Marginal Adaptation of Direct Class II Composite Restorations with Different Cavity Liners

Light-cured resin composites (RCs) benefit today by an increasing spectrum of indications; their application has been expanded to include the restoration of anterior tooth wear^1-6  and noncarious cervical lesions.^7-10  Therefore, RCs continuously face challenges associated with specific environments, such as high tensile stresses in cervical areas as well as increased expectations with regard to longevity. Long-term stability is considered to be affected to a significant degree by marginal integrity.^11-13 

The marginal seal of composite restorations represents an increasing concern. Over the years, in vitro evaluations of the performance of resin adhesives revealed that microleakage and gap formation, mainly at the dentin-composite interface, did not improve at the same rate as did bond strength values.^13-16  Independent of the bonding capacity of an adhesive system, it seems that adhesive restorations are far from assuring a perfect marginal seal, with degradation in time occurring regardless of the product used.¹⁷  Marginal integrity is considered to be the result of several parameters related to the forces created by curing contraction, as bond strength alone could not be correlated to marginal adaptation (MA) in a study undertaken by Kemp-Scholte and Davidson.¹⁸ 

A consensus can be found in the literature with regard to the effect of curing contraction in restoration performance. As a result of the confinement of the restoration, curing contraction has been shown to generate strain along the tooth restoration complex, leading to pre-stressed obturations, which are more prone to long-term degradation. The unfavorable effects of polymerization contraction forces can be evaluated effectively by marginal analysis, as marginal gaps offer proof of the forces generated by polymerization shrinkage or by thermo-mechanical strain, exceeding the bond strength.¹⁵  Other methods of assessing the effects of polymerization contraction are finite element analyses (FEA) of the stress distribution,^10,19-21  measurements of tooth deformation^22-27  and, recently, a three-dimensional (3D)-deformation analysis.²⁸ 

The phenomenon of force development in contracting materials was first described in the dental literature by Bowen.²⁹  Later, the subject of polymerization contraction stress was studied in depth in an attempt to clarify the complex interaction among the many factors involved. An interplay of polymerization shrinkage, elastic modulus (E-modulus), viscous flow capacity, and conversion degree of the material as well as adhesive ability, cavity configuration, and tensile stresses exerted on the restoration^{12,14,19,30-33}  is considered responsible for the marginal integrity. The combined action of several factors creates various stresses on the restoration-tooth interface, with consequences such as debonding or strain on the remaining tooth structure creating enamel fissures, fractures,³⁴  and cuspal movement.^22,25,35  Authors of studies^20,23,27  employing the FEA method also developed maps of polymerization stress accumulation, showing the distribution patterns of the forces and the effect on remaining tooth structures.

The dimensional modification due to shrinkage depends on the number, size, and functionality of the monomers as well as on the filler load. The volumetric contraction undergoes two phases: the phase before and the phase after the gel point. During the first phase, the resin retains its capacity to flow, and, therefore, it compensates the contraction forces by rearrangement of the molecules by preventing strain that would otherwise develop at the interfaces. During the second phase, however, the contraction forces are directed toward the bonded surfaces, leading to, among other things, defects in the MA. These defects occur when the elasticity of the material is not sufficient to compensate for the contraction forces, thus allowing for stress to develop at the interface of the tooth restoration complex. High strain capacity (reduced E-modulus), high viscous flow, and small amounts of polymerization shrinkage lead to reduced gap formation.¹⁹ 

Extensive research was done on the implications associated with the E-modulus of the restorative material. Flexibility is defined as the ability of a material to strain without becoming permanently deformed.³⁶  The higher the E-modulus and the polymerization shrinkage of the composite, the higher the contraction stress will be.^12,20,23,37  According to physical laws and in vitro observations, a low E-modulus is supposed to play an important role in stress relief, becoming, therefore, a desired property of the material.^19,20  It is assumed that an increased flexibility would lower stress values on the interface.^19,20  However, both the E-modulus and contraction are related to the filler content: a higher filler load would lead to a lower contraction but to an increase in stiffness as well.^{12,14,31,36,38}  A way to overcome this difficulty within material composition involves the use of a flexible material as liner under the restorative material, giving rise to the ‘elastic wall' theory. Based on the hypothetical benefits of increasing the compliance of the prepared cavity artificially, low-modulus intermediate layers such as flowable RCs and unfilled adhesives were presented as stress breaking materials.^33,36,38,39 

