The gingival location of secondary caries is by far the most common location of clinical restoration bonding failure, and occlusal secondary caries is found two times more frequently on resin-based composite restorations than on amalgam restorations.⁹ Secondary caries is the most common reason for replacement of resin-based composite restorations.^10–12 Under-filled or leaky margins at the cervical margin of Class II restorations can lead to an invasion of Streptococcus mutans¹¹ and the onset of secondary caries.^13,14 Gaps at the gingival margin between the composite and the tooth are frequently found.^15,16 This interface has been identified as the weak link in most restorative procedures.¹⁷ If the demineralized dentin at the gingival margin is not hybridized with adhesive resin, it can be vulnerable to hydrolytic breakdown and be susceptible to penetration by bacterial enzymes or other toxic substances.^8,18 This compromises the integrity of the dentin-adhesive bond and, ultimately, the restoration.^14,19

Many clinical problems of dentinal hypersensitivity, microleakage and recurrent caries are related to dentinal permeability. The density of tubules 1.0 millimeter above the cementoenamel junction, or CEJ, at the gingival cavity wall is 49 percent greater than that at the axial wall.²⁰ Compared with superficial dentin, deeper dentin has more tubules, and the tubule openings are larger.^6,21 Close to the dentinoenamel junction, or DEJ, under dry conditions, exposed tubules account for 1 percent of the total surface area; these values in deep dentin close to the pulp account for 22 percent of the surface area. When dentin is acid-etched, these values can increase to 13 percent at the DEJ and 34 percent close to the pulp.⁶ Cervical dentin also is 3.6 times more permeable than occlusal dentin in young adults.²¹ Greater amounts of fluid may flow from the pulp, and this pulpal fluid could compromise the adaptability of the restorative material placed at the cut surface of the dentin. This process could be especially detrimental in deeper preparations, in which the tubular diameter is much greater.²¹ This excessive surface moisture may result in voids at the resin-dentin interface.⁸ This has been described as "the overwet phenomenon."²² Extreme conditions of surface wetness or dehydration should be avoided.⁸ Failure of resin-based composite to bond to teeth at the dentin gingival margin may be due to a combination of factors such as the configuration factor (C-factor),^23–25 polymerization shrinkage^16,26 and wetness of dentin.^22,27–30 Most data on dentinal bonding have been obtained from laboratory studies, and no studies have compared the dentin microtensile bond strength of adhesive/resin-based composite bonded to gingival versus axial cavity walls under in vivo and in vitro conditions in Class II restorations.

We undertook a study to determine if the microtensile bond strength of adhesive bonded to axial and gingival dentin differs. We also aimed to determine whether the microtensile bond strength of adhesive to dentin placed under in vitro and in vivo conditions differs in human teeth.

	SUBJECTS, METHODS AND MATERIALS

TOP ABSTRACT SUBJECTS, METHODS AND MATERIALS RESULTS DISCUSSION SUMMARY REFERENCES

 
The research design was a 2 x 2 (condition—in vivo versus in vitro—versus location—axial versus gingival) controlled clinical trial. We used the results from a pilot study, the conditions in which were similar to the in vitro test condition, to determine sample size.³¹ A sample size of 18 per cell (N = 72) met the criteria for detecting a mean difference of 25 percent, with = .05 and power = 80 percent. Six microtensile samples from each tooth could be obtained, three from the mesial box and three from the distal box. Therefore, 12 teeth were required. We recruited subjects from an orthodontic school of dentistry clinic by means of brochures until we obtained enough samples.

Subjects. To be accepted for the study, a subject had to have two or four premolars planned for extraction to make space for orthodontic treatment. The premolars had to be on opposite sides of the same arch. Subjects had to be between the ages of 12 and 30 years, with no interproximal carious lesions, no previous restorations in the vital tooth and no periapical or periodontal pathology. They had to be able to follow instructions and be in good oral and general health. They could not have a heart condition requiring prophylactic antibiotics, allergies to products or medications used in the study or had tooth bleaching within the preceding four weeks. They could not be taking blood thinners and could not be diabetic. Their teeth had to be in occlusion with the opposing teeth. Each subject or his or her guardian signed an informed consent approved by the institutional review board of the University of Missouri–Kansas City.

