Injury to the growth plate is associated with growth disturbances, most notably premature cessation of growth. The goal of this study was to identify spatial changes in the structure and composition of the growth plate in response to injury to provide a foundation for developing therapies that minimize the consequences for skeletal development. We used contrast-enhanced microcomputed tomography (CECT) and histological analyses of a murine model of growth plate injury to quantify changes in the cartilaginous and osseous tissue of the growth plate. To distinguish between local and global changes, the growth plate was divided into regions of interest near to and far from the injury site. We noted increased thickness and CECT attenuation (a measure correlated with glycosaminoglycan (GAG) content) near the injury, and increased tissue mineral density (TMD) of bone bridges within the injury site, compared to outside the injury site and contralateral growth plates. Furthermore, we noted disruption of the normal zonal organization of the physis. The height of the hypertrophic zone was increased at the injury site, and the relative height of the proliferative zone was decreased across the entire injured growth plate. These results indicate that growth plate injury leads to localized disruption of cellular activity and of endochondral ossification. These local changes in tissue structure and composition may contribute to the observed retardation in femur growth. In particular, the changes in proliferative and hypertrophic zone heights seen following injury may impact growth and could be targeted when developing therapies for growth plate injury.
The growth plate, or physis, is the cartilaginous tissue responsible for longitudinal growth of bones. Damage to the growth plate occurs in up to 30% of sports injuries in children and increases the risk of growth disturbance [1,2]. This growth disturbance may take the form of premature fusion of the growth plate and growth arrest, leading to potential discrepancies in limb length and angular deformities [3,4]. Radiographic evidence indicates that growth disturbances are associated with formation of bone bridges at the injury site . Moreover, angular deformities occur when growth is inhibited in some, but not all, regions of the growth plate . While growth disturbance can be minimized with appropriate surgical intervention, these interventions are only successful in patients when the bone bridges affect less than 50% of the growth plate and when the location of the bone bridges is known precisely . Animal models have similarly indicated that whether growth disturbance occurs depends on the size of the injury . Thus, to develop effective treatments, it is imperative to understand how injury to the growth plate results in deleterious changes to the structure of the growth plate, including formation of bone bridges.
Thus far, researchers have used animal models in studies primarily focused on changes directly at the injury site and on the cellular and molecular processes that regulate formation of bone bridges. The injury response consists of four phases: inflammatory, fibrogenic, osteogenic, and remodeling . Throughout the formation of bone bridges, bone-related genes (runx2 and osteocalcin) and cartilage-related genes (collagen X, collagen II, and sox9) are upregulated at the injury site, indicating both endochondral ossification (formation of bone indirectly, via a cartilaginous template) and intramembranous ossification (direct bone formation) [7–10]. The amount of bone at the injury site increases over time, and during the later phases of the injury response, bone bridges form outside the injury site. However, conflicting reports exist on the presence of bone bridges in control (contralateral or age-matched) growth plates [11,12]. Prior studies also reported conflicting results for growth plate thickness in Sprague Dawley rats, with one study finding a reduction in thickness in injured versus contralateral growth plates , and another finding no difference in thickness between injured and control growth plates . These differences exist in spite of the similarity in animal age (5 versus 6 weeks), time points (56 days versus 60 days), and injury size (2 mm) and a difference only in whether the defect was introduced proximally through a cortical window or distally through the condyles. Histological evidence suggests that growth plate thickness increases immediately adjacent to the injury . This observation indicates that the effect of injury on physeal thickness may vary locally in ways not captured robustly by average measurements. Indeed, adjacent to the injury site, researchers have noted effects in the different zones of the growth plate: reserve, proliferative, and hypertrophic [7,11,12,14,15]. A central, unanswered question is how local changes in the structure, composition, and cellular activity in the growth plate contribute to disturbance of growth and formation of bone bridges within and outside the injury site.
The goal of this study was to quantify spatial changes in structure and composition of the growth plate that occur following injury and to investigate the association between these changes and growth disturbance. We used contrast-enhanced microcomputed tomography (CECT), a nondestructive method that enables quantitative, three-dimensional examination of cartilaginous and osseous tissues , to simultaneously quantify local changes in growth plate thickness, bone formation, and CECT attenuation (a parameter that is correlated with glycosaminoglycan (GAG) content [17,18]). We also investigated local changes in cellular activity in response to injury using histology.
