Abstract
Background: Damaged articular cartilage has a limited ability to repair. Operative removal of damaged cartilage and penetration into the subchondral bone to allow population of the defect with progenitor cells can result in filling of the defect with repair tissue. However, this repair tissue often degenerates over time because of its inability to withstand the mechanical forces to which it is subjected. We previously reported that recombinant human bone morphogenetic protein-2 (rhBMP-2) improves the repair of full-thickness defects of cartilage as long as six months postoperatively. We have now extended that study to examine the quality of the repair tissue at one year.
Methods: Full-thickness defects of cartilage were created in the trochlear groove of twenty-five adult New Zealand White rabbits. Eight defects were left empty, eight were filled with a collagen sponge, and nine were filled with a collagen sponge impregnated with five micrograms of rhBMP-2. The animals were killed at fifty-two weeks postoperatively, and the gross appearance of the healed defect was assessed. The repair tissue was examined histologically and was evaluated, according to a grading scale, by four individuals who were blinded with respect to the treatment. The tissue sections were immunostained with antibodies against type-I collagen, type-II collagen, aggrecan, and link protein. The residence time of the rhBMP-2 in the cartilage defect was evaluated in vivo with use of scintigraphic imaging of radiolabeled protein.
Results: One year after a single implantation of a collagen sponge containing five micrograms of rhBMP-2, the defects had a significantly better histological appearance than the untreated defects (those left empty or filled with a collagen sponge). The histological features that showed improvement were integration at the margin, cellular morphology, architecture within the defect, and reformation of the tidemark. The total scores were also better for the defects treated with rhBMP-2 than for the untreated defects, but in no instance was the repair tissue identical to normal articular cartilage. The thickness of the cartilage in the defects treated with rhBMP-2 was 70 percent that of the normal cartilage, an observation that was identical to that at twenty-four weeks postoperatively. Immunostaining demonstrated significantly less type-I collagen in the defects treated with rhBMP-2 than in the untreated defects. Immunostaining for other matrix components showed no difference among the treatment groups. The mean residence time of rhBMP-2 in the cartilage defects was eight days with an elimination half-life of 5.6 days. Detectable amounts of rhBMP-2 were present as long as fourteen days after implantation.
Conclusions: The problems associated with operative repair of cartilage include the formation of fibrocartilage rather than normal articular cartilage and the degeneration of that repair tissue over time. Our results demonstrate that the addition of rhBMP-2 to the operative site after creation of a full-thickness defect results in an improvement in the histological appearance and composition of the extracellular matrix at one year postoperatively. If these experimental results translate directly to the clinical situation, it is possible that the addition of rhBMP-2 to existing operative treatments for the repair of cartilage may improve the repair process and may help to maintain the integrity of the repair tissue.
Articular cartilage has a limited capacity to heal after injury. We previously reported that implantation of a type-I collagen sponge impregnated with recombinant human bone morphogenetic protein-2 (rhBMP-2) accelerates and improves the quality of repair in full-thickness defects of cartilage in adult New Zealand White rabbits as long as six months postoperatively. Other investigators have reported that repair tissue in full-thickness defects often appears to be well organized and intact at early time-points postoperatively but that this repair tissue, because it is biochemically and structurally different from normal articular cartilage, eventually degenerates due to its inability to withstand the biomechanical forces within the joint8. To examine the durability of the repair tissue formed after treatment with rhBMP-2, we examined such tissue one year after a single implantation of a collagen sponge containing rhBMP-2 into full-thickness defects of articular cartilage.
*One or more of the authors has received or will receive benefits for personal or professional use from a commercial party related directly or indirectly to the subject of this article. Funds were received in total or partial support of the research or clinical study presented in this article. The funding source was Genetics Institute.
†The Ohio State University College of Veterinary Medicine, Columbus, Ohio 43201.
‡Genetics Institute, Incorporated, 87 Cambridge Park Drive, Cambridge, Massachusetts 02140. E-mail address for E. A. Morris: emorris@genetics.com.
