_______________________________________________________________________________________________________________________________________________________________

Research Article_______________________________________________________________________________________________________________________________________________________________

The effect of toothbrushing on surface gloss of resin composites DORIEN LEFEVER, MSC, NIKOLAOS PERAKIS, DMD, MIGUEL ROIG, PHD, IVO KREJCI, DMD, PHD & STEFANO ARDU, DMD, PHD

ABSTRACT: Purpose: To determine the changes in surface gloss of different composite materials after laboratory toothbrushing simulation. Methods: 36 specimens were fabricated for each material and polished with 120-, 220-, 500-, 1200-, 2400- and 4000-grit SiC abrasive paper, respectively. Gloss measurements were made with a glossmeter (Novocurve) prior to testing procedures and then subjected to simulated toothbrushing for 5, 15, 30 and 60 minutes by means of an electric toothbrush with a pressure of 2N while being immersed in a 50 RDA toothpaste slurry. Four supplementary samples per group were analyzed under SEM immediately after polishing procedures and four samples after 60 minutes simulated toothbrushing in order to evaluate the causes of the gloss decrease. The tested resin composite materials were Filtek Supreme XTE, Durafill, HRi Enamel Plus, Miris 2, Empress Direct, Venus Diamond, Gradia Direct, Clearfil Photo Posterior and G-aenial. Natural enamel represented the control group. Statistical analysis was performed using Kruskal Wallis and Tukey post-hoc test, with a level of significance set at 0.05. Results: Resin composite initial gloss values ranged from 68.9 to 100.5 at baseline to 10.6 to 62.6 after 1 hour of brushing. Highest gloss values were obtained by Filtek Supreme XTE, followed by Empress Direct and Durafill. Lowest values were obtained by Clearfil Photoposterior, Miris 2, Enamel HRi and Venus Diamond. Natural enamel was the only substrate to maintain its gloss throughout the brushing procedure (110.4 after 60 minutes). SEM analysis revealed different patterns of surface degradation depending on the composite material. (Am J Dent 2012;25:54-58).

CLINICAL SIGNIFICANCE: None of the resin composites performed as well as natural enamel. Some restorative materials exhibited a decreased gloss due to toothbrushing, which might result in an esthetic problem. !: Dr. Dorien Lefever, Division of Cariology & Endodontology, Dental School, University of Geneva, Geneva, Switzerland. E-!: Dorien.Lefever@unige.ch

Introduction The esthetic restoration of the anterior dentition represents one of the greatest challenges in current daily practice. Continu-ous improvements regarding the esthetic properties of resin composites such as color match, translucency and opalescence have led to natural looking restorations. As a consequence, these resin composites are increasingly used as an alternative to porcelain-fused-to-metal (PFM) crowns and ceramic veneers for the restoration of severely compromised anterior teeth. In this indication, not only color match, translucency and opales-cence,1 but also surface gloss is of paramount importance. Surface quality of restorations is in fact one of the important factors that determine their clinical success. A smooth surface can improve longevity and esthetics of restorations by reducing plaque accumulation and surface staining, allowing successful mimicking of the tooth’s natural appearance.2,3 Directly related to surface quality is also the ability of the material to reflect direct light. This optical phenomenon is defined as gloss or reflective capacity. Differences in gloss between a restoration and sur-rounding enamel are clinically relevant as the human eye can easily detect differences in gloss even if their colors are matched. On the other hand, high gloss reduces the effect of a color difference, since the color of reflected light is more predominant than the color of the underlying composite material.4 Modern resin composites can achieve a high luster if appropiate polishing procedures are used.2 Gloss is an attribute of visual appearance that originates from the geometrical dis-tribution of light reflected by the surface, and is directly influ-enced by the surface roughness.3 However, the high gloss level obtained immediately after polishing procedures is not clinical-ly stable over time, leading to a matte surface. Previous stu-

