The SEM analysis was performed as previously described [16,17,18]. The decontaminated implant samples were fixed with 2% glutaraldehyde-RPMI 1640 immediately for 1 h at room temperature and washed with distilled water. Then, the samples were dehydrated with 100% t-butyl alcohol and freeze-dried. Finally, the samples were coated with platinum and examined using an emission-scanning electron microscope (JSM-6390LVZ; JEOL Ltd., Tokyo, Japan). An ordinal scale with the following variables was set to facilitate the evaluation of cleansability qualitatively by each method.

No effect: Surface was cleansed ineffectively and covered with an enormous amount of amorphous material, debris, and bacteria.

Fair: Surface was cleansed partially well but was far from the ideal and covered with a certain amount of amorphous material, debris, and bacteria.

Good: Surface was cleansed effectively but not perfectly and covered with a little amount or partially no amount of amorphous material, debris, and bacteria.

Excellent: Surface was cleansed ideally and covered with no or only a small amount of amorphous material, debris, and bacteria.

SEM images were evaluated by two examiners (M.Y. and S.K.) who were unaware of the aim of this study. Kendall’s coefficient of concordance was used to investigate the inter-examiner reliability in the evaluation of SEM images. There was no significant difference between the two examiners (w = 0.865, p < 0.01).

To examine bacterial amounts on the implant surfaces, the samples were vortexed at maximum power for 30 s in 1 ml phosphate-buffered saline solution. Resuspended bacteria were serially diluted in a phosphate-buffered saline solution and plated on Brain Heart Infusion agar (Becton Dickinson, Sparks, MD, USA). The number of CFUs was counted after overnight growth on the BHI agar at 37 °C in a candle jar.

Statistical significance of differences in bacterial CFU counts among the six groups, including the control group, was analyzed using the Steel-Dwass test (R version 3.4.0 (R Foundation for statistical Computing, Vienna, Austria)). Significant differences between rough and machined surface implants were analyzed using the Mann-Whitney U test (SPSS 23.0 (SPSS Inc., Chicago, IL)). p < 0.05 was considered statistically significant.