Methods. If a subject had four premolars to be extracted, two (one maxillary and one mandibular) teeth received in vivo restorations and the other two on the opposite side of the arch received in vitro restorations after the teeth were extracted. We randomly assigned treatment to teeth. One of the investigating clinicians cleaned the teeth with scalers and pumice to remove any plaque and debris. The subject received an injection of local anesthetic containing 1:100,000 epinephrine. Teeth were isolated with a rubber dam. The clinician prepared the tooth to receive a separate mesial and distal box preparation separated by an intact isthmus. Each preparation was cut with a new 169 carbide bur (Brasseler USA, Savannah, Ga.) and refined with a 56 carbide bur (Brasseler USA). The clinician broke the contact interproximally to a depth of approximately 4.0 mm. He made an attempt to keep the gingival margin 1.0 mm above the CEJ. The buccolingual width of the preparation was approximately 5.0 mm, and the axial depth was 1.5 to 2.0 mm. Enamel margins were not beveled. The clinician used a clear matrix band and a translucent plastic interproximal wedge (Premier Dental Products, King of Prussia, Pa.) to restore contact.

We used a halogen light-curing unit (XL 3000, 3M ESPE, St. Paul, Minn.) to cure all specimens. We measured light intensity before and after placement of the restoration using a light meter (Demetron, Kerr, Orange, Calif.) to confirm that a minimum light intensity of 500 milliwatts per square centimeter was present within three seconds. All materials were applied according to manufacturer’s directions.

The clinician acid-etched the preparation with a 35 percent phosphoric acid gel for 15 seconds on the enamel and dentin. The acid gel was washed off with an air/water spray for 15 seconds, and he dried the surface for five seconds with an oil-free, moisture-free air spray. The clinician tested for oil-free and moisture-free air by spraying on the surface of a mirror. He applied primer to dentin and enamel, then dried it for five seconds. Desiccation of the dentin was avoided. An adhesive (Scotchbond Multi-Purpose adhesive, lot no. 20001201, 3M ESPE) was added to the primed enamel and dentin and light-cured for 10 seconds. The clinician placed restorative material (Z100, lot no. 20001128, 3M ESPE) (shade A3) in four increments in layers no thicker than 2.0 mm. The clinician placed the first increment parallel to the gingival floor at a thickness of approximately 1.0 mm. He placed the next three increments at oblique angles until the preparation was restored to its original contours. Between increments and after final polishing, he light-cured the material for 40 seconds each from the buccal, lingual and occlusal directions. He finished the restoration with a high-speed handpiece under an air/water spray using a 12-bladed finishing bur and polished it with disks (Soflex, 3M ESPE). Occlusion was restored to its preoperative condition.

The clinician extracted the teeth 24 hours after placement of the in vivo restorations by exerting a light pressure on the roots with a forcep (on maxillary teeth, forcep #150 [Hu-Friedy, Chicago] and on mandibular teeth, #151 or #13 Ash forcep [Hu-Friedy]). The teeth were stored in a mixture of 0.9 percent sodium chloride solution (normal saline) and 0.002 percent sodium azide (Fisher Scientific, Versailles, Ky.) at 37 C. He removed excess tissue.

Twenty-four hours after the tooth was restored, the clinician sectioned each tooth (Figure 1) using a low-speed bone saw (Isomet, Buehler, Lake Bluff, Ill.) (no.11-1180) under water with a 4 x 0.012–square inch diamond rim blade (no. 11-4244, Buehler). He cut samples to form a flat rectangular plane with a cross-sectional area of 0.5 mm² with no notch at the adhesive junction. The thickness of each section was determined with a 1-micrometer-resolution digital micrometer (no. 20765A241, Mitutoyo, Aurora, Ill.). The clinician took three gingival sections from one of the proximal boxes, and three axial sections were taken from the remaining proximal box. Samples were attached to the opposing arms of a testing device (Bencor Multi-T, Danville Engineering, San Ramon, Calif.) using a cyanoacrylate adhesive (Zapit, lot no. DS99060012- 6/30/02, Dental Ventures of America, Corona, Calif.) and accelerator (Zapit, lot no. DS99060015-6/29/02, Dental Ventures of America). The clinician measured microtensile stress on a testing machine (Universal Instron, Model 1125, Instron, Canton, Mass.) using a 500-kilogram load cell at a crosshead speed of 1.0 mm/minute.

