Journal of Dental Research http://jdr.sagepub.

com/

Effects of Aged Fluoride-containing Restorative Materials on Recurrent Root Caries

C.-Y.S. Hsu, K.J. Donly, D.R. Drake and J.S. Wefel J DENT RES 1998 77: 418 DOI: 10.1177/00220345980770021101 The online version of this article can be found at: http://jdr.sagepub.com/content/77/2/418

Published by:
http://www.sagepublications.com

On behalf of:
International and American Associations for Dental Research

Additional services and information for Journal of Dental Research can be found at: Email Alerts: http://jdr.sagepub.com/cgi/alerts Subscriptions: http://jdr.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations: http://jdr.sagepub.com/content/77/2/418.refs.html

>> Version of Record - Feb 1, 1998 What is This?

Downloaded from jdr.sagepub.com by guest on October 30, 2011 For personal use only. No other uses without permission.

J Dent Res 77(2): 418-425, February, 1998

Effects of Aged Fluoride-containing Restorative Materials on Recurrent Root Caries

C.-Y. S. Hsu, K.J. Donly, D.R. Drake, and J.S. Wefel*
Dows Institute for Dental Research, College of Dentistry, University of Iowa, Iowa City, Iowa 52242, USA; *to whom correspondence and reprint requests should be addressed

Abstract. Accumulated evidence has illustrated that secondary caries is the major reason for the failure of amalgam and resin composite restorations. The purpose of this study was to assess the cariostatic effects of aged fluoride-containing restorative materials on the formation of secondary root caries. Fifty sound human molars were selected and randomly assigned to five material groups: non-fluoride-containing amalgam (NA), fluoride-containing amalgam (FA), non-fluoride-containing composite (NC), fluoride-containing composite (FC), and glass-ionomer cement (GIC). After standardized class V cavity preparations and placement of restorations, teeth within each group were randomly divided into two subgroups, "non-aged" and "aged". The aged subgroup was immersed in an inorganic buffer solution for 2 wks before being thermocycled. After being thermocycled and subjected to four cycles of caries formation in a bacterial model system, the teeth were sectioned. Depths of outer lesions and areas of lesions on the cavity walls were measured by polarized light microscopy. The results showed that the FA and GIC groups, whether aged or not, had significantly smaller outer lesion depth than the non-fluoride-containing NA and NC groups. After aging, the FA group demonstrated significantly greater lesion depth (p = 0.0002), while the GIC group exhibited no significant changes in lesion depths. The NA group had a greater wall lesion area than the NC group, while both demonstrated no significant changes following 2 weeks of aging. The FA and GIC groups had similar inhibition areas along the cavity walls, whereas both inhibition areas increased significantly after the aging process. It is concluded that the fluoride-containing amalgam and the glass-ionomer cement, even after a two-week aging process, can still elicit a significant preventive effect on recurrent root caries in an in vitro bacterial model system.
Key words: materials, fluoride, root caries, Streptococcus
mutans.

Introduction
Recurrent caries has been shown to be the most important factor in the failure of amalgam and composite restorations (Allan, 1977; Mjor, 1981; Boyd and Richardson, 1985; Qvist et al., 1990a,b; Kidd et al., 1992). It has been reported that approximately half of all restorative dentistry provided is in the form of restoration replacements, while 40% of replacements are attributed to recurrent caries (MacInnis et al., 1991). The observation of the low incidence of recurrent caries around silicate restorations, which contain fluoride, has prompted the idea of using fluoride-containing dental materials to prevent recurrent caries. In the past few decades, many in vitro studies have demonstrated the cariostatic effects of these fluoridated restorative materials. However, the amount of fluoride released from those fluoride-containing materials has also been shown to decrease significantly with time (Forsten, 1990). The residual cariostatic effect, caused by "aged" fluoridecontaining materials, has remained unclear. Recurrent caries has been categorized in two dimensions (Kidd et al., 1992), one on the tooth surface and the other along the cavity wall through microleakage. The advent of acid-etching techniques and resin-based bonding agents has minimized microleakage at enamel margins; however, microleakage along the root-restoration interface is still an unsolved problem (Mandras et al., 1993). Streptococcus mutans has been identified as one of the most important micro-organisms in the etiology of caries (Fitzgerald and Keyes, 1960; Emilson and Krasse, 1985; Loesche, 1986) and has been suggested to be of significance in the etiology of recurrent caries (Bentley et al., 1990; Gonzalez-Cabezas et al., 1995). Many of the cariostatic effects of fluoride-containing materials have been evaluated in chemical caries formation systems without the involvement of micro-organisms, which have been used in only a few published studies (Gilmour et al., 1990, 1993; Derand et al., 1991). The purpose of this study was to compare the cariostatic effects of "aged" fluoride-containing restorative materials with those of non-aged materials on the prevention

Received July 29, 1996; Last Revision May 7, 1997; Accepted June 7, 1997

418
Downloaded from jdr.sagepub.com by guest on October 30, 2011 For personal use only. No other uses without permission.