Different studies^18,20,37,40  measured stress relief when an unfilled resin with increased thickness was used as intermediate stress breaker, and an FEA study²⁰  was able to create a stress map of the tooth-restoration complex measuring different stress relief values for different thicknesses of the adhesive layers. However, studies evaluating the influence of flowable composites placed as intermediate layers on marginal integrity show conflicting results. Based on previous research, a general opinion prevails indicating that the behavior of a material as a stress absorber cannot be predicted using the E-modulus alone, as polymerization kinetics is a complex material-specific phenomenon.^12,38,41  Therefore, controversy is still present in the literature regarding the supposed positive effect of flowable composites on stress relief and marginal integrity.

Despite this high relativity, some hypotheses have been promoted and need further investigation. Accordingly, some authors^19,37,42  suggest the use of an extremely low E-modulus in order to obtain a significant effect, while others^37,42  suggest placing a pre-cured layer on top of the adhesive regardless of the composition (restorative or flowable RC).

The null hypothesis tested was that no difference in terms of percentage of continuous margin (% CM) exists among restorations with liners of different elasticity. The independent variables that were kept constant were loading condition (before and after loading), cavity lining (no lining; Optibond FL liner; Herculite XRV Ultra; and Premise Flowable) and tooth substrate (occlusal enamel [OE]; proximal enamel [PE]; and cervical dentin [D]). The only tested variable was MA, expressed in % CM.

Forty Class II occluso-mesial cavities, with margins extending 1 mm below the cemento-enamel junction, were prepared in extracted human third molars (for each tooth only one cavity was prepared). The teeth were stored in 0.1% thymol solution, and after scaling and pumicing, teeth were randomly assigned to four experimental groups (n=10) and mounted on custom-made specimen holders using a cold polymerizing resin (Technovit 40721, Haereus Kulzer, Wehrheim, Germany). Prior to the mounting, the apices were sealed with an adhesive system (Optibond FL, Kerr, West Collins, CA, USA). Diamond burs (80 μm; Universal Prep set, Intensiv SA, Lugano, Switzerland) were used under continuous water spray. The standardized dimensions for the proximal cavity were as follows: width = 5 mm, height = 6 mm (measured from the tip of the palatal cusp to the gingival margin), depth = 2 mm (mesio-distal width of the gingival floor). For the occlusal cavity, the dimensions were as follows: depth = 3.5 mm (measured from the tip of the palatal cusp) and width (vestibule-oral dimension) = 5 mm.

For all groups, a nano-hybrid composite (Herculite XRV Ultra) was used for the restoration. A three-step etch-and-rinse system (Optibond FL) with a proven bonding capacity, a 48% filler load, and a very low E-modulus was used as the adhesive. The flowable composite Premise Flowable was chosen for its low polymerization shrinkage relative to other flowable composites and for its high percentage of filler load, assuring good mechanical properties as the lining material is in direct contact with the oral environment (Figure 1). All of the above-mentioned materials were provided by the same manufacturer, therefore assuring compatibility. The properties of the mentioned materials are listed in Table 1.

Restoration was performed immediately after cavity preparation in order to avoid alteration of dental tissues. All cavities were beveled in the occlusal box. The teeth were randomly assigned to one of the four groups, as follows.

