| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
REVIEWS |
1 Leuven BIOMAT Research Cluster, Department of Conservative Dentistry, School of Dentistry, Oral Pathology and Maxillo-Facial Surgery, Catholic University of Leuven, Kapucijnenvoer 7, B-3000 Leuven, Belgium; and
2 Department TEWO, Laboratory of Dental Materials, University of Antwerp-RUCA, Groenenborgerlaan 171, B-2020 Antwerpen, Belgium;
* corresponding author, bart.vanmeerbeek{at}med.kuleuven.ac.be
ABSTRACT
The immediate bonding effectiveness of contemporary adhesives is quite favorable, regardless of the approach used. In the long term, the bonding effectiveness of some adhesives drops dramatically, whereas the bond strengths of other adhesives are more stable. This review examines the fundamental processes that cause the adhesion of biomaterials to enamel and dentin to degrade with time. Non-carious class V clinical trials remain the ultimate test method for the assessment of bonding effectiveness, but in addition to being high-cost, they are time- and labor-consuming, and they provide little information on the true cause of clinical failure. Therefore, several laboratory protocols were developed to predict bond durability. This paper critically appraises methodologies that focus on chemical degradation patterns of hydrolysis and elution of interface components, as well as mechanically oriented test set-ups, such as fatigue and fracture toughness measurements. A correlation of in vitro and in vivo data revealed that, currently, the most validated method to assess adhesion durability involves aging of micro-specimens of biomaterials bonded to either enamel or dentin. After about 3 months, all classes of adhesives exhibited mechanical and morphological evidence of degradation that resembles in vivo aging effects. A comparison of contemporary adhesives revealed that the three-step etch-and-rinse adhesives remain the ‘gold standard’ in terms of durability. Any kind of simplification in the clinical application procedure results in loss of bonding effectiveness. Only the two-step self-etch adhesives approach the gold standard and do have some additional clinical benefits.
KEY WORDS: artificial aging • dental adhesives • enamel • dentin
(1) INTRODUCTION
The major shortcoming of contemporary adhesive restoratives is their limited durability in vivo (Van Meerbeek et al., 1998). The most cited reasons for failure of adhesive restorations are loss of retention and marginal adaptation (Mjör et al., 2002; Mjör and Gordan, 2002). Hence, a valuable approach to prolong the clinical lifetime of adhesives might be to focus on improving the stability of the bond of these biomaterials to tooth tissue. The immediate bonding effectiveness of most current adhesive systems is quite favorable (Inoue et al., 2001b), regardless of the adhesive used. However, when these adhesives are tested in a clinical trial, the bonding effectiveness of some materials appears dramatically low, whereas the bonds of other materials are more stable (Van Dijken, 2000; Brackett WW et al., 2002). The objective of this review is to discuss the potential in vivo degradation processes involved, to critically review study designs to assess these phenomena, and, eventually, to find out how well laboratory tests can predict in vivo bond durability.
(2) CLASSIFICATION OF CONTEMPORARY ADHESIVES
The basic mechanism of bonding to enamel and dentin is essentially an exchange process involving replacement of minerals removed from the hard dental tissue by resin monomers, which, upon setting, become micro-mechanically interlocked in the created porosities. This interlock was first described by Nakabayashi et al. in 1982 and is commonly referred to as ‘hybridization’, or the formation of a ‘hybrid layer’. Based upon the underlying adhesion strategy, three mechanisms of adhesion are currently in use with modern adhesive systems (Fig. 1
; Van Meerbeek et al., 2001, 2003).
|
(2.2) Self-etch Adhesives
An alternative approach is based on the use of non-rinse acidic monomers that simultaneously condition and prime dentin, the so-called ‘self-etch’ adhesives. Regarding user-friendliness and technique-sensitivity, this approach seems clinically most promising. This approach eliminates the rinsing phase, which not only lessens the clinical application time, but also significantly reduces the technique-sensitivity or the risk of making errors during application. There are basically two types of ‘self-etch’ adhesives: ‘mild’ and ‘strong’ (Van Meerbeek et al., 2001). ‘Strong’ self-etch adhesives have a very low pH (< 1) and exhibit a bonding mechanism and interfacial ultra-morphology in dentin resembling that produced by etch-and-rinse adhesives. ‘Mild’ self-etch adhesives (pH of around 2) dissolve the dentin surface only partially, so that a substantial number of hydroxyapatite crystals remain within the hybrid layer. Specific carboxyl or phosphate groups of functional monomers can then chemically interact with this residual hydroxyapatite (Yoshida et al., 2004). This two-fold bonding mechanism (i.e., micro-mechanical and chemical bonding) is believed to be advantageous in terms of restoration durability. It has a micro-mechanical bonding component that may in particular provide resistance to abrupt de-bonding stress. The chemical interaction may result in bonds that better resist hydrolytic break-down and thus keep the restoration margins sealed for a longer period.
(2.3) Glass Ionomers and Glass-ionomer Adhesives
Glass-ionomers are still considered the only materials that self-adhere to tooth tissue (Yoshida et al., 2000). A short polyalkenoic acid pre-treatment cleans the tooth surface; it removes the smear layer and exposes collagen fibrils up to about 0.5–1 µm deep (Inoue et al., 2001a); therein, glass-ionomer components inter-diffuse and establish a micro-mechanical bond following the principle of hybridization (Lin et al., 1992; Van Meerbeek et al., 2001). In addition, chemical bonding is obtained by ionic interaction of the carboxyl groups of the polyalkenoic acid with calcium ions of hydroxyapatite that remained attached to the collagen fibrils (Yoshida et al., 2000). This additional chemical adhesion may be beneficial in terms of resistance to hydrolytic degradation. Consequently, a two-fold bonding mechanism is established, similar to that mentioned above for mild self-etch adhesives. The basic difference with the resin-based self-etch approach is that glass ionomers are self-etching through the use of a relatively high-molecular-weight (from 8000 to 15,000) polycarboxyl-base polymer. This limits their infiltration capacity, so that only shallow hybrid layers are formed. In addition, because of this high molecular weight, they cannot infiltrate phosphoric-acid-decalcified dentin. Consequently, such aggressive conditioners should not be used in conjunction with glass ionomers (De Munck et al., 2004).
