces of carbonatedapatite enamel crystals. The natural extension of the above information in relation to caries progression or inhibition is as follows. The plaque fluid and the fluid among the crystals will certainly not behave as an "infinite sink". The aqueous fluid among the crystals in a forming caries lesion will most likely have much higher concentrations of calcium and phosphate present, and be saturated with respect to the adjacent apatite crystal surface phases until acid diffuses in from the external plaque during a caries challenge. As the local pH falls, mineral components will dissolve-in particular calcium phosphate, carbonate, and fluoride-until saturation is again reached. Fluoride from within the crystal which appears in solution in this way will then inhibit further dissolution. More importantly, fluoride already present in this intercrystalline aqueous phase, or that diffused
in from the plaque at the same time as the acid, will have an inhibitory effect on dissolution in a manner similar to that observed in the experiments reported here. Further, fluoride in this aqueous phase between the crystals will contribute, together with calcium and phosphate, to the degree of saturation with respect to FAP, thereby inhibiting dissolution by thermodynamic considerations. This is best illustrated in the phase diagram of Fig. 3. Each line represents the boundary between solid and solution as calculated from published solubility product values for each mineral. Enamel apatite does not have a definite single solubility product (Ksp) because of its variablecomposition. The moresoluble mineral, however, can be represented by a line drawn based on pKsp = 104 reported by Patel and Brown (1975). Lines for FAP equilibria are shown for solution F concentrations of 0.1 and 1.0 ppm (5.3 and 53 umol/L). It can be seen that even in the presence of these low fluoride concentrations a crystal surface behaving like FAP would have an effective solubility several orders of magnitude lower than the original enamel apatite crystalat the same pH-for example, at pH = 5.0. Therefore, if enamel apatitecrystalsurfaces come in contactwith fluoride from topical sources and the fluoride adsorbs together with calcium and phosphate, producing a FAP-like surface, such as thatformed and tested in the experiments of Wong et al. (1987), this would at least partly explain the role of fluoride as an inhibitor of enamel demineralization when present during a caries challenge. In the present study, the pH-cycling de/remineralization model that we used was designed to simulate results found in vivo at highly susceptible sites around orthodontic brackets (Featherstone et al., 1986; O'Reilly and Featherstone, 1987). This model combines periods of acidchallenge and periods of lesion repair. It purposely eliminates salivary components exceptcalcium, phosphate, and fluoride and allows for daily, or twicedaily, fluoride product treatments. The model can also be used to study the effects of salivary organic components (Cooper and Featherstone, 1986), the effects of fluoride at low concentrations in the mineralizing solution (Featherstone et al., 1986), and root caries progression (Featherstone et al., 1989). Although care must be taken in the interpretation of results, this model has been used successfully to screen fluoride products as part of the American DentalAssociation (ADA) profile system (ADA Council on Dental Therapeutics, 1988), to predict the lack of interference of pyrophosphate with remineralization in calculus-control dentifrices (Featherstone et al., 1988). This latter conclusion has been supported by clinical trials (Lu et al., 1985; Triol et al., 1988). In the presentstudy, the net effect of daily fluoride solution treatments on the de/remineralization process was measured to establish whether the net mineral change was dependent on fluoride concentration in the treatment solutions. The inverse TABLE RELATIVE MINERAL LOSS, IlZ, IN pH CYCLING EXPERIMENT Fluoride Concentration (mg/L) in Treatment Solution
o 1 10 50 250 500
Upper window (SE")
IlZ" 3132 1845 836 578 470 51
(387) (366) (339) (293) (247) (234)
Lower window mean IlZ (SP)
4257 2616 980 1060 895 208
(605) (466) (296) (399) (123) (75)
""IlZ = mean relative mineral loss, volume percent x urn, for each test group. "SE = standard error of the mean.
Downloaded from jdr.sagepub.com at UCSF LIBRARY & CKM on February 25, 2015 For personal use only. No other uses without permission. © 1990 International & American Associations for Dental Research
FEATHERSTONE et ai.
624
J Dent Res February 1990 ,DO
2
\ \ \
3
80
\
\
SUPERSATURATED
c: 0
't
\
4
~
\
60
e,
C
5 """ ~
..•• u
X
4D
a,
DCPD