dies5-11 that investigated the influence of toothbrushing on sur-face gloss and surface roughness of resin composites showed a decrease in gloss and increase in roughness of various resin com-posites, as well as differences between the observed materials. This decrease in gloss is the result of degradation due to mechanical and/or chemical interaction with the oral environ-ment, resulting in changes in surface gloss and therefore deteriorated esthetics at long term. This surface degradation may be due to several factors: wear of fillers, degradation of the resin matrix or weakening of resin-filler bonding. These three factors lead to a roughening of the surface which is the main cause of a gloss decrease. Clinically, this kind of superficial degradation can cause esthetic problems especially in patients who present a high lip line. In this case, the different refraction index between natural tooth and resin composite on upper anterior teeth free of saliva can cause a severe esthetic problem. Newly developed resin composites have no information available on their surface gloss. Therefore, the present study evaluated the gloss behavior of newly developed resin com-posites with different filler range and mechanical characteristics such as Enamel HRi,a Gradia Direct,b Empress Directc and G-aenial,b (Table 1) immediately after polishing and after tooth-brushing simulation, compared to natural enamel and reference materials. The null hypothesis was that surface gloss of resin composites was not influenced by laboratory toothbrushing simulation.

Material and Methods Nine resin composites were selected, i.e. eight indicated by the manufacturer as anterior restorative materials and one for posterior restorations (negative control) (Table 1). Natural enamel was added as a positive control.

American Journal of Dentistry, Vol. 25, No. 1, February, 2012 Toothbrushing and composite surface gloss 55

Table 1. Description of the tested materials. _______________________________________________________________________________________________________________________________________________________________________________________________________________

Filler mean particle, Product Composite classification Filler type size and range Matrix composition Abbreviation _______________________________________________________________________________________________________________________________________________________________________________________________________________

Empress Direct Inhomogeneous microhybrid Ba-Al-fluorosilicate glass; Mean 550 nm Dimethacrylate EMP with homologous splinters barium glass filler (0.4 µm), range 150 nm-0.4µm mixed oxide (150 nm) Miris 2d Inhomogeneous microhybrid Silanized barium glass, Mean 0.6 µm Methacrylate MIR with homologous splinters amorphous silica splinters Range 0.02 µm-2.5 µm 15-20 µm (mean 1 µm) G-aenial Inhomogeneous microhybrid Strontium and lanthanoid Range 16 nm-17 µm UDMA with homologous splinters fluoride pre-polymerized Dimethacrylate co-monomer GAE fillers (16-17 µm) Filler 850 nm silica 16 nm fumed silica Gradia Direct Inhomogeneous microhybrid Silica 0.85 µm Mean 850 nm UDMA with heterologous splinters Prepolymerized filler Range 20 nm-20 µm Dimethacrylate co-monomer GRA Filtek Supreme Inhomogeneous microhybrid Silica filler (20 nm) Mean 0.6 µm bisGMA, UDMA, SUP XTEe with aggregated particles Zirconia filler 4-11 nm Range 20 nm-10 µm TEGDMA, bis-EMA Aggregated zirconia/silica cluster filler (0.6-10 µm) Enamel HRi Homogeneous microhybrid Glass filler (mean 1.0 µm) Mean 0.9 µm Diurethandimethacrylate HRI nano zirconium oxide Range 20 nm–2 µm bisGMA particles 20 nm 1,4-butandioldimethacrylate Durafillf Microfilled inhomogeneous Silicon dioxide (0.02-0.07µm) Mean 0.04 µm UDMA DUR splinter polymer (2-20 µm) Range 0.02-20 µm Venus Diamondf Fine hybrid Ba-Al fluoride glass Mean 0.6 µm TCD-DI-HEA, UDMA DIA nano-particles 5 nm-20 µm Range 5 nm-20 µm Clearfil Photo Coarse hybrid Silanated silica Mean 1-2 µm bisGMA CPP Posteriorg silanated barium glass TEGDMA silanated colloidal silica urethane tetramethacrylate _______________________________________________________________________________________________________________________________________________________________________________________________________________