(3) CLINICAL TRIALS
The ultimate test method for the assessment of bonding effectiveness remains a clinical trial. If the study objective is to investigate clinical effectiveness of adhesives, only studies involving non-carious class V adhesive restorations should be considered, for the following reasons (Van Meerbeek et al., 1998): (1) Such lesions do not provide any macro-mechanical retention, so that ineffective bonding will result in early restoration loss; (2) retention, i.e., evaluating if the restoration is present or not, is the only objective study parameter available; (3) any other type of cavity (I, II, III, and IV) will exhibit at least some macro-mechanical retention. Consequently, less-objective evaluation criteria, such as marginal integrity and discoloration, need to be quantified as a measure of bonding effectiveness; (4) class V restoration margins are located in enamel as well as in dentin; (5) lesions are commonly located on vestibular surfaces of anterior teeth and premolars, thus providing good access for the restorative procedure as well as for evaluation (visually, with the use of an explorer and magnifying glasses, and even by SEM, if epoxy replicas are made); (6) preparation and restoration of class V lesions are minimal and relatively easy, thereby somewhat reducing practitioner variability; (7) lesions are relatively widespread and occur on multiple teeth, facilitating patient selection and enabling balanced study designs to be developed; and (8) despite various cavity configuration factors of class V lesions (Feilzer et al., 1987; Carvalho et al., 1996), and thus resultant interfacial stress, the mechanical properties of the composite used are relatively unimportant (Browning et al., 2000; Tyas and Burrow, 2001, 2002; Van Meerbeek et al., 2005a, b). A drawback related to the use of non-carious class V lesions might be the substantial differences in the composition of the bonding surfaces. Non-carious class V lesions exhibit high degrees of sclerosis, and related high mineral content, as compared with intact or caries-affected dentin. Hybrid layer formation on this hypermineralized dentin is more difficult, and therefore less effective bonding is assumed (Van Meerbeek et al., 1994; Prati et al., 1999). Supporting this hypothesis is the observation of higher annual failure rates in the presence of sclerotic dentin (Van Dijken, 1994); however, this seems to be valid only for older adhesives (Van Meerbeek et al., 1998; Van Dijken, 2000).
The outcome of a clinical study is dependent not only on the patient and the materials used. ‘External’ factors—such as the skills of the operator, type of light source, isolation method, finishing instruments used, etc.—significantly influence the outcome. Also, in many studies, patient-related factors—such as age, oral hygiene, occlusal loading, and dentin sclerosis—are more influential than any material property (Bayne et al., 1991; Van Meerbeek et al., 1998). Therefore, an appropriate study design is paramount when clinical trials are conducted. The patient-related factors can be ruled out by the application of a balanced study design. In such a design, pairs of equal teeth (for instance, first and second premolars on the same side, left and corresponding right incisors, canines, or premolars, respectively) with similar lesions are chosen in each patient, and each tooth is assigned to one of the experimental treatments in a randomized way, so that the study hypothesis is tested at the patient level (Van Meerbeek et al., 1998). However, when it comes to a comparison of different studies, especially from different research groups, all external and patient-related factors must still be considered and can be dealt with only by randomization and an adequate number of patients, rather than restorations. In that way, the results from the statistical sample can be extrapolated to the population. The number of patients required for this purpose can be determined with statistical power analysis. If the obtained results are to be compared with data in the literature, a control group using a well-established adhesive technique is also required. Hence, a study that uses only a simplified adhesive, will demonstrate only if that adhesive complies with, for instance, ADA standards. Only from a well-controlled experimental design can one conclude that this new material competes with the best available material. Besides adequate study design, longer observation times (up to 5 years and longer) are needed. Unfortunately, most currently published data are obtained from short-term studies and are often published only as meeting abstracts, without any study details.
Different causes that may interact synergistically can lead to clinical failure in class V studies. Eccentric occlusal stress is one of the main causes for the development of abfraction lesions, as revealed by finite element analysis (Rees, 2002) and clinical evidence (Braem et al., 1992). The same stress that caused the lesions may also put tension on the restoration/tooth interface, resulting in increasing damage at the interface (Krejci et al., 1994; Van Meerbeek et al., 1998).
In addition to occlusal stresses, intra-oral temperature changes may also induce repetitive contraction/expansion stresses at the tooth-adhesive interface, due to the higher thermal contraction/expansion coefficient of the restorative material (as compared with that of tooth tissue) (Gale and Darvell, 1999).
The tooth-biomaterial bond may also degrade by exposure of the interface to water and/or human/bacterial enzymes present in saliva. Ingression of water into the hybrid layer (Sano et al., 1995) and subsequent leaching out of resin components are believed to lead to inefficient in situ polymerization (Eick et al., 1997) and degradation of resin components (Santerre et al., 2001; Finer and Santerre, 2004; Jaffer et al., 2002). Also, hydrolysis of hydroxyapatite-depleted or insufficiently resin-coated collagen fibrils compromises long-term bonding effectiveness (Hashimoto et al., 2000, 2002, 2003b).
Thus, biomaterial-tooth interfaces are subjected to chemical as well as mechanical degradation. Chemically, the most important reactions are hydrolysis and plasticizing of the resin components, which are both related to the ingression of water. Since this ingression is a diffusion-like process, the related degradation mechanisms will also be diffusion-dependent. Hydrolysis can break up covalent bonds, linking the different units of collagen fibrils as well as resinous polymers (Hashimoto et al., 2000). This process can be enhanced by enzymes released by bacteria (Santerre et al., 2001) and by the dentin itself (Pashley et al., 2004). Subsequently, the resultant breakdown products and residual monomers can leach out and so decrease the interfacial mechanics and allow more water to ingress. Water can also reduce the frictional forces between the polymer chains, which reduces the mechanical properties of the resin part and makes the polymer swell, a process known as the plasticization of resin. Also, repeated mechanical stress can affect interfacial integrity (De Munck et al., 2005a,c). During each chewing cycle (or other mechanical stress), the interface is loaded. At some sites, stress concentrations may exceed interfacial fracture toughness, which results in the initiation of a crack. In some cases, this may also cause catastrophic failure of the restoration. In other cases, this initiated crack can act as even a bigger stress concentrator, so that the subsequent mechanical loads will enlarge the crack, a process known as crack propagation. Catastrophic failure will occur when crack propagation has extended to a level affecting the structural integrity of the material.
In the clinical situation, however, it is difficult to discriminate the specific factor(s) that caused the restoration to fail. In contrast to in vivo research, in vitro testing can elucidate the specific factors that are most detrimental to long-term bonding effectiveness. Future research can then focus on improving these specific factors and thereby enhance bond durability.
(4) LABORATORY STUDIES
(4.1) Bond Strength
(4.1.1) Test methodologies
Bond strength tests are the most frequently used tests to screen adhesives. The rationale behind this testing method is that the stronger the adhesion between tooth and biomaterial, the better it will resist stress imposed by resin polymerization and oral function. Different bond strength tests have been developed (Pashley et al., 1995, 1999). Currently, the shear and micro-tensile bond strength (µTBS) test methods are the most used (Fig. 2). It is important to note that a bond strength value cannot be considered as a material property (Van Noort et al., 1989). The data recorded depend largely upon experimental factors such as, for example, the type of composite, stress rate, sample size and geometry, and the actual test method (Phrukkanon et al., 1998; Sudsangiam and Van Noort, 1999). Therefore, the absolute test values cannot be used to draw conclusions from, or be compared with, data gathered in other studies. Only relative study outcomes, in the sense of ‘A is better than B’, are a valid basis for further interpretation of the results. Nevertheless, bond-strength testing can reveal valuable clinical information, when gathered in a well-controlled design. For instance, by introducing an aging factor into the study design, one can assess the durability of adhesion. In this respect, the need for one or more controls to be included should be emphasized. In many studies (Tjan et al., 1996; Dietschi and Herzfeld, 1998; Pilo and Ben-Amar, 1999; Frankenberger et al., 2000; Meiers and Young, 2001; Cardoso et al., 2002), all specimens are ‘aged’ (by, for example, thermo-cycling) so that more ‘clinically’ relevant data can be obtained. But if a proper control is absent, the study provides no information on durability. Different artificial aging techniques can be used, depending upon the specific types of bond degradation that are being investigated.
|
), with one particular restorative composite (Z100, 3M ESPE).