View larger version (24K): [in this window] [in a new window]   Figure 1. Specimen preparation for testing on a testing instrument (Universal Instron 125, Instron, Canton, Mass.). Specimen preparation process from whole tooth to microtensile sample. A. Teeth restored with separate mesioocclusal and distoocclusal restorations. B. Each restored tooth is sectioned with saw to obtain three individual samples per tooth. C. Individual sample (three obtained per tooth). D. Individual sample further sectioned into halves (six halves obtained per tooth). E. Each half further sectioned to obtain an axial sample (three obtained per tooth) and a gingival sample (three obtained per tooth). F. Microtensile sample for testing obtained (six per tooth—three from the axial location and three from the gingival location). G. Each microtensile sample tested on Instron for microtensile adhesive bond strength.

We evaluated the mode of failure using environmental scanning electron microscopy, or SEM. We sputter-coated the fractured specimens with gold-palladium and observed them using a scanning electron microscope (XL30 ESEM-FEG Microscope, FEI, Hillsboro, Ore.) up to x 2,500 magnification at 15.0 kilovolts. We identified the mode of failure as a cohesive dentin failure, a cohesive composite failure or an adhesive failure. If any part of the adhesive layer was involved, we categorized the failure as adhesive. We photographed representative samples from each mode of failure and stored the images digitally.

	RESULTS

TOP ABSTRACT SUBJECTS, METHODS AND MATERIALS RESULTS DISCUSSION SUMMARY REFERENCES

 
The patient population consisted of five patients with a mean age of 16.2 years (age range, 13–26 years). We collected 14 premolar teeth that yielded 89 microtensile samples. During preparation of the samples with the low-speed saw, 19 samples debonded (17 gingival and two axial). We did not include these samples in the strength calculations. After SEM observation, we discarded four samples owing to presence of either Zapit adhesive or enamel at the adhesive joint, leaving 66 microtensile samples for data analysis. Table 1 provides summary data for bond strengths and the number of adhesive debonding incidents. In vitro axial samples had the highest strength, and in vivo gingival samples had the lowest strength. A two-way analysis of variance³² found microtensile bond strength values for in vivo samples significantly weaker than those for in vitro samples (F_1,66 = 18.716, P < .001), and gingival samples were significantly weaker than axial samples (F_1,66 = 25.849, P .001). The interaction between condition and location was not significant. The adjusted R^{</it>2 </it>}for the model was .40. The control group yielded a microtensile bond strength of 38.3 (10.9) megapascals (n = 19) with no debonding incidents during sample preparation. This value was similar to that found in other studies.^33–35

View larger version (139K): [in this window] [in a new window]   Figure 2. This dentin-side specimen from the in vivo axial group, whose bond strength was 44.1 megapascals, failed through the interface of adhesive, composite and dentin.

View larger version (119K): [in this window] [in a new window]   Figure 3. Backscatter image of a dentin side specimen from the in vivo axial group, whose bond strength was 10.4 megapascals. The specimen failed predominantly through the adhesive interface with a small amount of composite resin (CR) in the lower right-hand corner.

View larger version (176K): [in this window] [in a new window]   Figure 4. The magnified ( 800) dentin side of the specimen (the circled area in Figure 3) shows a layer of adhesive resin containing blisterlike spaces lying over the dentin surface.

View larger version (152K): [in this window] [in a new window]   Figure 5. The magnified (x 2,500) dentin side of the specimen (the circled area in Figure 4) shows a blisterlike space through the adhesive layer that polymerized on an overly wet surface. The moisture could have come from where the primer was applied (P), excess water after the acid was rinsed away, fluid from the dentinal tubules (DT) or some other form of moisture contamination. A listing (L), or peeling back, of the adhesive layer also can be seen.

	DISCUSSION

TOP ABSTRACT SUBJECTS, METHODS AND MATERIALS RESULTS DISCUSSION SUMMARY REFERENCES

Successful adhesive bonds to enamel are in the range of 25 to 30 MPa,^37–39 and bonds to dentin should exceed 17 MPa to resist forces generated by polymerization shrinkage.^4,40 In this study, the bond to the axial dentinal wall exceeded these values.

