Experimental Procedure
Fourteen adult mongrel dogs that each weighed approximately thirty kilograms were used in the study. Before inclusion in the study, the animals were examined roentgenographically to exclude those that had degenerative joint disease of the knee. Two four-millimeter-diameter defects were created in the trochlear groove of both femora of eight dogs and in the right femur of six dogs, for a total of forty-four defects. All operations were performed with the dogs under general anesthesia and with use of a sterile technique. A four-millimeter-diameter dermal punch was used to outline the defect. With use of loupe visualization, an attempt was made to remove all non-calcified cartilage from the defect by scraping with a customized curet, without damaging the calcified cartilage. The objective was to remove as much articular cartilage as possible without leaving residual articular cartilage that might interfere with healing or affect the validity of the histomorphometric measurements. Although no bleeding from the subchondral bone was noted during the operation, an evaluation that was performed immediately after the creation of six additional defects showed that the operative procedure did, in fact, sometimes penetrate the calcified cartilage. The two defects in each knee were placed approximately 1.25 and 2.25 centimeters proximal to the intercondylar notch, slightly lateral or medial to the midline. Before the capsule was closed, bleeding vessels were clamped and cauterized. The knee joint was closed by zero-point suturing.
In one treatment group, cultured autologous chondrocytes were implanted under a periosteal flap (sixteen defects in eight animals). In the second group, the defect was covered with a periosteal flap without implantation of autologous chondrocytes (twelve defects in six animals). In the third group (the control group), the defects were left empty (sixteen defects in eight animals). In eight dogs, the two defects in one knee were treated with cultured autologous chondrocytes and the two defects in the other knee were left untreated. The remaining six dogs had all twelve defects treated with periosteum only. The defects were evaluated after twelve or eighteen months of healing.
As mentioned, an additional six fresh defects were examined immediately after they had been created. These defects, which were identical to those used for the treatment groups, were created in three dogs from a related study immediately after the animals had been killed. The defects were evaluated in order to demonstrate the extent of the original lesion.
Treatment Protocol
In the group that was to receive cultured autologous chondrocytes, tissue was removed from the lateral and medial margins of the articular cartilage of the distal part of the left femur for use in the culture of the chondrocytes. At that time, two defects also were created in the trochlear groove of the left femur and were not treated; these were the control (empty) defects. After the procedure, the dogs were allowed free activity in the cages. The cartilage from the defects and the margins was sent to Genzyme Tissue Repair (Cambridge, Massachusetts), where the cells were isolated and expanded in culture. Three weeks later, the cultured autologous chondrocytes were provided for injection in a tuberculin syringe with a 22-gauge blunt-end needle. Defects for treatment were prepared in the contralateral femur as already described. Periosteum was obtained from the proximal part of the tibia, and a four-millimeter-diameter circular periosteal flap, with the cambium layer facing the base of the defect, was placed over the defect and sutured with use of 8-0 sutures to the articular cartilage surrounding the defect. Cultured autologous chondrocytes in culture medium were injected under the periosteum (2 x 106 cells per defect). The defect was sealed with autologous fibrin, as has been reported previously4. In brief, autologous fibrinogen was isolated by cryoprecipitation from the blood of the animal several weeks before the operation. At the operation, the frozen cryoprecipitate was thawed and one milliliter of the substance was drawn into a syringe. One milliliter of a solution of bovine thrombin (Parke-Davis, Morris Plains, New Jersey), at a concentration of five units per milliliter with 0.5 per cent calcium chloride and ten milligrams of aminocaproic acid per milliliter, was provided in a separate syringe. After the area of the defect was dried, the contents of both syringes were simultaneously injected into a sterile container, mixed, and spread to cover the entire surface of the periosteal flap and the immediately adjacent cartilage. The incision was closed, and the knee was immobilized with use of external fixation (IMEX Veterinary, Longview, Texas) for ten days to prevent dislodging of the graft or the reparative tissue.
In the group in which the defect was treated with periosteum alone, only the right knee was operated on and the procedure was the same as that for the group in which cultured autologous chondrocytes were implanted under a periosteal flap except that the cells were not used.