The pooled µTBS of all etch-and-rinse adhesives tested (Fig. 3a), when bonded to enamel, was 39 and 40 MPa for the three-step and two-step etch-and-rinse adhesives, respectively. As has been shown by Buonocore (1955), bonding to enamel requires only an acid-etch step followed by the application of a fluid resin, without the need for an intermediary primer step. Primers do not negatively influence bonding effectiveness and are mandatory when a ‘wet-bonding’ procedure is carried out. In general, self-etch procedures have resulted in significantly lower bonding effectiveness, although some two-step self-etch adhesives approached the bonding effectiveness of etch-and-rinse adhesives (De Munck et al., 2005b). A pooled µTBS of about 30 MPa was obtained for two-step self-etch adhesives. No significant difference was recorded in favor of either a ‘strong’ or a ‘mild’ self-etch approach. Because of the superficial interaction of self-etch adhesives with enamel, and thus less potential for micro-mechanical interlocking than a phosphoric-acid treatment, it can be hypothesized that an additional chemical bonding capacity to hydroxyapatite contributed to the actual bonding effectiveness of ‘mild’ self-etch adhesives. Among all adhesives tested, one-step self-etch adhesives produced the lowest µTBS (Fig. 3a). The glass-ionomer adhesive Fuji Bond LC (GC) performed equally as well as the two-step self-etch adhesives. However, during bond-strength testing, the glass-ionomer adhesive tended to fail in the glass-ionomer material itself rather than at the interface. Thus, the actual bonding effectiveness to enamel was never assessed (Inoue et al., 2001a).
|
). Statistical analysis of the pooled dentin µTBS data showed that three-step etch-and-rinse adhesives bonded significantly more strongly to dentin than did two-step etch-and-rinse and two-step self-etch adhesives. The latter two systems did not perform significantly differently from each other. Again, the significantly least favorable µTBS results were recorded for one-step self-etch adhesives, and the µTBS of these adhesives was not significantly different from that of the resin-modified glass-ionomer adhesive.
(4.1.3) Aging by storage
The most commonly used artificial aging technique is long-term water storage. The bonded specimens are stored in fluid at 37°C for a specific period. This period may vary from a few months (Shono et al., 1999) up to 4–5 years (Fukushima et al., 2001; De Munck et al., 2003b), or even longer. Most studies report significant decreases in bond strengths, even after relatively short storage periods (Burrow et al., 1996; Kato and Nakabayashi, 1998; Shono et al., 1999; Kitasako et al., 2000; Armstrong et al., 2001b, 2003; Meiers and Young, 2001; De Munck et al., 2003b; Giannini et al., 2003). Decrease in bonding effectiveness in this type of study is, first, supposed to be caused by degradation of interface components by hydrolysis (mainly resin and/or collagen). But, as previously mentioned, water can also infiltrate and decrease the mechanical properties of the polymer matrix, by swelling and reducing the frictional forces between the polymer chains, a process known as ‘plasticization’ (Ferracane et al., 1998; Santerre et al., 2001). Furthermore, some interface components, such as uncured monomers and break-down products of previous mechanisms, can elute and so weaken the bond (Hashimoto et al., 2002). The storage solution is usually water. To prevent bacterial growth during the storage period, investigators have added sodium azide (Burrow et al., 1996), chloramine (De Munck et al., 2003b; Armstrong et al., 2001a), or even antibiotics (Shono et al., 1999). To mimic the clinical situation more closely, artificial saliva solutions can also be used, but bond strength reductions obtained were similar to those obtained with pure water degradation (Kitasako et al., 2000). Even enzymes can be added to the storage medium. For example, esterases that can be produced by bacteria in vivo are able to catalyze the breakdown of resin components (Santerre et al., 2001; Finer and Santerre, 2004). Most degradation processes are diffusion-rate-dependent. Consequently, the length of the diffusion path is as important as the diffusion time itself. A way to exclude diffusion-dependent effects is to age micro-specimens so as to render the diffusion path as short as possible. For example, when small µTBS sticks were stored, a significant decrease in bond strength to dentin was detected within as few as 90 days (Shono et al., 1999; Armstrong et al., 2001b). Storing tiny µTBS sticks may thus be considered as a form of accelerated aging. Enamel-resin bonds, when produced by etch-and-rinse adhesives, are more stable over time (Frankenberger et al., 2000). They can seal off the path of water diffusion to the more vulnerable dentin-resin bond and so retard bond degradation (Hashimoto et al., 2002; De Munck et al., 2003b).
(4.1.4) Aging by thermo-cycling
Another widely used aging technique is thermo-cycling. The ISO TR 11450 standard (1994) indicates that a thermo-cycling regimen comprised of 500 cycles in water between 5 and 55°C is an appropriate artificial aging test. A literature review (Gale and Darvell, 1999) concluded that 10,000 cycles corresponds approximately to 1 year of in vivo functioning, rendering 500 cycles, as proposed by the ISO standard, as being very minimal in mimicking long-term bonding effectiveness. The artificial aging effect induced by thermo-cycling can occur in two ways: (1) Hot water may accelerate hydrolysis of interface components, and subsequent uptake of water and extraction of breakdown products or poorly polymerized resin oligomers (Miyazaki et al., 1998; Hashimoto et al., 2000); or (2) due to the higher thermal contraction/expansion coefficient of the restorative material (as compared with that of tooth tissue), repetitive contraction/expansion stresses are generated at the tooth-biomaterial interface. These stresses may lead to cracks that propagate along bonded interfaces, and, once a gap is created, changing gap dimensions can cause in- and outflow of oral fluids, a process known as ‘percolation’ (Gale and Darvell, 1999). In light of the first aging effect (diffusion-dependent hydrolysis and elusion), thermo-cycling should be applied to very small specimens, and any further preparation after aging is to be avoided. In that case, the most degraded interface is tested, corresponding to the vulnerable restoration margins. Aging of µTBS specimens might also be more appropriate than the larger shear-bond-strength specimens, because both resin composite and tooth tissue may protect the interfacial bond against thermal fluctuations. If the second aging effect (repetitive contraction/expansion stress) is considered, thermo-cycling should be applied to specimens in which stress similar to that in the clinical situation can be generated. In vivo stress will occur if the ratio of bonded to unbonded surfaces (the so-called ‘C-factor’) is high (Feilzer et al., 1987). Therefore, thermo-cycling of the restored tooth will result in the highest clinically relevant stress.
A recent meta-analysis (Leloup et al., 2001) of data published between 1992 and 1996 concluded that thermo-cycling has no significant effect on bond strength. Most studies included in the meta-analysis were carried out following the ISO standard of 500 cycles (mean number of cycles in the studies analyzed was 630). This number of cycles was probably too low for an aging effect to be obtained (Miyazaki et al., 1998; Gale and Darvell, 1999; Nikaido et al., 2002a). Also, specimen geometry has often not been taken into account. In most studies cited in that review, relatively large composite cylinders bonded to flat surfaces were thermo-cycled, prior to being pulled apart following a shear or tensile bond strength test protocol (Leloup et al., 2001). As a result, the surrounding tooth and composite must have thermally protected a large part of the interface (Watts et al., 1987). In addition, because of the low C-factor (about 1/6), little repetitive expansion/contraction stress must have been generated at the interface. Both reasons may explain why thermo-cycling did not affect bonding effectiveness in those studies. This hypothesis was confirmed by a study in which thermo-cycling of restored flat surfaces did not decrease the µTBS, whereas a similar aging protocol applied to restored cavities resulted in a significantly decreased µTBS (Nikaido et al., 2002a). Thermo-cycling of tiny µTBS specimens (diffusion path < 1 mm) resulted in a significant decrease in bond strength (Xie et al., 2002), thus supporting the assumption that thermo-cycling accelerates chemical degradation of the interface. Eventually, it can be concluded that thermo-cycling results in combined contraction/expansion stresses and accelerated chemical degradation. The relative contribution of each, however, is strongly dependent on the specific test set-up.
