Eur J Oral Sci 2000; 108: 555±560 Printed in UK. All rights reserved
Filler particle leachability of experimental dental composites
Karl-Johan M. SoÈderholm1, Mark C. K. Yang2, Ileana Garcea1 1
Department of Dental Biomaterials and the Department of Statistics, University of Florida, Gainesville, Florida, USA 2
SoÈderholm K-JM, Yang MCK, Garcea I. Filler particle leachability of experimental dental composites. Eur J Oral Sci 2000; 108: 555±560. # Eur J Oral Sci, 2000 We studied the eect of matrix selection, ®ller composition, and ®ller silanization on ®ller leachability after storage in distilled water or arti®cial saliva. We evaluated 2 matrix systems, 2 ®ller systems and 2 silane treatment procedures, combined into 8 dierent dental composite materials. A total of 128 batches were made, and 2 specimens per batch were prepared. Of these 2 specimens per batch, one was stored in distilled water and the other in arti®cial saliva, both at 37³C. We transferred the specimens each 30th day during a 3-yr period to new vials containing either freshly distilled water or newly mixed arti®cial saliva and analyzed the solutions the specimens had been stored in regarding Si, Ba and Al concentrations. The analyses revealed that storage solution, ®ller composition, and total time in the storage solution had strong eects on leachability. The average monthly leakage of the three elements was linear with time and higher in the arti®cial saliva. The Ba-containing ®ller leached Si faster in arti®cial saliva than in distilled water, and roughly twice as much as the quartz ®ller. The storage eect approached an order of magnitude, while the ®ller eect was roughly a factor of two. Filler leaching was linear over time.
Filler particles present in dental composites react with water and arti®cial saliva and leach dierent elements to the surrounding storage medium (1). In a previous study we found that the leaching pattern remained linear at least for 1-yr and was more pronounced for composites stored in arti®cial saliva than in distilled water (1). These ®ndings inspired us to extend our 1-yr study for another 2-yr period to determine whether these patterns remained the same over a longer time period. We needed to address that question because of a 3-yr clinical study that was run simultaneously with the same materials. According to a theory proposed by CHARLES (2), glasses containing glass-modifying ions such as Na leach both Na and Si when stored in water, while quartz remains quite stable. When these ions are released from the glass surface, the leaching process may induce a charge imbalance at the glass surface that could delay further ion release. However, if other ions of the same charge as those being released are present in the surrounding (e.g. in the saliva), these ions could help re-establishing a charge balance at the ®ller surface. Such a
Karl-Johan M. SoÈderholm, Dept. of Dental Biomaterials, Box J-0446, College of Dentistry, University of Florida, 1600 SW Archer Road, Gainesville, FL 32610-0446, USA Telefax: +1±352±3927808 E-mail:
[email protected] Key words: bisGMA; UEDMA; silane; artificial saliva; distilled water Accepted for publication September 2000
charge balance could facilitate a continuous release of the glass-modifying ions, explaining why we found that leaching was more pronounced in arti®cial saliva than in distilled water (1). The linear leaching behavior we found (1) could be explained by assuming that the release rate of ®ller elements from the ®ller surfaces is faster than the speed with which the elements are transported through the resin matrix. That behavior would result in a sustained release condition for the composite, and a linear ®ller element leaching pattern would be found. It has been known that dental composites stored in distilled water leach ®ller components (3). It has been known that glass modifying elements, such as Na, Ba and Sr, all increase the leaching rate (4, 5). It has been suggested that the above hydrolysis could explain why some composites in the past showed signi®cant wear (3). However, no conclusive evidence exists proving that hydrolytic degradation of the ®ller particles is a key factor in the clinical wear process of dental composites. In addition, since dental composites are surrounded by saliva, leaching data generated from storage in
556
SoÈderholm et al.
distilled water may be of little clinical relevance. Before we published our 1-yr leaching study cited above (1), it was not known that storage in arti®cial saliva resulted in an even faster leaching rate. Thus, the ®nding that ®ller leaching is signi®cantly higher in arti®cial saliva than in distilled water (1) could be of clinical signi®cance and explain some of the degradation mechanisms of dental composites. Because of the importance of exploring the long-term eect of ®ller degradation, the aims of this study were to extend our ®ller leaching study in distilled water and arti®cial saliva for 3 yr to determine whether the leaching pattern remains the same over a 3-yr period as it did under the ®rst year.