Processing of Tissues
The animals were killed with an injection of ten milliliters of pentobarbital sodium CII solution (0.6 milligram per milliliter). At that time, the defects were examined grossly and photographed. The distal part of each femur was excised and placed in 10 per cent neutral buffered formalin. Several hours later, each trochlea was carefully dissected with a fine-tooth coping saw and placed in formalin for four days. The trochlea was rinsed of formalin, dissected further into individual specimens approximately five millimeters thick, and placed into 15 per cent disodium EDTA decalcifying solution (pH 7.4). The sample was placed on a shaker at 4 degrees Celsius, and the decalcifying solution was changed three times each week for four weeks. The specimens were rinsed thoroughly, dehydrated, and embedded in paraffin at 60 degrees Celsius. Seven-micrometer-thick sections were prepared and were stored at 4 degrees Celsius.
Light Microscopy
Sections were stained with trichrome or safranin O-fast green. One section from the middle portion of each defect was analyzed. Light microscopy was performed with a Vanox AHBT research microscope (Olympus America, Melville, New York). Polarized light microscopy also was used to reveal the orientation of the collagen.
The quantified features included the specific types of tissue filling the defect, the integration of reparative tissue with the calcified cartilage and with the adjacent articular cartilage surrounding the defect, and the integrity of the calcified cartilage layer. The histological characteristics of the surrounding cartilage and bone also were described.
Quantitative Analysis
Quantitative analysis of the entire defect area of the selected sections was performed with the aid of an eyepiece reticle inscribed with a square grid of ten lines by ten lines. The geometry of the defect first was defined and then certain healing characteristics were determined as an areal or a linear percentage of the appropriate parameters. These percentages refer only to the representative histological cross section through the middle portion of the defect.
Grading was performed at a magnification of 100 times, with each grid-opening measuring sixty micrometers on each edge. The area of the defect was calculated as a rectangle of the base (b), which was equal to the distance along the base of the defect at the tidemark, and the height (h), which was equal to the mean height of the articular cartilage 600 micrometers lateral to each of the two edges of the defect. The height was recorded at a distance from the margins in this standard manner to reflect the original height of the defect; the height of the cartilage at the edge of the healing defect normally was slightly depressed relative to the original height of the cartilage, a finding associated with rounding of the surface at the edge of the defect. The mean dimensions of a defect in histological cross section were approximately 0.64 by 3.4 millimeters. The area of each specific type of tissue within the defect was measured by counting the number of grid-openings containing that specific type of tissue. The actual shape of the defect was a trapezoid, as the heights of the cartilage on either side were not necessarily identical. Dividing the area of the specific type of tissue by the total area of the defect yielded an areal percentage.
Additional quantitative measurements included the percentage of the total length of the base of the defect (0 to 100 per cent) that was bonded to the repair tissue and the percentage of the total length of the edges of the defect (0 to 100 per cent) over which repair tissue was integrated with the adjacent cartilage. Bonding to the calcified cartilage was demonstrated by polarized light. Perpendicular orientation of collagen fibers to the tidemark and apparent continuity of the fibers from calcified to non-calcified layers were used as criteria to assess the bonding of the reparative tissue to the calcified cartilage. However, the determination of bonding to the adjacent cartilage was more difficult. The final parameter recorded was the percentage of the length of the calcified cartilage that had a normal appearance (0 to 100 per cent).
Analysis of the Fresh Defects
Each of the six fresh untreated defects in the dogs from the related study was serially sectioned through approximately 60 per cent of its width. At least eight sections, spaced approximately 250 micrometers apart, were examined.
The amount of articular cartilage remaining in the defect was determined in three transverse sections made through the middle of the lesion. The parameters, related to the inability to remove all cartilage to the tidemark, were expressed as the percentage of the total area of the defect consisting of cartilage remaining in the corners of the defect, the percentage of the total area of the defect consisting of cartilage remaining elsewhere on the base of the defect, the percentage of the length of the base of the defect covered by residual articular cartilage, and the percentage of the base with damaged calcified cartilage. Calcified cartilage was described as damaged if there was evidence of thinning or there were fractures more than twenty micrometers wide or thirty micrometers deep from the tidemark. Smaller fractures, termed microfractures, were not considered damage.
Criteria for the Identification of Specific Types of Tissue
Four types of tissue (fibrous tissue, transitional tissue, hyaline cartilage, and bone) were distinguished on the basis of the appearance of their cells and matrix when examined by regular and polarized light microscopy. Safranin-O staining was used to identify the presence of sulfated glycosaminoglycans.
