To address these questions, we developed a canine gap model using highly porous titanium implants. We analyzed the independent gap treatment variable; DBM, devitalized DBM, and empty controls, together with the independent time variables of 4 and 12 weeks, using three-dimensional (3D) micro-CT and scanning electron microscopy (SEM) to quantify the dependent variables of volumetric healing of gaps and the extent of bone ingrowth and apposition.
We fabricated gap-type intramedullary implants from commercially pure titanium (Fig. ). Each implant was 4 cm in total length, had a 5-mm diameter porous core, and had 11-mm diameter solid end caps and central spacers. This resulted in two separate gap regions, 15 mm long and 3 mm wide, containing a volume of 1.13 cm2
. We fabricated the implant core using a powder metallurgy approach. Pure titanium powder was mixed with a polymeric binder and a foaming agent. The resulting powder mixture was molded and heated up to foam the mixture. After foaming, the binder was removed and the part was sintered to consolidate the structure. The final structure had 55% volume porosity and an average pore size of 375 μm [38
Fig. 1A–C (A) The photograph shows the titanium gap implant with porous core and solid end caps and middle spacer. End caps and spacers maintained the 3-mm gap between the reamed border of host bone and the porous implant core. (B) A SEM image illustrates the porous (more ...)
We custom-processed the DBM by exposing 100- to 500-μm canine cortical bone fibers to dilute HCl (Osteotech Inc, Eatontown, NJ). The content of the fibers was predominantly collagen matrix with embedded native bone morphogenetic proteins; the residual mineral content was less than 0.5%. We prepared the second graft material by exposing DBM to guanidine HCl to remove the matrix-embedded osteoinductive growth factors, creating a devitalized negative control. We combined both matrices with 100% glycerol as a carrier in a 1-to-4 ratio, producing graft materials with a putty-like consistency. By comparing total protein content using established protocols, we were able to compare growth factor content between the DBM and its devitalized form [2
]. Protein extraction solution consisted of guanidine HCl 4 mol/L, Tris 50 mmol/L, and a protease inhibitor cocktail with EDTA 1 mmol/L (Sigma-Aldrich, Oakville, Canada). Samples of each material underwent extraction for 48 hours at 4°C, followed by centrifuging at 12,000 rpm for 30 minutes at 4°C. We dialyzed supernatant with a 3.5-kDa membrane against deionized distilled H2
O for another 48 hours at 4°C. Subsequently, the supernatant was assayed using Quick Start™ Bradford protein assay (BioRad, Mississauga, Canada) and read at 595 nm using a BioRad spectrophotometer. Using this method, the total protein content of the DBM was 48.3 μg/mL compared with 3.2 μg/mL for the devitalized DBM, confirming the effectiveness of the devitalization process. We did not include autograft material in the experiment because it is rarely utilized in revision THA and would not have provided a clinically meaningful comparison.
Before starting the experiment, a power analysis was performed to estimate the sample size required. Estimates for the analysis were based on previous studies using gap implants with and without bone morphogenetic protein [33
]. To show a twofold difference in percentage volume healing between gaps left empty or filled with DBM or devitalized DBM, combined with SDs of 50%, an alpha error level of 0.05, and a beta level error of 30%, the estimated (unpaired) sample size was six.
We used 15 skeletally mature male and female mongrel dogs, weighing between 35 and 40 kg, for all experiments. The primary implant protocol consisted of initial bilateral surgery, during which we inserted gap implants into the proximal femora, followed 8 weeks later by secondary bilateral surgery, when we inserted implants into the proximal humeri. Harvest of the bones 4 weeks after the second surgeries provided implant-bone constructs from each animal at 4 weeks and 12 weeks. An additional (control) protocol consisted only of humeral surgery; we inserted gap implants into the proximal humeri and left them empty in situ for 12 weeks before bone harvest. This study had institutional approval of surgical and animal care protocols.
