Experimental Design
The study was approved by the local institutional animal care and use committee. Sixty-five four to six-week-old Yorkshire piglets weighing 5 to 8 kg were obtained from a local breeder. Ischemic osteonecrosis of the right femoral head was surgically induced by placing a ligature tightly around the femoral neck8, and the contralateral, normal hip of each animal was used as a control. Radiographic and histological analyses were carried out at two and four weeks (twelve animals at each time period) and at eight weeks (twenty-six animals) after the induction of ischemia (Fig. 1). In the eight-week group, seven animals received an intramuscular injection of calcein (20 mg/kg), two days before they were killed, to label new bone formation and to determine longitudinal growth.
Twelve additional animals were studied at twenty-four hours following the induction of ischemia to assess growth plate hypoxia (six animals) and chondrocytic viability (six animals). Three additional animals were studied at forty-eight hours following the induction of ischemia to determine the extent of vascular disruption in the epiphysis with use of micro-computed tomography imaging.
Micro-Computed Tomography Imaging
The micro-computed tomography method described by Duvall et al. for small animals9 was modified for this large-animal study. Normal and ischemic femoral heads from three animals were imaged forty-eight hours following the ischemia-induction surgery. Heparin was administered intravenously thirty minutes before the animals were killed. Inflow and outflow tubes were inserted into the distal aorta and the inferior vena cava, respectively. Four liters of saline solution were infused to clear the blood. MICROFIL (MV-117; Flow Tech, Carver, Massachusetts) was then infused until constant outflow of the compound was observed exiting from the vena cava tubing. After the infusion, the animals were placed into a -1°C freezer for ninety minutes for polymerization to occur.
Following polymerization, the femoral heads were retrieved and were fixed in 10% neutral buffered formalin. After fixation, femoral head cartilage was scored for orientation purposes and was scanned with use of a SCANCO Micro-CT 35 scanner (SCANCO Medical, Bassersdorf, Switzerland). A voltage setting of 55 kVp and a current setting of 145 µA were used. The resolution was set to create a 1024 × 1024-pixel image matrix and provide a resolution of 30 µm voxel. A second scanning was performed with use of the same settings following decalcification. All of the sectional images and three-dimensional reconstructed images were assessed for the presence of vessels in the osseous epiphysis, the growth plate, and the epiphyseal cartilage.
Radiographic Assessment
A Faxitron x-ray machine (Faxitron X-Ray, Wheeling, Illinois) was used to capture anteroposterior radiographs of the proximal part of the femur. The length of the femoral neck was determined with use of a previously described method10. Briefly, two parallel lines—one through the growth plate and the other through the base of the greater trochanter—were drawn, and the perpendicular distance between these two lines was measured to represent the femoral neck length.
Histological Preparation
The femoral heads were bisected in the midcoronal plane and were fixed in 10% neutral buffered formalin. The heads were decalcified in EDTA solution and processed for paraffin embedding and sectioning. Five-micrometer-thick sections were stained with hematoxylin and eosin or with safranin O and fast green to visualize the growth plate.
Quantitation of Elongation of the Femoral Neck with Calcein Labeling
The posterior halves of the femoral heads from the seven animals that had received calcein two days before being killed were left mineralized and prepared for plastic embedding. Ten-micrometer-thick sections were cut and were examined under a fluorescent microscope with the excitation wavelength of 495 nm and the emission wavelength of 517 nm. Digital images of the entire chondro-osseous junction of the growth plate were obtained at 100× magnification and were analyzed with use of Image-Pro software (Media Cybernetics, Bethesda, Maryland). The distance from the junction of the noncalcified and calcified cartilages to the band showing increased fluorescence in the primary spongiosa was considered to be the distance of elongation over the two-day period.