Fibrous tissue consisted of spindle-shaped cells with bipolar tapered ends and elongated nuclei. The cells were intimately associated with surrounding collagenous matrix and did not reside in lacunae. The matrix contained collagen fibers oriented parallel to each other, which formed bundles roughly parallel to the surface. The collagen fibers were clearly visible at low power under polarized light, and the matrix did not stain positively for glycosaminoglycans.
Transitional tissue had characteristics intermediate between fibrous tissue and hyaline cartilage. Some areas of matrix stained for glycosaminoglycans. The visibility of fibers in the matrix distinguished this tissue from hyaline cartilage. Cells were rounded and often resided in lacunae, thus distinguishing transitional tissue from fibrous tissue.
Hyaline cartilage was identified primarily by the appearance of its cells and matrix. No individual collagen fibers or bundles were visible in the matrix. The hyaline matrix also had a distinctive appearance under polarized light, with diffuse transmission through the matrix except where the large collagen bundles were seen inserting into the calcified cartilage at the tidemark. Cells displayed a spherical morphology (except those near the surface, which usually were elongated as in normal articular cartilage). Normally, the cells in hyaline cartilage have well developed lacunae, and pericellular staining by safranin O is more intense than interterritorial matrix staining. However, tissue lacking normal safranin-O staining or a cellular appearance was seen and graded as hyaline cartilage. Portions of hyaline cartilage that met the histological criteria of articular cartilage, including complete matrix staining and a columnar arrangement of cells, were recorded as a subset of hyaline cartilage.
In many specimens, the adjacent cartilage flowed into the peripheral regions of the defect, crossing an imaginary line rising normal to the base of the defect at the edge. This material was recorded separately as matrix flow and not as one of the four types of tissue.
Semiquantitative Grading
Semiquantitative grading of damage or degenerative changes in the articular cartilage and bone in the defects was performed, with scores ranging from 0 to 3 points. A score of 0 points indicated severe damage or degenerative change; 1 point, moderate degenerative change; 2 points, slight degenerative change; and 3 points, tissue similar to normal cartilage. The categories that were examined included the state of the subchondral bone and the effects on the adjacent cartilage, specifically staining for glycosaminoglycans, cloning, and the integrity of the surface and deeper tissues.
Statistical Methods
Comparisons were made on the basis of the averaged data from the two defects in each knee of all the animals in the treatment group. Two-way analysis of variance was performed to determine the effects of treatment and time on the types of tissue appearing in the defects. Student t tests were used to determine the significance of differences in results between selected groups. The level of significance was p < 0.05 for all tests.
Morphology and Characteristics of the Fresh Defects
In the six fresh defects that were examined immediately after they had been created in order to demonstrate the extent and characteristics of the original defects, the residual cartilage occupied a mean (and standard deviation) of 1.7 ± 2.1 per cent (range, 0.2 to 6.0 per cent) of the total area of the defect. Residual cartilage in the corners occupied 0.7 ± 0.6 per cent (range, 0.2 to 1.8 per cent) of the total area, and cartilage elsewhere along the base of the defect occupied 1.0 ± 1.6 per cent (range, 0 to 4.2 per cent). It also was found that 16 ± 26 per cent of the length of the base was covered by residual cartilage. The high coefficient of variation was due to an outlying value of 68 per cent (Fig. 1-A), one of 17 per cent, and four of less than 4 per cent (Figs. 1-B, 1-C, and 1-D).
Histological analysis of serial sections of the fresh defects revealed three types of injury of the calcified cartilage: fracture at the edge of the defect, thinning, and microfracture. In four of the six defects, at least one large fracture (fifty to 100 micrometers wide) was noted at the extreme edge. This damage was consistent with overpenetration by the dermal punch when the defect was circumscribed. Two of these fractures extended through the calcified cartilage but only superficially into the subchondral bone (Fig. 1-C). The second type of damage, thinning of the zone of calcified cartilage, presumably was due to scraping of the base of the defect with the curet. Thinning of the calcified cartilage was inconsistent, varying from specimen to specimen (Figs. 1-A, 1-B, 1-C, and 1-D) and even within a given specimen. An average of 22 ± 14 per cent (range, 2 to 39 per cent) of the calcified cartilage was thinned or fractured. Finally, each of the five defects that contained little or no residual articular cartilage showed microfracturing of the surface of the calcified cartilage (Fig. 1-C), with a range of one or two microfractures per slide to ten or more per slide. None of the microfractures penetrated the underlying bone.
