This controlled, randomised, double-blind (patient and examiner) clinical trial with three groups (Filtek P60, Surefil and Suprafill) and similar allocation rate (split-mouth design) was approved by the local ethics committee of Universidade Federal de Pernambuco/PE-Brazil (protocol n. 43327415.3.0000.5208). The trial was registered at ReBEC (http://www.ensaiosclinicos.gov.br) under RBR-8gcd45 and followed the recommendations of the Consolidated Standards of Reporting Trials-CONSORT.

The study was held at the dental school of the Universidade Federal de Pernambuco (Recife, Pernambuco, Brazil). Patients older than 21 years (mean age 25.67±6.32) that attended the dental clinics were recruited for the study beginning in July 2002 [27]. The sample size was 20 occlusal restorations per group (60 in all). After 14 years, 7 patients were included in this study. Twenty-one restorations were evaluated between July and December 2016.

Inclusion criteria were as follows: 1) need for occlusal restorations in at least three posterior teeth with primary caries, or the need to replace a restoration; 2) good general health; 3) sound or adequately restored opposing teeth [27]. The exclusion criteria were as follows: 1) periodontal disease; 2) teeth with endodontic treatment; 3) teeth with painful symptomatology. After 14 years, patients who performed any modification in the teeth selected for the research (replacement of the restoration, extraction, total crowns or any other procedure that would impede the clinical evaluation) were also excluded.

A total of 60 restorations were done on 20 patients by the same operator. Each patient received 3 restorations, each one with one of the 3 evaluated restorative materials. The allocation sequence was determined following the quadrants (upper left, upper right, lower right and lower left). In the same quadrant, the restorations always began with the most distally located tooth. The split-mouth design assured that all patients received restorations from all three experimental groups [Table/Fig-1]. Brown sealed envelopes were used to assure allocation concealment and randomness. Patients and evaluators were blinded to which restorative material was placed [Table/Fig-2].

The restorative materials used are described in [Table/Fig-1]. Two packable (Filtek P60 and Surefil) and one microfill (Suprafill) resin composites were evaluated in this study. The microfill Suprafill served as a control group.

The restorative procedures were performed following the protocol: 1) cavity preparation using high-speed diamond burs and low-speed round burs; 2) rubber dam isolation; 3) prophylaxis with a brush and pumice stone (ultrafine/SS White, São Cristovão, Brazil); 4) bonding agent application; and (5) restorative procedures according to manufacturer’s instructions [Table/Fig-1]. In shallow and medium cavities, no pulp protection procedures were performed. In deep cavities (at least 0.5 mm of remaining dentin between the cavity floor and pulp chamber), glass ionomer cement was applied as a liner before the bonding agent. In very deep cavities (0.5 mm or less of remaining dentin), calcium hydroxide cement followed by glass ionomer cement was applied before the bonding agent. In very deep cavities with pulp exposure, Ca(OH)₂ solution was used, followed by the application of Ca(OH)₂ powder over the exposure, followed by the cement. Glass ionomer cement was then applied before the bonding agent. All photopolymerisation procedures were done using halogen light of 600 mW/cm² (Optilight Plus/Gnatus, Brazil) within the times recommended by the manufacturers.

Two examiners were previously calibrated (Kappa=0.88) using the USPHS (United States Public Health Service) criteria [Table/Fig-3]. Consensus answers were given for each parameter, where satisfactory restorations were scored as Alpha (ideal) or Bravo (clinically acceptable). Charlie and Delta scores were considered unsatisfactory therefore failure. The evaluation was performed with the aid of a dental mirror and explorer probe. For the colour match, the evaluation was done under natural light. The evaluation occurred at the baseline, 12 months and after 14 years.

The indirect evaluation was performed through replicas made of cast at baseline and after 14 years of placement. The impression of the restorations was obtained using condensation silicone (Precise SX Kit/Dentsply-Milford/USA) which was then poured with type IV gypsum (Troquel/Polidental-Cotia/Brazil). All cast models were analysed using spectral domain OCT at 930 nm central wavelength (SR-930, Callisto Imaging System, Thorlabs). The system has a 1.7 mm/1.3 mm imaging depth with 7.0 μm/5.3 μm axial resolution (air/water) and 8 μm lateral resolutions. The cast replicas were positioned over the translational stage of the scanner stand, in an individualised silicone device used for image standardisation [Table/Fig-4]. A 6-mm scanning range was done in the lingual-buccal direction in the 2-D mode. The mean values per group were calculated for further statistical analysis.