J Dent Res 77(2) 1998

of recurrent root caries in a bacterial caries model system.

Secondary Caries

419

Materials and methods

Specimen and materials selection The teeth used in this study were all from residents of the State of Iowa, USA, and were collected from oral surgeons who extracted the teeth for reasons other than for the present study. Therefore, the teeth fall into exempt category 4, and the human subjects' approval is not required. Non-carious molars in a 0.1% thymol solution were selected, disinfected in a 10% formalin solution for 2 wks, and stored in a 0.1% thymol solution. After being examined with a lOX magnifying lens, Figure. Three major patterns of demineralization along cavity walls evaluated with polarized light teeth with visible defects or microscopy. Three types of lesions along the cavity wall can be categorized based on three different white-spot lesions were disdirections that the inner border of demineralization may take toward the cavity wall (the DF line). If the carded. The remaining 50 inner border of the demineralization extended straight to the cavity wall at 90 (the ABE line), the lesion teeth were randomly divided exhibited "no effect". If the inner border of demineralization curved upward (the ABC line), the lesion exhibited an "inhibition area" (the BCDE area). If the inner border of demineralization curved downward into 5 groups, corresponding (the ABF line), the lesion exhibited a "wall lesion" represented by the BEF triangular area. to the five different restorative dental materials. These materials were TytinTM (Kerr, until all groups completed the thermocycling procedure. Romulus, MI, USA), Synalloy fluoreTM (Dentoria, Cachan, France), Z-l0tTM (3M, St. Paul, MN, USA), HeliomolarTM (Vivadent, Schaan, Liechtenstein), and VitremerTM (3M, St. Paul, Artificial recurrent caries production MN, USA) (Table 1). Standard Class V preparations were placed Streptococcus mutans 10449 (serotype C) was initially cultured in on both mesial and distal surfaces of each tooth, with dimenTrypticase Soy Broth supplemented with 0.5'Y, Yeast Extract sions of 2 mm in depth, 5 mm in length, and 2 mm in width (TSB-YE) and incubated overnight in a 5K1 CO2 incubator at (half above and half below the cemento-enamel junction). All 37C. A 2-mL quantity of this primary culture was used to teeth were prepared by means of a high-speed handpiece with a inoculate a 250-mL flask containing 150 mL TSB-YE #330 bur in an air turbine under adequate water cooling. Five supplemented with 1% glucose and 1K, sucrose. Each subgroup teeth in each group, randomly selected as the aged subgroup, of five teeth was suspended in the medium by orthodontic were restored with the dental materials according to manuwires at approximately the same level. Each flask was then facturers' instructions. The dentin bonding systems used in this covered by an elastic cup with overlying tin foil and incubated study are summarized in Table 2. After the restorative proceat 37C and in 5% CO2. Media were changed every 48 hrs. After dures were completed, the five groups of teeth were separately completion of a total of four incubation cycles, all teeth were placed into an inorganic buffer solution (20 mM NaHCO3, 3 removed from the bacterial broth and separately put into a test 1 a two-week for mM NaH2PO4 H20, and mM CaCl2 2 H20) tube containing 10 mL of 5.0 N sodium hydroxide (NaOH) to aging period, with the solution changed every other day. The stop the demineralization process. five remaining teeth in each group were restored with the same materials after the aged subgroups finished the aging process. Then, all teeth were covered by acid-resistant nail varnish, Lesion measurement and data collection except for a zone approximately I mm wide around the restorations. After the nail varnish dried, all groups were separately We used a Silverstone-Taylor hard-tissue microtome (series thermocycled in 55 + 5C and 10 + 5C distilled water baths for 1000 Deluxe, Sci Fab, Littleton, CO) to cut each tooth to obtain 3 500 cycles with a 32-second dwell time in each bath and a 14or 4 sections of 100 + 20 pm thickness. Sections were imbibed in second interval between baths. The groups that finished water and viewed with a polarized light microscope. Standard thermocycling first were stored in a 100% humidity chamber 35-mm polarized light photomicrographs were taken. The
Downloaded from jdr.sagepub.com by guest on October 30, 2011 For personal use only. No other uses without permission.

420

Hsu et al.