Material and methods Experimental dental composite matrices were made of either bisGMA (lot # 307-45; Esschem Co., Essington, PA, USA)/TEGDMA (lot # 30744-2; Esschem Co.) (50:50 by weight) or UEDMA (lot # PB 2037; Esschem Co.)/TEGDMA (50:50 by weight). To each matrix resin we added 0.35 wt.-% d-2,3-bornanedione (lot # 8074101297; Eastman Kodak, Rochester, NY, USA) and 0.7 vol.-% 2-(dimethylamino) ethyl methacrylate (lot # 06622FF; Aldrich Chemical Company, Milwaukee, WI, USA) to make them light curable, and then added either 53 vol.-% quartz or barium glass ®ller particles, both from Industrial Corporation (Lionville, PA, USA) to the light curable resin. The ®ller particles consisted of either crystalline quartz or barium glass ®ller particles (52% SiO2, 32% BaO, 8% B2O3, 8% Al2O3). The mean particle size of the ®ller particles was 3¡2 mm. Before the ®ller particles were incorporated into the monomer matrices, they were silane treated. This treatment consisted of mixing 75 g ®ller in a 3.75 ml g-methacryloxypropyltri-methoxysilane (lot # 51H0280; Sigma Chemical Co., St. Louis, MO, USA)/toluene (185 ml) (lot # 920352; Fisher Scienti®c, Fair Lawn, NJ, USA) solution for 24 h. After completed mixing, the silane solution was decanted and the ®ller particles were washed in toluene for 1 h, whereupon the toluene was decanted and the ®ller particles ®ltered. After 24 h of air-drying, the ®ller particles were transferred to an oven and dried for another 24 h at 60³C. After completing that drying, half of the ®ller particles were removed from the oven and cooled to room temperature, while the remaining half of the ®ller particles were dried in vacuum for 1 h at 110³C, and then cooled to room temperature. After several days of room temperature storage in a desiccator, the ®ller particles were mixed with the matrix
monomers to yield composites containing 53% ®ller by volume. Eight distinct materials were produced (two resins, two ®ller compositions, and two silane treatments), and for each of these eight combinations, sixteen individual batches were made. From each of the 8616 individual batches, 2 samples were made. One of these samples was stored in distilled water and the other in arti®cial saliva described below. Consequently, a total of 256 specimens were studied. The samples used for the study of leachable components were made by inserting the material into a circular mold, 15 mm in diameter and 0.5 mm high, covered on both sides with a Mylar> strip and a glass slide. The ®lled mold was then placed on a white paper sheet and cured with a Translux A lamp (Kultzer, South Bend, IN, USA). The specimen was cured for 40 s at the center and at eight peripheral locations, each being displaced 45 degrees relative to the previous cured location. Both surfaces were cured that way, and two specimens were made from each batch. The two specimens from each batch were stored in either distilled water or in arti®cial saliva used to simulate the salt composition of saliva. The arti®cial saliva (6) was made of the following composition: 100 ml each of 25 mM K2PO4, 24 mM Na2HPO4, 150 mM KHCO3, 100 mM NaCl, 1.5 mM MgCl2. To this were added 6 ml of 25 mM citric acid and 100 ml of 15 mM CaCl2. The pH was then adjusted to 6.7 with NaOH or HCl and the volume made up to 1 l. To avoid bacterial growth, 0.05% (by weight) thymol was added to the arti®cial saliva. All chemicals were of analytical grade and bought from Fisher Scienti®c Company (Fair Lawn, NJ, USA). Each specimen was placed in polystyrene vials (Dilu-Vial; FisherBrand, Fisher Scienti®c Company) containing 10 ml of either distilled water or arti®cial saliva, and kept in an oven at 37³C for 30 d. Each week the vials with the specimens were turned up