The cavities were encircled with a metallic matrix band after conditioning enamel and dentin with a 37% H₃PO₄ gel (Gel Etchant, Kerr) for 45 seconds and 15 seconds, respectively. Dental tissues were rinsed and nearly dried with a light air pressure spray (two to three seconds) before the application of the two adhesive components (Primer and Adhesive), following the manufacturer's instructions. The bonding resin was light-cured (Bluephase, Ivoclar Vivadent, Schaan, Liechtenstein) for 40 seconds. This step was preceded by the buildup of the proximal wall with a unique vertical increment of less than 1-mm thickness (centripetal layering technique). The vertical increment of the proximal box was cured for 40 seconds from the occlusal direction. The matrix band was then removed, and the composite resin was applied using a three-increment layering technique for the occlusal cavity (Figure 1). Each increment in the occlusal cavity was 1.5 mm thick and was individually cured for 40 seconds (light intensity produced by the curing device = ∼1000 mW/cm²).

The same steps were performed as for group A. After polymerization of the adhesive system, a thin lining layer of Premise Flowable (1-mm thickness) was applied on the bottom of the occlusal cavity and the proximal box (exposed to external environment) and polymerized before the buildup of the proximal wall. The occlusal cavity was further filled, after removal of the matrix band, using the same composite, Herculite XRV Ultra, and the same layering technique as described for group A (Figure 1).

The same steps were performed as for group A until cavity restoration. After polymerization of the adhesive system, a thin lining layer of the restorative composite Herculite XRV Ultra (1-mm thickness) was applied on the bottom of the occlusal cavity and the proximal box and was polymerized before buildup of the proximal wall. The occlusal cavity was further filled after removal of the matrix band using the same composite, Herculite XRV Ultra, and the same layering technique as described for group A (Figure 1).

The same steps were performed as for group A until cavity restoration. After polymerization of the adhesive system, a thick layer of the adhesive from the Optibond FL adhesive system (1-mm thickness) was applied as lining material on top of the same Optibond FL adhesive system and was polymerized before the buildup of the proximal wall. The occlusal cavity was further filled after removal of the matrix band using the same composite, Herculite XRV Ultra, and the same layering technique as described for group A (Figure 1).

Seventy-two hours after polymerization, the restorations were finished and polished using flexible disks with different grain sizes (SofLex PopOn, 3M-ESPE, St Paul, MN, USA). Immediately after completion of the polishing procedure, impressions were made of each restoration with a polyvinyl-siloxane impression material (President Light Body, Coltene, Alstätten, Switzerland). Subsequently, epoxy replicas were prepared for the computer-assisted quantitative marginal analysis in a scanning electron microscope (SEM; XL20, Philips, Eindhoven, The Netherlands) at 200× magnification. Continuous margins were assessed separately in D (% CM in D), PE (% CM in PE), and OE (% CM in OE), as percent of the total length of the analyzed margins. Total margin length represents the data from all of the margins of a Class II cavity, which is composed of OE, PE, and D. Additionally, enamel fractures (EF) parallel to the margins were documented in the SEM.

After storage at 37°C in water in the dark for two months, the restored teeth were loaded with repeated thermal and mechanical stress in a chewing machine. Thermo-cycling was carried out in flushing water with temperature changing 600 times from 5°C to 50°C and back with a dwell time of two minutes at each temperature. The mechanical stress comprised 200,000 load cycles transferred to the center of the occlusal surface with a frequency of 1.7 Hz and a maximal load of 49 N. Epoxy replicas were prepared for the computer-assisted quantitative marginal analysis in a SEM, using the same parameters as were used before the loading procedure.

Statistical analysis was done using the Shapiro-Wilks test, and differences between groups were statistically evaluated using analysis of variance and the Duncan post hoc test at p<0.05. The difference between results related to loading conditions (before and after the fatigue test) was analyzed using the statistical method of the paired t-test. MA was the only dependent variable.

Only the means and standard deviations for the percentages of CM and EF are reported, as ‘overfilled margins' and ‘underfilled margins' did not exceed 5% in any group. The scores for the MA are displayed separately for PE, D, and OE in Tables 2 through 5.