The calculated bond strengths were those that survived specimen preparation; 36 percent (17) of the gingival samples debonded and 5 percent (two) of the axial samples debonded during the preparation of the specimens in the laboratory. This debonding rate was similar to the results of another study.²⁵ A pilot study using a two-step bonding system also yielded bonding failures of 46 percent for the gingival samples and 10 percent for the axial samples.⁴¹ If the number of gingival debonding incidents in our study had been counted as zero and included in the data analysis, then the in vivo gingival group would have had a mean (SD) bond strength of 10.9 (12.2) MPa and the in vitro gingival group a bond strength of 21.7 (19.2) MPa. These strengths and their high variability may not be reliable enough to resist the forces of polymerization shrinkage.

The eta-squared statistic describes the proportion of total variability attributable to a factor.³² In our study, 29 percent of variability was attributable to location and 23 percent was attributable to condition. The entire model, adjusted for the number of terms, accounted for 40 percent of the variance. The 60 percent of variance not accounted for by the experimental model may be due to operator variability, moisture content at the cavity wall location, decreasing curing-light intensity at the depth of the gingival wall, in vivo patient stresses placed on the restoration and the stresses placed on the in vivo tooth group during the extraction procedure.

Adhesion is highly influenced by the operator,⁴² who can commit errors during different stages of the adhesion procedure.⁴³ There will always be a subjective analysis of what is "too wet" or "too dry." Rinsing and drying steps are difficult to standardize under clinical conditions.^43,44 Evidence of voids that probably are the result of moisture contamination can be seen in Figures 3 through 5. Failures at low bond strengths (< 17 MPa) showed blisterlike spaces (Figures 3–5) similar to those described by Tay and colleagues²² or voids representing incomplete wetness or phase separation of the adhesive on the surface of the dentin.⁴⁵ These spaces, approximately 5 to 40 µm in diameter, could be found throughout the adhesive layer in a number of specimens. The spaces could be the result of excessive moisture on the dentin substrate due to inadequate drying of the primed surface before the adhesive resin was applied,²² moisture from the dentinal tubules after acid-etching with a strong acid,^18,44,46 excess moisture after rinsing acid with water from the dentinal surface or other extrinsic moisture. We saw these voids quite often in the in vivo samples that fractured at low bond strengths. This type of failure also was present in some of the in vitro samples, suggesting that the dentin was too wet from the water-based primer³⁰ for the adhesive to adequately hybridize the decalcified dentin.

A wet-bonding technique can guarantee efficient resin interdiffusion only if all the remaining water on the dentin surface is eliminated and replaced by monomers and adhesives during the subsequent adhesive steps.⁴⁷ The term bonding to "moist" or "wet" dentin should not be loosely interpreted and must be more clearly defined.²² It may take more than the five seconds of drying after the primer is applied to prevent an overwet phenomenon from occurring.³⁰ Bond strengths doubled after extra time was allowed for the primer to evaporate.³⁰ It may be necessary to blot the surface with tissue paper or a dry applicator sponge after application of the acid etchant and primer to obtain a surface that is wet, but not too wet, before applying the adhesive resin.⁴⁷ It may be that surface resin blisters can be eliminated through a reduction in the water content of the bonding system.⁴⁸ Gaps along the tooth-restorative interface can lead to severe leakage into the dentinal tubules, allowing the movement of fluid and bacteria through the interface and into the tubules, which can be manifested clinically as postoperative sensitivity and recurrent caries.⁴⁸ If the dentin at this location is not hybridized, it can become vulnerable to hydrolytic breakdown and susceptible to penetration by bacterial enzymes or other toxic substances.^8,18

Fluid from dentinal tubules on the axial wall of the cavity preparation may account for less water than fluid from dentinal tubules on the gingival wall.

A "window of opportunity"²⁹ for achieving a dentinal surface that is wet but not overly wet, and dry but not overly dry, is difficult to find. This window for bonding to in vivo gingival walls may be extremely narrow, and that may explain why the bond strength is weaker. A dentinal surface that is dry but not overly dry may be a better surface to which to bond than an overwet surface. A recent in vivo study found that the level of residual moisture on human dentin pulpal floors of Class I preparations did not influence the microtensile bond strengths.⁴⁹ It has been suggested that the hydrostatic pulpal pressure, the dentinal fluid flow and the increased dentinal wetness in vital versus nonvital dentin could affect the intimate interaction of certain dentin adhesives with the dentinal tissue.⁵⁰ Other reasons for lower bond strengths at the gingival wall include working in a humid in vivo environment, failure of primer solvent to evaporate completely,³⁰ pooling of solvents and adhesives, increased permeability of dentin, difficulty of isolating the in vivo gingival margin from sulcular fluids, and inability of air drying to remove excess water. The result of this contamination could be inadequate polymerization curing or defects within the hybrid zone.^22,27