All animal experiments were conducted under a protocol approved by the Institutional Animal Care and Use Committee (IACUC) at the Boston University School of Medicine (Boston, MA). A pinhole defect was created in the distal femoral growth plate of 48 4-week-old, male, C57BL/6 J mice (Jackson Labs, 16.4 ± 2.0 g at time of procedure). The surgical method was adapted from a method for inserting an intramedullary pin prior to induction of a closed femoral fracture . An incision was made medial to the patella to displace the patellar tendon and expose the femoral condyles. A 0.51 mm-diameter pin (4.7 ± 0.1% of growth plate area 7 days following surgery) was inserted between the condyles in a retrograde fashion to create the defect (Fig. 1(a)), and the incision was closed with grade-5 sutures. Mice were randomly assigned to groups that were euthanized via carbon dioxide asphyxiation at 7, 21, or 42 days following injury. Injured and contralateral femora were harvested, measured with calipers from the tip of the greater trochanter to the lateral condyle, and frozen at−80 °C in saline-soaked gauze. The length of a subset of femora (n = 25) was also measured from plain-film radiographs that were subsequently digitized (ImageJ, National Institutes of Health, Bethesda, MD) . The lengths measured via calipers and from the digital images were highly correlated (p < 0.0001, R2 = 0.71). Five mice were excluded from analysis because the growth plate was damaged during harvest, and five mice were excluded because the defect did not penetrate the growth plate, resulting in group sizes of 12, 13, and 13 for the day-7, 21, and 42 groups, respectively.
A 3-mm length of the distal femur, which included the growth plate, was scanned in air using a desktop microcomputed tomography (μCT) system (μCT40, Scanco Medical, Brüttisellen Switzerland; 6 μm/voxel, 70 kV, 114 mA, and 200 ms integration time). Scans were performed before (pre- incubation) and after (postincubation) incubation in 27 mg iodine/mL solution of CA4+ (pH 7.4, 350–450 mOsm), penicillin-streptomycin, and protease inhibitor for 14 h [16,21].
Following μCT scanning, femora were incubated in phosphate-buffered saline for 24 h to clear contrast agent. This 24 h period was selected based on a preliminary study where three femora were scanned before incubation in CA4+, after incubation in CA4+ for 8 h and after incubation in phosphate-buffered saline for 24, 48, 72, and 96 h to measure clearance of contrast agent over time. The results showed consistent staining with safranin O following a 24 h incubation period. Femora were fixed in paraformaldehyde for 1 week, decalcified in EDTA, dehydrated in ethanol, embedded in paraffin, and sectioned into 7 μm-thick sagittal slices. Sections were stained with hematoxylin, Fast Green, and safranin-O to visualize cell morphology and tissue types.
CECT Image Processing and Analysis of Cartilage.
The growth plate cartilage was segmented using a previously published protocol [21,22]. Stacks of images from the pre- and postincubation μCT scans were aligned using an affine registration constrained to rigid body transformation (Amira 5.2.2, Visage Imaging, Andover, MA). The pre-incubation image was subtracted from the postincubation image, and cartilage was identified by applying a threshold on the subtracted image. These two steps segment voxels containing cartilage from those containing mineralized tissue and noncartilaginous soft tissue, respectively. The cartilage threshold was based on a four-part Gaussian fit of the intensity histogram of the subtracted image . The threshold was determined such that more than 50% of all voxels with intensities above the threshold are expected to be cartilage. Contiguous voxels above this threshold located within the growth plate region were manually selected, eliminating any noncontiguous voxels, which generally represented noise, registration error, or articular cartilage. Thickness of the growth plate was calculated using distance-transformation methods (Scanco Medical). CECT attenuation was measured as an estimate of GAG content [17,18]. The average thickness and CECT attenuation of the growth plate were measured within a 1 mm diameter volume of interest (VOI) centered at the injury site (“near”; Fig. 2(d) inset) and also within the rest of the growth plate (far). For comparison, near and far VOIs were defined at matched sites in the contralateral growth plate. Matched sites in the contralateral growth plate were identified by mirroring the contralateral growth plate and registering it to the injured growth plate.
μCT Image Processing and Analysis of Bone.
To assess bone bridges, the tissue mineral density (TMD) and bone volume fraction (BV/TV) within the growth plate were measured from the pre-incubation scan within a manually defined VOI on sagittal slices. In the injured growth plate, this VOI was further split into a 0.75 mm-diameter VOI at the injury site (near) and the rest of the growth plate (far). The threshold for segmenting mineralized tissue in all VOIs (531 mgHA/cm3) was determined using an iterative selection method .
Comparison of CECT to Histology.
Two-dimensional histological sections were manually registered with the 3D, subtracted μCT images. The top and bottom boundaries of the growth plate were defined manually on five histological sections per femur at approximately 50 μm intervals (25–40% of the medial–lateral width of the growth plate) and at corresponding locations on corresponding μCT slices. Growth plate thickness was defined as the distance between the top and bottom boundaries of the growth plate in both modalities. The histology- and CECT-measured thicknesses were reported as an average of all measurements from each limb. The normalized red content was used to represent intensity of safranin-O staining . The median red content was computed for a 24 × 24 μm2 (360 × 360 pixel2) area at the midpoint between the top and bottom boundaries of the growth plate. The median CECT attenuation was computed for a 24 × 24 μm2 (4 × 4 voxel2) area centered between the top and bottom boundaries of the growth plate.
