2.1

Specimen fabrication and sample allocation

Sixty-four extracted human central incisors visually free of carious lesions or fractures were included. Immediately after extraction and before the examination, the teeth were stored in a 0.5% chloramine solution at 4 °C. The teeth were randomly assigned to eight groups (n = 8, each group): Four main groups of different preparation designs were divided into two subgroups each with different thicknesses of the laminate veneers [0.2 – 0.5 mm (T1) and > 0.5 – 1.2 mm (T2)]: non-prep (NP), minimally invasive (MI, no dentine exposed), semi-invasive (SI, 50% dentine exposed) and invasive (I, labial preparation, 100% dentine exposed) ( Fig. 1 , Table 1 ).

Preparation designs of veneers: (a) non-prep, (b) minimally-invasive, (c) semi-invasive and (d) invasive preparation.

All teeth were fixed in a full arch plaster model within a socket made by a paste/paste silicone (Zetaplus, Zhermack GmbH, Marl am Dümmer, Germany) without neighboring teeth. A silicone key was produced to transfer the crown shape of the tooth and thus to be able to check a uniform and standardized reduction of the tooth structure during the preparation. The position of the preparation margins was previously marked with a waterproof pen. Veneer preparations of the teeth were made using a preparation set for veneers (#TD1356A.314, Komet Dental/Gebr. Brasseler GmbH & Co. KG, Lemgo, Germany). All teeth were shortened incisally by 1.5 mm. The incisal plateau was angulated palatally towards the apical. The proximal extension was defined by the contact points, and cervically of these, so-called “elbows” were created as lateral surfaces, which could extend orally slightly beyond the contact point.

The extent of dentine exposure of the various preparation designs was determined visually on dry surfaces. In the minimally invasive group, the extension was restricted to the enamel. In the semi-invasive group, the preparation was 50% in enamel and about 50% in dentine and the cervical margins ended in dentine. In the invasive group, nearly 100% of dentine was exposed. The cervical preparation margin was always within the root dentine. Sharp edges or corners within the preparation line were rounded using flexible discs (Soflex, 1981 M, Medium 10 – 40 μm, 3 M Oral Care, Seefeld, Germany). Care was taken to ensure that the finishing line itself was always visible and sharply defined. The impressions were made using the double mix impression technique with a silicone impression material (Silagum light body, DMG, Hamburg, Germany). For each preparation design, two veneer restorations were made from a leucite-containing veneering ceramic in conventional layering technique (Ceramic IPS InLine, Ivoclar Vivadent AG, Schaan, Liechtenstein) in a commercial dental laboratory by a master dental technician according to the manufacturer’s instructions.

2.2

Adhesive luting of restorations

The intaglio surfaces of the veneers were etched for 120 s using a 5% hydrofluoric acid gel (Porcelain Etch, Ultradent, South Jordan, UT, USA) sprayed with water for 20 s and dried using oil-free compressed air. The veneers were cleaned in an ultrasonic bath with 96% ethanol for 3 min. After air-drying, a silane coupling agent (Monobond Plus, Ivoclar Vivadent AG) was applied and air-dried after 60 s. Preparation areas of the specimens were etched with 35% phosphoric acid (Ultra-etch, Ultradent, South Jordan, UT, USA) according to the manufacturer’s recommendations and immediately afterward conditioned with a three-step etch-and-rinse adhesive (Optibond FL, Kerr, Orange, CA, USA). All veneer restorations were adhesively luted using a dual-cure composite resin (Variolink Veneer, Ivoclar Vivadent AG) and light-cured by a Valo unit (Ultradent, South Jordan, UT, USA) in standard mode with a documented light intensity of 1250 mW/cm ² at a distance of approximately 5 mm for 40 s from labial, proximal, and incisal aspects. The specimens were polished with silicone tips (Brownie/Greenie, Shofu Dental, Ratingen, Germany) and flexure discs (Sof-Lex, 3 M Espe, Seefeld, Germany) before water storage at 37 °C for 21 days.

2.3

Cyclic fatigue loading

Specimens were thermocycled (2000 cycles between 5 and 55 °C) and mechanically loaded to simulate the intraoral stress situation. For this purpose, the samples were mounted with acrylic resin (Technovit, Kulzer Dental, Hanau, Germany) in a holder that allowed reproducible positioning in the chewing simulator (SD Mechatronic GmbH, Feldkirchen-Westerham, Germany) and to the antagonist. The test specimens were first loaded with 50 N for 2 million cycles and then a further 1 million cycles with 100 N at a frequency of 1.7 Hz and an opening of 5 mm without lateral movement. In between, the teeth were stored in 0.5% chloramine solution.

