Tissue Restoration After Implantation of Polyglycolide, Polydioxanone, Polylevolactide, and Metallic Pins in Cortical Bone: An Experimental Study in Rabbits

Commercially available bioabsorbable PGA and PLLA pins (Bionx Implants, Tampere, Finland) and poly-p-dioxanone pins (OrthoSorb; Johnson and Johnson Orthopedics, New Brunswick, NJ) were used. According to the manufacturer, the molecular weight of the PGA was between 50,000 and 200,000 daltons. For PLLA, the viscometric mean molecular weight of the raw material was 700,000 daltons. After processing, the molecular weight of the PLLA was 50,000 daltons and the degree of crystallinity was 50%. The manufacturers sterilized the self-reinforced PGA pins and PDS pins using ethylene oxide and the self-reinforced PLLA pins using gamma radiation at doses of 25 kGy. Commercially available metallic, stainless steel Kirschner wires (Synthes, Solothurn, Switzerland) were used in the control group and to secure the fixation in the experimental animal groups. Based on the manufacturer’s information, the K-wires were “implant quality 316 L stainless steel” in the United States and “implant quality DIN 1.4441 stainless steel” in Europe. All bioabsorbable pins were 2.0 mm in diameter and 40 mm in length. All stainless steel Kirschner wires were also 2.0 mm in diameter, and their length was determined by the length of the intramedullary cavity of the femur.

Surgery was performed in 80 skeletally mature New Zealand rabbits of both sexes with a mean weight of 3.5 kg (range 2.6–4.4). The rabbits were anesthetized with subcutaneous injections of medetomidine (300 μg/kg Domitor; Orion Pharma, Turku, Finland) and ketamine (25 mg/kg Ketalar; Parke-Davis, Barcelona, Spain) [25]. Under standard aseptic conditions, a lateral longitudinal incision was made in the right distal thigh and knee, the patella was dislocated medially, and the distal portion of the femur was exposed. On the distal third of the femur, a standardized, anterolateral, semicircular cortical bone defect was created using an oscillating saw (Fig. 1). The length of the semicircular defect was 10 mm and its depth was half of the lateromedial and anteroposterior thickness of the femur. The distal margin of the defect was located 30 mm from the articular surface of the distal lateral condyle of the rabbit knee. The loose cortical bone quadrant was removed, and the semicircular defect on the distal third of the femoral diaphysis was left open. A longitudinal channel, 4.0 mm in diameter, was drilled centrally to extend through the intercondylar portion into the intramedullary canal. Two implants were driven into the drill channel and the intramedullary canal (Fig. 1). To prevent the defect from advancing to a fracture, one of the implants in every operated knee was a metallic Kirschner wire reaching approximately to the level of the lesser trochanter of the femur. Previous reports indicate that without external support a bioabsorbable implant alone is not sufficient for fixation of a femoral shaft osteotomy in rabbits [26]. The 80 left femurs served as intact controls and were not operated on. The incision was closed in layers with 3-0 PGA sutures (Dexon; Davis & Geck, Gosport, UK), and the rabbits were returned to their cages. Postoperatively, all rabbits were fed ad libitum and allowed to use their limbs freely.

The disarticulated and dissected femora were radiographed in the anteroposterior and lateral views (target-tube distance, 100 cm; exposure factors, 40 kV, 5 mA, and 0.03 seconds) before removing the metallic Kirschner wires (Fig. 2). The distal third of each femur was fixed in 50% alcohol and dehydrated in increasing concentrations of alcohol. After dehydration was complete, the specimens were embedded in methylmethacrylate. For microradiography, 80-μm-thick longitudinal sections were cut using a Leitz 1600 saw microtome (Ernst Leitz, Wetzlar, Germany). The microradiographs were made using a Faxitron 43855A cabinet X-ray system (Hewlett-Packard, McMinnville, OR) and Kodak professional plates (type 649-0, CAT 1690718; Eastman Kodak, Rochester, NY). The exposure conditions were 50 kV, 9 mA, with a 12-min exposure and direct specimen-film contact. The radiographs and microradiographic specimens were scanned into digital format for further analysis.

For histologic and histomorphometric analysis, 5-μm-thick sections were cut with a Polycut-S-microtome (Reichert-Jung, Nussloch, Germany) in the oblique sagittal plane, parallel to the long axis of the pin through the anterolateral midpoint of the cortical defect area, and stained using the Masson-Goldner trichrome method.