In summary, one defect showed no thinning of the calcified cartilage but had much more residual cartilage than the other defects (Fig. 1-A). Two defects showed thinning of the calcified cartilage; a portion of the calcified layer (as much as 700 micrometers along the base) had been completely removed, with consequent damage to the underlying bone (Fig. 1-B). The depth of the remaining three defects generally extended to the tidemark, with little residual cartilage, and thinning (if any) was limited to the superficial portions of the calcified cartilage (Fig. 1-D). Some of these specimens had a corner fracture, and all had microfractures.
Effects of Implantation of Cultured Autologous Chondrocytes
All animals walked normally after removal of the external fixation. When the joints were opened, the synovial tissue appeared normal on gross examination, although there was evidence of slight hypertrophy or, less commonly, atrophy in certain joints. In one animal, the knee with two empty defects had substantial effusion with turbid synovial fluid and much more pronounced synovial hypertrophy at eighteen months. The patella appeared laterally displaced, and denuded cartilage on the trochlea and patella was consistent with patellar subluxation. These defects were excluded from subsequent analysis.
No apparent differences were found among the three groups. All defects were distinguishable grossly from the surrounding cartilage, and there was variability in the filling and appearance. Repair tissue usually had a whiter, more opaque, and less glistening appearance than normal cartilage.
Histological Findings
No discernible differences were seen in the histological appearance of the repair tissue among the three treatment groups. Histological examination confirmed both the variable filling of the defects observed grossly and the over-all similarity of appearance of the lesions among the three groups (Figs. 2, 3, 4, 5, 6-A, 6-B and 6-C). Several defects were relatively unfilled, while others were filled with repair tissue to or above the level of the surrounding cartilage. In each treatment group, the repair tissue filling the eighteen-month-old defects displayed limited safranin-O staining of glycosaminoglycans as well as loss of staining in the adjacent cartilage (Figs. 2, 6-A, 6-B, and 6-C). The predominant types of tissue filling the eighteen-month-old defects in each group were transitional tissue and hyaline cartilage (Figs. 3, 4, 5, 6-A, 6-B and 6-C).
Hyaline cartilage, especially that most similar to articular cartilage, formed preferentially at the base of the defect, most frequently in the corners, and was well bonded to the calcified layer. Much of the hyaline cartilage in the defect showed decreased staining with safranin O.
Only five defects contained osseous repair tissue. These included one defect that had healed for eighteen months after implantation of cultured autologous chondrocytes (Fig. 2, A), two that had healed for twelve months after coverage with periosteum alone, as well as one that had healed for twelve months and one that had healed for eighteen months after no treatment. The tidemark partially reformed at an elevated level within one defect. In two others, it appeared that a new, elevated, calcified layer was forming, but the structure was abnormal.
Histomorphometric Evaluation of the Tissue Filling the Defects
The percentage of the total area of the defect filled by different types of tissue varied considerably from defect to defect, but the mean values were similar in all treatment groups (Fig. 7 and Table I). With use of two-way analysis of variance with respect to treatment and time, we could detect no significant differences among the three groups, with the numbers available. Analysis of variance, however, indicated several differences between the defects that had twelve months of healing and those that had eighteen months. The significant trends from twelve to eighteen months included increasing amounts of fibrous tissue (p = 0.029) and transitional tissue (p < 0.010). In addition, an increase in the total amount of repair tissue and a decrease in the amount of hyaline cartilage were found in all treatment groups at eighteen months compared with the amounts found at twelve months, but with the numbers available these trends did not meet our criteria for significance (p = 0.052 and p = 0.087, respectively).
For completeness, the results at twelve and eighteen months were divided into critical healing categories (total filling and amount of hyaline cartilage) and examined with t tests for differences among the treatment groups. The t test was chosen over analysis of variance to take advantage of the paired comparison made possible by the presence of empty control defects and defects implanted with cultured autologous chondrocytes in the contralateral limbs of the same animal. The t tests for the group that had treatment of the defect with periosteum alone were unpaired. With the numbers available, we could detect no significant findings with the t test at either time-period.