(4.1.5) Aging by occlusal loading
Mechanical loading may also affect resin adhesion to tooth structure. Again, to simulate this stress in vitro, it is important that one imposes stress similar to that occurring in vivo (Krejci et al., 1994). One possibility is to ‘age’ restored cavities in a chewing simulator and afterward measure the bonding effectiveness (Nikaido et al., 2002a; Frankenberger et al., 2003a,b). A better alternative is to study dynamic mechanical phenomena, such as crack initiation and propagation, in well-controlled fracture toughness or fatigue test set-ups.
(4.1.6) In vivo degradation studies
The above-mentioned factors are expected to decrease bonding effectiveness, since they are in vitro equivalents of in vivo phenomena. Clinically, however, all these factors interact simultaneously and, consequently, may accelerate degradation. One way to overcome this mismatch between laboratory and clinical testing is to age specimens in diverse ways at the same time (Nikaido et al., 2002a), or to age restorations in the oral cavity itself. Because of ethical reasons, the latter type of aging can be performed only on teeth already scheduled for extraction (Ferrari et al., 1997; Iracki et al., 2003), or on primary molars that require restorative treatment and that can be collected after exfoliation (Hashimoto et al., 2000). Alternatively, animal studies can be conducted in which, after a certain period of oral functioning, the teeth are extracted and bond strengths measured (Sano et al., 1999; Takahashi et al., 2002).
The µTBS of in vivo resin-dentin bonds slowly decreases over time (Hashimoto et al., 2000). Based on fractographic analysis, it has been concluded that bond degradation might be caused by deproteinization of collagen fibrils in demineralized non-protected dentin (Sano et al., 1999; Hashimoto et al., 2003a). It has been hypothesized that a rapid method of simulating this degradation is to expose the resin-dentin interface to a deproteinizing agent (Yamauti et al., 2003). A solution of NaOCl in water is a non-specific deproteinizing agent. As an accelerated aging test, the bond strength of µTBS beams immersed in 10% NaOCl for one hour was determined (Yamauti et al., 2003). After this short period, the µTBS significantly decreased, rendering this procedure a very rapid test method. Failure analysis also revealed that the drop in µTBS correlated with specific dissolution of the hybrid layer, similar to in vivo failure patterns. A drawback of this method is that, because of the non-specific properties, NaOCl also causes the mechanical properties of the dentin substrate itself to deteriorate (Sim et al., 2001). Unfortunately, no data are as yet available to link this test directly to the clinical situation.
Of all the abovementioned tests, the fastest method to age tooth-biomaterial interfaces is to immerse µTBS specimens in NaOCl. A more commonly used and validated method is the storage of micro-specimens in water. After relatively short periods (from 90 days to 6 months) of storage in water, all classes of adhesives exhibited at least some decrease in bonding effectiveness (Shono et al., 1999; Armstrong et al., 2001b, 2003; Xie et al., 2002) and a change in interfacial morphology similar to that of in vivo-aged specimens (Sano et al., 1999; Hashimoto et al., 2000; Takahashi et al., 2002).
(4.2) Fracture Toughness
In fracture mechanics, the initiation and propagation of cracks in materials are studied. Fracture toughness can be defined as a measure of the material’s resistance to crack propagation. In contrast to fracture strength—which is dependent on different factors, such as specimen geometry, surface roughness, and test configuration—fracture toughness is an intrinsic material property, independent of the type of test used. Conventional bond strength tests have a non-uniform stress distribution, as revealed by finite element analysis (Van Noort 1989, 1991; DeHoff et al., 1995; Phrukkanon et al., 1998). This accounts, in part, for the large variation in values obtained by different labs for micro-tensile as well as shear bond strengths (Van Noort et al., 1989, 1991). Therefore, interfacial fracture toughness measurements should, in theory, provide more consistent results and are gaining popularity. Most research groups make use of an adapted short-rod test (Tam and Pilliar, 1994; Ruse et al., 1996; Tam and Yim, 1997; Armstrong et al., 1998, 2001a; Tantbirojn et al., 2000; Tam et al., 2001; Destoop, 2003), since it does not require fatigue pre-cracking for the formation of a sharp crack tip, and there is no need to perform any crack length measurement. Moreover, this test can be applied to small specimens, such as human teeth (Armstrong et al., 1998). In all studies on fracture toughness, failure analyses revealed predominantly interfacial failures, in agreement with observed clinical failure modes (Tam and Pilliar, 1994). However, it has been hypothesized that the cured bonding resin of the adhesive may be a weak link at the interface, since its fracture toughness is in the same range as the interfacial fracture toughness (Khajotia et al., 1997). This is corroborated by the observation that the interfacial fracture toughness is higher for filled than for unfilled adhesives (Tam et al., 2001; Destoop, 2003).
Specimen preparation for fracture toughness measurements is difficult, and no standard procedure for dental adhesives is available. One point of concern is the presence of ‘resin flashes’ extending out of the chevron, if the required notch is prepared with Teflon tape. These ‘flashes’ may increase the values measured (Van Noort et al., 1991; Tantbirojn et al., 2000), but can be avoided if groove cuts are prepared afterward (Armstrong et al., 2001a). However, the prepared grooves are more prone to micro-cracks, which may act as crack initiators and so lower the interfacial fracture toughness. Despite these differences in test set-ups, some consistent data have been reported by different research groups. For example, for Scotchbond MP (3M ESPE), a fracture toughness value of 0.45 MPa 
m has been recorded (Tam and Yim, 1997), which is in the same range as the value 0.38 MPa m, obtained by Destoop (2003). Also, for All Bond 2 (Bisco), similar values of 0.88 and 0.80 MPa m were recorded (Tam and Yim, 1997; Armstrong et al., 1998). The three-step etch-and-rinse adhesive OptiBond FL, in contrast, produced high values in one study (0.82 MPa m; Armstrong et al., 2001a), somewhat lower in another (0.63 MPa m; Destoop, 2003), and could not be measured in a third study (Tam and Yim, 1997) because of very low bonding effectiveness.
Similar to bond strength tests, all kinds of ‘aging’ factors can be applied prior to fracture toughness measurements. However, in contrast to µTBS tests, it is not possible to prepare multiple specimens from the same tooth. Unfortunately, only a few studies have implemented such aging variables. Armstrong et al.(2001a) determined the µTBS and fracture toughness of a resin-dentin interface after 1 and 6 months of water storage. In contrast to µTBS, the fracture toughness did not decrease over that period.