Quantitative Histological Analysis of Growth Plate Height and Zones
A representative histological section from the midcoronal plane of the femoral head was used. Digital images of growth plates were obtained at 40× magnification with use of a Nikon Eclipse E800M microscope (Nikon, Tokyo, Japan). Eight, nine, or ten images were made to capture the entire growth plate. The height of the growth plate as well as the heights of three separate zones (epiphyseal, proliferative, and hypertrophic) were measured with use of Image-Pro software. The zones were chosen on the basis of the cellular morphology (Fig. 2). First, the proliferative zone was outlined by tracing the region of the growth plate where the chondrocytes were beginning to flatten out and to stack in longitudinal columns. The junction of the proliferative zone and the hypertrophic zone was outlined by tracing the region where the cells began to enlarge in size and to cluster. After the proliferative zone had been defined, the epiphyseal zone was defined as the zone between the start of the proliferative zone and the epiphyseal bone. The overall height of the growth plate and the heights of the three zones were measured with use of a continuous, two-line distance measurement function of the Image-Pro software.
Assessment of Growth Plate Hypoxia
Hypoxyprobe-1 (Bioscience Research Reagents [formerly Chemicon International], a Millipore company, Temecula, California) was used to detect the areas of hypoxia in the proximal part of the femur11,12. Hypoxyprobe-1 consists of pimonidazole hydrochloride, which forms adducts with thiol groups in proteins, peptides, and amino acids under severe hypoxic conditions (Po2 of <10 mm Hg). Since the recommended protocol for administration of Hypoxyprobe-1 is for small animals, a modification was made for our large-animal model. We used a local injection method to deliver the agent directly into the femoral head and the hip joint13. A high concentration of 200 mg/kg of Hypoxyprobe-1 solution was administered to each head; this concentration was based on the average weight of the femoral head (8 g) as determined with previous measurements. One milliliter of Hypoxyprobe-1 solution was injected into the osseous epiphysis before and after the placement of the ligature around the femoral neck. An additional 1 mL of Hypoxyprobe-1 solution was injected into the hip joint. Intraosseous and intra-articular injections of Hypoxyprobe-1 were performed, with use of fluoroscopic guidance, on the contralateral (control) side immediately after the surgery on the ischemic side. A radiopaque contrast medium was injected before the Hypoxyprobe-1 was administered, to ensure correct placement of the needle in the osseous epiphysis and the hip joint. The femoral heads were retrieved twenty-four hours following surgery and were processed for Hypoxyprobe-1 detection with use of a monoclonal antibody supplied in the kit. Peroxidase reaction with diaminobenzidine (DAB) as the chromogen was used to reveal Hypoxyprobe-1 adducts.
Assessment of Chondrocytic Viability
Immediately after the femoral heads were retrieved, they were immersed in an icy saline solution. They were bisected and cut into 150-µm-thick sections with use of a 0.22-mm-diameter-wire saw while being irrigated with a cold saline solution. The sections were decalcified overnight in a cold EDTA/Tris buffer solution. On the following day, a tetrazolium-formazan reduction reaction for the demonstration of lactate dehydrogenase activity was performed in an incubation solution containing nitroblue tetrazolium at 37°C for four hours, as previously described14,15. The sections were fixed in 10% formol saline solution, dehydrated in a series of ethanols, and mounted in a xylene-based medium. For negative controls, we used an incubation solution without substrates and porcine bone that had been heated to kill the cells in the bone.
Bromodeoxyuridine Staining
Bromodeoxyuridine (BrdU) labeling (Detection Kit II; Boehringer Mannheim, Mannheim, Germany) was used to determine the number of positive and negative-staining cells in the proliferative zone of the ischemic and the contralateral, normal side. Five animals were studied at two weeks and four animals were studied at four weeks after the induction of ischemia. The bromodeoxyuridine was administered intravenously (50 mg/kg) one day before the animals were killed. The femoral heads were stained according to the manufacturer's protocol. Digital images of the proliferative zone were obtained at 200× magnification. For each growth plate, twenty-five to thirty fields were examined by moving the stage of the microscope systematically from left to right to cover the whole growth plate without overlapping the field of view. In each field, the numbers of positive and negative-staining cells were counted.
Statistical Analysis
A paired t test was used to analyze the difference between the ischemic and the contralateral, normal sides at each time point. To determine the difference between the findings at the three time points, analysis of variance was used followed by a post hoc testing called Fisher protected least significant difference, which adjusts for the possibility of finding significance by chance in multiple comparisons. A p value of <0.05 was considered significant. Values are given as the mean and standard deviation.