Bonding to the calcified cartilage was related to the amount of hyaline cartilage found at the base of the defect. The amount of bonding ranged from 41 to 60 per cent of the length of the base, excluding the empty defects evaluated after twelve months of healing, which had an outlying value of 89 per cent (Table II). With the numbers available, we could detect no significant difference among the groups (p = 0.44) with use of two-way analysis of variance. A prerequisite for bonding was an intact layer of calcified cartilage, which ranged from 65 to 98 per cent, and we could detect no difference among the treatment groups with two-way analysis of variance (p = 0.86). Integration with the adjacent cartilage was universally poorer than attachment to the base, ranging from 16 to 32 per cent (Table II). With two-way analysis of variance, we could detect no difference among the treatment groups (p = 0.90).
Abnormal Changes in Bone and the Articular Cartilage Surrounding the Defect
In all groups, reactions at the base of the defects sometimes resulted in disruption of the layer of calcified cartilage, penetration into the subchondral bone, and resorption of the subchondral bone, leading to decreased bone support under the defect (Table III, Fig. 6-C). Moderate or severe loss of bone was noted in five of the sixteen defects in which cultured autologous chondrocytes had been implanted, two of the twelve defects treated with periosteum alone, and three of the fourteen empty defects.
Most samples of the articular cartilage adjacent to the defects showed cloning of chondrocytes and variable loss of safranin-O staining. Loss of safranin-O staining was greater in the defects that were sutured: moderate or severe loss occurred in eleven of the sixteen defects in which cultured autologous chondrocytes had been implanted and in seven of the twelve defects that had been treated with periosteum alone compared with only two of the fourteen empty defects (Table III, Figs. 6-A, 6-B, and 6-C). The area immediately adjacent to the defect was similar, with respect to cloning of chondrocytes and integrity of the tissue, among the treatment groups (Table III). About three-quarters of the defects in each group showed moderately or slightly increased cloning (Table III). The integrity of the surface layers was normal or only slightly changed in more than three-quarters of the defects in each group, while the deep tissue appeared normal in at least two-thirds of the defects in each group (Table III).
Suturing often affected the adjacent articular cartilage more than 600 micrometers from the defect (Fig. 6-B). In certain cross sections, suturing of the periosteal flap was found to create clefts, completely separating the superficial layers from the deep layers of articular cartilage. Regions surrounding suture marks often displayed increased cloning and decreased safranin-O staining, and they occasionally showed complete disruption of normal architecture.
Many specimens exhibited degenerative changes in the adjacent cartilage that may have been attributable to the operation or to the procedure to obtain chondrocytes for culture but not necessarily to the suturing. In the group that had implantation of cultured autologous chondrocytes and the group that had treatment with periosteum alone, these changes included cloning and loss of safranin-O staining beyond the suture marks (several millimeters from the edge of the defect). In addition, almost all of the empty defects, which had been subjected to the procedure to obtain chondrocytes but not to suturing, were associated with at least slight cloning in the cartilage far from the defect. Three of those eight knees, including the one excluded from grading, showed severe cloning, loss of safranin-O staining, and disruption of the normal collagen architecture across at least several millimeters of the articular surface. Furthermore, in two of those knees, pannus formed over the cartilage surface and in some locations invaded the superficial zone. In the knee that was excluded, the pannus completely covered the defects.
In contrast to the findings in the rabbit model1,3, with the numbers available, we detected no significant effects of treatment with cultured autologous chondrocytes on the types and the amount of repair tissue. This may be explained by the differences in the species, the age of the animals, the lack of retention of the cells in the defect due to possible displacement of the periosteal flap, and the location of the defect (the patella compared with the trochlea).
The finding of hyaline cartilage in the untreated control group in the canine model differed from the findings in the rabbit model1. In the rabbit model, in which a similar technique of areal analysis of sections made through the center of the defect was used, untreated defects were examined at twelve weeks only. At that time, these defects had significantly less reparative tissue and it was of significantly lower quality compared with the tissue in the defects treated with cultured autologous chondrocytes. At one year, sites that had been treated with cultured autologous chondrocytes were filled with significantly more reparative tissue than those that had been covered with the periosteal flap only (87 compared with 31 per cent). In the present canine model, treatment with cultured autologous chondrocytes was associated with the least filling after twelve months and after eighteen months, although with the numbers available we could detect no differences among the groups. In our study, a mean of 36 to 76 per cent of the area of the defect was filled with new repair tissue, a range that was similar to that seen in the study in the rabbit model1.