For quantitative histomorphometric analysis of the tissue–implant interface, three microscopic fields measuring 0.9 × 1.2 mm within 10 mm from the center of the cortical bone defect (Fig. 3) were digitally microphotographed (Nikon Eclipse 80i microscope, Nikon DS-Fi1 digital camera; Nikon, Tokyo, Japan). The microscopic fields were standardized so that the implant-occupied area or implant channel comprised 50% of the field. Sample fields located at the corresponding area of the intramedullary canal from the intact control femora were also microphotographed. The digital microphotographs, radiographs, and microradiographs were morphometrically analyzed using IP Lab software (Scanalytics, Fairfax, VA). Fractions of the following components were measured in the microphotographs: trabecular (cancellous) bone, implant channel, bone marrow fat, fibrous connective tissue, and hematopoietic tissue. Trabecular bone was defined as a meshwork of interconnecting rods and plates of calcified bone matrix, which was stained green with Masson-Goldner staining. The total area of the trabecular bone included both calcified trabeculae and osteoid lining of trabeculae, which were stained red with Masson-Goldner. Fat cells were defined as large unilocular cells (diameter 50–60 μm) with unstained cytoplasm. Connective tissue was defined as fibers, and the bundles of collagen were stained red or orange with Masson-Goldner. Hematopoietic tissue was defined as dense infiltrates of mature blood cells, megakaryocytes, and their precursors.

The projection area of the external callus at the anterolateral bone defect was measured from the radiographs. The callus area was defined by following the contours of the callus on the outer surface and the original periosteum in the inner surface. The callus area was normalized and reported as a percentage of the total radiographic area of the femoral bone in the anteroposterior view. Maturation and calcification of the callus were studied quantitatively on bone microradiographs. The trabecular bone area fraction of the callus was measured and reported as a percentage of the whole callus area. Polarizing microscopy was used to study degradation of the bioabsorbable implants. The presence or absence of birefringent polymeric material was examined in the histologic specimens.

Histomorphometric measurements at each time point were pooled and reported as means for each animal. A nonparametric Kruskal–Wallis test was used to compare differences between the implant groups at each time point and within the implant groups over time. The difference was considered significant at a probability level of less than 0.05. Statistical analyses were performed with StatView 5.0 software (SAS Institute, Cary, NC).

In the intact control specimens, the intramedullary canal consisted of fat (56%) and hematopoietic tissue (44%) (Table 1). Neither connective tissue nor trabecular bone was present in any of the control specimens.

In all implanted femora, a zone of collagenous connective tissue surrounded the intramedullary implants at 3 weeks (Table 1, Figs. 4 and 5). Formation of bone trabeculae at the tissue–implant interface was also observed in all implant groups.

At the 6-week follow-up, collagenous connective tissue surrounded (walled off) the implant at the tissue–implant interface (Table 1). There was no significant difference in trabecular bone area between the groups at the 3- and 6-week follow-ups.

At 12 weeks, the PGA group had formed the most intramedullary bone compared to the other implant groups (Table 1). The differences between groups were not statistically significant.

At 24 weeks, bone and connective tissue formation began to decline in the PDS and metallic implant groups (Table 1). The differences between the groups were not statistically significant. At the end of the follow-up, 52 weeks after surgery, bone and connective tissue had formed in the intramedullary canal in all implant groups, in contrast to the intact femora (Table 1, Fig. 5). Hematopoietic cells were observed sporadically within the bone marrow fat in all implant specimens, and their proportion was not correlated with the implant material.

Three weeks after surgery, the callus area of the cortical bone defect was largest in the PDS implant group (Table 2, Fig. 2). Callus formation tended to increase for up to 6 weeks in the PLLA and metallic implant groups and for up to 24 weeks in the PGA group. The callus area was not statistically different between the groups. Maturation of the callus was assessed microradiographically by studying the calcified bone trabeculae (Fig. 2b). The area fraction of the mineralized trabeculae ranged from 12% to 51%. There was no statistically significant difference in the trabecular bone area fraction of the callus between the implant groups.

The shape of the implant was histologically unchanged in the PGA, PDS, and PLLA samples at 3–6 weeks. At 12 weeks, phagocytosing macrophages and birefringent intracellular fragments were observed at the tissue–implant interface in the PGA and PDS samples. At 24 weeks, no PGA material was detected in any of the samples. The shape of the PDS pin could not be identified, but single implant fragments and macrophages were observed at the implant area. At 52 weeks, no birefringent material could be detected in any of the PDS specimens. The PLLA implants were histologically unchanged at all follow-up points for up to 52 weeks. Inflammation, as indicated by accumulation of lymphocytes or polymorphonuclear cells, was not observed in any femur.