Source of Funding
The study was funded by Shriners Hospital for Children. This was an internal funding source that provided funding for salary and supplies for the study.
All femoral heads with induced ischemia displayed radiographic and histological changes consistent with ischemic injury to the osseous epiphysis. All infarcted femoral heads had a moderate-to-severe deformity by eight weeks (Fig. 3). One of the twelve femoral heads analyzed at four weeks and six of the twenty-six analyzed at eight weeks had severe disruption of the growth plate that precluded quantitative histological assessment of the growth plate zones. Severe disruption is defined as damage of >50% of the growth plate as seen on a midcoronal section of the femoral head. These femoral heads had fibrous or osseous tissue bridging the growth plate or diffuse resorption of the growth plate. The results pertaining to these femoral heads with severe growth plate damage have been reported previously10. Thus, in the present study, we focused on the remaining femoral heads (those without severe disruption of the growth plate).
Radiographic Assessment
The length of the femoral neck on the ischemic side was compared with that on the contralateral, normal side at two weeks (twelve femoral heads per side), four weeks (eleven per side), and eight weeks (twenty per side) after the induction of ischemia. The length of the femoral neck increased over time on both the normal and the ischemic side (Fig. 3, C). However, the neck length on the ischemic side was significantly less than that on the contralateral, normal side at two (p = 0.007), four (p = 0.0003), and eight weeks (p = 0.04). At eight weeks, the mean neck length was 2.21 ± 0.29 cm on the normal side compared with 2.05 ± 0.34 cm on the ischemic side, which was a 7% decrease in the mean neck length on the ischemic side.
The widths of the femoral necks on the ischemic and normal sides were compared at two weeks (twelve femoral heads per side), four weeks (eleven per side), and eight weeks (twenty per side) after the induction of ischemia. The width of the femoral neck increased over time on both the normal and the ischemic side (Fig. 3, D). However, the increase in the width on the ischemic side was significantly greater than that on the normal side at all time points (p < 0.05); this indicated broadening of the neck, which is consistently found in patients with juvenile osteonecrosis. At eight weeks, the mean neck width was 2.36 ± 0.15 cm on the normal side compared with 2.59 ± 0.24 cm on the ischemic side, which was a 10% increase in the mean neck width on the ischemic side.
Calcein Labeling
Seven animals were injected with calcein, two days before they were killed, to compare the elongation occurring in the proximal femoral growth plate on the ischemic side with that on the normal side. Calcein labeling showed elongation on the ischemic side, but its rate was significantly lower than that on the contralateral, normal side (p = 0.04). The mean elongation on the ischemic side was 134 ± 49 µm compared with 166 ± 61 µm on the normal side. This was a 19% decrease in elongation, over the two-day period, on the ischemic side.
Histological Assessment
Histological assessment revealed that the growth plate height decreased significantly from two to eight weeks on the normal side (p = 0.01), demonstrating the normal developmental thinning of the growth plate as animals mature. In contrast, the height on the ischemic side did not change from two to eight weeks, demonstrating the absence of the developmental change that normally occurs with aging (Fig. 4).
Comparison of the ischemic and contralateral, normal sides at two weeks (eight femoral heads per side), four weeks (seven per side), and eight weeks (nine per side) after the induction of ischemia showed the growth plate height on the ischemic side to be significantly greater at four (p = 0.008) and eight weeks (p < 0.0002). At four weeks, the difference in the growth plate height between the ischemic and the contralateral, normal side was attributable to the difference in the height of the epiphyseal zone, which was 351 ± 95 µm on the ischemic side compared with 262 ± 38 µm on the normal side (p = 0.04). At eight weeks, the most noticeable difference (96 µm) was seen in the proliferative zone (p < 0.00001). The mean height of the proliferative zone on the normal side decreased significantly from 304 ± 30 µm at two weeks to 243 ± 50 µm at eight weeks (p = 0.01). However, the height of the proliferative zone on the ischemic side showed no decrease over time, with the heights being 315 ± 42 µm at two weeks, 316 ± 69 µm at four weeks, and 339 ± 45 µm at eight weeks.