Before loading, the MA in cervical D was superior (p<0.05) for groups C and D compared to groups A and B. In PE and OE, all groups showed the same behavior (p>0.05) and it was noticed that the % CM in PE was relatively low compared to the % CM in OE (Tables 2 through 5).

After two months of water storage and after thermal and mechanical stressing, the MA decreased considerably in all four groups at a statistically significant level. A paired-samples test showed significant differences (p=0.000) between both testing intervals before and after loading in all of the marginal segments tested (occlusal, proximal, and terminal).

Significant differences between groups were found for the criteria % CM in D and % CM in PE. In D, the superiority of the group D was clearly outlined (p<0.05) compared to all of the other three groups (Table 2). In PE, significant superiority was found only for group C compared to group A, with no statistically significant difference between the other liners. The % of enamel fractures (% EF) was high after loading, the application of the liners showing no statistically significant improvement of the MA (Table 5).

All restorations, both with and without lining material, presented discontinuous margins. The null hypothesis could be rejected, as statistically significant differences could be demonstrated between groups.

The present study was designed to test the classical presumption about the positive effect of flowable composite liners (Premise Flowable in this experiment) and to gain a better understanding of the influence of the E-modulus on marginal sealing of composite restorations before and after artificial mechanical and thermal aging. For this purpose, the study was designed to include a group with a pre-cured layer of an extremely elastic material, such as the adhesive of the adhesive system Optibond FL, as well as a group with a thin lining (1-mm thickness) of the same restorative composite used for the filling (Herculite XRV Ultra). The significant difference between the E-modulus of the three different cavity linings could therefore add more to the knowledge of marginal sealing of composite restorations than could be gained by using only groups with different flowable composites as liners.

SEM is a widely used method with which to evaluate MA. Direct observation by SEM is difficult as a result of the vacuum procedure utilized during SEM, which causes cracks. By using a replica method, artificial gap formation can be avoided.⁴³  The specimens were polished after 72 hours and stored for two months before thermo-mechanical stressing in order to take into consideration the complex post-polymerization processes, such as ongoing contraction and stress relaxation by water absorption.^44,45 

While several other studies assessed the MA of restorative RC with and without a flowable liner through the dye penetration method, no marginal evaluation by SEM was done on a bonding product placed as liner (in this case the Adhesive of the Optibond FL adhesive system). Also unlike previous studies, our experiment was not designed to compare the effect of different flowable RC liners but rather to assess the influence of an E-modulus within a broad range (2-3 GPa and 9.4 GPa) (Table 1).

A group with the Herculite XRV Ultra used as a 1-mm pre-cured layer was introduced with the purpose of determining to which degree the improvement in MA is due to the pure layering procedure rather than to the properties of the lining materials.

Knowing the unpredictable effect on MA of a flowable composite placed as a liner, regardless of its E-modulus,^12,38,41,46  we preferred to use a flowable RC with rather optimized mechanical properties (high filler load), which could be beneficial to clinical use. Among other flowable RCs, the Premise Flowable has one of the lowest polymerization shrinkage rates, and the filler load is close to that of a restorative RC (72.5%).

The composite placement technique was chosen in order to minimize the polymerization shrinkage stress, especially at the cervical D wall. Therefore, a unique vertical layer was considered to have a minimal contact with the walls of the proximal box and a favorable ratio of unbonded to bonded surface.

The cavity configuration, the adhesive, and the restorative composite were kept constant, with the focus remaining on those parameters related to lining materials.

In dentin, the lowest values of marginal continuity were obtained for groups A (no liner) and B (Premise Flowable liner). The application of a thin liner (1 mm) of Herculite XRV Ultra led to significantly superior MA in dentin only before loading and only compared to the control group, but this finding was not significantly different compared to group D (p>0.05). This result is probably due to the benefits of placing a pre-cured layer, as previously described.^37,42  It is worth noting that, while not always statistically significant, the group with Herculite XRV Ultra lining showed generally better MA than did the control group. However, group D was significantly better than all of the groups after loading, indicating that layering is not the only factor contributing to the stress relief (Table 2).