Animals
Twenty-five eight-month-old female New Zealand White rabbits were used for this study. The rabbits were anesthetized by intramuscular injection of eighty milligrams of ketamine per kilogram of body weight and eight milligrams of xylazine per kilogram of body weight. In all rabbits, the right knee served as the experimental joint and the contralateral limb served as the control. The knee joint was approached by means of a medial parapatellar incision. One full-thickness cylindrical defect, three millimeters in diameter and three millimeters deep, was then created in the articular cartilage down to bleeding subchondral bone in the trochlear groove of the femur with use of a low-speed dental drill (Kavo American, East Zurich, Illinois) equipped with a diamond-impregnated stainless-steel burr (Brasseler USA, Savannah, Georgia). The animals were divided into three groups. In eight rabbits the defect was left empty, in eight rabbits the defect was filled with a commercially available collagen sponge (bovine type-I collagen [Helistat; Integra Life Sciences, Plainsboro, New Jersey]) onto which a buffer (thirty-millimolar L-glutamic acid, 2.5 percent glycine, 0.5 percent sucrose, and 0.01 percent Tween 80, pH 4.5) had been lyophilized, and in nine rabbits the defect was filled with the same type of collagen sponge onto which five micrograms of rhBMP-2 (in the same buffer) had been lyophilized. Before implantation, the collagen sponge was cut to the approximate size of the cartilage defect.
Penicillin G procaine, at a dosage of 40,000 international units per kilogram of body weight, was administered twenty-four hours preoperatively, at the time of the operation, and forty-eight hours postoperatively. The rabbits were allowed unrestrained movement within their cages immediately after recovery from anesthesia.
Processing and Evaluation of Tissue
Histological Analysis
Fifty-two weeks postoperatively, the rabbits were killed with an injection of pentobarbital sodium, and the defects were photographed and examined grossly. The area of the defect was cut out of the bone and was fixed in 4 percent paraformaldehyde in phosphate-buffered saline solution (pH 7.4) for twenty-four hours at 4 degrees Celsius, decalcified in 20 percent (weight per volume) EDTA in phosphate-buffered saline solution for eighteen days, and embedded in paraffin. Histological sections through the center of the defect either were stained with safranin O-fast green and evaluated with the use of a histological grading system previously described7 (Table I) or were prepared for immunostaining.
Histological Grading
The sections were evaluated blindly by four of us (R. S. S., R. Z., S. S. G., and E. A. M.). Composite scoring ranged from 0 points (normal cartilage) to 31 points (no repair tissue). This modified scale included evaluation of all relevant aspects of repair of a full-thickness defect of articular cartilage7 (Table I). The composite scores as well as the scores for individual categories were compared among the treatment groups. Statistical analysis of the scores was performed with use of the Student t test.
Immunohistochemical Analysis
Immunohistochemical studies of the histological sections from the defects were carried out with use of polyclonal antibodies against type-I and type-II collagen (Southern Biotechnology Association, Birmingham, Alabama) and monoclonal antibodies against aggrecan and link protein (Developmental Studies Hybridoma Bank, University of Iowa, Iowa City, Iowa). The sections were deparaffinized in xylene and then passed through decreasing gradations of ethanol. Endogenous peroxidase activity was blocked with 1 percent hydrogen peroxide. The sections were then immersed in 20 percent normal horse serum to block nonspecific reactions and were incubated in a moisture chamber with primary antibody, or 1 percent normal horse serum in phosphate-buffered saline solution as a control, at 4 degrees Celsius overnight. An avidin-biotin-peroxidase reagent (Vectastain ABC; Vector Laboratories, Burlingame, California) and diaminobenzidine substrate solutions (Vector Laboratories) were used for visualization of the immunoreactivity in accordance with the manufacturer's instructions.