Specimen preparation - Thirty-six disc-shaped specimens mea-suring 10 mm in diameter were made of each of nine com-posites (Table 1), resulting in a total of 324 samples, by covering the resin composite with a transparent matrix strip and gently pressing it with a glass slide to the thickness of 1 mm. The resin composites were light cured for 40 seconds from a distance of 1 mm by using a curing light (L.E.Demetron IIh) at a light intensity of 1100 mW/cm2 as measured with a radiome-ter (L.E.D.radiometerh). One additional group, made of 36 slices of natural tooth enamel (ENML) obtained from freshly extracted human anterior teeth, was added to the restorative material groups as a positive control. This enamel was sub-jected to the same polishing protocol and the same testing procedures as the composite materials. The surface of all speci-mens was then polished with 120-, 220-, 500-, 1200-, 2400- and 4000-grit SiC abrasive paper. Polishing was performed for 60 seconds for each grit of abrasive paper under water cooling at a constant force of 2 N.7 After storage in artificial saliva at 37°C for 24 hours, initial surface gloss measurements were made for each specimen. Gloss measurements - Surface gloss was measured by means of a glossmeter (Novo-Curvei) according to Heintze et al.12. This device measures the amount of light reflected from the surface of an object, which is then translated into a numerical scale. The measuring principle of this device is based on a light beam that strikes the object at a 60° angle. The intensity of the reflect-ed light is measured and compared to the reference value. The device has a measuring window of 2 mm x 8 mm over which the specimen is placed and then covered with a black film container to avoid external light exposure during the measure-ment. Each time before a new measurement was made, the

glossmeter was calibrated by comparing the results with a calibration plate provided by the manufacturer, which has a reference value of 94.0 and by checking the zero point to exclude negative values. Simulated toothbrushing - After baseline gloss measurements, each specimen (n=36 per material) was then subjected to consecutively 5, 15, 30 and 60 minutes of brushing with an electric toothbrush (Triumph Professional Carej) fixed on a cus-tom made holder, applying a standardized force of 2 N. The specimens were immersed in a toothpaste slurry consisting of 3g 50 RDA toothpaste (Signalk) mixed with 0.3 mL of distilled water. After each treatment the slurry was renewed and speci-mens were thoroughly cleaned of any treatment material resi-due both manually and in an ultrasonic bath with distilled water for 10 minutes in order to remove eventual smear layer created on their surface. Surface gloss measurements were made subsequently using the NovoCurve device as described above. SEM evaluation - Four specimens of each material before and four samples after 1 hour of simulated toothbrushing were gold sputtered and then analyzed by scanning electron microscopy (SEM Phillips XL 20l) in order to investigate possible surface changes. Statistical analysis - Statistical analysis was performed with SPSS 17.0 software. As the distribution of data was not normal (Kolmogorov-Smirnov test) non parametric methods were used. A Wilcoxon signed-rank test was run for each paired group, i.e. before vs. after treatment (P= 0.05). Furthermore, to detect whether the results were material dependent, a Kruskal-Wallis test was run. Tukey post-hoc test was used to detect differences among group means.

56 Lefever et al American Journal of Dentistry, Vol. 25, No. 1, February, 2012 Table 2. Mean gloss values at baseline and after each brushing cycle of 5, 15, 30 and 60 minutes (GU), !GU and the group of statistical significance of each material. ________________________________________________________________________________________________________________________________________________________________________________________________________________________

Baseline After 5 minutes After 15 minutes After 30 minutes After 60 minutes Statistical significance Material (± SD) (± SD) (± SD) (± SD) (± SD) "GU (after 60 minutes) ________________________________________________________________________________________________________________________________________________________________________________________________________________________

HRI 88.2 (±7.1) 29.0 (±12.3) 15.9 (± 5.6) 12.8 (±2.9) 10.6 (±2.2) 77.7 F DIA 82.5 (±4.3) 53.3 (±12.8) 17.2 (± 8.7) 12.1 (±5.7) 12.5 (±5.5) 70.1 FE GAE 79.5 (±4.1) 26.7 (± 7.9) 18.4 (± 7.4) 16.6 (±5.1) 13.4 (±4.8) 66.1 FE CPP 68.9 (±6.2) 33.1 (± 9.6) 22.1 (± 6.9) 19.0 (±8.2) 14.8 (±5.7) 54.2 E MIR 81.1 (±4.7) 53.3 (±11.7) 42.6 (± 9.0) 29.6 (±9.0) 18.9 (±5.9) 62.2 D GRA 78.7 (±2.9) 50.0 (±12.3) 31.8 (±11.1) 21.6 (±6.5) 20.8 (±7.2) 58.0 D DUR 84.2 (±5.8) 77.5 (± 6.0) 67.9 (± 6.9) 56.8 (±7.6) 48.0 (±6.5) 36.2 C EMP 86.7 (±3.8) 83.6 (± 5.3) 73.1 (± 7.9) 65.1 (±7.3) 51.1 (±7.4) 35.6 C SUP 100.5 (±4.2) 83.2 (± 6.3) 74.0 (± 7.1) 68.2 (±8.9) 62.6 (±7.5) 37.9 B ENML 113.2 (±4.0) 111.3 (± 6.3) 116.6 (± 2.1) 111.9 (±5.8) 110.4 (±1.4) 2.8 A ________________________________________________________________________________________________________________________________________________________________________________________________________________________

*"GU is the difference in gloss values between the initial and the final values. It is calculated according to the following formula: GUinit - GUfin where init and fin are the respective values at the baseline and at the end of the experimental phase. Fig. 1. Boxplot of GU values of all materials immediately after polishing.