J Dent Res 77(2) 1998

Table 1. Five material groups used in this study Group

Abbreviation
NA

Brand
Tytin

Comments

Manufacturer
Kerr, Romulus, MI, USA

Non-fluoride amalgam

High-copper; spherical
1% SnF2 High-copper

Fluoride amalgam

FA

Synalloy fluore
Z-100

Dentoria, Cachan, France

3M, St. Paul, MN, USA

Non-fluoride resin composite

Fluoride resin composite

NC
FC

Combined with Scotchbond Multi-purpose

9% YbF3 combined with Syntac system

Heliomolar
Vitremer

Vivadent, Schaan, Liechtenstein

Glass-ionomer cement
a

GIC

FASa glass with tri-cure system

3M, St. Paul, MN, USA

FAS stands for the fluoroaluminosilicate.

demineralization areas were measured by projection of the photomicrographs onto a sonic-digitizing pad. Those areas were traced by means of a digitizing pen (Graf/Pen Sonic Digitizer; Science Accessories Corp., Southport, CT) onto a 23 mm x 34 mm template. The lesion depth and area of each section from the same lesion were measured, and the averaged data represented an observation for each lesion. We determined the averaged depth of each outer lesion by measuring a lesion area of 200 pm in length and dividing it by 200. The wall lesion area or inhibition area was defined by three peripheral lines: (1) the vertical line representing the cavity wall or the toothrestoration interface; (2) the inner border of the demineralization area curving down or up, axially, to meet the cavity wall; and (3) the imaginary line following the horizontal portion of the inner border of the demineralization area and extending straight to the cavity wall at 900 (Fig.). Statistical analysis For the outer lesion depth, the mean and variance in each group were calculated, and the interaction between the material and aging effects were assessed. Since Bartlett's test indicated that
Table 2. Dentin bonding systems

the variances were unequal, we used the Kruskal-Wallis nonparametric test to evaluate significant differences among the five material groups, within aged or non-aged subgroups. All post hoc pairwise comparisons among the five material groups were done by means of Wilcoxon rank-sum tests with an alpha level of 0.005 based on the Bonferroni correction. Paired comparisons between the aged and non-aged groups were evaluated by means of the Wilcoxon signed-rank tests with an alpha of 0.01 according to the Bonferroni correction. For the inhibition area and the wall lesion area, a two-way ANOVA was used to assess material and aging effects with an alpha of 0.05.
Results

After the first 24-hour incubation, the culture media became turbid, and a thin layer of plaque-like substance covered all the wires and the varnished and unvarnished tooth surfaces. Under the light microscope, Gram stains revealed the abundant presence of Streptococcus mutans. During the incubation period, this covering substance continuously grew thicker. At the end of the incubation period (8 days), the thickness of this layer was approximately 1 mm.

Group
NA (Tytin) FA (Synalloy fluore) NC (Z-100) FC (Heliomolar)

Bonding System
None None Scotchbond MP Scotchbond MP Scotchbond MP Syntac

Component
None None Etchant Primer Adhesive Primer Adhesive

Content
None None 10% maleic acid HEMAa/Vitrebond copolymer

HEMAa/Bis-GMAb
4% maleic acid/TEG-DMAc/Acetone

Syntac Heliobond
GIC (Vitremer)
b
d

PEG-DMAd/Glutaraldehyde Bis-GMAb
HEMAa/ethanol/Vitrebond copolymer/photoinitiators

Primer

HEMA = Hydroxyethyl methacrylate. Bis-GMA: Bisphenol-A glycidylmethacrylate TEG-DMA: Triethylene glycoldimethacrylate PEG-DMA: Polyethylene glycoldimethacrylate
Downloaded from jdr.sagepub.com by guest on October 30, 2011 For personal use only. No other uses without permission.

J Dent Res 77(2) 1998

Secondary Caries
Wall Lesion
24 24 48 (91%)
0 0 0 (0%)

421

Table 3. Numbers of sections with various types of lesions on the root ca[vity walls Material
NA

Subgroup

Inhibition Area
0 0 0 (0%) 27 26 53 (98%) 0 0 0 (0%) 5 10 15 (27%) 28 27 55 (98%)

Non-aged Aged Total

Non-aged Aged Total

FA

NC

Non-aged Aged Total

Non-aged Aged Total

10 16 26 (46%)
7 16 23 (43%)

FC

GIC

Non-aged Aged Total

0 0 0 (0%)