The intensity of the immunohistochemical staining was quantified with use of a computerized image-analysis system. Each image was captured by a digital camera (DCS 460; Eastman Kodak, Rochester, New York). The intensity of illumination and the magnification of the image were kept constant. Areas containing the repair tissue and normal adjacent cartilage that were of interest were identified, and average pixel density per unit area was measured with use of computer-assisted image-analysis software (Optimas, version 6.2; Optimas, Bothell, Washington). To normalize for differences in staining intensity between different sections (an expected finding because of differences in reaction conditions and tissue-processing), the results were expressed as the ratio of staining intensity in the repair tissue to that of the normal adjacent cartilage in the same histological section. The mean ratio for sections of untreated defects was compared with the mean ratio for sections from defects treated with rhBMP-2 with use of the Student t test.
Measurement of Residence Time of 125I-rhBMP-2 in Cartilage Defects
Iodination of the rhBMP-2 was performed with use of the Iodogen iodination reagent (Pierce, Rockford, Illinois) method, and the iodinated protein was combined with noniodinated protein at a ratio of 1:500. The protein was then lyophilized into the collagen sponge in a manner identical to that used for the samples evaluated histologically. After measurement of radioactivity in a radioisotope dose calibrator (CRC-15R; Capintec, Pittsburgh, Pennsylvania) to obtain a baseline measurement, the collagen sponge containing the iodinated rhBMP-2 was implanted into a full-thickness articular cartilage defect, three millimeters in diameter and three millimeters deep, in the right trochlear groove of twelve New Zealand White rabbits. An image of the right femorotibial joint was then made with use of a full field-of-view gamma camera (ZLC 7500 Orbital; Siemens, Hoffman Estates, Illinois) for three minutes or more. The imaging was repeated at three hours; at one, two, and three days; and at one and two weeks postoperatively. The counts emanating from the region of interest over the trochlear groove were compared with those from a known radioactive source to enable calibration between scans. Pharmacokinetics were calculated with use of WinNonLin software (Pharsight, Mountain View, California).
No postoperative complications occurred in any of the experimental animals.
Macroscopic Findings
Gross observation revealed no synovial effusion or signs of synovitis in any joint. Small-to-moderate-sized osteophytes were observed in 12.5 percent of the joints with an empty defect, 62.5 percent of the joints treated with a collagen sponge, and 66.7 percent of the joints treated with rhBMP-2 in a collagen sponge. All osteophytes were localized on the medial aspect of the trochlear groove at the site of the incision. All defects were filled with reparative tissue, and the edges of the defects were distinguishable from the surrounding normal cartilage. The repair tissue was white or a mixture of white and light pink. One empty defect and one treated with a collagen sponge alone had exuberant tissue, which projected slightly above the level of the adjacent cartilage. The surface of the repair tissue in the defects treated with rhBMP-2 was smoother and more regular than that in the empty defects or those treated with a collagen sponge.
Grading
All histological sections were independently graded by four individuals who had no knowledge of the treatment, and the mean score was calculated for each histological section (Table II). The mean scores were then combined to determine the mean, median, minimum, and maximum scores for each category in the scoring system as well as the total scores for each treatment group. No significant difference was found between the scores for the empty defects and those for the defects treated with a collagen sponge and buffer with respect to any of the individual categories or the composite score. This finding is consistent with our previous work, which showed no difference between the empty defects and the defects treated with a collagen sponge and buffer with respect to histological appearance or grades at four, eight, and twenty-four weeks postoperatively7. Therefore, the scores for these groups were combined and the combined group was referred to as untreated defects for comparison with the defects treated with rhBMP-2. The scores for the defects treated with rhBMP-2 were significantly superior (p < 0.05) to those for the untreated defects in categories describing integration at the margin (category 2), cellular morphology (category 4), architecture within the defect (category 5), and reformation of the tidemark (category 8). The total scores were also significantly better (p < 0.05) for the defects treated with rhBMP-2 than for the untreated defects, indicating an overall better histological appearance (Figs. 1-A and 1-B). The mean and standard deviation as well as the median, minimum, and maximum scores for each histological category were calculated, and the results demonstrated individual-to-individual variation (Table II). For the defects treated with rhBMP-2, the mean scores in the category of integration at the margin ranged from 0.25 point (indicating a confluent margin with only an area of decreased cellularity) to 1.75 points (indicating a discontinuity or gap between the repair tissue and the normal adjacent cartilage on one side of the histological section). Seven of the nine defects showed no histological signs of discontinuities separating the repair tissue and the existing adjacent cartilage. In contrast, the scores for the empty defects and the defects filled with a collagen sponge with respect to integration at the margin ranged from 1 point (indicating acellular margins) to 3 points (indicating discontinuities on both sides of the histological section). Only four of the sixteen defects that were empty or treated with a collagen sponge had no histological signs of discontinuities between the repair tissue and the adjacent normal cartilage.