side down 5 times to induce some stirring. After the 30-d storage period, the specimens were transferred to new vials containing 10 ml of the same storage medium/solution as they had been stored in during the previous month. In addition to the vials with the composite specimens, vials containing only distilled water and arti®cial saliva or distilled water and arti®cial saliva with known concentrations (0.1, 1.0 and 5.0 mg/ml) of Si, Ba and Al were stored as controls for the same time periods. After each storage period, the liquids were analyzed regarding Si, Ba and Al concentrations. Induction coupled plasma emission spectroscopy (ICAP 6lE; Thermo Jarrell Ash, Franklin, MA, USA) was used to determine the amount of Si, Ba and Al in the test solutions. The wavelength that
Filler leachability was used for each element was; Si~251.6 nm, Ba~533.6 nm, and Al~309.3 nm. After the samples had been analyzed, an analytical quality control procedure was conducted to determine the reliability of the analysis. This analysis was conducted within the 0 to 250 mg/ml range for the three elements, and each month 10 dierent standards were run to determine the reliability between measured versus true value. At the end of the 12th, 24th, and 36th months, the specimens were sonicated (Biosonic, Whaledent International, NY, USA) for 5 min before removal to remove a precipitate that had built up on the surface of the specimens stored in arti®cial saliva. This was done to free the surface of any barrier to leaching, and make it possible to collect and analyze the precipitate. The reason sonication was done on a yearly rather than a monthly basis was simply because it was ®rst after many months of storage as we discovered the semitransparent precipitate on the surface of the arti®cial saliva stored specimens (1). At that point in time we decided to proceed with a yearly sonication to make it possible to compare the leaching data at least on a yearly basis. Regarding the precipitate, we analyzed that using a Phillips Model 1710 X-ray diraction machine. X-ray diraction analysis was performed with CuKa1 of 1.54056 AÊ operating at 40 kV and 20 mA. The identi®ed peaks were analyzed by using a software package capable of ®nding suitable matches in the databank generated by the JCPDS international center for diraction data. The 36 monthly concentrations were condensed into twelve 3-month data points per sample. It was apparent that the measured standard deviation was approximately proportional to the mean. A log transformation was used to process the data, and to avoid the problem of taking logarithms of 0 mg/ml, a limit of detection was used to replace 0 mg/ml measurements. These limits were 0.1 mg/ml for Al and Si and 0.01 mg/ml for Ba). By using the natural logarithm scale, the parameter estimates from the statistical analysis could be interpreted directly as symmetric relative dierences (7) of geometric
557
means, which are easily converted to the multiplicative eects. However, even if the log transformation had not been used, the statistical conclusions would have been the same. The basic data structures being analyzed were: Each combination of treatments including storage (2 levels; arti®cial saliva and distilled water), resin (2 levels; bisGMA and UEDMA), silane (2 levels; H and NH) and ®ller (2 levels; B and Q) was assigned to a sample. There were 256 samples each with 36 months observations. Since sonication did not change the leaching pattern, the 36 months results were considered as repeated measurements and the basic sampling unit is the sample. SAS GLM procedure was used with sample as a random eect, i.e., sample was nested in the treatment. All the eects were tested by the error at the sample level.