Applying the flowable RC as a liner did not achieve better MA. Before loading, group B was significantly worse than group C. After loading, results showed no difference in MA by using the flowable liner compared to the results achieved by layering the composite (group C) or compared to the control group results (group A). Both before and after loading, group B was significantly worse than group D. A possible explanation, presented by other authors^12,38,41  as well, indicates that the benefits of the elasticity of the flowable liner were probably surpassed by its contraction.

After thermo-cycling, a significant improvement of MA in dentin was obtained only by placing a 1-mm-thick layer of Optibond FL on top of the same bonding product (Table 3). A clear positive effect on MA was observed in dentin before and after loading. This positive influence could be explained by the very low E-modulus of the Adhesive of the Optibond FL system, in the range of 1 to 3 MPa, as was predicted by some authors^20,33,40  and in accordance with other in vitro studies^18,36,37  showing a positive effect of thick layers of unfilled resins used as stress breaking liners under restorative composites. One FEA analysis⁴⁷  revealed an interesting finding in Class V cavities, in which only a very low E-modulus (2 GPa) was shown to be capable of absorbing tensions along the tooth restoration interface and thus preserving the cohesive and adhesive integrity of the tooth restoration complex.

In this study, the adaptation to PE was low and decreased significantly after loading. Some considerations are to be made regarding the evaluation of marginal integrity in PE, as follows:

• In all groups, the % of CM in PE after loading was low compared to OE (Tables 3 and 4). The most important cause is probably the lack of beveling PE, leading, therefore, to a weak adhesion prone to EF.⁴⁸  The comparison between groups reveals no differences in MA before loading as well as an almost significantly higher % EF for group B after loading (p=0.05), although at a high standard deviation. While the % CM in PE was generally very low, the only significant difference between groups was found after thermo-cycling in group C, which scored higher % CM compared only to group A. It should be noted that no significant difference was found among the three different liners. It would appear that in PE, the most influential on MA was the lack of beveling of the margin. Less important seems to be the type of liner, which was not placed directly on the PE margins. In this particular configuration, the Optibond FL liner seems to have only a local effect in its area of direct application, which is the cervical D margins.

• Contraction seems to contribute as well to the degradation of the marginal integrity. The paired-samples test showed significantly higher values for the % EF and significantly lower % CM in proximal cavities after thermo-cycling compared to before thermo-cycling. It is well documented²³  that areas pre-stressed by polymerization shrinkage show failure under occlusal load. Forces due to contraction created a residual strain within the tooth, leading, after loading, to the formation of cracks in the PE, a finding that is in accordance with those of the FEA analysis of Versluis and others.²⁷ 

• Moreover, as a result of the cavity geometry, the axial enamel in a mesio-occluso-distal (MOD) cavity compensates less for deformation under load, compared to OE margins.²⁷  Under thermo-mechanical load, sound teeth distribute applied stresses more homogeneously compared to the structurally modified teeth, which show a complex biomechanical behavior due to the interruption of tissue continuity.^23,27,33  Therefore, the high percentage of noncontinuous margins in PE observed after loading could also result from a differential behavior, under load, of the PE walls compared to the OE margins when the continuity of a tooth is interrupted by a MOD cavity.

Therefore, the interaction between shrinkage stress distribution patterns^27,28  and biomechanical balance^23,27  of an occluso-proximal cavity under loading condition, together with strength of adhesion (bevel), creates a complex and differential behavior of proximal and occlusal tooth structures.

Shrinkage stress development is a dynamic, nonlinear process, depending on the complex interaction of many material-dependent, cavity-dependent, and restoration placement–dependent factors. Contraction stress in a tooth restoration complex reveals a distribution pattern with forces acting differently along the interface.²⁷  In this study, the highest percentage of noncontinuous margins was found in cervical D, which is in accordance with the stress distribution pattern depicted by FEA studies.^20,23,27  Polymerization stress maps showed shrinkage stress along the interface, with maximum stress accumulation at the cervical zone, in our case the dentin-composite interface, also known to be the most problematic area for adhesion.