No defect received a score of 0 points (indicating normal or nearly normal chondrocyte morphology) in the category of cellular morphology. For the defects treated with rhBMP-2, the scores ranged from 1 point (mostly round cells, with 25 to 75 percent of the repair tissue demonstrating columns in the radial zone) to 2 points (mostly round cells, with less than 25 percent of the repair tissue demonstrating columns in the radial zone). In comparison, the maximum score for both the empty defects and those filled with a collagen sponge was 4 points (indicating only 50 percent round cells and little or no evidence of column formation).
The category of architecture within the defect describes voids within the defect that are not associated with fibrillation or discontinuities of the surface. Five of the nine defects treated with rhBMP-2 had normal architecture within the defect (a finding that accounted for the median score of 0 points in this category for the defects treated with rhBMP-2) compared with only four of the sixteen defects that were empty or filled with a collagen sponge.
Although the scores in the four categories described were significantly improved by treatment with rhBMP-2 on a collagen sponge, the defects treated with rhBMP-2 did not have a score of 0 points in any category. This finding indicated that the histological appearance of the defects treated with rhBMP-2 was better than that of the defects that were not treated with rhBMP-2 but was not that of normal cartilage.
Replacement of subchondral bone averaged 100 percent (indicating that the subchondral bone had been replaced to the level of the original tidemark) in the empty defects, 88 percent in the defects treated with a collagen sponge, and more than 100 percent (above the original tidemark) in the defects treated with rhBMP-2; the thickness of the repair cartilage in the rhBMP-2-treated defects was 70 percent that of the normal cartilage. These differences were not significant.
Semiquantitative Immunohistochemical Analysis
Quantitative immunostaining with use of antibodies against type-II collagen, link protein, and aggrecan revealed no differences between the defects treated with rhBMP-2 and the untreated defects with respect to the ratios between staining of the repair cartilage and that of normal adjacent cartilage (Table III). A similar distribution was observed for immunostaining for aggrecan and safranin-O staining of the extracellular matrix on adjacent histological sections.
The defects treated with rhBMP-2 demonstrated significantly less (p < 0.05) relative staining density for type-I collagen than did the untreated defects (Figs. 2-A and 2-B). The ratio of the staining density for type-I collagen of repair cartilage to that of normal adjacent cartilage was 2.58 ± 0.96 in the defects treated with rhBMP-2 and 19.02 ± 5.01 in the untreated defects. An inverse relationship was observed between type-I-collagen immunostaining and safranin-O staining of the extracellular matrix (Figs. 3-A and 3-B).
Residence Time of rhBMP-2
Scintigraphic imaging of the collagen sponges that had been impregnated with iodinated rhBMP-2 and then implanted in cartilage defects revealed that the protein exhibited a two-phase elimination, with an elimination half-life of 5.6 days. The mean residence time of rhBMP-2 was eight days. Gamma radiation was no longer detectable after fourteen days, indicating minimum residual protein (Fig. 4).