Histological features of demineralization lesions In the histological evaluation of caries-like lesions, outer lesions vary only in lesion depth, while the lesions along the cavity walls may vary in shape as well as in size of the demineralized area. The Fig. illustrates three major types of observed demineralization patterns. In the first pattern, the inner border of demineralization follows the horizontal orientation of the distal and middle portions of the lesion and meets the cavity wall at a 900 angle. If the proximal portion of the inner border of a lesion curved downward to meet the cavity wall, we called this type of lesion a "wall lesion". If the proximal portion of the inner border ran upward to meet the cavity wall, we called this wedgeshaped region an "inhibition area". Table 3 lists the numbers and the percentages of the tooth sections with various types of lesions on the root cavity walls. It is interesting that the FC group showed both wall lesions and inhibition areas on their root cavity walls, while the NA and NC groups never
Table 4. Aging effects on root outer lesion depths (pm)

demonstrated inhibition areas, and the FA and GIG groups never showed wall Total INo Effect lesions. In 28 sections of the non-aged FC group, 16 27 3 sections have neither wall 26 2 lesions nor inhibition areas, 53 (100%) 5 (9%) 7 sections showed wall lesions, and 5 sections 28 1 inhibition areas. In the 26 26 0 sections of the aged FC 54 (100%) 1 (2%) group, 10 sections demonstrated inhibition areas and 28 18 16 wall lesions. The number 29 13 of sections with both wall 57 (100%) 31 (54%) lesions and inhibition areas appeared to be increased in 28 16 the aged FC group. In the 26 0 aged NC group, the number 54 (100%) 16 (30%) of sections with "no effect" decreased and that of 29 1 sections with wall lesions 27 0 increased when compared 56 (100%) 1 (2%) with the non-aged subgroup. The FA and GIC groups both had 98% sections exhibiting inhibition areas, while the NA group had 91% sections demonstrating wall lesions.
Outer lesions The p-values of pairwise comparisons between the aged and non-aged groups are shown in Table 4. The two-week aging process exhibited a variable effect on demineralization of all groups (Table 4). The FA and FC groups had deeper lesions after aging, while the NA and GIC groups exhibited no significant change. Interestingly, the non-aged NC group had deeper lesions on the root surfaces after aging. The rankings of the outer lesion depths for the five material groups are listed in Table 5. The order, from the deepest to the shallowest groups, is "NC, NA, FC, GIC, FA" for both the non-aged and aged groups. Those groups having means of outer lesion depth that were not statistically different from each other were assigned the
Table 5. Material effects on root outer lesion depths (pm)

Non-aged
Material NA FA NC FC GIC
a

Aged
Mean SD

Mean SD

Aging Effect p valuea

0.0140 0.0002 0.0002 0.0073 0.1212

Subgroup Material
NC NA FC GIC FA
a

Non-aged
Mean SD 141.8 5.8 120.9 3.7 103.5 3.9 101.8 7.0 52.0 3.9

Aged
Rankinga
A B C C D
Mean SD

Ranking
a a

120.9 3.7 52.0 3.9 141.8 5.8 103.5 3.9 101.8 7.0

129.6 11.5
79.9 5.4 121.6 6.2 122.6 15.7 106.0 5.8

121.6 6.2 129.6 11.5 122.6 15.7 106.0 5.8 79.9 5.4

ab b
c

Statistical significance of the aging effects was evaluated by means of Wilcoxon signed-rank tests. P value stands for the statistical significance of comparisons between the aged and non-aged groups restored with the same materials, considered significant when p < 0.01.

Means of groups which did not significantly differ from each other were assigned the same letters (upper case for the nonaged groups and lower case for the aged groups).

Downloaded from jdr.sagepub.com by guest on October 30, 2011 For personal use only. No other uses without permission.

422

Hsu et al.

J Dent Res 77(2) 1998

Table 7. Area of inhibition areas in lesions on root cavity walls (pm x PM)

Table 6. Area of the wall lesion on the root surface (pm x pm)

Material
NA NA NC NC
a

Treatment

N
10 10 10 10

Mean SDa

Material

Treatment

N
10 10 10 10

Mean SDa
60.4 19.3 112.8 20.1 48.7 24.8 101.1 20.9

Non-aged Aged Non-aged Aged

28.3 15.1 36.0 16.6 10.2 14.3 19.1 24.9

Two-way ANOVA revealed no significant interaction between material and aging effects, a significant material effect (p = 0.0045), and an insignificant aging effect (p = 0.1602).

FA FA GIC GIC
a

Non-aged Aged Non-aged

Aged

Two-way ANOVA revealed no significant interaction between material and aging effects, a significant aging effect (p = 0.0001), and an insignificant material effect (p = 0.0923).

same ranking, denoted by letters in a descending order. The p-values in each pairwise comparison were considered statistically significant when less than 0.005, after the Bonferroni correction.