2.4

Verification of gap detection by OCT

Additional four specimens as validation objects were provided with two veneer preparation designs (2x SI and 2x I) and a ceramic thickness of 0.2 – 0.5 mm with an artificial adhesive defect at the interfaces ceramic/composite and tooth/composite. Half of the preparation area of these specimens was neither etched nor bonded, and the corresponding portions of the ceramic surfaces were not etched and silanized. After thermocycling (500 cycles, 5 / 55 °C, W15, Haake, Karlsruhe, Germany), specimens were imaged using SD-OCT (Telesto II, Thorlabs GmbH, Dachau, Germany). The four samples were then embedded (Stycast, Emerson & Cuming, Westerlo, Belgium) and cut along the reference planes in 6 slices of 200 μm each (Leitz™ 1600 sawing microtome, Ernst Leitz Wetzlar GmbH, Wetzlar, Germany). Before the analysis by scanning electron microscopy (SEM, Phenom G2 pro, Phenom-World BV, Eindhoven, Netherlands), the specimens were prepared according to the laboratory standard. Samples had been conditioned (HCl: 2%, 10 s, 20 °C; NaOCl: 10%, 30 s, 20 °C) with pure water rinsing after every step (60 s, 20 °C). Subsequently, the samples had been dehydrated by ascending alcohol series, stored in hexamethyldisilazane (10 min, 20 °C, HMDS, Carl Roth GmbH, Karlsruhe, Germany) and air-dried. Each slice was mounted on an aluminum specimen stub (12.5 mm Ø, Plano GmbH, Wetzlar, Germany) and gold-coated (2 nm, Sputter Coater MSC1, Ingenieurbüro Peter Liebscher, Wetzlar, Germany). The SEM images were compared with the OCT-B scans regarding the visualization of the adhesive defects at the interfaces.

2.5

OCT imaging of gap formation

After thermocycling and mechanical loading, the specimens were 3-dimensionally imaged by SD-OCT (see above). OCT is based on low-coherence interferometry. SD-OCT is a variant of Fourier domain OCT, using a broadband, low coherence light source. The underlying principle has already been described in detail elsewhere [ , ]. The reflected light from the sample and reference arm interferes with each other, and the resulting signal is recorded. Adjacent structures or phase boundaries with different refractive indices and different light absorption generate detectable signals as grey value differences in 2D B-scans (e.g., interfaces between hard tooth tissues or to the pulp, gaps, bubbles, cracks, or porosities) [ ]. Table 1 summarizes the specifications of the equipment used.

The images obtained by OCT were analyzed based on the platform Image J v1.45 (Wayne Rasband, National Institutes of Health, Bethesda, MD, USA). About 250 B-scans were extracted from the 3D volume dataset, and 35 equally distributed vestibulo-oral oriented B-scan images were assessed.

An OCT-signal indicated an interfacial adhesive defect, which was defined as a signal peak resulting in a bright line extending along with the interfaces between air and ceramic, composite, or air and tooth or composite [ , ]. As previously described [ ], the lengths of the gap-induced OCT signals at the ceramic-composite and tooth-composite interfaces were measured through 35 B-scans per tooth. The length was measured in pixels using ImageJ. The length of the defects on the respective interfaces was put in relation to the total length of this interface and expressed as a percentage of the total length. The assessment was performed by a calibrated examiner, trained by a physicist with several years of practical experience in OCT applications in dentistry.

2.6

Statistical analysis

The statistics program used was Prism (GraphPad Software Inc., La Jolla, CA, USA). Weighted mean values (SD) and medians of the percentage of interfacial gap formation were determined for the subgroups. Since the values of teeth were not generally normally distributed (Kolmogorov-Smirnov test, Shapiro-Wilk test), the group comparisons were carried out non-parametrically using the Friedmann/Wilcoxon test (interfaces VC vs. TC within the subgroups) or the Kruskal-Wallis/Mann-Whitney U -test (interfaces between the subgroups). For all tests, the significance level of α = 0.05 with two-tailed testing was chosen. Due to the exploratory nature of this study, raw p-values are presented, and no correction for multiple testing has been made.