Our finding that cultured autologous chondrocytes had no effect on the healing of defects in the distal part of the canine femur is not entirely inconsistent with the results reported in a clinical study by Brittberg et al.2. They found that treatment with cultured autologous chondrocytes was less successful in articular defects in the patellofemoral region than in those in the tibiofemoral articulation. Among the important differences between the canine and the human conditions is the thickness of the articular cartilage (approximately 0.7 millimeter for canine cartilage compared with three millimeters for human cartilage). The thickness of the cartilage affects the number of cells that can be injected and retained in a lesion. Although the findings of a short-term study of a canine model provided evidence that some injected chondrocytes had been retained in the defect six weeks after implantation4, the number of cells retained in the canine lesion relative to that retained in a defect in human cartilage has not been studied. Another important difference may be the weight-bearing patterns allowed postoperatively. Although weight-bearing by patients can be carefully controlled, it was necessary to rely on external fixation to limit motion and loading of the joint in the canine model.
Because of the variability in the morphology of the fresh, untreated defects, the depth of the lesion may have been an important factor in determining which defects healed. We found that, although no bleeding was detected intraoperatively, variable damage occurred to the calcified cartilage. As several of the fresh defects showed evidence of slight penetrations through the calcified cartilage, some of the defects were technically full-thickness lesions, directly communicating with the underlying bone. The variability in depth also may have led to healing by different mechanisms—that is, the defects with penetration into the subchondral bone may have filled in partly with tissue from subchondral vasculature and marrow spaces, and these contributions would have been absent initially in the partial-thickness lesions.
The variability in the extent of the lesion seemed to increase during the course of healing. In certain experimental defects, large portions of the calcified cartilage and subchondral bone were resorbed and eroded, forming deep osteochondral lesions. Furthermore, on the basis of the simultaneous observations of bone formation in the area of certain defects and of an intact but elevated tidemark, we suspected that damage to the calcified cartilage occurred early in the healing period and that the bone subsequently healed imperfectly—that is, with a change in the location of the tidemark—by the end point of the study. It was not possible to relate the pattern of healing in a specific defect with the pattern of injury to the zone of calcified cartilage and bone at the initial operation or to explain the reason for the changes in the underlying bone.
Fibrous tissue, transitional tissue, and hyaline cartilage (including the more specialized articular cartilage) represented a full spectrum of reparative tissues. In many instances, one tissue was found blending into another, with a gradual change or mixture of cell and matrix characteristics. In other instances, the boundary between the types of tissue was found to be more abrupt. It is likely that there was some pattern of conversion of one type of tissue to another during the course of regeneration and possibly during degeneration as well. Additional observations are needed to characterize this pattern. A study to assess the mechanical and chemical regulators that may be important in this process also would be of interest.
The amount of the subset of hyaline cartilage that met the histological criteria for articular cartilage was comparable with the amount of residual articular cartilage found in the fresh defects. Therefore, it was not possible to determine if any of the articular cartilage in the specimens examined at twelve and eighteen months had been newly synthesized. However, the amount of hyaline cartilage after one year was substantially higher than the residual amount in the fresh defects, thus indicating the formation of new tissue.
Our observations indicate that suturing a periosteal flap to the defect was detrimental to the adjacent cartilage. Problems found in the adjacent cartilage included disruption of the tissue, loss of proteoglycans, and increased cloning of chondrocytes. Although the damage was pronounced in the dogs, the effects may be less pronounced in humans because of the increased thickness of the cartilage.
Although some regeneration of hyaline cartilage was found at eighteen months, there was substantially less than that found at twelve months. Thus, the situation was not considered likely to improve. In addition, the presence of degenerative changes at both times suggested that all treatments had undesirable effects on the joint. Due to the similar profiles of reparative tissue types after the longest time-interval in this model, changes after an even longer time are expected to be similar in all treatment groups.
NOTE: The authors appreciate the consultation on tissue-typing criteria with Dr. Andrew E. Rosenberg.