Results Initial gloss values of each composite material and changes from baseline after each cycle of brushing are shown in Table 2. Resin composite gloss at baseline ranged from 68.9 (CPP) to 100.5 (SUP) GU (gloss units). After 1 hour of toothbrushing simulation gloss values ranged from to 10.6 to 62.6 (SUP) GU. The highest !GU was detected for HRi, followed by DIA, GAE and CPP. EMP showed the lowest !GU. Natural enamel provided the best baseline and post-treatment gloss values, resulting in the lowest !GU when compared to the tested resin composite materials (Figs. 1, 2). SEM images of the tested materials before and after 60 minutes of simulated toothbrushing are shown in Figs. 3A-D.

Discussion A visual gloss evaluation may include many errors due to subjectivity. Therefore, a numeric quantitative approach such as a glossmeter device is mandatory to achieve an objective evaluation. Furthermore, the glossmeter used in this study (Novo-Curve) was specifically chosen because it is able to measure surface gloss of a restricted area. The composites evaluated in this study were chosen as they represented a group of resin composites relatively recently introduced into the market. They were compared to natural enamel (positive control) and Clearfil PhotoPosterior (nega-tive control). According to the resin composite classification of

Fig. 2. Boxplot of GU values of all materials after 1 hour of simulated toothbrushing.

Ardu et al,13 the investigated materials represented different classification categories, in order to investigate the possible influence of the composition of the resin composite on the surface gloss. Natural enamel was chosen as positive control because it is supposed to be the ideal natural substrate. Clearfil PhotoPosterior on the other hand was considered as negative control due to its mean filler size which suggests indication for posterior restorations only.13 The samples were prepared under standardized conditions. Pre-roughening of the specimens was found necessary to eliminate voids in the external layer of the composite samples. In most studies14-18 pre-roughening was performed either with diamond or tungsten carbide burs to mimic clinical procedures. However, Heintze et al12 claimed that pre-roughening with dia-mond burs resulted in an inhomogeneous surface texture and consequently in increased scattering of the results. In order to obtain a standardized force, a calibration session was initiated prior to the application of the polishing system, using an elec-tronic laboratory scale to measure the force applied (2 N) during the polishing steps.19 To overcome the possible influ-ence of the type of illumination and angle of the observer, a glossmeter with 60° angle of illumination was used for all measurements.20 As previously reported,7,21 toothbrush abrasion of composite materials varied according to the type of composite, type of toothpaste22 and the nature of the toothbrush employed.23 There-

American Journal of Dentistry, Vol. 25, No. 1, February, 2012 Toothbrushing and composite surface gloss 57

Fig. 3A-D. SEM images of Clearfil PhotoPosterior and Filtek Supreme XTE before and after 60 minutes of simulated toothbrushing. A. Filtek Supreme XTE before simulated toothbrushing. B. Filtek Supreme XTE after 60 minutes toothbrushing. C. Clearfil PhotoPosterior before simulated toothbrushing; D. Clearfil PhotoPosterior after 60 minutes of toothbrushing. The surface integrity was substantially maintained. Only small superficial defects are present, which might be due to loss of hard zirconium particles. Decrease in surface quality with resin loss and appearance of macrofillers can be observed.