Lesions along the cavity wall As mentioned above, the demineralization along the cavity wall can be a wall lesion, an inhibition area, or a simple "noeffect" lesion (Table 3). Since we found no sections in the NA and NC groups demonstrating inhibition areas and no wall lesions in the FA and GIC groups, we compared quantitatively only the wall lesions in the NA and NC groups and the inhibition areas in the FA and GIC groups. The two-way ANOVA test was used to evaluate the material and aging effects. The root wall lesion areas measured in the NA and NC groups revealed a statistically significant material effect (p = 0.0045) but not a significant aging effect (p = 0.1602) (Table 6). The results indicated two noteworthy points. First, the root cavity walls in the NA group demonstrated a greater wall lesion area than those in the NC group. Second, the two-week aging period resulted in a larger wall lesion area in both the NA and NC groups, although the difference between the aged and non-aged subgroups did not reach a statistically significant level. The inhibition areas measured in root wall lesions of the FA and GIC groups elucidated a significant increase of the inhibition area along the cavity wall after aging (p = 0.0001), but no significant difference was found between these two material groups (p = 0.0923) (Table 7).

agents and the degree of microleakage between restorations and tooth structures. It has been previously shown that if there is no microleakage, there will be no wall lesion (Derand et al., 1991). Although not the only contributing factor, microleakage is a necessary condition for "wall lesion" formation. The presence of cariostatic agents may reduce or eliminate caries formation via the inhibition of demineralization and/or bacterial activities, such as colonization and acid production. Cariostatic agents include mercury, silver, zinc, copper, and fluoride (0rstavik, 1985; Forss et al., 1991; Lyttle and Bowden, 1993). Among them, fluoride is one of the major controlled variables in this study.

Discussion
In this study, demineralization on the outer surfaces appeared to be caused by acid production from Streptococcus mutans, which colonized the tooth surface, while the demineralization along the cavity wall was caused by a combination of the primary acid attack and an additional acid attack through microgaps between the cavity wall and restoration. Both types of acid attack can be countered by cariostatic agents deposited on the outer surfaces, at the cavity walls, or in the microgaps. In general, the outer lesion depth was mainly affected by the local concentration of cariostatic agents on the surface, while the demineralization along the cavity wall was influenced by both the cariostatic

Fluoride effects The ability of restorative materials to release fluoride and the ability of the adjacent tooth structure to take up fluoride are both important factors influencing the cariostatic effect of fluoride. Vitremer (GIC) contains the highest percentage of fluoride and has the greatest releasing ability. Next is the fluoride-containing resin composite (Heliomolar), while Synalloy fluore (FA) has the least fluoride. However, in addition to fluoride, amalgam contains other cariostatic metal ions, such as Hg, Ag, and Zn. As demonstrated in a previous study, glass-ionomer cements release the greatest amount of fluoride, followed by the fluoridated amalgam group, which released more fluoride than the fluoridated composite group in the first 6 wks (Forsten, 1990; Ulukapi et al., 1996). The FA group exhibited a smaller demineralized lesion on the outer surface compared with the NA group, probably due to fluoride release (Table 5). Using a gel-type caries formation system, Skartveit et al. (1991) compared demineralization adjacent to Synalloy and Synalloy fluore restorations. They demonstrated a significant inhibition of demineralization in the fluoridated amalgam group along the cavity walls, but not on the outer surfaces, both in root and in enamel. A possible explanation is that the cavity wall is closer to the fluoride-releasing amalgam and acquires more fluoride than the outer surface (Tveit and Hals, 1980). In a previous in vitro study, we investigated the effects of the same five restorative materials used in this study on bacterial colonization. The results showed that the FA group had a significantly greater inhibition of colonization than the other material groups, including the NA (Tytin) group (Hsu et al., 1994). It is also possible that acid production by

Downloaded from jdr.sagepub.com by guest on October 30, 2011 For personal use only. No other uses without permission.

