2.4 Immunohistochemical analysis

Sections were de‐waxed and incubated in DIVA antigen retrieval solution at 60°C overnight. The sections were incubated with a primary antibody for 30 minutes followed by incubation with Envision HRP labeled polymer for 30 minutes. Positive cells were detected using DAB substrate. The chosen antibodies were CD3 1:200 (T‐lymphocyte), CD20 1:400 (B‐lymphocyte), CD138 1:50 (plasma cell, clone MI15), CD68 1:200 (macrophage, clone PG‐M1). Counterstaining was performed with hematoxylin.

2.5 Histological analysis

The prepared serial sections were digitized via the Panoramic SCAN device and investigated by a Panoramic Viewer (3DHISTECH, Budapest, Hungary). Every sample was filled and covered with 20 randomly selected ROIs (Region of Interests) (Fig. 1). The ROIs comprised a size of 500 × 800 µm at a magnification of ×15.5. Pictures of the ROIs were taken and antibody positive cells in the ROIs were counted using ImageJ (NIH, Bethesda, USA). The results were assessed by a trained investigator (JM). The investigator was masked to clinical patient characteristics.

2.6 Statistical analysis

For descriptive analyses, the mean, median, and standard deviation were computed. Box and stacked plots were used for graphical presentations.

Linear mixed regression models were used to check for differences among the log‐transformed cell counts with regard to implant material (ceramic and titan), as well as differences among the cell counts with regard to cell type (CD3, CD20, CD68, and CD138) within each material. All of the aforementioned linear mixed models were adjusted for region of interest. To correct for the multiple testing problem, the results of pairwise comparisons were adjusted by the method of Scheffe. The calculations were performed with the statistical software STATA 15.1.