Experimental Design
This research was approved by the Institutional Animal Care Committee of Utrecht University. In nine adult female Dutch milk goats, a standardized defect in the medial femoral condyle was created in both knees. After ten weeks, the defects were randomly treated with either marrow stimulation (microfracture) or the use of an oxidized zirconium press-fit implant. The animals were killed twenty-six weeks after treatment (thirty-six weeks after creation of the defect). Macroscopic evaluation of the articular compartments was performed immediately before creation of the defect as well as before and after treatment.
Implant osseointegration was measured by automated histomorphometry, and defect healing after microfracture was scored histologically. Tibial and femoral cartilage quality was scored after treatment by macroscopic, histologic, and biochemical analysis.
Animals
Nine adult female Dutch milk goats with a mean age (and standard deviation) of 2.0 ± 0.21 years and a mean weight of 57.1 ± 10.1 kg were used for surgery. The number of animals needed for this study was determined by a power analysis. The power was 0.8, and a was 0.05. The data used for this analysis were obtained from biochemical analysis as described previously20. The minimum effect considered meaningful was 15%. Given the two samples with normal distribution and equal variance, a double-sided power analysis was performed. The sample size was calculated to be eight animals per group. On the basis of these calculations, the number of animals used was nine (including one reserved for drop out). Food and water were given ad libitum. General health and care conditions were monitored by the laboratory animal welfare officer.
Implants
Implants were custom-manufactured to our design specifications by Smith and Nephew (Memphis, Tennessee). The oxidized zirconium components were produced from a wrought zirconium alloy (Zr-2.5% Nb) that was oxidized by thermal diffusion to create a zirconia surface, which is approximately 5 µm thick, and then polished (Ra < 0.03 µm). The size of the implant was 13.5 mm in length with a 5.0-mm-diameter articulating surface. The articulating shape of the implant was designed after a study on goat cadaver knees and was tested in a pilot study (Fig. 1).
Surgery
After the goats were acclimatized for at least three weeks in the animal care facility, a topical fentanyl patch was given as pain medication one day prior to surgery. The goats were weighed preoperatively. Surgery was performed on both knees, with the goat under general inhalation anesthesia with use of an isoflurane (2% in air) gas mixture (Abbott Laboratories, AST Pharma, Oudewater, The Netherlands) preceded by detomidine hydrochloride sedation (Pfizer, Capelle aan de IJssel, The Netherlands) and antibiotic prophylaxis consisting of intravenous administration of 500/100 mg amoxicillin-clavulanate potassium (Augmentin; GlaxoSmithKline, Brentford, Middlesex, United Kingdom). All surgical procedures were carried out under aseptic conditions and by the same surgeon (R.J.H.C.), who had gained specific experience during the course of a pilot study. During the first procedure, the cartilage defect was created; the medial femoral condyle was exposed through a medial parapatellar incision, without dislocating the patella. After the joint was inspected and the location for implantation was determined, a 5-mm-diameter drill was used to create a standardized full-thickness cartilage defect (2 mm deep) that did not penetrate the subchondral bone layer. Following lavage, the joint was closed in three layers. Postoperative pain relief was provided by buprenorphine (Schering-Plough, Maarssen, The Netherlands). Until five days postoperatively, ampicillin (Albipen LA; Intervet, Boxmeer, The Netherlands) was given. After ten weeks, a second arthrotomy was performed and the defects were randomly treated by microfracture or by placement of an oxidized zirconium implant.
During the microfracture technique, a 1.5-mm-diameter drill was used to perforate the subchondral bone layer at the middle of the defect until blood appeared (at a depth of 4 mm).
To place the oxidized zirconium implant, the defects were inspected and drilled deeper (3.0 mm) and, subsequently, the implants were placed flush to the surrounding cartilage surface by tapping the implant into place with use of a specially designed tamp with a polyethylene head, as previously described9. The threads on the implants were designed for osseointegration. After insertion, the implants were visually inspected and manually tested for loosening.
Postoperatively, the goats were allowed to bear full weight, without casting or bandages. After a subsequent period of twenty-six weeks, the animals were weighed and killed with use of an overdose of pentobarbital (Euthesate; Willows Francis Veterinary, Crawley, United Kingdom).
Radiographs
Fluoroscopy was used at several stages of the experiment. Preoperatively, fluoroscopy was used to confirm the normal anatomy and the size of the bone. Immediately after surgery, the osteochondral defect and/or the positioning of each implant were visualized and surgical complications such as fracture were sought. After the animals were killed (thirty-six weeks after creation of the defect), the joints were checked for malpositioning, loosening (sclerotic zones and radiolucency surrounding the implant), and movement of the implant from the original position or for other complications.