J Dent Res 77(2) 1998

Secondary Caries

423

Streptococcus mutans may be inhibited by fluoride, as shown by Seppa et al. (1992). Collectively, then, we suggest the involvement of at least two possible mechanisms in the demonstrated cariostatic effects of Synalloy fluore on recurrent caries: inhibition of demineralization and inhibition of bacterial colonization and/or acid production. As for lesions on the wall, the FA group demonstrated inhibition areas, while the NA group showed wall lesions (Tables 6, 7). This result is in agreement with those of other studies that demonstrated reduced demineralization along the cavity wall adjacent to the restoration of Synalloy fluore (Skartveit et al., 1991). Because the high-copper Tytin (NA) would have a corrosion rate similar to that of the highcopper Synalloy fluore (FA), microleakage may exert only a minimal influence on the difference in the lesion along the wall between the NA and FA groups. Therefore, fluoride released from the Synalloy fluore (FA) seems to be the major factor leading to less demineralization along the cavity wall in the FA group than in the NA group. A cariostatic effect of fluoride-containing resin restorations along cavity walls has been elucidated in other studies (Dijkman and Arends, 1992; Rebitski and Donly, 1993). Compared with the NC group, the FC group has demonstrated a significantly smaller outer lesion depth before aging, but the difference in lesion depth between them became statistically insignificant after the aging process. Thus, the fluoride effects may decrease due to the aging process. As for lesions on cavity walls, some tooth sections of the FC group demonstrated inhibition zones which were not seen at all in those of the NC group. Interestingly, some sections showed inhibition areas, but others, even some belonging to the same tooth, exhibited wall lesions in the FC group. Forty-three and 46% of the sections in the FC and NC groups, respectively, had wall lesions (Table 3). This finding seemed to corroborate the observation that perfect marginal sealing was not obtained, even by the use of new-generation bonding agents (Mandras et al., 1993). Another possible reason could be that the fluoride-free dentin bonding system might adhere to the cavity wall so strongly that it, partially or completely, interfered with the diffusion of fluoride from Heliomolar (FC) into the adjacent tooth structure. This hypothesis has been mentioned in other published articles (Kerber and Donly, 1993; Prati et al., 1994).
Aging effect Most fluoridated restorative materials have been shown to release a smaller amount of fluoride after aging (Forsten, 1990; Dijkman et al., 1993). In this study, we changed the inorganic buffer solution and the culture media every other day. The loosely attached fluoride, or other cariostatic agents, may be gradually lost during media exchange through the whole two-week aging period. However, the fluoride which continuously diffused into the juxtaposed cavity wall may be cumulatively incorporated into the porous root dentin during and after aging. The concentration of fluoride ions released from amalgam decreases during the aging process (Forsten, 1990), as do those of the other cariostatic ions mentioned

previously (Lyttle and Bowden, 1993). The FA groups showed more demineralization on the outer surfaces after the two-week aging process (p = 0.0002) and greater inhibition areas compared with the non-aged groups (p = 0.0001). However, the increase in outer lesion depths of the NA groups (p = 0.014) and wall lesion areas (p = 0.16) did not reach the statistically significant level of ox = 0.01 and 0.05, respectively. These results suggested that the inhibitory effects of the metal ions on demineralization in the NA group are insignificantly or only slightly reduced after a two-week aging process. The outer lesions of the NC and FC groups both changed after aging (p = 0.0002 and 0.0073, respectively). However, they are in opposite directions (Table 4). The decrease in the local fluoride concentration after aging in the FC group could explain the finding of deeper lesions after aging. It remains unclear why the aged NC group exhibited shallower lesions on the outer surfaces. For the lesions along the cavity walls, the NC group showed a greater number and area of wall lesions after aging; however, it did not reach statistical significance (Table 7). We speculate that this could be a temporary balance between the effects of the hygroscopic expansion and the hydrolytic degradation on microleakage. Interestingly, the FC group showed an increased percentage of tooth sections with both wall lesions and inhibition areas after aging (Table 3). Hydrolytic degradation of the adhesive bonding system in the FC group (Syntac) may result in microleakage and thus enhance demineralization in some teeth but may also have facilitated the diffusion of fluoride from restorations into the cavity walls. The ultimate outcome of aging effects on demineralization may be dictated by the total net effects of these two factors, which may interact with each other. Furthermore, some bonding between bonding agents and tooth surfaces may break after aging and thermal cycling procedures, while some bonding may remain intact. Even in the same tooth sample, albeit not often seen, some sections did have inhibition areas while others had wall lesions. The factors which influence the bonding integrity of dentin bonding agents at microscopic levels warrant further investigation. It is well-accepted that the fluoride-releasing ability of GIC decreases with aging. Different materials have exhibited various aging effects on the materials' fluoridereleasing ability (Forsten, 1990). Crim (1993) showed a good marginal adaptation of one light-activated GIC (Fuji II LCTM), even after the six-month aging process. Prati et al. (1994) demonstrated no significant increase of microleakage adjacent to VitrebondTM (3M, St. Paul, MN, USA) after two months. Our findings suggest that there are no significant differences in outer lesion depths between the GIC-aged and non-aged subgroups. On the cavity walls, the increased accumulation of fluoride did enhance the inhibitory effect on demineralization and caused a significant increase in the inhibition areas after the two-week aging process

(p = 0.0001).
Material effects
In general, the higher the concentration of cariostatic agents,

Downloaded from jdr.sagepub.com by guest on October 30, 2011 For personal use only. No other uses without permission.

424

Hsu et al.