Results The quality control analysis revealed that the analyses followed the relationship Si-conc.~0.991 6 Si-measuredz0.098, Al-conc.~1.0016Al-measuredÿ0.008 and Ba- conc.~0.9956Ba-measured ÿ0.019. The R2 values for both Si and Ba were 40.999 while that of Al was 0.997. The analytical results of standards stored for 1 month in the oven with the composite samples are shown in Table 1. As seen from that table, the measured values are generally 10±20% lower than the standard values. Regarding the leaching samples, the detailed analysis for the silicon (Si) results is given, while only the signi®cant aluminum (Al) and barium (Ba) results are given. Table 2 shows the complete analysis of variance table for Si leaching under the logarithmic transformation. From Table 2, we see that all high-way interactions were not signi®cant, while storage, resin, and ®ller all had signi®cant eects. Also, there were interactions between storage and resin and storage and ®ller. We also found that leaching
Table 1 Analytical results (Mean value¡standard deviation) of standards expressed in mg/ml. These standards had been stored for 1 month at 37³C in either arti®cial saliva or distilled water Standard
Storage medium
Ba
Al
Si
Sample size/element
0.1 0.1 1.0 1.0 5.0 5.0
Arti®cial saliva Distilled water Arti®cial saliva Distilled water Arti®cial saliva Distilled water
0.07¡0.05 0.03¡0.05 1.00¡0.20 0.91¡0.10 4.35¡0.24 4.57¡0.22
0.08¡0.03 0.04¡0.03 0.91¡0.13 0.57¡0.10 4.05¡0.37 4.08¡0.30
0.08¡0.07 0.07¡0.07 0.98¡0.19 0.83¡0.16 4.42¡0.29 4.24¡0.36
n~36 n~36 n~36 n~36 n~36 n~36
558
SoÈderholm et al. Table 2
Analysis of variance table for Si leaching based on SAS GLM procedure. The error sum of squares (SS) is the sample level. All the other SS are SAS Type III SS Source Storage Resin Storage*Resin Silane Storage*Silane Resin*Silane Storage*Resin*Silane Filler Storage*Filler Resin*Filler Storage*Resin*Filler Silane*Filler Storage*Silane*Filler Resin*Silane*Filler Storage*Resin*Silane*Filler Error
DF
Type III SS
F Value
Pr4F
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
2584.5868 23.9077 17.1900 0.1885 0.1710 0.2018 0.0078 538.2827 2.9256 0.3057 0.0041 0.1557 0.0027 0.1037 0.8806
5329.41 49.30 35.45 0.39 0.35 0.42 0.02 1109.94 6.03 0.63 0.01 0.32 0.01 0.21 1.82
0.0001 0.0001 0.0001 0.5335 0.5532 0.5195 0.8994 0.0001 0.0148 0.4280 0.9272 0.5715 0.9402 0.6443 0.1791
241
116.8770
over time can be represented with the simple linear equation y a bt e (1) where y is element leaching in mg/ml/quarter (3-month), a is the intercept, b is the slope measured in mg/ml/quarter, t is time measured in quarter, and e is the error term in ®tting. If b diers signi®cantly from 0, then we can conclude that the leaching rate changes with time. Otherwise, leaching rate is constant in the 36 months. Also, if a diers signi®cantly from 0, then we can claim that this element leaches; otherwise, there may be no leaching. Table 3 presents the storage-resin interaction for Si leaching. For easy interpretation, the data is the original data without log transformation Table 2 shows that arti®cial saliva produced stronger leaching, but there was no signi®cant leaching in distilled water (Table 3). The resin eect and the storage-resin interaction were both small. Though statistically signi®cant dierent in the log data, there were no practically signi®cant dierences. This conclusion is consistent with the huge dierence in the F values of the three eects in Table 2. Table 4 presents the storage-®ller interaction for Si leaching. Again, the distilled water produced no signi®cant leaching, but ®ller B (barium) produced more leaching than Q (quartz) ( p~0.03). Moreover, leaching with ®ller B increased with time. Filler Q had a constant leaching rate. Table 5 shows the eects for Al leaching. Only signi®cant dierences ( p¦0.05) are presented in the table. Detailed analysis (Table 4) shows that only Ba glass-®lled samples stored in arti®cial saliva
Table 3 Storage and resin effect to Si leaching (mg/ml). Symbols a and b are de®ned in Equation 1 as a~slope and b~intercept
BisGMA ( p-value) UEDMA ( p-value)
Arti®cial saliva
Distilled water
a
b
a
b
3.5¡0.34 (0.0001) 3.2¡0.33 (0.0001)
0.14¡0.05 (0.01) 0.18¡0.04 (0.003)
0.13¡0.23 (0.57) 0.37¡0.25 (0.16)
0.074¡0.03 (.04) 0.072¡0.03 (0.06)