(4.3) Fatigue Resistance
Typically, bonding effectiveness to tooth tissue is measured statically—for example, by shear bond or µTBS testing. In the clinical situation, however, tooth/composite bonds are seldom subjected to such acute tensile/shear stresses, but are subjected to cyclic sub-critical loads during function. Although each of these cycles alone will be insufficient to provoke failure, they will induce damage by generating cracks that grow with time and eventually result in marginal deterioration and loss of the restoration. Fatigue can be defined as the failure of mechanical properties after repeated applications of stress, at a level well below the ultimate fracture strength of the material or interface. Consequently, fatigue tests provide information on the ability of a material or interface to resist the development of cracks as a result of a large number of cycles (Baran et al., 2001).
Strength values are often looked at as indicators of structural performance. However, strength values are more determined by conditional factors such as stressing rate, stress concentrations, flaws, and test geometry rather than by material properties. Consequently, information regarding the material’s microstructure, the test set-up, testing environment, and processing history is of critical value in the interpretation of strength data (Kelly, 1995). Especially, the importance of initiation of micro-cracks, flaws and other stress-raisers is often overlooked. Since a tooth-composite interface is complex and consists of materials with various properties, many potential initiation sites are present. A brittle (enamel) or semi-brittle material (dentin) is attached to an elastic-brittle resin composite (Jameson et al., 1993; Baran et al., 1999) with more elastic, ductile layers in between the adhesive resin and the hybrid layer (Van Meerbeek et al., 1993). At all these interfaces, stress is easily built up and so can cause defects and flaws. Also, the heterogenic and anisotropic features of the involved materials are a source for initiation sites, with loosely cohering enamel prisms, porosities, and large composite glass fillers as obvious examples. Static bond strengths may not adequately demonstrate the effects these defects may have on bond durability, as can fatigue tests. Therefore, theoretically, in vitro fatigue testing of dental adhesives should serve as a good predictor of the in vivo mechanical performance of adhesives. Despite these inherent advantages, there are very little data on fatiguing of tooth-biomaterial interfaces (Dewji et al., 1998). Even fewer studies address the fatigue resistance of contemporary adhesives bonded to enamel or dentin. In addition, to date, no standard fatigue test exists. Some research groups (Ruse et al., 1995; Dewji et al., 1998; Frankenberger et al., 2003b) have applied cyclic stresses to standard shear bond strength specimens to determine the endurance limit of adhesives. Although such tests have drawbacks similar to those of static shear bond strength testing (Van Noort et al., 1989; Sudsangiam and Van Noort, 1999), analysis of the data seems to provide a better insight into the in vivo behavior of adhesives (Ruse et al., 1995). An alternative approach is a push-out set-up (Frankenberger et al., 1999, 2003b): Standard conical cavities in dentin discs are restored with an adhesive and composite. Subsequently, these are pushed out in a (static or) cyclic way. This fatigue test yielded results similar to those obtained by static tests. One advantage of this push-out design is that it takes into account the effect polymerization stress might have in the clinical situation.
Another approach is based upon the classic rotating beam experiment (Wiskott et al., 1994). With a miniaturized version of this test set-up, it is possible to apply tension-compression cycles to tooth-biomaterial interfaces (Fig. 4). An advantage of this method is that the specimens are very similar to standard µTBS specimens, thereby facilitating comparison with previous static results (De Munck et al., 2005c). By this methodology, it was concluded that resin-tooth interfaces are vulnerable to progressive damage by subcritical loads, as was substantiated by SEM analysis that revealed typical fatigue fracture patterns. This set-up also allows for the preparation of multiple specimens per tooth, enabling balanced study designs to rule out the ‘tooth’ factor (De Munck et al., 2005a).
|
(4.4.1) Microleakage
Microleakage is defined as the clinically undetectable passage of bacteria, fluids, molecules, or ions between a cavity wall and the restorative material applied to it (Kidd, 1976). All resin-based restorative materials shrink and induce stress at the interface, which may lead to gap formation and interfacial stress. All current adhesives appear incapable of sealing the restoration margins and thus preventing microleakage (Pilo and Ben-Amar, 1999; Bouillaguet et al., 2000; Grobler et al., 2000; Hilton, 2002a). Many techniques have been used to assess microleakage, and the results vary considerably (Hilton, 2002b).
The use of organic dyes as tracers is one of the oldest and most common methods of detecting leakage in vitro. In general, this method for detecting microleakage involves the placement of a restoration in an extracted tooth, followed by immersion of the tooth in a dye solution after the unfilled parts have been coated with a waterproof varnish. After a certain time interval, the specimens are removed, washed, and sectioned for visual examination to measure the extent of dye infiltration around the filling (Hilton, 2002b). Many dyes can be used with different particle sizes and affinity to substrates, but this does not seem to influence the test results significantly (Hilton, 2002b). In a well-controlled study design, it is theoretically possible to use the same aging factors as mentioned before. The main disadvantage of microleakage evaluation is that it is a qualitative method, which can be made semi-quantitative by the application of a non-parametric scale (Castelnuovo et al., 1996). Generally, the results obtained in each study group differ only slightly, rendering interpretation of the results difficult and reducing the sensitivity of the test. Compared with bond-strength tests, the effect of artificial aging methods such as water storage and thermo-cycling on microleakage is minimal (Wendt et al., 1992; Chan and Jones, 1994; Gwinnett and Yu, 1994; Gale and Darvell, 1999; Wahab et al., 2003). However, microleakage seems more affected by mechanical load cycling (Abdalla and Davidson, 1990; Davidson and Abdalla, 1994; Jang et al., 2001; Kubo et al., 2001; Mitsui et al., 2003), even more pronounced by additional thermo-cycling (Hilton, 2002b).
A quantitative method to assess microleakage is to measure the flow of fluid along the interface (Pagliarini et al., 1996) or from the pulp to a sealed dentin surface (Derkson et al., 1986; Bouillaguet et al., 2000; Del-Nero et al., 2000). The advantage of this method is that it is fully quantitative, and specimens can be followed longitudinally because of its non-destructive nature. The main disadvantage is that the nominal values are usually very low, so that the actual leakage path is sometimes unclear. Leakage may even occur through the dental substrate itself and so falsely increase the interfacial leakage values.
(4.4.2) Nanoleakage
Sano et al.(1994) revealed that leakage can occur between the hybrid layer and intact dentin, even when no gaps can be observed. This was morphologically assessed with extremely small Ag ions (Fig. 5). Because of the morphological and non-quantitative nature of this assessment, one must be very careful in interpreting the results of nanoleakage experiments, since appropriate controls are often absent (Agee et al., 2003), and the results do not correlate with, for example, bond-strength tests. It is hypothesized that this tracer infiltration represents potential voids in the hybrid layer or within demineralized sub-micron spaces that have not effectively been filled by adhesive resin (Sano et al., 1995). These voids are so small that bacteria may not be able to enter. Nevertheless, they make the bond more susceptible to hydrolytic degradation and bacterial by-products such as acids and enzymes (Paul et al., 1999).
|
). One can also assess nanoleakage quantitively by measuring the dye penetration depth using confocal laser scanning microscopy (Dörfer et al., 2000) or SEM (Li et al., 2002a,b). The same artificial aging factors as used for other tests can also be applied to nanoleakage assessments. However, nanoleakage seems little affected by thermo-cycling (Li et al., 2002a) or by mechanical load-cycling (Li et al., 2002b). Water storage of the specimens, in contrast, resulted in increased nanoleakage, which even correlated well with a decrease in µTBS for some adhesives (Okuda et al., 2001, 2002).