General Macroscopic Articular Evaluation
After opening of the joint space prior to creation of the defects (the healthy joint), prior to application of the treatments (the defect baseline), and post mortem, the joint was inspected and evaluated by two observers (R.J.H.C. and M.H.P.v.R.) according to the scoring system guidelines as described by O'Driscoll et al.21-23. This scoring system evaluates the joint with respect to range of motion, intra-articular fibrosis, and cartilage appearance. Since it was not possible to score the so-called restoration of contour and cartilage erosion of the graft for the implant groups, the macroscopic articular evaluation parameters were modified (scores 0 to 6 were used instead of 0 to 10). A score of 6 represents a healthy joint, whereas a score of 0 represents a severely degenerated, fibrillated, and stiff knee joint.
Macroscopic Scoring of Articular Cartilage Compartments
After removal of soft tissues, macroscopic evaluation, as described by Mastbergen et al.24, was performed on coded high-resolution photographs for each compartment separately (the medial plateau, which was directly in contact with the defect or implant, and the lateral plateau and condyle, as part of the knee joint but not directly in contact with the defect or implant) by two observers (R.J.H.C. and M.H.P.v.R.), who were blinded to the source of the photograph. This scoring system evaluates the cartilage surface with respect to fibrillation and grooves. The scores of the two observers were averaged, and outliers with a difference of >1 point were scored again, until consensus was reached. The higher the score, the more the cartilage is damaged.
Cartilage Explants Harvest
Cartilage tissue samples were obtained from predefined locations under aseptic conditions (Fig. 2). First, two full-thickness cartilage tissue samples were cut from the underlying subchondral bone layer, one just anterior to and one just posterior to the greatest weight-bearing area of each femoral condyle and tibial plateau. Subsequently, each cartilage sample was cut into three or five separate samples (depending on location), of which two and four samples, respectively, were used for biochemical assays. One sample was used for histological analysis. Furthermore, full cross-sectional osteochondral samples were harvested for histological analysis from the middle of each compartment (the medial and lateral tibial plateau, and the medial and lateral femoral condyle), including the defect area treated by microfracture or containing the implant.
Evaluation of Biocompatibility and Osseointegration
After forty-eight hours of fixation in 10% buffered formalin, the tissue slices from the medial femoral condyle including the implants were dehydrated by means of 70% to 100% ethanol and were embedded in polymethylmethacrylate. Approximately 10 to 20-µm-thick sections were sawed in a longitudinal direction through the middle of the implant with use of the Leica saw microtome (SP1600; Leica Microsystems, Rijswijk, The Netherlands) and subsequently stained with basic fuchsin and eosin.
Histomorphometry was performed with a personal computer-based system equipped with the KS400 software (version 3.0; Carl Zeiss Vision, Oberkochen, Germany) as described previously9. In short, the percentage of bone surrounding the implant was calculated and the percentage of the implant circumference in contact with bone was determined.
Histologic Evaluation of Cartilage
Immediately after harvesting, all tissue samples used for histological analysis were fixed in 10% buffered formalin for forty-eight hours. Subsequently, all osteochondral tissue samples were decalcified with use of Luthra solution (3.2% 11 M HCl and 10% formic acid in distilled water) for forty-eight hours. Nondecalcified and decalcified tissues were then dehydrated by means of 70% to 100% ethanol, were immersed in xylene, and were embedded in paraffin. Embedded tissues were cut into 5-µm-thick paraffin sections and stained with safranin O and fast green according to the Osteoarthritis Research Society International (OARSI) guidelines25. The histological sections were blinded and presented at random to two observers (R.J.H.C. and M.H.P.v.R.).
Osteochondral sections from the entire lateral femoral condyle and the tibial plateau from medial to lateral were evaluated with use of the OARSI Cartilage Histopathology Assessment System (OOCHAS)25 to determine the degree of cartilage degeneration for each compartment separately. A score of 0 points represents normal cartilage, whereas a score of 24 points represents a severely degenerated joint surface. The degree of filling of the cartilage defects after microfracture in these decalcified sections was also scored, at random, with use of the criteria suggested by O'Driscoll et al.22,23, by two different observers (R.J.H.C. and M.H.P.v.R.) who were blinded to the treatment.
In addition, the histologic and histochemical grading system as described by Mankin et al.26 was applied to the nondecalcified cartilage samples. Since bone was not included in these specimens, the tidemark between cartilage and bone was not present in our cartilage samples. Also, because of the dissection method, cartilage samples could not be covered with pannus. Therefore, the maximum score that could be obtained was 11 instead of 14, because of the omission of the criteria for "pannus," "clefts to calcified zone," and "tidemark crossed by blood vessels," as described by Lafeber et al.27.