produced Al leaching. All the other 3 combinations showed no signi®cant leaching. The results presented in Table 5 show how the dierent variables aected Ba leaching signi®cantly. Detailed analysis showed that Ba leaching is not time dependent. As a consequence, there is no time (t) factor left from Equation 1. Instead, if the average leaching per 3 month (y) is represented as a function of storage and resin, the algebraic expression for y in a Ba glass-®lled sample is: y~2.05 z 1.06 6 storage ÿ 0.21 6 resin z e (2) where storage~1 if stored in arti®cial saliva, and ~0 for storage in distilled water, resin~1 for bisGMA and ~0 for UEDMA. The p-values for the intercept, storage and ®ller coecients are respectively 0.0001, 0.0001 and 0.045. The standard deviation for e is 0.583. The computerized X-ray diraction analyses suggested that the Ba-containing composites were covered with precipitates consisting of Ba or Ca silicide (compounds 31-0162 and 32-0180) or Ba or Si phosphide (databank ®le numbers 23-0828, 27-0609 and 44-1124), while the quartz containing
Filler leachability
559
Table 4 Storage and ®ller effect to Si and Al leaching (mg/ml). Symbols a and b are defined in Equation 1 as a~slope and b~intercept Arti®cial saliva
Distilled water
a
b
a
b
Si values below Filler~B ( p-value) Filler Q ( p-value)
4.14¡0.47 (0.0001) 2.79¡0.41 (0.0001)
0.32¡0.06 (0.0005) 0.00¡0.05 (0.99)
0.38¡0.34 (0.29) 0.12¡0.14 (0.42)
0.11¡0.04 (.04) 0.04¡0.02 (0.09)
Al values below Filler~B ( p-value) Filler~Q ( p-value)
0.53¡0.10 (0.0004) 0.011¡0.032 (0.72)
0.031¡0.014 (0.05) 0.005¡0.004 (0.24)
0.016¡0.030 (0.60) 0.017¡0.016 (0.31)
0.002¡0.004 (0.63) 0.000¡0.002 (0.95)
Table 5 Analysis of variance for Al and Ba leaching (mg/ml). See Table 2 for description. Only signi®cant (p¦0.05) are presented Source
DF
Type III SS
F Value
Pr4F
Al values below Storage Filler Storage*Filler Error
1 1 1 241
728.61472 653.11219 542.30300 76.052
2308.88 2069.62 1718.48
0.0001 0.0001 0.0001
Ba values below Storage Resin Filler Error
1 1 1 241
91.6169 13.1675 5207.3732 76.052
59.18 8.51 3363.96
0.0001 0.0039 0.0001
composites were covered by Ca silicide (databank ®le number 26-0324). Discussion By comparing the analytical results of the standards (Table 1) with the values one would expect to ®nd after one month, it is reasonable to assume that the analyzed leaching values underestimate the leaching values. The underestimated values do not seem to be related to an instrumental error since the quality control results show that the instrumental readings are almost identical with the values of fresh standards. Thus, the lower values of the control samples measured after one month of storage suggest that these samples have been depleted during storage regarding their original concentrations. Such depletion could be due to that some ions form complexes with other ions and precipitate or that the ions interact with the vials. Consequently, the results we presented have a tendency to underestimate the amount of elements leached from the sample. Since the
leaching process is continuous, we did not make any attempt to correct for these changes. Such a correction would most likely be unreliable because we are still lacking information about all leaching and precipitation mechanisms that are involved in the process. Despite inherent analytical problems, this study shows that Ba-containing ®ller particles leach more Si than quartz particles. These ®ndings support the hypothesis that glass modi®ers often lower the hydrolytic stability of ®ller particles (3, 4). The results also show that the ®ller eect is less pronounced than the eect of storage medium, even if the ®ller eect is substantial. The amounts of leached Si from the two ®ller types was such that Ba-containing ®ller leached less Si after storage in distilled water compared to quartz-containing composites stored in arti®cial saliva. Leaching accelerates slightly during the ®rst few months in arti®cial saliva, implying that the autocatalytic reaction proposed by CHARLES (2) needs a certain OH concentration before it reaches its fullest potential. Signi®cantly more ®ller material was leached in arti®cial saliva than in distilled water, suggesting that the oral environment is likely to cause more pronounced ®ller degradation than indicated by storage in distilled water. This ®nding is important, because it implies that in vitro wear studies, conducted after composites have been stored in distilled water, will underestimate the wear rate of samples that should rather be stored in arti®cial saliva. The higher ®ller leaching found in arti®cial saliva suggests that an ion exchange mechanism occurs at the ®ller surface. Thus, when the initial leaching of positive ions occurs, the remaining Si-O-Si structure becomes negatively charged. The negatively charged ®ller particle surfaces will then reduce the amount of cations leaving the ®ller surfaces. However, if positive ions, present in the storage
560
SoÈderholm et al.