Histological assessment of morphological changes in the growth plates showed endothelial and perivascular cellular death in the cartilage canals in the epiphyseal zone. The rest of the growth plate appeared normal. Specifically, the growth plates did not have any evidence of cartilage degradation, disorganization of chondrocyte arrangement, or loss of matrix staining.
Micro-Computed Tomography Imaging
To confirm that our animal model produced total disruption of the epiphyseal vasculature, micro-computed tomography assessments were performed on three femoral heads forty-eight hours following the surgery. A complete absence of MICROFIL was observed in all three osseous epiphyses on the ischemic side, indicating total disruption of the epiphyseal vasculature.
Growth Plate Hypoxia
To investigate the effect of disruption of blood flow to the femoral head on the growth plate, a marker for hypoxia was used to determine the extent of hypoxia in six animals. At twenty-four hours following the induction of ischemia, diffuse staining of marrow cells was found consistently in the osseous epiphysis. Positive staining was observed in the epiphyseal zone of the growth plate in five of the six ischemic femoral heads studied (Fig. 5). No staining was observed in the proliferative or hypertrophic zones or in the metaphysis of these heads. In the remaining animal, the staining extended to the proliferative zone and light staining was observed in the region of the primary and secondary spongiosa in the metaphysis. No positive staining for ischemia was observed on the contralateral, normal side of any animal or on the negative control slides where the primary antibody was omitted or non-immune IgG antibody was used.
Chondrocytic Viability and Bromodeoxyuridine Staining
Cellular viability was assessed with use of lactate dehydrogenase staining at twenty-four hours following the induction of ischemia in six animals. On the normal side, lactate dehydrogenase staining was positive, indicating viable cells, throughout the height of the growth plate. Osteocytes and marrow cells in the osseous epiphysis also stained positively (Fig. 6). On the ischemic side, a complete absence of staining, indicating cellular death, was observed in the osseous epiphysis. Osteocytes in the trabecular bone and cells in the marrow space did not stain for lactate dehydrogenase activity. In the growth plate, chondrocytes in the epiphyseal zone bordering on the osseous epiphysis and endothelial cells in the cartilage canals located in the epiphyseal zone consistently showed absence of lactate dehydrogenase staining. The area of absent staining was similar to the area of severe hypoxia noted above. Chondrocytes in the proliferative and hypertrophic zones showed normal staining.
Bromodeoxyuridine staining was positive in the proliferative zone of the growth plates on both the normal and the ischemic side. Quantitation of bromodeoxyuridine-positive and negative cells in the proliferative zone showed 9% ± 2% positive cells on the normal side compared with 11% ± 5% positive cells on the ischemic side at two weeks (p = 0.48) and 10% ± 6% positive cells on the normal side compared with 7% ± 3% positive cells on the ischemic side at four weeks (p = 0.50).
Growth disturbance of the proximal femoral growth plate following ischemic osteonecrosis can have serious consequences with regard to the function of the affected hip by producing proximal femoral deformities and later degenerative osteoarthritis. Using a multiple-assessment approach, we performed a detailed study of the effects of disruption of the epiphyseal vasculature on the proximal femoral growth plate in a large-animal model. We found diffuse disruption of the growth plate in six of twenty-six ischemic femoral heads at eight weeks. The majority of the growth plates on the ischemic side remained intact with evidence of elongation occurring over time, although the elongation occurred at a slower rate than it did on the contralateral, normal side. In addition to the shortening of the femoral neck, significant broadening of the femoral neck was observed. Both shortening and broadening of the femoral neck are the deformities consistently found in patients with Legg-Calvé-Perthes disease, indicating that our large-animal model is clinically relevant. The relatively low rate of growth plate disruption found in this study is also in keeping with the <30% rate of premature growth arrest reported in patients with Legg-Calvé-Perthes disease5,6.