(4.5) Marginal Analysis
Several in vitro studies have tested the performance of adhesives by evaluating the marginal gap formation around restorations placed in extracted teeth (Roulet et al., 1989). This method assumes that if the forces generated by polymerization shrinkage or by thermo-mechanical strain exceed the bond strength, an observable gap will be formed at the margin of the restoration. Although there is no clear correlation between in vitro gap formation and interfacial failures observed in vivo, it is reasonable to assume that this marginal gap formation is clinically relevant (Roulet, 1994). The main advantage is that the original specimens can be used for longitudinal observation if epoxy replicas are used, rather than the original specimen. For that reason, measurements can be repeated after thermo-cycling, for example (Krejci et al., 1993), or after long-term water storage (Blunck and Roulet, 2002). Even in vivo Class V restorations can be used for similar longitudinal marginal analysis studies (van Dijken and Hörstedt, 1997). A drawback is that it is a labor-intensive and time-consuming procedure.
(5) DURABILITY OF CURRENT ADHESIVES
The purpose of this section is to assess, based upon the in vivo and in vitro methodologies discussed above, the durability of resin-tooth interfaces. Because it is of no use to compare the results obtained with different adhesives in different studies, the durability was assessed per adhesive system. For some contemporary adhesives, data on bond durability were gathered from published, controlled, in vitro, and in vivo studies, as listed in the Table. In an attempt at better interpretation of laboratory results in light of in vivo functioning of adhesives, in vitro data were linked to the results of Class V clinical trials, since they provide direct and objective information on bond durability (see above). Because of the different adhesion strategies, etch-and-rinse, self-etch, and glass-ionomer adhesives were analyzed separately.
|
(5.2) Etch-and-Rinse Adhesives
From the ‘traditional’ three-step etch-and-rinse adhesives (Fig. 1), simplified two-step etch-and-rinse adhesives have been developed that combine the primer and adhesive resin into one application step. Because the three- and two-step etch-and-rinse adhesives produced by the same manufacturer mostly have similar compositions regarding solvent and adhesive monomers, they were considered together. Since the primer solvent within etch-and-rinse adhesives is a major factor affecting handling (Tay et al., 1998) and performance (Carvalho et al., 2003), the etch-and-rinse adhesives were categorized according to their respective solvent. In each category, the adhesive studied most frequently was chosen to represent its class of adhesives.
(5.2.1) Ethanol-based adhesives
The ethanol/water-based OptiBond adhesives (Kerr) are the most frequently studied commercial representatives of this category. The shear bond strength of the two-step version, OptiBond Solo (and later OptiBond Solo Plus), is significantly decreased by water storage (Meiers and Young, 2001; Giannini et al., 2003), as well as by thermo-cycling (Miyazaki et al., 1998). Also, the µTBS of OptiBond Solo decreased after 4 years of water storage, when the dentin interface was directly exposed to the storage medium (De Munck et al., 2003b). Semi-quantitative marginal analysis (by SEM) after 1 year’s water storage also revealed that the percentage of ‘continuous’ (gap-free) dentin margins decreased over time (Blunck and Roulet, 2002). The bonding effectiveness of the three-step etch-and-rinse version, in contrast, was not affected by water storage, thermo-cycling, and/or mechanical loading (Blunck and Roulet, 2002; De Munck et al., 2003b, 2005c; Frankenberger et al., 2003b; Krejci et al., 1994; Shirai et al., 2005). Only when very small (cross-sectional areas of 0.4–0.6 mm2) µTBS specimens were aged was a significant decrease in µTBS recorded for this three-step etch-and-rinse adhesive (Armstrong et al., 2001a,b). The µTBS of other types of adhesives decreased at least to the same extent (Armstrong et al., 2003). It seems reasonable to link this reduced in vitro durability of Optibond Solo to its reduced infiltration/hybridization capacity (Van Meerbeek et al., 1999). Such a sub-optimal hybridization might explain, to a large extent, why the hybrid layer produced by the two-step version is more prone to hydrolytic degradation than that produced by the three-step etch-and-rinse version (De Munck et al., 2003b). Despite these in vitro differences, both the two- and the three-step versions seem to function rather well in vivo (Boghosian, 1996; Swift et al., 2001a,b; Van Meerbeek et al., 2001, 2003, 2005a, Van Meerbeek et al., b).
(5.2.2) Acetone-based adhesives
The representatives for this adhesive category were the three- and two-step adhesives, All-bond 2 and One-Step (BISCO), respectively. The bonding effectiveness of the two-step version decreased following thermo-cycling (Miyazaki et al., 1998; Xie et al., 2002) and water storage (Shono et al., 1999; Okuda et al., 2001). Just as with the other two-step etch-and-rinse adhesives, bond degradation could be prevented to a certain extent by a surrounding enamel seal (Hashimoto et al., 2002). Only one study investigated the in vitro durability of the three-step etch-and-rinse adhesive, All-bond 2 (Gwinnett and Yu, 1994). In that study, the shear bond strength decreased significantly after 6 months of water storage, whereas no increase in microleakage could be detected over the same period. Although excellent in vitro bonding effectiveness can be obtained with acetone-based adhesives, the high technique-sensitivity of these types of adhesives may explain the less-than-optimal long-term results (Tay et al., 1996a; Van Meerbeek et al., 2001, 2003). These findings have been confirmed by clinical research. Almost 50% of the Class V restorations restored with One-step debonded during a three-year observation period (Burrow and Tyas, 1999; Van Dijken, 2000; Baratieri et al., 2003). All-Bond 2 performed somewhat better, with retention rates of 98% after 1 year (Ianzano and Gwinnett, 1993) and 72% after 3 years (McCoy et al., 1998).
