This study was conducted in accordance to the international standard fatigue test (ISO 14801:2007) for endosseous dental implants in vitro (Fig. 2). Eighty-four implants (PW Plus implant system, PW Plus Company) were prepared, which had a diameter of 3.75 mm and a length of 10 mm, with the Octatorx-cone implant-abutment connection. All were embedded in individual epoxy resin blocks (Chockfast Orange Resin, Shannon Industrial Estate, Co. Clare, Ireland). The platforms of the implants were 3 mm above the level of the upper border of the resin block. The straight abutment of the implant sample was attached to a hemispherical metal cap, which was relined with Chockfast resin on the inner side (Fig. 3). The abutment was attached to the implant component which is mounted with a digital torque gauge (Tohnichi torque gauge, model BTGE50CN) and the abutment retaining screw was tightened to 30 N cm, (according to the company’s recommendation) using the digital torque gauge (Fig. 4). After 10 min, all abutment screw samples were retightened at the same torque (30 N cm) and left unloaded for 10 min. Six samples were randomly selected as the control group (group 0) to measure and record the initial RTVs of the abutment screws using the digital torque gauge. The remaining 78 samples were randomly divided into 13 experimental groups of 6 and underwent different numbers of mechanical loading cycles in the Electropuls E1000 dynamic test instrument (Instron, Fig. 5), which delivered dynamic loading forces between 15 and 250 N with a frequency of 15 Hz. The implant samples were mounted in an angled steel holder so that the axis was at a 30 ± 1° angle to the loading direction. The loading force was applied to the hemispherical metal cap of the implant samples, with no lateral constraint (Fig. 6).

The numbers of loading cycles for each experimental group were 50,000, 100,000, 150,000, 200,000, 400,000, and continually increasing 200,000 cycles per group to the last group at 2,000,000 cycles. After the cyclic loading, all implant samples were measured to determine the RTVs of the abutment screw using the digital torque gauge. The RTVs were recorded and the mean percentage loss of the RTVs in each experimental group was calculated by comparison with the initial RTVs. The schematic experimental procedures are shown in Fig. 7.