[Table/Fig-4]:

Individualised silicone device for indirect cast replica analysis.

The restoration wear was measured by analysing the distance (μm) between the cavosurface angles (buccal and lingual) and the deepest point of the central pit which served as reference points. For each point, the difference between baseline measurements and after 14 years of clinical use, was calculated. The amount of lost material was not analysed as the restorations were three-dimensional structures, rather the increase in the distance between the reference points was measured. [Table/Fig-5] illustrates schematically how the images were obtained and how the distance measurements were taken.

[Table/Fig-5]:

Schematic diagram of OCT analysis.

(a) Cast replica (Baseline and 14 anos); (b) positioning of the replicas in the central portion of the OCT and movement with micrometric buttons; (c) measurement between the cavosurface angles (buccal and lingual) and the deepest point of the central pit (line red and green).

Data were inserted into Excel and analysed descriptively. Inferential analysis was performed with non-parametric Friedman’s tests for the comparison between the resin composites. Wilcoxon test was applied for paired data in the comparison between wear assessments through OCT. The margin of error used in the decision of the statistical tests was 5% and the software used for the statistical analysis was the SPSS (Statistical Package for the Social Sciences), version 23.

Of 20 patients, contact was possible with 10 patients, but only seven accepted to participate in this study, accounting for 21 restorations [Table/Fig-6]. The sample loss can be explained due to the long-term evaluation; addresses and telephone contacts were modified. Those who were contacted did not respond in the same manner to the motivational stimulus for the recall visit. No follow-up was done between 1 and 14 years.

[Table/Fig-6]:

Consort flow diagram.

The distribution of the restorations and the restored teeth are summarised in [Table/Fig-2]. Restorations were equally distributed within posterior teeth as shown by Fisher’s-exact test (p=0.925). The results of the direct clinical evaluations are detailed in [Table/Fig-7]. Differences were not observed between the evaluated resin composites regarding each of the analysed parameters (p>0.05). However, for each resin composite, differences were observed between the observation periods through Friedman’s test (p<0.001), except for caries incidence. Baseline and 1-year results did not show any differences. Differences were observed when the 14-year data were compared. Colour match, anatomic form and marginal adaptation showed statistically significant results within the observation periods for all three resin composite. For marginal discolouration, only Filtek p60 showed differences within observation periods, at the threshold of significance (p=0.046).

After 14 years, failures (Charlie scores) were observed in all three groups for the colour match: 3 Filtek P60 restorations, 2 Surefilland 1 Suprafill. Suprafill also received 1 Delta score for the same parameter. Anatomic form and Marginal adaptation both showed failures for one restoration each of Filtek P60 and Suprafill.

The mean wear is shown on [Table/Fig-8]. For each resin composite, Wilcoxon signed rank test did not identify significant differences between buccal and lingual cusps (p>0.05). Kruskal Wallis test showed significant differences (p=0.022) between the results for the buccal cusps, with the highest values observed for Surefill (303.49 μm) and the lowest for Filtek P60 (43.05 μm). The restoration wear and evaluation images can be observed in [Table/Fig-9].

[Table/Fig-9]:

Cast replicas photographs and 2D-OCT image evaluation at baseline and 14-years of follow-up for the evaluated resin composites.

Restorative materials are continuously formulated and marketed. Therefore, long-term clinical studies evaluating their efficacy and durability should be encouraged [1,11]. Randomised clinical trials are strong scientific evidence [4]. This study aimed to evaluate the clinical performance of three resin composites indicated for posterior teeth restorations and their wear after 14 years.

After 14 years, an overall success rate of 89.52% was observed. Overall success rates are calculated based on the number of acceptable restorations (Alpha or Bravo ratings) during the whole evaluation period (94 acceptable restorations in a total of 105). In the one-year evaluation, the overall success rate of 100% (105/105) was observed. The annual failure rate, however, informs the percentage of possible failures (Charlie or Delta ratings) per year in this study was 0.74% (11 failed restorations out of 105, divided by 14 years). High clinical success rates (over 90%) were observed even after an extended evaluation time (more than 8 years) [28,29]. An 18-year clinical follow-up of composite resin occlusal restorations also observed promising results with low annual failure rates (2.1%) [6].