The results reported in the present study demonstrate that a single implantation of five micrograms of rhBMP-2 into full-thickness defects in articular cartilage improved the histological appearance of the repair cartilage compared with that of empty defects and defects filled with a collagen sponge. This improvement was still evident as long as one year after implantation. The use of a commercially available type-I collagen sponge had no influence on the repair of the defect, as no difference was observed between the empty defects and the defects treated with a collagen sponge with respect to the gross or histological appearance or the immunostaining of the extracellular matrix.
Investigators previously have observed that full-thickness defects of articular cartilage often fill with repair tissue. However, this tissue is biochemically and structurally different from normal hyaline articular cartilage and often degenerates over time, perhaps because of the inability of the repair tissue to withstand the mechanical forces within the joint8. This finding has also been noted clinically in patients who respond positively for a short period after operative intervention but then have a painful relapse as the repair tissue degenerates3. Because of these observations, we were interested to see whether our previous findings7 of improvement in the repair of the cartilage defects at one, two, and six months after treatment with rhBMP-2 would be maintained for a full year after treatment. Our results demonstrated that the improvement was still significant at this late time-point. A comparison with the previous data from our laboratory revealed that the histological appearance of defects treated with rhBMP-2 was maintained between six months (a score of 8.6 ± 4.1 points) and twelve months (a score of 9.0 ± 2.71 points).
The specific categories that were improved by treatment with rhBMP-2 included integration at the margin, cellular morphology, architecture within the defect, and formation of a new tidemark. Integration between existing and repair cartilage has been identified as a problem during the process of cartilage repair8. In all of the treatment groups the margins of the defect were clearly identifiable and acellular, but in the defects treated with rhBMP-2 the discontinuities and gaps that frequently characterize the edges of defects were much less common, occurring in 22 percent of those defects compared with 75 percent of the defects that were not treated with rhBMP-2. Chondrocytes have a characteristic morphology, consisting of a large cell with a round cytoplasmic membrane. As identified by our grading scale, the defects treated with rhBMP-2 had a greater percentage of round cells with typical chondrocyte morphology. Although in vitro experiments by several investigators have demonstrated conflicting results5,6,10, we noted that areas containing cells with this morphology stained more intensely with safranin O and anti-aggrecan antibody, allowing us to conclude that cells with rounded morphology were synthesizing and secreting a greater amount of proteoglycan. It should be noted that our scoring system was not able to detect this finding, as the scores for safranin-O staining of the extracellular matrix for the defects treated with rhBMP-2 and those not treated with rhBMP-2 were not significantly different, with the numbers available.
Treatment with rhBMP-2 also improved the results in the category termed architecture within the defect, which refers to voids found within the repair cartilage and bone that do not reach the surface of the repair tissue. This improvement may have been due to the acceleration of subchondral bone replacement, as was also noted in our previous study7, which may provide increased physical support to the newly forming articular surface in the treated defects. Finally, the score for reformation of the tidemark was significantly improved for the defects treated with rhBMP-2. It is notable that there was evidence of reformation of the tidemark in most defects, leading to the hypothesis that the physical or chemical conditions at the surface of the joint prevent the wave of endochondral ossification from extending to the surface of the joint. Although the new cartilage above the tidemark in the defects treated with rhBMP-2 was 70 percent of the thickness of the normal adjacent cartilage, this reduction in thickness was the same as that seen at six months postoperatively in defects treated with rhBMP-2 in our previous study7, indicating that there was no additional movement of the tidemark toward the articular surface after six months.
Categories that did not show improvement at one year after treatment with rhBMP-2 were filling of the defect, safranin-O staining of the extracellular matrix, architecture of the surface (fibrillation and fissures), and the amount of bone repair. In this model of cartilage repair, the defects were completely filled (category 1) in almost all histological sections. As noted earlier, although cellular morphology was improved by treatment with rhBMP-2, no detectable improvement was seen on safranin-O staining of the extracellular matrix. Although safranin-O staining may not be sensitive enough to detect small increases in proteoglycan content, immunostaining was also not able to detect a difference in aggrecan content of the extracellular matrix.