J Dent Res 77(2) 1998

the adjacent root structure after 2 wks, there would be some value for further studies with a longer and more vigorous aging scheme for better simulation of the oral condition.

the greater the potential cariostatic effect. It has been reported that there is a linear relationship between the logarithm of in vitro fluoride release and the inhibitory effect on enamel demineralization (Dijkman et al., 1993). The outer lesion depths of the fluoride-containing groups (GIC and FA) were always smaller than those of the non-fluoride-containing groups (NC and NA). Table 5 shows that the order of "NC 2 NA > FC = GIC > FA" remains the same for outer lesion depth, whether aged or non-aged. It is interesting to note that the "aged" NC, NA, and FC groups have no statistically significant differences between their outer lesion depths. This may be due to the decreased availability of cariostatic agents on the root surfaces after the aging process in the NA and FC groups. There are two possible reasons why the FA groups had smaller outer lesions than the GIC groups. The first is the potential synergistic cariostatic effects of fluoride and other metal ions in amalgam. Another possibility is that the larger microleakage in the FA group, compared with that of the GIC group, may facilitate the leaking out of those cariostatic agents. The comparable inhibition areas along the cavity wall in the FA and GIC groups, aged or not, seem potentially to support both possibilities. Recently, the properties of GICs to take up and release fluoride have been substantiated (Creanor et al., 1994). Therefore, in the oral cavity of an individual who frequently uses fluoride products (e.g., toothpaste), the long-term cariostatic effects of GICs on adjacent tooth structures may transcend those of the FA restorations.
Experimental designs Since the cavity varnish has been shown to be quickly washed away in the clinical situation around amalgam restorations, and the sealing effect is only temporary (Dutton et al., 1993), we chose not to apply cavity varnish in our amalgam groups. The frequent presence of wall lesions in the NA groups and inhibition areas in the FA groups seemed to support the concept that sealing capacity is not important enough to outweigh the advantage of a release of incorporated fluoride from amalgam restorations in terms of the development and rapidity of the secondary caries attack in a short-term experiment (Heintze and M0rnstad, 1980). Regarding the GIC restorations, the mean gaps between the GIC and dentin have been reported to be about 8 to 10 pm (Sidhu, 1994). The fact that the predominant type (98%) of section in the GIC group demonstrated inhibition zones in spite of the microgaps seems to substantiate the aforementioned notion as well (Heintze and M0rnstad, 1980). Although various fluoridated materials have different rates and durations of fluoride release, most of the major fluoride release usually takes place in the first day or week. This holds true for fluoridated amalgam (Fazzi et al., 1977; Shannon and Miller, 1980), fluoridated resin composite (Swift, 1989; Ulukapi et al., 1996), and glass-ionomer cements (Diaz-Arnold et al., 1995; Musa et al., 1996). Although we recognize that amalgam corrosion products may not be present, we used a two-week aging period to circumvent the initial burst of fluoride release. Since this study indicated a significant cariostatic effect of glassionomer cements and fluoridated amalgam restorations on

Acknowledgments
The authors would like to thank Ting-ru Hsu for her statistical assistance. This investigation was financially supported in part by Research Grants P50 DE-11134 from the National Institute of Dental Research, National Institutes of Health, Bethesda, MD 20892.

References
Allan DN (1977). A longitudinal study of dental restorations. Br Dent J 143:87-89. Bentley CD, Broderius CA, Drake CW, Crawford JJ (1990). Relationship between salivary levels of mutans streptococci and restoration longevity. Caries Res 24:298-300. Boyd MA, Richardson AS (1985). Frequency of amalgam replacement in general dental practice. Can Dent Assoc J 51:763-766. Creanor SL, Carruthers LM, Saunders WP, Strang R, Foye RH (1994). Fluoride uptake and release characteristics of glassionomer cements. Caries Res 28:322-328. Crim GA (1993). Effect of aging on microleakage of restorative systems. Am J Dent 6:192-194. Derand T, Birkhed D, Edwardsson S (1991). Secondary caries related to various marginal gaps around amalgam restoration in vitro. Swed Dent J 15:133-138. Diaz-Arnold AM, Holmes DC, Wistrom DW, Swift EJ Jr (1995). Short-term fluoride release/uptake of glass ionomer restoratives. Dent Mat 11:96-101. Dijkman GE, Arends J (1992). Secondary caries in situ around fluoride releasing light-curing composites: A quantitative model investigation on four materials with a fluoride content between 0 and 26 vol%. Caries Res 26:351-357. Dijkman GE, de Vries J, Lodding A, Arends J (1993). Long-term fluoride release of visible light-activated composites in vitro: A correlation with in situ demineralization data. Caries Res 27:117-123. Dutton FB, Summitt JB, Chan CN, Garcia-Godoy F (1993). Effect of a resin lining and rebonding on the marginal leakage of amalgam restorations. 7 Dent 21:52-56. Emilson CG, Krasse B (1985). Support for and implications of the specific plaque hypothesis. Scand J Dent Res 93:96-104. Fazzi R, Vieira DF, Zucas SM (1977). Fluoride release and physical properties of a fluoride-containing amalgam. J Prosthet Dent 38:526-531. Fitzgerald RJ, Keyes PH (1960). Demonstration of the etiologic role of Streptococci in experimental caries in the hamster. J Am Dent Assoc 61:9-19. Forss H, Jokinen J, Spets-Happonen S, Seppa L, Luoma H (1991). Fluoride and mutans Streptococci in plaque growth on glass ionomer and composite. Caries Res 25:454-458. Forsten L (1990). Short- and long-term fluoride release from glass ionomers and other fluoride-containing filling materials in vitro. Scand J Dent Res 98:179-185. Gilmour SM, Edmunds DH, Dummer PM (1990). The

Downloaded from jdr.sagepub.com by guest on October 30, 2011 For personal use only. No other uses without permission.

