Intravenous administration of anti-vascular endothelial growth factor humanized monoclonal antibody bevacizumab improves articular cartilage repair

Twenty Japanese white rabbits (female, 16-18 weeks old, weighing approximately 3 kg) were used in this study. The rabbits were anesthetized by exposure to sevoflurane and O₂ gas. After receiving a medial parapatellar incision to both legs, each patella was dislocated laterally and an osteochondral defect (diameter, 5 mm; depth, 3 mm) was created on the patellar groove of the femur in both legs using a drill and a biopsy punch (Kai Industries, Seki, Japan). The bottom of the subchondral bone was shaved to a plane using the biopsy punch until bleeding was seen from the marrow. Rabbits were classified into two recipient groups: Group B, with administration of bevacizumab (10 rabbits; 100-mg intravenous injection administered on the day of surgery and 2 weeks later); and controls (10 rabbits; defect only). After recovery from surgery, all animals were allowed to walk freely in their cages without any splints.

Rabbits were killed 1 and 3 months postoperatively by an overdose of intravenous anesthetic. The distal part of the femur was excised and fixed with 4% paraformaldehyde for 7 days. Each specimen was decalcified in a solution of 10% ethylenediaminetetraacetic acid (EDTA) in distilled water (pH 7.4) for 2-3 weeks, then embedded in paraffin wax and sectioned perpendicularly (4.5-m sections) through the center of the defect. Each section was stained with safranin O for glycosaminoglycans.

Immunohistochemistry was performed as described previously [23, 33]. Briefly, sections were deparaffinized according to standard procedures. Sections were treated with 0.005% proteinase (type XXIV; Sigma-Aldrich Co., St. Louis, MO, USA) for 30 min at 37°C for antigen retrieval. For chondromodulin-I (ChM-I), primary goat polyclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA) diluted 1:200 in phosphate-buffered saline (PBS) and 1% bovine serum albumin (BSA) was placed on the section overnight at 4°C. For VEGF, a primary mouse monoclonal antibody (Upstate, Lake Placid, NY, USA) diluted 1:50 in PBS and 1% BSA was placed on the section overnight at 4°C. Slides were washed with PBS after incubation for 1 h at room temperature with biotin-conjugated goat antimouse secondary antibody for VEGF and with biotin-conjugated donkey antigoat secondary antibody for ChM-I. Next, slides were treated with horseradish peroxidase-labeled streptavidin for 1 h and then soaked in 0.05% solution of diaminobenzidine in Tris HCl buffer (pH 7.6) containing 0.005% hydrogen peroxide. Finally, slides were counterstained with Mayer's hematoxylin. Safranin O-stained sections were scored by two individuals under blinded conditions, according to a modified O'Driscoll [34] International Cartilage Repair Society (ICRS) grading scale [26] (Table 1).

Results are presented as the means ± standard deviation (SD). Histological score was analyzed by the Mann-Whitney U test. Values of P < 0.05 were considered statistically significant for any differences.

Operations were uneventful, and all rabbits immediately resumed normal cage activity. In Group B, major infection was identified in one knee at 1 month after surgery and in three knees at 3 months. These infected knees were omitted from the study. As a result, nine knees at 1 month and seven knees at 3 months were assessed from Group B, compared to 10 knees in controls at both 1 and 3 months.

At 1 month after surgery, defects in both Group B and the controls were filled with reparative cells. In Group B, the repair site appeared to be filled with cartilaginous tissue, which was stained with safranin O (Figures 1a and 1b). Lateral integration was well bonded at both sides of the surrounding cartilage. The surface of the repair site showed several smooth fibrous cell layers, and rounded chondrocytes formed inside the repair tissue in a convex pattern. The lower portion of the repair tissue contained hypertrophic chondrocytes that were remodeling the subchondral bone. Thus, the defects showed sequential construction of abundant inhomogeneous extracellular matrix from the surface by fibrous cells (or fibrocartilage cells), rounded chondrocytes and hypertrophic chondrocytes. Similarly, in the controls, defects consisted of inhomogeneous extracellular matrix from fibrous cells (or fibrocartilage cells), rounded chondrocytes and hypertrophic chondrocytes; however, no tendency toward uniform constitution was apparent. The defect was not filled with repaired tissue (Figures 1c and 1d). At 3 months in Group B, the repair site maintained a cartilage phenotype and was well integrated with the surrounding cartilage (Figure 2). Repaired tissue showed a columnar organization. The surface of the repair site retained a smooth and convex formation similar to the surrounding cartilage. In the controls at 3 months, the repair site had been replaced with fibrous tissue and bone (Figure 3), and in the surrounding adjacent cartilage there was evidence of osteoarthritic changes with loss of cartilage.

At 1 month, Group B showed intense positive results for ChM-I at the bottom of the repair tissue in the remodeling hypertrophic chondrocyte layer, representing the border between cartilage and bone (Figure 4a). ChM-I had accumulated in the interterritorial space of the repaired matrix (Figure 4a). In the controls, no ChM-I was observed in the repair tissue (Figure 4b). Conversely, the remodeling hypertrophic chondrocyte layer was intensely positive for VEGF in both Group B and in the controls (Figures 4c and 4d).

