Specimen pre-treatment
Human maxillary incisors were selected from a tooth bank and stored at room temperature in a 0.5% chloramine solution. To ensure the use of teeth of comparable dimension within the groups, mesio-distal (MD) and facial-lingual (FL) dimensions were measured at the level of the cemento-enamel junction (CEJ). Tooth size was calculated as the product of MD × FL. Extremely small or large teeth were excluded. Specimens were randomly distributed into 4 groups (n = 12) by means of a ten-digit random table to either no change to crown-to-root ratio (control, group I), apical root resection (group II), extrusion (group III) and surgical crown lengthening (group IV). The crowns of teeth in groups #I and #II were cut 2mm coronally to the CEJ, and in groups III and IV at CEJ level. Root canals were enlarged using the X-Smart (Dentsply DeTrey, Konstanz, Germany) and NiTi-files to size F2 (Protaper, Dentsply DeTrey) and rinsed with 3% sodium hypochlorite. Root canals were filled by corresponding size F2 of gutta-percha (Protaper, Dentsply DeTrey) and sealer (AH 26 Plus Jet, Dentsply DeTrey).
The roots of the specimens of groups #I and #II were blocked out with wax 2mm and specimens of group III and IV 4 mm below the CEJ. To imitate a periodontium and physiological tooth mobility, roots of the teeth were covered with a layer of silicone (Mollosil Plus, Detax, Ettlingen, Germany) as described elsewhere [13]. All teeth and implants (group V, n = 12) were embedded in acrylic resin (Technovit 4004, Kulzer, Wehrheim, Germany). To prevent overheating, the teeth were immersed in water for 5 minutes during resin polymerization.
Tooth-based restoration, group I-IV
Post space preparation was performed 8mm within the root canal in one sequence as described by the manufacturer. All restorative steps were performed using the Dentsply Core & Post System (CTS, Dentsply DeTrey).
The etch-and-rinse and bonding procedure was performed according to the manufacturer’s instruction. The root canal and the coronal tooth surface were etched with 36% phosphoric acid (Conditioner 36, Dentsply DeTrey) for 15 sec. After water rinsing and air-drying, the adhesive was applied and left for 20s (XP Bond, Self-cure Activator, Dentsply DeTrey, 1:1 ratio, mixed for 2s). A glass-fiber post (size 2 (red), Ø 1.25mm, X-Post, CTS, Dentsply DeTrey) was treated with adhesive and was luted with core build-up material in a staged procedure using a dual curing core-and-post composite (Core-X™ flow, Dentsply DeTrey). The core was built up by means of a strip crown (upper central incisor, Frasaco, Tettnang, Germany) and polymerized from the incisal, palatal and facial aspect for 20s each. All teeth were prepared with a circumferential 0.8mm shoulder and 6° convergence angle to meet all-ceramic crown requirements. To achieve an equal crown length the core build-ups were similar in length for group I to III (4mm) and for group IV (6mm). The margin was located 2mm below the core build-up in dentin to ensure proper ferrule design. Specimens were scanned with an intraoral scanner (Trios, 3Shape, Kopenhagen, Dänemark), models were milled out of polyurethan. Crowns were constructed digitally (Dental Designer, 3Shape, Copenhagen, Denmark), subsequently milled in wax (Organical Multi, R + K CAD/CAM, Berlin, Germany), transferred to lithiumdisilicate (IPS e.max, Ivoclar Vivadent, Schaan, Germany), and glazed (IPS e.max Ceram Glaze, Ivoclar Vivadent). Crown width was 2mm incisally, 1.5mm in the middle, and 0.8mm at the preparation margin. Crown height was 8mm, except in group IV (10mm) (Figs. 1 and 2).
Implant-based restoration, group V
Dental implants (length 12 mm, diameter 4.1 mm, Straumann Bone Level, Freiburg, Germany) were restored with titanium alloy base (RC Variobase Abutment, diameter 4.5mm, length 3.5mm, Straumann, Germany) with lithiumdisilicate abutments (IPS e.max, Ivoclar Vivadent, Schaan, Liechtenstein). 12 identical lithium-disilicate abutments were modeled in wax (Dental Designer, 3Shape, Copenhagen), milled (Organical Multi, R + K CAD/CAM, Berlin), and transferred in lithiumdisilicate (IPS e.max, Ivoclar Vivadent). Abutment measures were equivalent to cores of group I. Abutments were luted on the alloy bases with self-adhesive luting composites (IPS E.max Abutment Solution Cem Kit,, Ivoclar Vivadent). Abutments were screwed in with 35Ncm. Crowns were etched 20s with fluoric acid (Vita Ceramics Etch, Vita, Bad Säckingen, Germany), cleaned with water and isopropanol, silanized (Monobond Plus, Ivoclar Vivadent, Schaan) for 60s, and self-adhesively luted (SmartCem Dentsply DeTrey). Final light curing was performed for 20s from each restoration side.
Loading protocol
Thermal and mechanical loading (TML) was performed (parameters: 6,000 thermal cycles, 5°C / 55°C, 2min each cycle; dist. water; 1.2x106 mastication cycles with 50N) to simulate five years of clinical service [14]. The restorations were loaded under 135°, 3mm below the incisal edge, on the palatal surface of the crown. After TCML tooth mobility was measured three times for each specimen by means of a periotest device perpendicular to tooth and implant axis (Periotest Classic, Medizintechnik Gulden, Germany). Specimens were statically loaded in a universal testing machine (Zwick 1446, Zwick, Ulm, Germany; v = 1mm/min) until failure. Failure detection was set at a 10% loss of the maximum applied force. A 0.3mm thick tin foil was positioned between the steel piston and the palatal crown surface to reduce excessive stress concentrations.
Calculation of crown-to-root ratio (RCRR)
The length of the prosthetic crown LC was defined with 8mm for groups I to III and 10mm for group IV. The distance crown margin to crestal alveolar bone was 2mm in order to simulate biologic width (LBW). The sum of LC + LBW was defined as effective crown height CE. The root length LR is the distance from the apex to the CEJ. The effective root length RE is defined as length of the root LR within alveolar bone [15]. In groups II, III and IV, RE was 2mm smaller than LR due to simulated apical root resection, extrusion or crown lengthening, respectively (RE = LR − 2mm). In groups II and III, the effective crown height CE was equivalent to that of group I. For group IV, CE of group IV was calculated from LC + LBW + 2mm of crestal bone resection.
Statistical analysis
The number of cycles until failure during dynamic loading (TML) was compared with log-rank statistics. Non-parametric Kruskal-Wallis and Mann-Whitney-U test as post hoc were applied to determine differences between group median values of the crown-to-root ratios (RCRR), as well as maximum load capability Fmax after linear loading. Differences in the frequency of the failure modes between the groups were evaluated by Chi-Square tests. Data were pooled and categorized into four patterns: crown fracture, crown fracture with additional core build-up loosening, fracture at the coronal or middle third of the root (root fracture). Subsequently, failures were classified as “restorable” (crown fracture, crown fracture with core loosening) and “not restorable” (i.e. root fracture). We conducted both a complete case analysis of maximum load capability, i.e. excluding specimens that failed during TML for analysis of Fmax, and a sensitivity analysis assigning those specimens that failed Fmax=0. All statistical tests were two-sided at α = 0.05.