(5.2.3) Water-based adhesives
Some supposedly lower-technique-sensitive adhesives use water-based primers. The most-studied commercial representatives are the three- and two-step etch-and-rinse adhesives, Scotchbond MP and ScotchBond 1, respectively (3M ESPE). No difference has been detected in short-term bonding effectiveness between the three- and two-step versions (Dunn and Söderholm, 2001; Armstrong et al., 2003; De Munck et al., 2003b). The bonds produced with the three-step version resisted dynamic loading in a fatigue test (Ruse et al., 1995; Frankenberger et al., 1999, 2003b). In the long run, the bonding effectiveness of Scotchbond MP is prone to some degradation by thermo-cycling as well as by water storage (Krejci et al., 1994; Blunck and Roulet, 2002; Armstrong et al., 2003; De Munck et al., 2003b; Hashimoto et al., 2003a). After in vivo functioning, the µTBS of ScotchBond MP decreased due to hydrolytic break-down of the interface (Hashimoto et al., 2000). It has been hypothesized that this reduced durability is related to the incorporation of a high-molecular-weight polyalkenoic-acid co-polymer (De Munck et al., 2003b). Phase separation was shown to occur, with the co-polymer being filtered out by the collagen network and deposited as a distinct gel on the surface of the collagen network (Van Meerbeek et al., 1996a; Eliades et al., 2001). In the extreme case, the gel may hinder adequate resin-interdiffusion, leading to hybrid layers consisting of collagen mainly infiltrated by the low-MW 2-hydroxyethylmethacrylate (HEMA) that was polymerized to linear poly-HEMA chains, and of residual water (solvent) that was insufficiently removed (and/or kept in situ due to HEMA). This rather poorly infiltrated and polymerized hybrid layer is more susceptible to degradation. In most studies, the two-step derivative, Scotchbond 1 (Single Bond outside Europe), seemed to lose much more of its bonding capacity, regardless of the artificial aging factor applied (Miyazaki et al., 1998; Okuda et al., 2001; Blunck and Roulet, 2002; Nikaido et al., 2002b; Armstrong et al., 2003; De Munck et al., 2003b; Giannini et al., 2003; Shirai et al., 2005). However, the results from clinical studies with Scotchbond 1 do not always support the laboratory longevity studies. In one study, this two-step etch-and-rinse adhesive showed an excellent performance in a Class V clinical trial after 6 months of service (Perdigão et al., 2001), while in a similar study the adhesive did not even fulfill the minimum ADA requirements (Brackett MG et al., 2002; Brackett WW et al., 2003). This considerable variability in bonding effectiveness is probably due to an operator-dependent technique-sensitivity. Over- or under-drying of acid-etched dentin, a very technique-sensitive step (Kanca, 1992; Tay et al., 1996a,b), cannot explain this varying clinical effectiveness, since this particular adhesive is relatively insensitive to the amount of drying (Van Meerbeek et al., 1998; Perdigão et al., 2001). Conversely, in vitro research indicated that, with increasing thickness of the bonding layer (Zheng et al., 2001), bond strengths decrease, probably because of incomplete solvent evaporation when thick layers of this water-based adhesive are applied. In the clinical study in which Scotchbond 1 performed rather poorly, each layer was only lightly air-dried (Brackett MG et al., 2002). This may have resulted in poor solvent evaporation and thick adhesive layers. The application of a solvent-free bonding layer in a third step may overcome problems associated with residual solvent and fulfills the need (in light of the ‘elastic bonding’ concept) for a thick bonding layer at the same time (Kemp-Scholte and Davidson, 1990; Van Meerbeek et al., 1993). This hypothesis is corroborated by the excellent clinical results (up to 3 years) for the three-step version, Scotchbond MP (Van Meerbeek et al., 1996b; Browning et al., 2000; Ozgunaltay and Onen, 2002).
(5.2.4) Conclusions regarding etch-and-rinse adhesives
Class V clinical trials have shown that most etch-and-rinse adhesives fulfill the ADA requirements, when the adhesive was applied correctly. Some adhesives are more difficult to apply and thus are less preferred in demanding clinical situations:
Consequently, three-step ethanol-water-based etch-and-rinse adhesives are still regarded as the ‘gold standard’ in terms of bond durability, especially in demanding cavity preparations that have exposed dentin margins.
(5.3) Self-etch Adhesives
Self-etch adhesives make use of acidic monomers that simultaneously etch and prime dentin. Without the need for rinsing, the application time of self-etch adhesives is shorter and the technique-sensitivity lower. In its most conventional form, the application of the self-etch primer is followed by the application of a hydrophobic bonding resin (Fig. 1). Just as was the case with etch-and-rinse adhesives, simplified adhesives that combine the (self-etch) primer with the adhesive resin were developed to create what is known as the one-step self-etch adhesives or so-called ‘all-in-one’ adhesives.
Because of the limited amount of in vitro and in vivo data available on these recently developed products, however, it is impossible to evaluate all one-step and two-step, mild and strong self-etch adhesives separately. Nevertheless, the bond produced by mild two-step self-etch adhesives to dentin seems quite stable in vitro: (1) No significant decrease in µTBS occurred after 1 year of function in vivo (Sano et al., 1999; Takahashi et al., 2002), although SEM fracture analysis revealed an increased porosity within the adhesive resin; (2) up to 30,000 thermal cycles did not significantly decrease the shear bond strength (Miyazaki et al., 1998); (3) combined thermal and occlusal loading resulted in a small but statistically insignificant decrease of µTBS (Nikaido et al., 2002a); and (4) long-term water storage decreased the µTBS of two-step self-etch adhesives (Armstrong et al., 2003; Shirai et al., 2005), but not more than other adhesive systems.
There is little information available regarding the durability of one-step self-etch adhesives. In general, their short-term bonding effectiveness is disappointing (Inoue et al., 2001a; De Munck et al., 2003a, 2005b), and must certainly compromise their long-term usefulness. An example of these one-step self-etch adhesives is the ‘strong’ self-etch adhesive Adper Prompt (3M ESPE), and its predecessor Prompt L-Pop, for which some information on bond durability is available. In two in vitro studies, the µTBS of water-stored specimens was so low that most specimens did not survive specimen processing (Armstrong et al., 2003; Shirai et al., 2005).
There is even less information available on the durability of self-etch adhesives bonded to enamel. In one study, the bond strength decreased after thermo-cycling (Miyazaki et al., 2000). However, this reduced bonding capacity seemed to have little influence on the in vivo bonding to enamel (Peumans et al., 2003b).
Despite the limited amount of in vitro data, it can be concluded that the bonds obtained by ‘mild’ two-step self-etch adhesives seem quite durable, in contrast to ‘all-in-one’ adhesives that produce less durable bonds in vitro. These findings are corroborated by clinical research: The ‘strong’ one-step self-etch adhesive, Prompt L-Pop, performed very poorly, with a retention rate of 65% after 1 year (Brackett WW et al., 2002), in contrast to the ‘mild’ two-step self-etch adhesive, Clearfil SE Bond, that exhibited excellent results for up to 2 years (Peumans et al., 2003b; Türkün, 2003).
(5.4) Glass Ionomers
Currently, glass ionomers are the only self-adhesive restorative materials, due to ionic-bond formation between the carboxyl groups of polyalkenoic acid with hydroxyapatite (HAp) at the tooth surface (Yoshida et al., 2000). Currently, only one resin-modified glass-ionomer adhesive is commercially available (FujiBond LC, GC) to bond resin composites to enamel and dentin. Though the initial bond strength was quite favorable, the µTBS decreased over time as a result of water storage (De Munck et al., 2004). This correlated well with the decrease in µTBS of almost 50% after 1 year of in vivo function (Takahashi et al., 2002). This degradation process must probably be ascribed to decreased material properties rather than to decreased bonding potential. The clinical bonding effectiveness seems not to be affected, however, since the retention rates in clinical Class V trials (up to 5 years) are very high and approximate those achieved with the three-step etch-and-rinse ‘gold standard’ (Tyas and Burrow, 2001, 2002; Loguercio et al., 2003; Peumans et al., 2003a). Despite the esthetic disadvantages (Gladys et al., 1999), resin-modified glass-ionomer restoratives provide a viable alternative for Class V restorations (Gladys et al., 1999; Folwaczny et al., 2000; Van Dijken, 2000; Brackett MG et al., 2002).
(6) CONCLUSION
Non-carious Class V clinical trials remain the ultimate test method for the assessment of bonding effectiveness, but in addition to their high costs, they are time- and labor-consuming, and they cannot identify the true cause of clinical failure. Therefore, several techniques to predict bond durability in vitro have been developed. Most in vitro durability studies mimic one of the in vivo degradation factors involved, to disclose its effect on the general degradation process, in contrast to the clinical situation, where all these factors are operational simultaneously.