Biochemical Cartilage Evaluation
For femoral condyles and tibial plateaus, the cartilage proteoglycan content, synthesis, and release were determined, respectively, for eight and sixteen explants obtained from predefined locations. Two to four full-thickness cartilage slices, anterior and posterior from the greatest weight-bearing area, were weighed aseptically (range, 5 to 20 mg) and cultured individually in ninety-six-well round-bottom microtiter plates. Explants were cultured in Dulbecco modified Eagle medium (74-01600; Gibco, Breda, The Netherlands), containing 0.81 mM SO42— and 24 mM NaHCO3, supplemented with ascorbic acid (85 µM), glutamine (2 mM), penicillin (100 IU/mL), streptomycin sulfate (100 µg/mL), and 10% pooled goat serum, at 5% CO2 and 37°C. Proteoglycan synthesis was determined at day 0 by measuring 35SO42— incorporation, and explants were incubated for a subsequent three-day period in the absence of label to determine proteoglycan release and proteoglycan content. Conditioned medium and cartilage samples were stored at -20°C until analysis.
Proteoglycan Synthesis
Proteoglycan synthesis was determined by measuring 35SO42— incorporation into glycosaminoglycans in the presence of 10 µCi/mL of 35SO42— (Na235SO4, carrier-free; NEX-041-H; DuPont, Dordrecht, The Netherlands). Four hours after the addition of 35SO42—, the samples were washed three times for forty-five minutes in fresh culture medium (at 37°C) and were cultured for another three days. Cartilage samples were digested in 3% papain buffer (dissolved in 0.5 M phosphate buffer, 20 mM N-acetylcysteine, and 20 mM Na2-EDTA, pH 6.5) at 65°C for two hours. Papain digests were diluted to the appropriate concentrations for analysis of the proteoglycan synthesis rate and proteoglycan content as well as DNA content.
The 35SO42— incorporation into glycosaminoglycans was measured by counting the cartilage papain digests and conditioned medium with use of a liquid scintillation counter (Tri-Carb 1900CA; Packard Instrument, Downers Grove, Illinois). The sum of 35SO42— radioactivity in the cartilage digests and conditioned medium after three days of culturing represented 35SO42— incorporation on day 0. The rate of 35SO42— incorporation was expressed as nanomoles of 35SO42— incorporated per hour per gram wet weight of the cartilage tissue.
Proteoglycan Content and Release
Precipitation of glycosaminoglycans with alcian blue stain (A5268; Sigma-Aldrich, Zwijndrecht, The Netherlands) in the papain digests and conditioned medium was determined as a parameter for total proteoglycan content and release, respectively. Staining for glycosaminoglycan was measured as the change in absorbance at 620 nm, with use of chondroitin sulfate (C4384; Sigma-Aldrich) as a reference. The results are expressed as milligrams of glycosaminoglycan per gram wet weight of the cartilage explants, and the initial glycosaminoglycan content on day 0 was calculated from the total amount of glycosaminoglycan released into the medium and the total glycosaminoglycan content of the explants after seventy-two hours.
Cellularity
The DNA content of the cartilage tissue digests, as a measure of cell content, was determined in the cartilage digests with use of the fluorescent stain Hoechst 33258 (382061; Calbiochem, La Jolla, California). Calf thymus DNA (D4764; Sigma-Aldrich) was used as a reference. The results are expressed as milligrams of DNA per gram wet weight of cartilage.
For all biochemical assays, the results of the four or eight cartilage explants per specific location in the joint were averaged, and this average was used for statistical analysis.
Statistical Analysis
The values are given as the mean and the standard error of the mean. For the nonparametric data (i.e., macroscopic articularevaluation parameters and macroscopic cartilage score), a Friedman test was used. For the parametric data (i.e., the histologic cartilage score, grade on the histologic and histochemical system and the OOCHAS system, glycosaminoglycan content and release, 35SO42— incorporation, and DNA content), a paired t-test was used to analyze the differences between treatments at each location.
All results were adjusted for multiple testing with use of a Bonferroni correction. For all analyses, p = 0.05 was defined as a significant difference.
Source of Funding
D.B.F. Saris was supported by the Netherlands Organisation for Health Research and Development (NWO). This research was supported by the Dutch Arthritis Association and the Anna Foundation. The implants were provided by Smith and Nephew, Memphis, Tennessee.