Our study shows that, despite the total disruption of the epiphyseal vasculature, the region of severe hypoxia (Po2 of <10 mm Hg) and loss of cellular viability was limited to the layer of growth plate cartilage bordering on the epiphysis. The proliferative and hypertrophic cell zones of the growth plate appeared to remain viable, as demonstrated by positive lactate dehydrogenase staining of the chondrocytes. Chondrocytes in the proliferative zone showed continued proliferation, as demonstrated by positive bromodeoxyuridine staining. The fact that growth of the femoral neck took place on the ischemic side, albeit at a slower rate than it did on the normal side, further validates the idea that the growth plate is viable and functioning following total disruption of the epiphyseal vasculature.
The results of our study are in contrast with those of studies of small-animal models of disruption of the epiphyseal vasculature, including the classic study by Trueta and Amato2. In that study, diffuse damage to the growth plate with complete growth arrest was observed following disruption of the epiphyseal vasculature in young rabbits. These studies have led to the hypothesis that the nutritional flow through the growth plate is unidirectional, with the epiphyseal vasculature being the source of nutrients. Our findings do not support this simplistic view of growth plate nutrition. Recent studies in which multiphoton microscopy was used to analyze the real-time dynamics of solute transport in the growth plate cartilage indicated that the pattern of solute transport is not unidirectional and the epiphyseal vasculature is not the sole source of nutrients for the growth plate cartilage16,17. In murine proximal tibial growth plates, the contributions of the epiphyseal, metaphyseal, and circumferential perichondrial vasculatures varied according to the solute size. The smaller tracer molecules (up to 10 kDa) entered through all three vasculatures, with equal permissiveness from the epiphyseal and metaphyseal sides. The perichondrial plexus appeared to be a more important means of entrance for the larger molecules. No such data on nutritional flow to the growth plates of larger animals are available, to our knowledge. However, if the data from the murine study can be extrapolated to our large-animal model, the fact that nutrients can access the growth plate through the remaining metaphyseal and, possibly, perichondrial vessels would explain why the majority of the growth plates remained viable after total disruption of the epiphyseal vasculature. It also would explain why only a small layer of the growth plate bordering on the epiphysis became severely hypoxic and why the chondrocytes in the proliferative zone continued to undergo cellular division.
It is unclear why one of the twelve growth plates analyzed at four weeks and six of the twenty-six growth plates analyzed at eight weeks had diffuse growth plate disruption. One possible explanation is variation in the vascular anatomy of the proximal growth plate. It is possible that some growth plates contain a greater number of cartilage canals. This may be due to individual variation or to a delay in growth plate maturation since, in large animals, the number of cartilage canals present within a growth plate decreases with maturation. Disruption of blood flow in the cartilage canals may result in greater ischemic damage to those growth plates that have a greater number of these canals. It is also possible that the placement of the suture ligature around the femoral neck in some animals affected not only the epiphyseal vasculature but also the perichondrial or metaphyseal vasculature. One other factor that may have affected the growth plates is mechanical loading with compression-type damage of the growth plate. In the piglet model, it is known that the indentation stiffness of the infarcted femoral head and the modulus of elasticity and yield strength of its components (cartilage and bone) decrease significantly from two to eight weeks following induction of ischemia18,19. Alteration of mechanical load transmitted to the growth plates due to these changes may have negatively affected some of these animals. Additional studies to determine the mechanisms underlying the extent of growth plate damage would provide useful insight into ways to prevent this serious complication associated with ischemic osteonecrosis.
This study had some limitations. In order to minimize the number of animals that we used, we did not evaluate a truly normal control group of animals that had not been operated on. Instead, we used the femoral head from the contralateral limb that had not been operated on as a control. Another limitation of the study is that our histological assessments were limited to the midcoronal segments of the femoral heads. Because of the large size of the animals that we used and as a result of limited resources, we were not able to obtain and examine the histological sections of the whole femoral head.
In summary, our study provides a comprehensive assessment of the morphological and functional changes in the proximal femoral growth plate following total disruption of the epiphyseal vasculature. The findings provide new insight into the effects of disruption of the epiphyseal vasculature on the proximal femoral growth plate with respect to growth plate hypoxia, morphology, and function. The study showed that disruption of the epiphyseal vasculature does not necessarily produce massive growth plate damage. This study of piglets brings into question the validity of the hypothesis that there is a unidirectional flow of nutrients from the epiphyseal to the metaphyseal direction in the growth plate in humans. 