When in vitro and in vivo bonding effectiveness data were correlated, some clear associations were apparent. Adhesives that performed less well in several independent laboratory studies also appeared to be less clinically effective. Consequently, in contrast to common belief, clinical effectiveness of adhesives can be predicted. This analysis also revealed that, despite the fact that adhesives are sensitive to mechanical fatigue phenomena, the major factor affecting durability in vivo is hydrolysis of interface components, such as collagen and resin, and subsequent elution of the breakdown products. Currently, the most validated method for the assessment of this degradation process in vitro is the storage of micro-specimens in water. Within about 3 months, all classes of adhesives exhibited mechanical and morphological evidence of degradation that resembles in vivo aging. Consequently, the hydrolytic stability of cured adhesives is of crucial importance. The best way to achieve this goal is to apply a solvent-free, neutral-pH, hydrophobic adhesive resin layer in a separate step. This has been confirmed by the inferior in vitro and in vivo results obtained with two-step etch-and-rinse and one-step self-etch adhesives. These adhesives, because of their hydrophilic nature, act as semi-permeable membranes, attract water, and degrade faster than hydrophobic adhesives.
A comparison of contemporary adhesives reveals that the three-step, ethanol-water-based etch-and-rinse adhesives remain the ‘gold standard’ in terms of adhesion durability. Any kind of simplification in the clinical application procedure results in a loss of bonding effectiveness. Only two-step self-etch adhesives most closely approach this standard, and have additional clinical benefits such as ease of manipulation and reduced technique-sensitivity.
(7) RECOMMENDATIONS FOR FUTURE RESEARCH
The purpose of this section is to point out some shortcomings in the currently available adhesive systems that, if improved, may enhance bond durability. A topic already mentioned is the hydrolytic instability of the various components at the tooth-biomaterial interface. Some adhesive monomers (e.g., phenyl-P) are known to hydrolyze over time and so compromise long-term bonding effectiveness (Yoshida et al., 2004). Also, hydrolysis of exposed collagen fibrils, even enhanced by host-derived metalloproteinases, is an issue (Pashley et al., 2004). A solution to this problem might be achieved, for instance, by: (1) only partially exposing these fibrils, as do mild self-etch and glass-ionomer adhesives; or (2) incorporating the appropriate degradation inhibitors into the bonding procedure.
Micro-mechanical interlocking is believed to be a prerequisite to the achievement of a strong mechanical bond as shown by µTBS testing. Additional chemical adhesion may be beneficial in terms of durability, since it ensures an intimate adaptation of both the substrate and biomaterial components, thereby preventing nanoleakage (Sano et al., 1999). Two components of the tooth substrate can be used to bond to collagen fibrils and hydroxyapatite crystals. Binding to collagen fibrils can be achieved only by hydrogen bonds; however, these are relatively weak and unstable, especially in an aqueous environment. However, the calcium available in hydroxyapatite can serve as as a receptor for stronger ionic bond formation (Yoshida et al., 2004).
Most current durability studies focus on the adhesion to dentin and consider the enamel bond as strong and stable. Indeed, it has been well-established that strong and durable bonds to enamel can be produced by the well-established phosphoric-acid etch technique, followed by the application of a fluid hydrophobic adhesive resin. Contemporary two-step etch-and-rinse adhesives and self-etch adhesives, however, make use of resins that are more hydrophilic and conditioners that are less acidic (at least for the self-etch adhesives). Though the use of these new formulations raised the dentin bonding effectiveness of some new adhesives to a level comparable with that of the ‘gold standard’ three-step etch-and-rinse approach, their bonds to enamel may be compromised. Also, a tight protective enamel-resin seal may not necessarily be accomplished, since some of these adhesives act as semi-permeable membranes. Unfortunately, these are only speculations, since nearly no research has been performed to determine the durability of contemporary simplified adhesives bonded to enamel.
A trend in adhesive dentistry is to provide simpler and faster adhesives. Therefore, many manufacturers have launched an ‘all-in-one’ adhesive. None of the contemporary ‘all-in-one’ adhesives, however, can compete with the more traditional multi-step adhesives. Reasons for this inadequate performance are numerous, but the most important are that: (1) they are too hydrophilic and act, even after polymerization, as semi-permeable membranes; (2) because of the high solvent concentration, it is impossible to obtain an adhesive resin layer of adequate thickness and void from residual solvent; (3) during solvent evaporation, the monomer/water ratio may change and subsequently result in phase separations and blistering; and (4) the acidic components of these adhesives may also adversely interact with the initiator system of the composite and so weaken the bonding complex. If these problems are properly addressed, it must be possible to produce a universally applicable and reliable adhesive that is easier and faster to manipulate.
All current resin composites shrink by about 2–4%. The resultant polymerization shrinkage stress puts the immature bond under severe tension, which may lead to early failures of, especially, simplified adhesives. A non-shrinking composite may allow the bond to mature and so improve the short- as well as the long-term bonding effectiveness.
The ultimate goal is to develop self-adhesive restorative biomaterials that no longer need an adhesive for bonding to tooth tissue. Although glass ionomers and their derivatives can be considered as self-adhering restoratives, they lack other clinically relevant properties, such as sufficient mechanical strength, wear resistance, polishability, and, thus, esthetics. In brief, the concept of ‘minimally invasive dentistry’ may require a different set of restorative materials with physical and chemical properties adapted to the biomechanical needs of these new techniques.
ACKNOWLEDGMENTS
This paper is based in part on a thesis submitted to the Faculty of Medicine, Katholieke Universiteit Leuven, in partial fulfillment of the requirements for the degree of Doctor of Medical Sciences.
Received February 25, 2004; Accepted August 27, 2004
REFERENCES
Abdalla AI, Davidson CL (1990). Effect of mechanical load cycling on the marginal integrity of adhesive Class I resin composite restorations. J Dent 24:87–90.
Agee KL, Pashley EL, Itthagarun A, Sano H, Tay FR, Pashley DH (2003). Submicron hiati in acid-etched dentin are artifacts of desiccation. Dent Mater 19:60–68.[ISI][Medline]
Armstrong SR, Boyer DB, Keller JC, Park JB (1998). Effect of hybrid layer on fracture toughness of adhesively bonded dentin-resin composite joint. Dent Mater 14:91–98.[ISI][Medline]
Armstrong SR, Keller JC, Boyer DB (2001a). Mode of failure in the dentin-adhesive resin-resin composite bonded joint as determined by strength-based (µTBS) and fracture-based (CNSB) mechanical testing. Dent Mater 17:201–210.[ISI][Medline]
Armstrong SR, Keller JC, Boyer DB (2001b). The influence of water storage and C-factor on the dentin-resin composite microtensile bond strength and debond pathway utilizing a filled and unfilled adhesive resin. Dent Mater 17:268–276.[ISI][Medline]
Armstrong SR, Vargas MA, Fang Q, Laffoon JE (2003). Microtensile bond strength of a total-etch 3-step, total-etch 2-step, self-etch 2-step, and a self-etch 1-step dentin bonding system through 15-month water sto
