A gap-free restoration is considered as a key factor for the optimal clinical performance of resin-composite restorations. [8, 9, 10, 11, 12] Despite the fact that there is no threshold gap-size for the occurrence of secondary caries, the presence of marginal defects provides potential pathways for bacteria penetration and as a consequence, jeopardizes longevity and clinical success of adhesive restorations. 
According to the results of the present study, the null hypothesis was rejected since both marginal and internal adaptation of resin-composite restorations on dentin walls were found to be negatively affected by the presence of saliva, when contamination took place after the adhesive application. Reapplication of adhesive restored adaptation and could be considered as the decontamination technique of choice. On the other hand, the saliva contamination that occurred after acid-etching and before adhesive application did not adversely affect the adaptation in a significant level. Air-drying for 5 seconds appears to be sufficient in such cases, with no need for water-rinse of the operation field.
An important outcome of the present study was that %DM and MG values were not significantly affected when saliva contaminated the etched dentin and following removed either with air or with water. These findings are in agreement with other studies which showed that adhesion was not negatively affected when the duration of contamination was between 5 and 15 seconds.     [16, 17, 18] Notably, in 2-step etch and rinse adhesives, saliva contamination did not prevent hybrid layer formation or the resin penetration into the dentin tubules.  For extended contamination time (60 seconds), re-etching of the contaminated dentin effectively removes contamination, but it appears to also deteriorate the quality of adhesion due to the collapse of denatured collagen fibrils, increased thickness of the collagen layer and inadequate hybridization.  
Original and internal adaptation. These results are consistent with a similar study on contamination, where it was shown that salivary contamination after light-curing of the adhesive reduced the shear bond strength of a two-step etch-and-rinse adhesive system to about 50% of control values.  It was hypothesized that the absorbed glycoproteins onto the poorly polymerized adhesive surface, inhibit adequate copolymerization. Another possible explanation suggests that rinsing and air-drying provides a collapsed layer of resin deprived collagen which cannot be effectively penetrated. [20, 22]
Conversely, several studies concluded that there was no adverse effect on bond strength values of two-step self-etching adhesives when contamination occurred after light-curing. [23, 24] This can be possibly explained by the simultaneous demineralization and infiltration achieved by these systems. Self-etching adhesives contain acidic monomers combined with hydrophilic monomers that simultaneously etch and prime dentin. Hydrophilic and acidic dentin adhesives are considered to be less sensitive to the presence of saliva and researchers have concluded that bond strength values to contaminated dentin are not negatively affected.  
As far as the width of the formed gaps is concerned, mean values ranged between 0 and 30.74 μm, which are similar to the ones reported elsewhere.  At 0.5 mm in-depth level, groups 2 and 4 present particularly high SD values. At those groups, two extreme gaps were measured, which could be possibly attributed to false technique application. The presence of too much moisture on the substrate can create small, blister-like voids which appear like gaps in the interface and which have an impact on the performance of the adhesive material. Incomplete hybridization of the demineralised dentin leaves exposed collagen fibrils and causes nanoleakage through gaps of 20-100 nm size. 
Numerous factors are involved to the debonding phenomenon either during bond development or during the oral service of a restoration. Although several studies [4, 7]  [5, 14] have previously investigated the effect of saliva contamination on the adhesive interface, there is no sufficient evidence regarding the effect of saliva contamination on the adaptation of adhesive restorations. Parameters such as the extent of the debonded margin and the width of the marginal gap determine the quality of the adhesive interface and the ability of the adhesive system to prevent marginal debonding.  The marginal adaptation of the restorations is not necessarily similar to the internal adaptation to the cavity walls; therefore, in the present study, the internal adaptation was additionally evaluated as a factor with crucial contribution to the long term good clinical performance of the restorations.  In the current study, the cavity configuration factor was the same for all samples and only one adhesive system and resin-composite were used, to avoid interference by the aforementioned factors. Consequently, differences could be attributed to the different stages of contamination and decontamination techniques applied.
As already stated, the vast majority of previous studies on saliva contamination evaluate the effect of contamination on bond strength. [4, 7, 11] In the present study, marginal adaptation and gap formation were evaluated in order to obtain information concerning the sealing ability of the adhesive system on the contaminated dentin. Under the clinical point of view, saliva contamination is possible to occur either after the etching procedure and before the adhesive application or after the completion of the adhesive placement. Both conditions were investigated in the current study. As an experimental procedure, marginal and internal adaptation is considered to be laborious and time-consuming technique. It often yields false negative results and requires a large number of specimens in each group because of the high variability of the values. Despite the above limitations, it is thought to be more clinically relevant to evaluate the capability of an adhesive to maintain the tooth-restoration interface sealed, rather than to compare bond strength values. [27, 28]
Direct optical microscopic observation of the specimens was preferred to evaluation of replicas by a scanning electron microscopy (SEM) since the first method allows measurements directly on dentin specimens instead of replicas, providing reliable phase identification and allowing evaluation of the same specimen at different levels. It is a simpler, less time-consuming and less destructive method, which avoids any potential drawbacks of the replica technique. 
In this study, human saliva was chosen as the contaminant, in order to achieve clinically relevant conditions, contrary to artificial saliva and substitutes which were used in other studies. [19, 24] SEM observations have shown that high-molecular-weight proteins diffuse into dentin tubules, altering the surface characteristics and competing with hydrophilic monomers during the hybridization process. [5, 25] Fresh whole saliva is considered an acceptable material to be used for contaminating testing as artificial saliva, which is deprived of macromolecules, may confound the results. 
Saliva was left undisturbed for 5 seconds prior to the application of any decontamination technique, with the hypothesis that this is an average time for the clinician to notice and deal with the contamination. The time span during which the contaminant interacts with the dentin surface is of significant importance, as longer contamination time results in lower bond strengths than drying the saliva quickly.  This observation was related to evaporation of water and formation of a thick film of glycoprotein on the contaminated surfaces. The duration of contamination varies significantly among studies and it is a factor that should be particularly considered when comparing their findings.