fore, absolute values of the results of this study cannot be com-pared with other reports. In the present study, the specimens were brushed during 60 minutes, which may correspond to the amount of toothbrushing that is carried out over a period of 2 years, if it is assumed that the ideal brushing time is 120 seconds three times a day which is equal to a tooth surface brushing of 6 seconds a day. However, various studies showed that the actual mean brushing, even if variable, is about 120 seconds per day, which corresponds to 2 seconds per day per tooth surface.24,25 Therefore, our brushing simulation may correspond to a clinical simulation of 6 years. After polishing procedures at baseline, natural tooth showed the highest gloss. This highlights the fact that, so far, no resin composite is able to mimic reflectivity of the natural tooth. However, Filtek Supreme XTE, a micro-hybrid with aggre-gated clusters of SiO2 and ZrO nano particles, showed the highest luster among the tested materials (Fig. 3A). This could be due to the fact that only microfillers are present in this material. Enamel HRi, Empress Direct and Durafill showed all a higher reflectivity than the coarse hybrid but lower than Filtek Supreme XTE. All three restorative materials contained small filler particles (20 nm-0.4 µm). According to Lee et al26 not only the filler size, but the resin matrix system as well as the shape of the fillers influence initial gloss of materials. Light reflectivity seems, therefore, to be related to mean filler size and to the homogeneity of the filler-matrix complex. Higher filler size and lower homogeneity of the filler-matrix complex result in less light reflectivity. The present study clearly showed that, except for the natural tooth group, the surface gloss of all the materials was signifi-

cantly reduced by simulated toothbrushing. This phenomenon, as reported in other studies,9,11 was material dependent. Gloss decrease is influenced by the resin matrix, filler particles and the silanization between both. Whenever fillers are much harder than the surrounding resin, this may result in a rough surface after toothbrushing simulation. The abrasion of the softer resin might cause a lack of support of the filler, which finally de-taches from the matrix, resulting in a concavity in the surface (Photohole effect). This phenomenon may be the cause of the poor gloss values observed for Clearfil PhotoPosterior (Fig. 3D) and Venus Diamond. On the other hand, when the material con-sists of softer fillers, such as Filtek Supreme XTE, the abrasion will occur in a more uniform way, therefore resulting in a smoother surface and thus a higher gloss (Fig. 3B). Silanization, on the other hand, is another factor of para-mount importance.27 A higher quality of silanization might explain the lower !GU for Empress Direct when compared to other materials of the same family (inhomogeneous microhybrid resin composite with homologous splinters), as claimed by the manufacturer. Within the limitation of this study, natural enamel demon-strated to be the best substrate in respect to gloss stability and behavior throughout toothbrushing simulation. In fact, no artifi-cial material showed comparable behavior. The null hypothesis was therefore rejected. In future studies, the influence of specific parameters and parameter combinations should be evaluated in order to determine their exact influence on gloss. a. Micerium, Avegno, Italy. b. GC, Leuven, Belgium.

58 Lefever et al c. Ivoclar Vivadent, Schaan, Liechtestein. d. Coltène Whaledent, Alstätten. e. 3M Espe, St. Paul, MN, USA. f. Heraeus Kulzer, South Bend, IN, USA. g. Kuraray, Tokyo, Japan. h. Kerr, Orange, CA, USA. i. Rhopoint Instrumentation Ltd., Bexhill on Sea, UK. j. Oral B Braun GmbH, Kronberg/Ts., Germany. k. Unilever, Thyngen, Switzerland. l. Philips, Eindhoven, The Netherlands. Acknowledgements: To Mrs. Marie Claude Reymond for her help with the SEM images. Disclosure statement: The authors declared no conflict of interest. The manufacturers generously provided the products tested in this study. Dr. Lefever is Assistant, and Dr. Krejci is Head, Department of Cariology and Endodontology; Dr. Ardu is Head of Programming and Treatment Planning Unit, University of Geneva, Geneva, Switzerland. Dr. Perakis is in private practice, Bologna, Italy. Dr. Roig is Head, Department of Restorative Dentistry, International University of Catalunya, Barcelona, Spain.

References 1. Vanini L. Light and color in anterior composite restorations. Pract

Periodontics Aesthet Dent 1996;8:673-682. 2. Lu H, Roeder LB, Lei L, Powers JM. Effect of surface roughness on stain

resistance of dental resin composites. J Esthet Restor Dent 2005;17:102-108. 3. van Dijken JW, Ruyter IE. Surface characteristics of posterior composites

after polishing and toothbrushing. Acta Odontol Scand 1987;45:337-346. 4. Keyf F, Etikan I. Evaluation of gloss changes of two denture acrylic resin

materials in four different beverages. Dent Mater 2004;20:244-251. 5. Heintze SD, Forjanic M, Ohmiti K, Rousson V. Surface deterioration of

dental materials after simulated toothbrushing in relation to brushing time and load. Dent Mater 2010;26:306-319.