Bonding resins that contain hydroxyethyl-methacrylate, or HEMA, and bisphenol glycidyldimethacrylate, or Bis-GMA, can undergo phase separation in the presence of water,⁴⁵ reducing the conversion level of the adhesive system by 50 percent.³⁰ HEMA is a hydrophilic resin and can readily mix with water, but Bis-GMA is hydrophobic³⁰ and, instead of mixing with water, forms resin globules in the presence of water.²² Bonding resins containing Bis-GMA, therefore, could be quite technique-sensitive^30,51 in the presence of a small amount of water contamination.^30,50,51

Good technique may not guarantee strong bonds. Deep dentin has a higher water content than does superficial dentin owing to the larger diameter and number of tubules per unit area.²⁴ This water may dilute the organic solvents of some bonding systems, causing monomers to leave the soluble phase and form resin globules in water.^24,52 Water from either oral fluids or dentinal tubules could weaken the properties of the HEMA-based primers. In one study, a 9 percent volume of water added to a HEMA/Bis-GMA bonding resin under wet conditions severely weakened the resin by 64 percent.⁵¹ Fluid from dentinal tubules on the axial wall of the cavity preparation may account for less water than fluid from dentinal tubules on the gingival wall. Reduced strengths of bonds to the gingival wall also could be explained by the fact that the density of tubules 1.0 mm above the CEJ at the gingival cavity preparation wall is 49 percent greater than that at the axial wall.²⁰ There thus will be less intertubular dentin to which to bond at the gingival wall than at the axial wall.

Perhaps a shorter period of etching with 35 percent phosphoric acid on in vivo dentin would provide enough demineralization but reduce the wetness from the dentinal tubules. Alternatively, a weaker acid for dentin and a stronger acid (35 percent phosphoric acid) for enamel might be more advisable. Clearly, more research is required to determine the optimal etchants and etching times for conditioners used on dentin.²⁵

The direction of the dentinal tubules may be one reason why bond strengths are not uniform inside a cavity.⁵³ Resin bonded to parallel-oriented tubules had higher bond strengths in vitro (approximately 11–13 MPa) than did resin bonded to perpendicularly oriented tubules.⁵³ In another study, however, when resin-based composite was bonded to occlusal and gingival margins in wedge-shaped in vitro Class V preparations and tested for regional bond strengths, the researchers found no difference between the regions.⁵⁴ Etching a parallel-oriented tubule might open the tubule, allowing all of the tubule fluid to be washed away, whereas the fluid in a perpendicular tubule might continue to seep after the smear layer has been removed. One of the authors of this study, in a separate SEM observation (J. Purk, D.D.S., Ph.D., unpublished data, 2001) of the dentinal walls of Class II preparations, observed more parallel-oriented tubules on the axial wall, which corresponds to higher bond strength in the present study. The tubule direction also may determine the intrinsic wetness of the surface.⁵³ Tubule direction alone might account for the reduced bond strengths to the gingival wall in the absence of an overly wet substrate.

In vitro microtensile bond strengths may overestimate bond strengths and may be useful only for comparing products. In vitro testing appears to be inadequate for accurately predicting in vivo bond strengths. It is the use of adhesive bonding materials under in vivo conditions that is the most relevant.^49,55

	SUMMARY

TOP ABSTRACT SUBJECTS, METHODS AND MATERIALS RESULTS DISCUSSION SUMMARY REFERENCES

 
To our knowledge, this study is the first to measure the microtensile bond strength of adhesive/composite bonded to the dentin at the gingival cavity preparation wall of human premolars under clinical conditions. Bonding to gingival cavity preparation walls was weaker than was bonding to axial cavity preparation walls. Bonding to teeth under in vivo conditions yielded much weaker microtensile bond strengths than did bonding under in vitro conditions. Bonding of resin-based composite to dentin at the gingival wall under in vivo conditions is weaker than that in previously reported in vitro studies. The in vivo bond of the resin-based composite to dentin at the gingival wall may not be strong enough to resist the forces of the composite’s polymerization shrinkage. This lower strength may be explained by the presence of excessive wetness on the gingival wall. Further in vivo studies are necessary to improve the fragile adhesive bond to the gingival wall in Class II resin-based composite restorations.