medium, diuse through the matrix and interact with the negatively charged Si-O-Si surfaces, it will become easier for some of the original ®ller cations to leave the ®ller particles. Some of these cations will then diuse through the matrix into the storage medium and cause an increase in detectable elements. Such an ion exchange mechanism implies that the original ®ller ions are substituted by ions from the solution, and that these ions may ``heal'' the otherwise degraded ®ller surface. Therefore, future studies are needed to determine whether there is a correlation between ®ller leachability in arti®cial saliva and wear of composites. Before such studies have been conducted, it is impossible to know if the ®ndings presented in this study are as important as they might imply. However, since leaching was an order of magnitude greater in arti®cial saliva than in distilled water, the above ®ndings imply that leaching from ceramic materials, particularly ®ller particles of composites, may be more pronounced clinically then earlier believed. Another dierence between this study and a clinical situation was that we shook our specimens only weekly, while clinically composites would be cycled. The X-ray diraction analysis after 1, 2 and 3 yr of the precipitate formed on the specimens stored in arti®cial saliva was inconclusive. The overlapping nature of several spectra made it impossible to conclude which precipitates that had formed. The computerized identi®cation reveals the most likely compounds, but cannot exclude the possibility that some of the multiple peaks belong to a compound that cannot be identi®ed by the program. An interesting aspect is that the formed precipitates may contribute to a reduction in leakage around dental composite restorations as they precipitate over time and seal contraction gaps. However, considering the uncertainties regarding composition and sealing ability of these precipitates, further research is needed regarding these precipitates and their sealing abilities. The conclusion drawn from this study is that Ba-containing ®ller particles leach more than
quartz ®ller particles, and that leaching from composites is faster in arti®cial saliva than in distilled water. That ®nding can be explained by assuming that a negatively charged silica network surface is formed because of ®ller leaching, and that the negatively charged surface restricts the release of cations. The presence of cations, transferred from the arti®cial saliva, will neutralize these charges and facilitate cation release from the silica surface. From a clinical viewpoint, the above ®ndings suggest that there is a continuous release of elements from the ®ller particles, and that there are no signs that the element release decreases over a 3-yr time period. Because of potential eects on both clinical wear performance and biological behavior caused by elements being released, more research is needed to determine the eects of dierent saliva compositions on the ®ller leachability of dental composites. Acknowledgements ± We would like to thank the Analytical Research Laboratory (Dr. A. E. Hanlon, Mr. J. M. Bartos and Mr. D. L. Moon), Gainesville, FL, for their excellent support with the ICAP analyses. We would also like to thank NIH/ NIDCR for supporting us with Grant DE 09292.
References 1. SOÈDERHOLM K-JM, MUKHERJEE R, LONGMATE J. Filler leachability of composite stored in distilled water or arti®cial saliva. J Dent Res 1996; 75: 1692±1699. 2. CHARLES RJ. Static fatigue of glass I. J Appl Phys 1958; 29: 1549±1553. 3. SOÈDERHOLM K-JM. Leaking of ®llers in dental composites. J Dent Res 1983; 62: 126 ±130. 4. SOÈDERHOLM K-JM. Degradation of glass ®ller in dental composites. J Dent Res 1981; 60: 1867±1875. 5. SOÈDERHOLM K-JM, ZIGAN M, RAGAN M, FISCHLSCHWEIGER W, BERGMAN M. Hydrolytic degradation of dental composites. J Dent Res 1984; 63: 1248±1254. 6. ARVIDSON K, JOHANSSON EG. Galvanic currents between dental alloys in vitro. Scand J Dent Res 1985; 93: 467± 473. 7. TOÈRNQVIST L, VARTIA P, VARTIA YO. How should relative changes be measured? Am Stat 1985; 39: 43± 46.