We evaluated the repair site using a modified version of the grading system developed by O'Driscoll, Keeley and Salter [34]. In this system, 11 histologic categories were evaluated and scored: tissue morphology (Ti), matrix staining (Matx), structural integrity (Stru), cluster formation (Clus), tidemark opening (Tide), bone formation (Bform), histologic appraisal of surface architecture (SurfH), histologic appraisal of the degree of defect filling (FilH), lateral integration of defect-filling tissue (Latl), basal integration of defect-filling tissue (Basl) and histologic signs of inflammation (InfH). The total score ranged from 11 (no repair) to 45 (normal articular cartilage) (Table 1).

At 1 month, inside the repair tissue in Group B, Ti was mostly hyaline cartilage in seven of nine cases, with high cellularity of rounded chondrocytes and mostly fibrocartilage in two of nine cases. Conversely, in the controls, 4 of 10 cases showed mostly hyaline cartilage, 4 of 10 cases were mostly fibrocartilage and 2 of 10 cases were mostly noncartilage. For Matx in Group B, six of nine cases were strong, two of nine cases were moderate and one case showed slight staining. In the controls, 5 of 10 cases were strong, 3 of 10 cases were moderate and 2 of 10 cases showed slight staining. In Group B, Stru of the defect filling revealed the beginning of columnar organization of chondrocytes in six of nine cases and no organization of chondrocytes in three of nine cases. In the controls, 3 of 10 cases showed the beginning of columnar organization, 4 of 10 cases showed no organization and 3 of 10 cases showed cysts or disruptions. In both groups, Clus was not observed except in one instance. In one instance, there was a small amount of Clus in both groups. Also, in both groups, Tide was opened in all instances. In Group B, subchondral Bform was not recognized in four of nine cases, slightly recognized in four of nine cases and strongly recognized in one of nine cases. In the control group, subchondral Bform was not recognized in 8 of 10 cases and was slightly recognized in 2 of 10 cases. SurfH in Group B was normal in five of nine cases and showed slight fibrillation or irregularity in four of nine cases; in the control group, 1 of 10 cases was normal, 6 of 10 cases showed slight fibrillation or irregularity, 2 of 10 cases showed moderate fibrillation or irregularity and 1 of 10 cases showed severe fibrillation or disruption. FilH in Group B was complete in six of nine cases and nearly complete in three of nine cases. In controls, FilH was complete in 2 of 10 cases, nearly complete in 4 of 10 cases, moderate in 2 of 10 cases and nearly empty in 2 of 10 cases. Latl in Group B was bonded on both sides in seven of nine cases and bonded at one end or partially at both ends in two of nine cases. In controls, Latl was bonded on both sides in 1 of 10 cases, bonded at one end or partially at both ends in 6 of 10 cases and not bonded in 3 of 10 cases. Basl was good in all cases for both groups. In Group B, no inflammation was observed in all cases for InfH. The control group showed no inflammation in 5 of 10 cases, slight inflammation in 4 of 10 cases and strong inflammation in 1 of 10 cases. As a result, at 1 month, the total score was significantly higher for Group B than for the controls. In terms of individual scores, SurfH, FilH and Latl for Group B were significantly higher than in controls.

In Group B, Ti at 3 months showed mostly hyaline cartilage in six of seven cases. In one of seven cases, tissue was mostly fibrocartilage. Conversely, in the controls, 3 of 10 cases were mostly hyaline cartilage, with mostly fibrocartilage in 1 of 10 cases and exclusively noncartilage in 6 of 10 cases. For Matx, six of seven cases in Group B showed strong staining and one of seven cases showed moderate staining; in the controls, 3 of 10 cases were strong, one instance was moderate and 6 of 10 cases were nonstaining. In Group B, Stru of the defect filling revealed tissue similar to healthy mature cartilage in three of seven cases, beginning columnar organization of chondrocytes in three of seven cases and no organization of the chondrocytes in one of seven cases. In the controls, only one instance was similar to healthy mature cartilage, 2 of 10 cases were beginning columnar organization, 1 of 10 cases had no organization and 6 of 10 cases showed severe disintegration. In Group B, Clus was not observed. In controls, 2 of 10 cases showed no clusters, 2 of 10 cases showed some clusters and 6 of 10 cases showed abundant cluster cells or nonchondrocytes. Tide in Group B was complete in three of seven cases, nearly complete in two of seven cases, half degree in one of seven cases and nearly absent in one of seven cases; in the controls, 1 of 10 cases was complete, 2 of 10 cases were nearly complete, 1 of 10 cases was half degree and 6 of 10 cases were not recognized as containing a calcified cartilage layer. Bform was recognized in both groups, except for one instance in each. In one instance, there was a slightly Bform in both groups. SurfH in Group B was normal in five of seven cases and showed slight fibrillation or irregularity in two of seven cases; in the controls, 1 of 10 cases was normal, 1 of 10 cases showed slight fibrillation or irregularity, 2 of 10 cases showed moderate fibrillation or irregularity and 6 of 10 cases showed severe fibrillation or disruption. FilH in Group B was complete in all instances. In the controls, FilH was complete in 4 of 10 cases, moderate in 1 of 10 cases, nearly empty in 1 of 10 cases and almost empty in 4 of 10 cases. Latl in Group B were bonded at both sides in six of seven cases and bonded at one end in one of seven cases. In the controls, Latl was bonded both sides in 1 of 10 cases, bonded at one end or partially both ends in 2 of 10 cases and not bonded in 6 of 10 cases. Basl was good in all cases in both groups. For InfH, no inflammation was observed in any cases in either group. At 3 months, the total score for Group B was significantly higher than that for the controls. In terms of individual scores, Ti, Matx, Stru, Clus, SurfH, FilH and Latl were significantly higher for Group B than for the control group (Table 2).