The results of the present study confirm recent findings indicating that a thick adhesive layer with extremely low modulus could relieve stress at the interface, and they reaffirm the unpredictable stress-absorbing effect of flowable composites as liners. Research on properties of flowable composites also showed substantial differences in E-modulus and volumetric shrinkage, indicating flowable composites to be a very inhomogeneous group of materials.^{12,38,41,46, 50-54}  One study³⁸  situated the E-modulus in a wide range, with values between 6.5 GPa and 12.5 GPa for 12 different flowable composites. It seems that the performance of flowable composites as a stress buffer remains unpredictable, as polymerization kinetics tend to be material specific and because the optimization of the elasticity and shrinkage, with opposing effects, is difficult to achieve. In other words, despite their low E-modulus, the contraction stress produced by some flowable composites could be high because of their high volumetric shrinkage, which approaches 6% in some products.^12,38,41  Other studies^22,24,55-64  evaluating the influence of flowable composites placed as intermediate layers,⁴⁹  on marginal integrity show conflicting results as well.

Based on the present research, it seems that a liner with a very high elasticity applied in a thick layer could redistribute shrinkage stress and contribute to maintaining integrity along the tooth restoration interface, as indicated by previous research.^{18,20,33,36,37,40} 

Although the % CM after load in cervical D was higher in group D, the % EF in PE was not significantly lower compared to the other groups, indicating that enamel did not benefit from the same stress relief as did the interface. Our results correlate as well with those of other studies^16,18,50,51  reporting that a layer of low modulus composite can lead to redistribution of the forces and some strain relief along the interface but does not improve bond strengths and therefore does not reduce the overall negative influence of curing contraction. In the present study, the liner was applied only on the cervical margin and pulpal wall, not on the buccal and lingual proximal walls. It seems that the liner of Optibond FL was efficient in reducing and redistributing stress at the interface predominantly in the area of its application, where the marginal seal is also known to be the most problematic because of the concentration of stress, which is not uniformly distributed along the cavity walls.^20,21,23  This area is also marked by high variation in the bond strength along the bonded surface.^37,52  In return, no benefit of the stress-relieving effect was found on the enamel, where the bevel seems to play a more important role in assuring a good bonding stability after the fatigue test.

Any attempt to predict the contraction stress development of a material during polymerization based on E-modulus or polymerization shrinkage value alone seems to be misleading.^{12,14,27,38,53}  This hypothesis is confirmed as well by the present study, in which the flowable composite used as liner did not allow for the expected stress relief, while the performance of the Optibond FL liner was more heavily influenced by its very low E-modulus and was less affected by the high shrinkage. Product-specific properties such as flow capacity and pregel and postgel shrinkage render the stress-relieving effect unpredictable.

However, using only a composition-based approach to relieve shrinkage stress may still be too simplistic. Ensuring optimal dentin bonding and using incremental layers of composite in appropriate configurations are factors within the control range of the practitioner, which could be of importance in assuring a better marginal seal.⁵⁴  Accordingly, marginal quality could be optimized by factors such as guidance of the shrinkage vectors, reducing the ratio of bonded to free unbonded restoration surfaces, and minimizing the mass of in situ-cured composite.

For the materials used in this study, no improvement of MA was found when the flowable composite was used as a liner, both before and after thermo-mechanical loading. The application of a layer of the same restorative composite used for the restoration with the same thickness as the flowable composite (1 mm) had a positive effect before loading, probably as a result of the benefit of placing a precured layer on the bottom of the cavity before stratification of the restoration. After thermo-mechanical loading, a significant improvement of MA in D was obtained only by placing a 1-mm-thick layer of the Adhesive of the Optibond FL system on top of the same product used as adhesive. This positive effect may be explained by its very low E-modulus, in the range of 1 to 3 MPa.

The authors of this manuscript certify that they have no proprietary, financial, or other personal interest of any nature or kind in any product, service, and/or company that is presented in this article.