For the immunohistochemical analysis, the semi-thin sections were incubated over night with primary antibodies at 4 °C (Santa Cruz Biotechnology Inc., Santa Cruz, Calif., USA). The antibody dilutions used were alkaline phosphatase (ALP) 1:100, osteopontin (OPN) 1:100, osteocalcin (OSC) 1:100, and tatrate resistant acid phosphatase (TRAP) 1:100.

The obtained semi-thin sections were evaluated with a motorized (Märzhäuser, Wetzlar-Steindorf, Germany) light microscope connected to a digital camera and a PC-based image-capture system (BX51, DP71, Olympus Corporation, Tokyo, Japan). Photographs were obtained at ×5 and ×20 magnifications (Fig. 4).

From the obtained images, areas within the biopsies occupied by bone, biomaterial and connective tissue were identified using a pen computer (Cintiq companion, Wacom, Düsseldorf, Germany), coloured (Photoshop, Adobe, San José, CA, USA) and digitally measured using an automated image-analysis system (CellSens, Olympus Corporation) (Fig. 5).

The obtained histological sections were observed in a light microscope using 5x magnification. In the centre of each trephine biopsy, a rectangular region of interest (ROI) with a size of 30,000,000 to 32,000,000 pixels was defined and standardized photographs were obtained. The intensity of the antibody staining in the images was analysed using the software ImageJ, which by evaluating the antibody staining intensity in the area of interest allows for assessing quantitatively the specific marker (ImageJ®, IHC Profiler plugin). With this tool, the specimens were categorized into four groups: high positive (HP), positive (P), low positive (LP), and negative (N). To reduce false positives, only the HP and P values were considered for evaluating the percentage of positiveness for each immunohistochemical marker (Fig. 6).

Data were entered into an Excel (Microsoft Office 2011) database and proofed for entry errors. The software package (IBM SPSS Statistics 21.0; IBM Corporation, Armonk, NY, USA) was used for the analysis. A subject level analysis was performed for each outcome measurement reporting data as mean values, standard deviations, medians, 95% confidence intervals (CI), and frequencies. Shapiro–Wilk goodness-of-fit tests were used to assess the normality and distribution of data. Descriptive analysis of the histological and immunohistochemical outcomes was carried out by reporting means and standard deviations and comparisons between these histological outcomes between patients with subsequent implant loss versus patients with successful implant outcomes were evaluated using the paired sample t test or U Mann-Whitney if the distributions were non-normalized. Results were considered statistically significant at p < 0.05.