Colour match was inadequate in 33.3% after 14 years [Table/Fig-7]. In one-year, no restoration received unsatisfactory evaluations (14 alpha and 7 bravo) for the same parameter, indicating that time is a determinant factor for colour match success. Resin composite restorations are very susceptible to pigmentation when exposed to exogenous dyes [30]. The pigmentation is related to the degree of water sorption and the hydrophilic or hydrophobic characteristics of the resin matrix [30,31]. There is compatibility of the polymer phase of the resin composites with yellow dyes, which can adsorb and penetrate the organic matrix of the resin composites [31,32]. Also, inorganic filler detachment from the surface of the resin composites causes irregularities that favors the penetration of the pigments [32]. Foods with low pH may cause further damage to the materials’ surface, increasing the susceptibility to staining [33]. Furthermore, the translucency decreases over the years [22,34].

The discolouration of the cavosurface margin was satisfactory in all restorations at 1 and 14-year evaluation. Piva F et al., evaluated three resin composites in 34 primary molars for six months using USPHS criteria [35]. Filtek^TM P60 presented less satisfactory results concerning margin discolouration when compared to two micro hybrid resins composite (Charisma and Filtek Z250) [35]. The authors attributed these results to possible differences within the inorganic phase of the material, type, concentration and size could have accounted for greater difficulty in polishing [36,37].

Secondary caries was also not observed throughout the observation period, which is in agreement with previous studies with two and three years of follow-up [24,25,38]. The execution of restorations with a rubber dam, previous prophylaxis and incremental technique has been reported to reduce the caries incidence and problems related to handling in clinical evaluation of 36 months [39]. A systematic review and meta-analysis by Heintze SD et al., observed that caries adjacent to the restoration was infrequent and that restorations that were placed with a rubber dam showed fewer caries at the margins than restorations that were placed without a rubber dam [40].

Anatomicform and marginal adaptation parameters had two failures at 14-years, one Filtek P60 and one Suprafil restoration for each parameter. The one-year evaluation showed that all restorations were acceptable. The Surefil/Dentsply Caulk resin composite was the only one that received satisfactory evaluations after 14 years for these parameters. In eight-year clinical evaluation, only one composite resin restoration of 213 was classified as a failure for the anatomic form parameter [28]. While in another study an extended evaluation time, for the same parameter, 81% and 65% of resin composite restoration received an alpha score, for five and eight-year evaluation, respectively [29].

Restorative materials wear is influenced by the composition of the resin composite [41]. Materials with higher inorganic filler concentration tend to exhibit higher mechanical resistance and lower wear values [7]. Fillers size, shape and hardness, as well as the organic matrix ratio, can influence the wear resistance [7,42,43]. In this study, differences were only observed between materials within the buccal cusp. However, buccal triturating slopes were equally distributed among groups, in a sense that the restored tooth could not help to explain these results.

Optical coherence tomography is an efficient tool for the quantitative evaluation of restorative materials wear through replicate models. In addition, it is considered a non-destructive, non-invasive, with high resolution, fast and easy image acquisition [15,17,20]. In clinical settings, with direct observation of the restorations, OCT has also been able to detect the presence of air bubbles, cracking points, fractures, resin step shifts, gaps and possible agglomerates of inorganic particles [42].

Loss of contact with patients was a limiting factor in the present study. However, considering the split-mouth study design, the losses occurred equally for all three study groups. Intra oral Optical Coherence Tomography evaluations could favor clinical evaluations in future studies.

The condensable/packable resin composites (Filtek^TM P60/3M-ESPE and Surefil/Dentsply Caulk) and microhybrid (Suprafill/SS White) presented a satisfactory performance after 14 years. These materials can be safely indicated for clinical use in posterior restorations. The cast replica analysis by OCT showed accurate and precise evaluations and is a viable method of follow-up evaluations. The OCT findings were consistent with the clinical observations.