Fibrillation of the surface is considered to be evidence of inadequate mechanical properties of cartilage and could presage future degeneration of the tissue. It is notable that although treatment with rhBMP-2 resulted in a lower mean and median score for architecture of the surface, these differences were not significant. Furthermore, the maximum score for all treatment groups was the same, as all groups had defects that showed marked fibrillation and disruption of the surface. By one year postoperatively, the subchondral bone had been replaced in all treatment groups and therefore no differences were seen.
The section-to-section variability between immunostained histological slides can be quite large because of different environmental conditions, if tissue-processing and immunostaining are performed at different times. We tried to reduce this variability by internally controlling each section by comparing the density of staining in the repair tissue with that of normal adjacent cartilage on each slide. We also used computer-assisted image analysis to perform this comparison, extending our attempt to evaluate the immunostaining objectively and reproducibly2. The ratios of staining of repair tissue to that of normal adjacent cartilage may correspond to the ratio of absolute amounts of each matrix component. However, it is likely that the huge difference in physical characteristics between repair tissue and normal cartilage affects the accessibility of the antibody to the antigen. Therefore, the most valid comparison was considered to be the difference in the ratio of the repair tissue to normal adjacent cartilage and these ratios were used to compare the defects treated with rhBMP-2 with the untreated defects. The results of immunostaining with use of antibodies against type-II collagen, aggrecan, or link protein did not reveal a difference between the defects treated with rhBMP-2 and the untreated defects with respect to these ratios. Therefore, an increase in matrix composition of components that are considered to be desirable in cartilage cannot be the factor responsible for the improvement in the histological appearance of the defects treated with rhBMP-2. We previously reported an increase in type-II-collagen immunostaining in defects treated with rhBMP-2 compared with untreated defects at eight weeks postoperatively7. It is possible that treatment with rhBMP-2 accelerated the differentiation of the repair tissue and, over time, the untreated defects achieved a similar ability to synthesize and secrete type-II collagen.
A 7.4-fold decrease in the ratio of staining density for type-I collagen was observed in the defects treated with rhBMP-2 compared with the untreated defects. Other investigators have reported a large increase in the ratio of type-I to type-II collagen in the extracellular matrix of repairing cartilage defects as long as one year postoperatively1,4,9, and it is evident from our results that this ratio would be significantly decreased in rhBMP-2-treated defects because of a reduction in type-I-collagen content of the extracellular matrix.
An advantage of immunostaining over biochemical analysis of the repair tissue was the ability to demonstrate the anatomical variability of matrix components, and it was evident from our results that there was a substantial degree of variability throughout the sections. For this reason, we calculated our ratios from the mean density over the entire area of the repairing defect. Adjacent sections were stained with safranin O, enabling a direct comparison between histological-histochemical appearance and localized immunostaining. It was apparent that type-I-collagen immunostaining varied inversely with the safranin-O staining, showing that areas deficient in proteoglycan had an increased type-I-collagen content. In addition, these were the focal areas most prone to fibrillation of the surface. These characteristics define the descriptive term fibrocartilage. As mentioned previously, our observations also revealed that safranin-O staining, aggrecan immunostaining, and a round appearance of the cytoplasm of the cells had a positive association.