6. Heintze SD, Forjanic M. Surface roughness of different dental materials before and after simulated toothbrushing in vitro. Oper Dent 2005;30:617-626.

7. Ardu S, Braut V, Uhac I, Benbachir N, Feilzer AJ, Krejci I. Influence of mechanical and chemical degradation on surface gloss of resin composite materials. Am J Dent 2009;22:264-268.

8. Neme AL, Frazier KB, Roeder LB, Debner TL. Effect of prophylactic polishing protocols on the surface roughness of esthetic restorative materials. Oper Dent 2002;27:50-58.

9. Neme AM, Wagner WC, Pink FE, Frazier KB. The effect of prophylactic polishing pastes and toothbrushing on the surface roughness of resin composite materials in vitro. Oper Dent 2003;28:808-815.

10. dos Santos PH, Consani S, Correr Sobrinho L, Coelho Sinhoreti MA.

American Journal of Dentistry, Vol. 25, No. 1, February, 2012 Effect of surface penetrating sealant on roughness of posterior composite

resins. Am J Dent 2003;16:197-201. 11. Tanoue N, Matsumura H, Atsuta M. Wear and surface roughness of current

prosthetic composites after toothbrush/dentifrice abrasion. J Prosthet Dent 2000;84:93-97.

12. Heintze SD, Forjanic M, Rousson V. Surface roughness and gloss of dental materials as a function of force and polishing time in vitro. Dent Mater 2006;22:146-165.

13. Ardu S, Braut V, Uhac I, Benbachir N, Feilzer AJ, Krejci I. A new classification of resin-based aesthetic adhesive materials. Coll Antropol 2010;34:1045-1050.

14. Lu H, Roeder LB, Powers JM. Effect of polishing systems on the surface roughness of microhybrid composites. J Esthet Restor Dent 2003;15:297-303.

15. Roeder LB, Powers JM. Surface roughness of resin composite prepared by single-use and multi-use diamonds. Am J Dent 2004;17:109-112.

16. Kaplan BA, Goldstein GR, Vijayaraghavan TV, Nelson IK. The effect of three polishing systems on the surface roughness of four hybrid composites: A profilometric and scanning electron microscopy study. JProsthet Dent 1996;76:34-38.

17. Hoelscher DC, Neme AM, Pink FE, Hughes PJ. The effect of three finishing systems on four esthetic restorative materials. Oper Dent 1998;23:36-42.

18. Tate WH, Powers JM. Surface roughness of composites and hybrid ionomers. Oper Dent 1996;21:53-58.

19. Wilder AD, Jr., Swift EJ, Jr., May KN, Jr., Thompson JY, McDougal RA. Effect of finishing technique on the microleakage and surface texture of resin-modified glass ionomer restorative materials. J Dent 2000;28:367-373.

20. Da Costa J, Ferracane J, Paravina RD, Mazur RF, Roeder L. The effect of different polishing systems on surface roughness and gloss of various resin composites. J Esthet Restor Dent 2007;19:214-224.

21. Kanter J, Koski RE, Martin D. The relationship of weight loss to surface roughness of composite resins from simulated toothbrushing. J Prosthet Dent 1982;47:505-513.

22. Goldstein GR, Lerner T. The effect of toothbrushing on a hybrid composite resin. J Prosthet Dent 1991;66:498-500.

23. Wictorin L. Effect of toothbrushing on acrylic resin veneering material. II. Abrasive effect of selected dentifrices and toothbrushes. Acta Odontol Scand 1972;30:383-395.

24. Macgregor ID, Rugg-Gunn AJ. Toothbrushing duration in 60 uninstructed young adults. Community Dent Oral Epidemiol 1985;13:121-122.

25. Saxer UP, Barbakow J, Yankell SL. New studies on estimated and actual toothbrushing times and dentifrice use. J Clin Dent 1998;9:49-51.

26. Lee YK, Lu H, Oguri M, Powers JM. Changes in gloss after simulated generalized wear of composite resins. J Prosthet Dent 2005;94:370-376.

27. Peutzfeldt A. Resin composites in dentistry: The monomer systems. Eur J Oral Sci 1997;105:97-116.