Cone beam computed tomographic (CBCT) scans were obtained from each patient for the surgical planning of implant removal. Linear measurements of the defect length (DL) at affected implants were made by drawing a vertical line, following the long axis of the implant, from the implant shoulder (IS) to the bottom of the defect (BD) at buccal and oral aspects in the CBCT data sets (Implant viewer, version 1.84, USA). Radiological evaluations were performed by one experienced and calibrated examiner (M.G.).

Implant removal in all patients was performed by the same clinician (M.B.) following a standardized surgical procedure. Patients were prescribed 500 mg of oral amoxicillin, 3 times a day during 1 week, starting 1 day before surgery. Patients were locally anesthetized (2% lidocaine, 1:100,000 epinephrine). The prostheses of the implants were removed. Surgical incision was performed, and full-thickness mucoperiosteal lingual and buccal flaps were raised; bone-implant biopsies were obtained using a trephine bur under copious irrigation with sterile saline. The implants were removed with the minimal required trephine diameter and without compromising the soft-tissue component for ethical reasons. Specimens were immersed immediately in 10% neutral buffered formalin. Following the removal of granulation tissue from the defect area, mucoperiosteal buccal and lingual flaps were repositioned and fixed with single sutures.

Specimens were dehydrated in ascending series of alcohol and embedded in light-cured resin (Technovit 7200, VLC+BPO, Heraeus, Kulzer Co., Wehrheim Germany). Undecalcified thin ground sections were prepared along the longitudinal axis (bucco-lingual direction) with a high precision diamond-coated band saw and precision grinding equipment (Exakt, Apparatebau, Norderstedt, Germany) [18]. Histological sections were stained with Levai-Laczko dye and scanned with an Olympus BX61VS microscope using a digital virtual light microscopy system (dotSlide 2.4; Olympus, Tokyo, Japan) at a resolution of 0.321 μm/pixel.

Histomorphometrical analysis and microscopic observations were performed by one calibrated examiner (M.G.). Digitalized histological sections (bucco-lingual direction) were analyzed using a morphometric software (Definiens Developer XD 2.0, Definiens AG, Munich Germany). The following landmarks were identified in the stained sections (buccal aspect) (Fig. 1): implant shoulder (IS), level of the alveolar bone crest (BC), bottom of the defect (BD), and the most apical extension of the residual bone tissue (A). Linear measurements were performed by drawing a vertical line following the long axis of the implant: defect length (DL), as measured from IS to BD (mm), residual bone (RB) as measured from BC to A, residual BIC (i.e., the length proportion of the implant surface that was in direct contact with mineralized tissue) as measured as a percentage of the distance from BD to A. The density of the RB was measured as a percentage by dividing the bone area (B.Ar) by the total area of tissues (T.Ar) surrounding the implant.