Although it is possible to hypothesize release characteristics of a protein from a delivery matrix with use of in vitro methods, we were able to calculate the release of rhBMP-2 in vivo with use of scintigraphy. The limit of detectability of the gamma camera is 0.4 microcurie, which would correspond to 0.06 microgram of rhBMP-2 in our system if soft-tissue attenuation of signal were ignored. Therefore, we cannot determine the concentration of rhBMP-2 below this threshold, and it is possible that concentrations of less than that amount would still have substantial biological activity in vivo. Within this limit of detectability, our results demonstrated that rhBMP-2 had a half-life of 5.6 days and there was no detectable protein after fourteen days. It is interesting that an effect that can be demonstrated as long as one year after a single implantation of rhBMP-2 results from a protein that is so short-lived at the site of activity. Explanations may include initiation of differentiation cascades by rhBMP-2 during the initial weeks of the repair or the recruitment of more of the necessary cells into the defect during the initial weeks, or both. It is also possible that small, undetectable amounts of rhBMP-2 are maintained in the surrounding extracellular matrix and newly forming tissue and that these small amounts have biological activity over an extended time-period.
Our results demonstrate the longevity of rhBMP-2-induced improvement in a model of repair of a full-thickness defect of articular cartilage and emphasize the importance of continuing the investigation into growth-factor-induced modulation of the cartilage repair process.
Amiel, D.; Coutts, R. D.; Harwood, R. L.; Ishizue, K. K.; and Kleiner, J. B.: The chondrogenesis of rib perichondrial grafts for repair of full thickness articular cartilage defects in a rabbit model: a one year postoperative assessment. Connect. Tissue Res.,18: 27-39, 1988.1827
1988
[PubMed]
Ben-Hur, H.; Thole, H. H.; Mashiah, A.; Insler, V.; Berman, V.; Shezen, E.; Elias, D.; Zuckerman, A.; and Ornoy, A.: Estrogen, progesterone and testosterone receptors in human fetal cartilaginous tissue: immunohistochemical studies. Calcif. Tissue Internat.,60: 520-526, 1997.60520
1997
Buckwalter, J. A., and Mankin, H. J.: Articular cartilage. Part II: Degeneration and osteoarthrosis, repair, regeneration, and transplantation. J. Bone and Joint Surg.,79-A: 612-632, April 1997.79-A612
1997
Furukawa, T.; Eyre, D. R.; Koide, S.; and Glimcher, M. J.: Biochemical studies on repair cartilage resurfacing experimental defects in the rabbit knee. J. Bone and Joint Surg.,62-A: 79-89, Jan. 1980.62-A79
1980
Hauselmann, H. J.; Fernandes, R. J.; Mok, S. S.; Schmid, T. M.; Block, J. A.; Aydelotte, M. B.; Kuettner, K. E.; and Thonar, E. J.: Phenotypic stability of bovine articular chondrocytes after long-term culture in alginate beads. J. Cell Sci.,107: 17-27, 1994.10717
1994
[PubMed]
Sailor, L. Z.; Hewick, R. M.; and Morris, E. A.: Recombinant human bone morphogenetic protein-2 maintains the articular chondrocyte phenotype in long-term culture. J. Orthop. Res.,14: 937-945, 1996.14937
1996
[PubMed]
Sellers, R. S.; Peluso, D.; and Morris, E. A.: The effect of recombinant human bone morphogenetic protein-2 (rhBMP-2) on the healing of full-thickness defects of articular cartilage. J. Bone and Joint Surg.,79-A: 1452-1463, Oct. 1997.79-A1452
1997
Shapiro, F.; Koide, S.; and Glimcher, M. J.: Cell origin and differentiation in the repair of full-thickness defects of articular cartilage. J. Bone and Joint Surg.,75-A: 532-553, April 1993.75-A532
1993
Vachon, A. M.; McIlwraith, C. W.; Powers, B. E.; McFadden, P. R.; and Amiel, D.: Morphologic and biochemical study of sternal cartilage autografts for resurfacing induced osteochondral defects in horses. Am. J. Vet. Res.,53: 1038-1047, 1992.531038
1992
[PubMed]
von der Mark, K.; Gauss, V.; von der Mark, H.; and Muller, P.: Relationship between cell shape and type of collagen synthesised as chondrocytes lose their cartilage phenotype in culture. Nature,267: 531-532, 1977.267531
1977
[PubMed]