J Dent Res 77(2) 1998

Secondary Caries

425

production of secondary caries-like lesions on cavity walls and the assessment of microleakage using an in vitro microbial caries system. I Oral Rehabil 17:573-578. Gilmour AS, Edmunds DH, Newcombe RG, Clark MF (1993). An in vitro study into the effect of a bacterial artificial caries system on the enamel adjacent to composite and amalgam restorations. Caries Res 27:169-175. Gonzalez-Cabezas C, Li Y, Noblitt TW, Gregory RL, Kafrawy AH, Stookey GK (1995). Detection of mutans Streptococci in secondary carious lesions using immunofluorescent techniques and confocal laser scanning microscopy. Caries Res 29:198-203. Heintze U, Mornstad H (1980). Artificial caries-like lesions around conventional, fluoride-containing and dispersed amalgams. Caries Res 14:414-421. Hsu CY, Wefel JS, Drake DR, Donly KJ (1994). Colonization of dental materials by Streptococcus mutans (abstract). J Dent Res 73 (Spec Iss):417. Kerber LJ, Donly KJ (1993). Caries inhibition by fluoridereleasing primers. Am J Dent 6:216-218. Kidd EA, Toffenetti F, Mjor IA (1992). Secondary caries. Int Dent 1 42:127-138. Loesche WJ (1986). Role of Streptococcus mutans in human dental decay. Microbiol Rev 50:353-380. Lyttle HA, Bowden GH (1993). The level of mercury in human dental plaque and interaction in vitro between biofilms of Streptococcus mutans and dental amalgam. J Dent Res 72:1320-1324. Mandras RS, Retief DH, Russell CM (1993). Quantitative microleakage of six dentin bonding systems. Am J Dent 6:119-122. Maclnnis WA, Ismail A, Brogan H (1991). Placement and replacement of restorations in a military population. Can Dent Assoc J 57:227-231. Mjor IA (1981). Placement and replacement of restorations. Oper Dent 6:49-54. Musa A, Pearson GJ, Gelbier M (1996). In vitro investigation of

fluoride ion release from four resin-modified glass polyalkenoate cements. Biomaterials 17:1019-1023. 0rstavik D (1985). Antibacterial properties of and element release from some dental amalgams. Acta Odontol Scand
43:231-239. Prati C, Tao L, Simpson M, Pashley DH (1994). Permeability and microleakage of Class II resin composite restorations. I Dent 22:49-56. Qvist J, Qvist V, Mjor IA (1990a). Placement and longevity of amalgam restorations in Denmark. Acta Odontol Scand 48:297-303. Qvist V, Qvist J, Mjor IA (1990b). Placement and longevity of tooth-colored restorations in Denmark. Acta Odontol Scand 48:305-311. Rebitski G, Donly KJ (1993). Dentin pretreatment and caries inhibition by a fluoride-releasing resin. Am I Dent 6:204206. Seppa L, Torppa-Saarinen E, Luoma H (1992). Effect of different glass ionomers on the acid production and electrolyte metabolism of Streptococcus mutans Ingbritt. Caries Res 26:434-438. Shannon IL, Miller JT (1980). Fluoride supplemented dental amalgam. Oper Dent 5:11-13. Sidhu SK (1994). Marginal contraction gap formation of lightcured glass ionomers. Am J Dent 7:115-118. Skartveit L, Wefel JS, Ekstrand J (1991). Effect of fluoride amalgams on artificial recurrent enamel and root caries. Scand J Dent Res 99:287-294. Swift EJ Jr (1989). Fluoride release from two composite resins. Quintessence Int 20:895-897. Tveit AB, Hals E (1980). Inhibitory effect of a fluoridecontaining amalgam on development of cavity wall lesions in vitro. Acta Odontol Scand 38:29-39. Ulukapi H, Benderli Y, Soyman M (1996). Determination of fluoride release from light-cured glass-ionomers and a fluoridated composite resin from the viewpoint of curing time. J Oral Rehabil 23:197-201.

Downloaded from jdr.sagepub.com by guest on October 30, 2011 For personal use only. No other uses without permission.