Surgery and Animal Health
Surgery was performed without complications. All goats were able to load the limbs and move the knees without any limitations. The goats showed a mean maximal weight loss (and standard deviation) of 0.9% ± 1.7% after the first surgery and 2.1% ± 4.1% after the second surgery. After the second surgery, visual inspection and manual palpation did not show any fibrotic changes and all implants were mechanically stable and located in their original position (Fig. 3). Preoperatively, fluoroscopy confirmed normal knee joint anatomy for all goats. After thirty-six weeks of follow-up, fluoroscopy showed no signs of malpositioning, loosening, or other complications.
General Macroscopic Articular Evaluation
The mean (and standard error) of the modified macroscopic articular evaluation parameters was 5.9 ± 0.11 for the uninvolved, healthy joints (Fig. 4). The scores decreased to 3.9 ± 0.16 at ten weeks after creation of the defect (the defect baseline) (p < 0.05). When the animals were killed at twenty-six weeks after surgical treatment, the scores had increased to 4.89 ± 0.31 for the implants (p < 0.05) and, although not significantly different from the implant-treatment value, to 4.33 ± 0.33 for microfracture treatment. On macroscopic evaluation, no significant difference was seen between the microfracture treatment and treatment with an implant.
Macroscopic Evaluation of Articular Compartments
Macroscopically, twenty-six weeks after treatment of the defect, all compartments showed mild degeneration, ranging from slightly fibrillated to fibrillated with shallow grooves. When the compartments were analyzed separately, no significant difference was detected between the knees treated with microfracture and those treated with an implant, although in all compartments the mean scores for the implant-containing knees were lower (Fig. 5).
Histologic Cartilage Repair Score
As determined by the cartilage repair score, the healing of the microfracture-treated defects was extensive although not complete (Fig. 6, A). The average score (and standard error) was 18.38 ± 0.43 points of a maximum possible score of 24 points.
Biocompatibility and Osseointegration
Five of the nine analyzed implants showed limited bone-implant contact, and the other four implants showed moderate bone-implant contact, with the remainder of the implant surface being covered by fibrous tissue (Fig. 6, B). This resulted in a mean bone-implant contact (and standard error) of 14.6% ± 5.4% and bone formation surrounding the implant of a mean of 40.3% ± 4.0%.
Histologic Cartilage Evaluation
According to the OOCHAS scores of the osteochondral tissue samples, all compartments showed slight (intact surface with hypertrophy and/or edema) to moderate (simple to complex fissures with proteoglycan washout) cartilage degeneration (Fig. 7). In general, the lower scores showed mild proteoglycan washout and minor fissures, whereas the higher scores showed moderate proteoglycan washout (i.e., less intense safranin-O staining) and moderate fissures. Fibrillations were rarely seen histologically. The microfracture treatment was associated with significantly more degeneration of the medial tibial plateau compared with the implant treatment (p < 0.05). The lateral tibial plateau and lateral femoral condyle showed no significant difference between the two treatments (Fig. 8).
Analysis of the scores on the histologic and histochemical grading system for the cartilage samples taken from locations anterior and posterior to the greatest weight-bearing area revealed that all compartments showed slight degeneration (surface irregularities with slight reduction of safranin-O staining), with no difference found between treatments (p > 0.05).
Proteoglycan Content and Release
The glycosaminoglycan content (in micrograms per milligram of wet weight) was significantly less for the medial tibial plateau cartilage in joints treated by microfracture (mean and standard error, 22.41 ± 2.10) compared with those treated with implants (34.87 ± 3.76) (p < 0.05) (Fig. 9, A). No differences were seen at the other locations in the joint. The glycosaminoglycan release after seventy-two hours of culturing was significantly higher in the medial tibial plateau in the knees treated with microfracture (mean and standard error, 9.90% ± 1.09%) compared with implant-treated knees (6.18% ± 0.49%) (p < 0.05) (Fig. 9, B). No differences were seen at the other locations in the joint.
Metabolic Cartilage Activity
The 35SO42— incorporation (in nanomoles per hour per gram of wet weight) after four hours showed significantly less incorporation in the medial tibial plateau cartilage in knees treated with microfracture (mean and standard error, 7.35 ± 1.27) compared with cartilage from knees treated with an implant (14.83 ± 2.12) (p < 0.05) (Fig. 9, C). The other locations showed no difference between the treatments.
Cellularity
Cartilage DNA content (in micrograms per milligram of wet weight) after seventy-two hours of culturing was significantly lower in the medial tibial plateau in knees treated with microfracture (mean and standard error, 0.035 ± 0.005) compared with those treated with implants (0.049 ± 0.003) (p < 0.05). The other locations showed no difference between treatments.