In the primary implant protocol including nine dogs, we filled the proximal gap of each implant with either DBM or devitalized DBM (negative control), while leaving the distal gap empty (control), for both femora and humeri. In three dogs, we used DBM on one side and used devitalized DBM on the contralateral side in both femora and humeri. In a second group of three dogs, we used DBM on one side without implanting devitalized DBM on the contralateral side. In a third group of three dogs, we used devitalized DBM on one side without implanting DBM on the contralateral side. This protocol resulted in six data points for each material at 4 weeks (proximal humerus) and 12 weeks (proximal femur) (Table ). The secondary implant protocol, designed to assess healing within proximal empty gaps as further verification of the critical gap size, consisted of six dogs with six empty gaps in the humeri at 12 weeks (Table ).
Allocation of 15 adult mongrel dogs into the various experimental groups
We induced anesthesia using sodium pentobarbital and maintained it with 3% isoflurane and oxygen. We administered intravenous cefazolin in two 0.5-g doses in sterile water, once before and after surgery. The left hindlegs (or forelegs) were shaved and prepared in standard fashion. After making a small incision over the proximal end of the femur (or humerus) and splitting the gluteus medius (or rotator cuff), we drilled a 5-mm pilot hole into the intramedullary canal of the femur (or humerus), a procedure much like a blind intramedullary rodding. We progressively enlarged the hole in 0.5-mm increments to 11 mm using reamers. After filling an implant gap with 1.8 g (± 10%) of DBM or devitalized DBM using manual pressure to ensure dense and uniform filling, we inserted the implant into a hollow, thin-walled stainless steel tube with an inner diameter of 11.3 mm. Then, we inserted the tube at the proximal opening of the reamed hole, placed an inserter rod into the other tube end, and gently tapped the implant down the intramedullary canal of the bone until it was positioned within metaphyseal cancellous bone. This insertion technique ensured containment of graft material within the implant gap and controlled placement of each implant. The surgical wound was closed in layers using Vicryl® (Ethicon, Inc, Somerville, NJ) sutures. Analgesia consisted of a 100-mg fentanyl patch placed before surgery and intramuscular buprenorphine (0.02 mg/kg) immediately after surgery and twice over the following 16 hours. We repeated the procedure on the contralateral femur (or humerus) for the primary implant protocol. All dogs returned to normal weightbearing activities 1 to 2 days after surgery; there were no infections or postoperative complications. Postoperative management included environmental enrichment and daily exercise for 1 hour. At the end of each experiment, euthanasia was induced with an overdose of intravenous pentobarbital (120 mg/kg).
We scanned the femora and humeri with a high-voltage, high-resolution X-Tek micro-CT scanner (Model XTH 225; Nikon Metrology, Leuven, Belgium) [19
]. We imaged the bone samples in water at 135 kV and 65 μA, with a 0.5-mm copper filter and 1000-ms exposure time to obtain 18-μm-thick serial images of the complete bone-implant construct. Using CT Pro 2.0 (Metris, Derby, UK), we were able to reconstruct raw images and used the resulting 3000 serial CT images of each specimen to quantify gap healing using image analysis software ORSVisual v18.104.22.1681 (Objects Research Systems, Montreal, Canada). Bone that formed within the 3-mm implant gaps was measured and expressed as a volume percentage of the gap. In similar fashion, we measured the density of host cancellous bone immediately adjacent to each implant gap (proximal and distal) to compare with and normalize the gap healing data.
We subsequently embedded specimens in acrylic and imaged undecalcified transverse serial sections through the gap regions using backscattered SEM (three sections/images per gap). Using ImageJ software v.1.6.0_12 (National Institutes of Health, Bethesda, MD) and Bone Ingrowth Macro (SA Hacking, ImageJ Macro PTBI v1.0), we analyzed digital images for quantitative measurements of the mean percentage of filling of the implant pores with new bone and the mean percentage of bone apposition to the perimeter of the implant. We estimated the effect of empty controls compared to devitalized DBM controls and DBM on gap healing, bone ingrowth, and apposition using a Bayesian hierarchical (random effects) linear model. Differences between outcomes were expressed with 95% credible intervals (95% CrIs). Independent variables in the model included host bone, number of weeks, and treatment. We ran models both with and without an interaction term between weeks and treatment. A random intercept term was used to account for possible baseline differences between dogs. All analyses were conducted using WinBUGS v.1.4.3 (MRC Biostatistics Unit, Cambridge, UK).