Measurements of Zone Heights.
Three histological sections near the injury site and three histological sections at an equivalent location in the contralateral were selected. The boundary between the resting and proliferative zone was defined as the point at which the first flattened chondrocyte of the proliferating column appeared, and the boundary between the proliferative and hypertrophic zone was defined as the point at which the first round chondrocyte appeared . Zone heights were measured at 20 locations per section, and the average values for each limb were the averages of these 60 measurements. On each slice, the four measurements adjacent to the injury site (within 0.25 mm of the injury) were designated as “near” while all other measurements were designated as “far.” At each location, the proportional height of each zone was determined as the fraction of the total growth plate height at that location. The average value of proportional zone height for each limb was the average of the 60 measurements of proportional zone height. The reliability of zone height measurements was assessed by having an independent observer perform measurements on a subset of twelve samples (two animals at each time point, injured and contralateral). The intraclass correlation coefficient (ICC) with bias and interaction was calculated for each zone height: resting (0.88), proliferative (0.78), hypertrophic (0.93), total (0.98), proportional resting (0.39), proportional proliferative (0.46), and proliferative hypertrophic (0.74).
Immunohistochemistry was performed on subset of samples processed for histology (3 postoperative day 7, 4 postoperative day 21, 3 postoperative day 42) without antigen retrieval. Slides were treated with peroxidase 1 for 10 min, washed with tris-buffered saline, and blocked with Rodent Block M for 30 min, followed by a second wash with tris-buffered saline (Biocare Medical, Concord, CA). Slides were incubated with rabbit polyclonal anti-collagen X (1:1000, Abcam, Cambridge, MA, ab58632) for 2 h at room temperature, followed by rabbit-on-rodent HRP-polymer (Biocare Medical, Concord, CA) for 30 min at room temperature. Slides were washed with tris-buffered saline, incubated with chromagen dab (3,3’-diaminobenzidine tetrahydrochloride) for 5 min, washed with de-ionized water, stained with hematoxylin for 3 min, and washed with tris-buffered saline (Biocare Medical, Concord, CA).
Differences in femur length between injured and contralateral and across time points were determined using a repeated-measures analysis of variance (ANOVA) with side (injured versus contralateral) as the within-subjects factor and time point as the between-subjects factor. Differences in growth plate thickness and intensity were determined using a repeated-measures ANOVA with side (injured versus contralateral) and location (near versus far) as the within-subjects factors and time point as the between-subjects factor. Differences in the volume fraction and mineral density of bone bridges, and in the height of each of the three zones, were determined using a repeated-measures ANOVA with side/location (injured near versus injured far versus contralateral) as the within-subjects factor and time point as the between-subjects factor. When a significant main effect or interaction was found, appropriate posthoc tests (Tukey honestly significant difference test or pair wise comparison) were performed with appropriate Bonferroni corrections. Correlations between histological and CECT measures of the growth plate were determined using an analysis of covariance with time point and side (injured versus contralateral) as covariates. Correlations between the percentage difference in femur length between injured and contralateral sides and measures of growth plate thickness, attenuation, and bone bridge formation were also examined using Pearson correlation analysis.
Longitudinal Growth was Disturbed by Injury.
Both injured and contralateral femora exhibited longitudinal growth from day 7 through day 42 (p < 0.0001). However, injured femora were 0.9%, 1.6%, and 2.0% shorter than contralateral femora at days 7, 21, and 42, respectively (p = 0.009; Fig. 1(b)). No abnormalities in gait were observed following the surgery.
Thickness was Locally Increased at the Injury Site.
The structure of the growth plate defined by CECT qualitatively matched that observed with histology (Fig. 2(a)). The histologically measured thickness of the growth plate was highly correlated to the CECT-measured thickness on corresponding sections (R2 = 0.94, p < 0.0001; Fig. 2(b)).
Thickness varied spatially within the growth plate, and over time and was affected by injury (Fig. 2(c)). Although thickness decreased over time in both regions and regardless of injury (p < 0.0001), this decrease was least pronounced at the injury site: thickness in the near VOI of the injured growth plate was higher than that in the far VOI (p < 0.0001) and in the near VOI of the contralateral growth plate (p < 0.0001; Fig. 2(d)). None of the measures of growth plate thickness were correlated with the differences in femur length between injured and contralateral sides (p > 0.528).
Injury Abrogated the Age-Related Decrease in CECT Attenuation.
CECT attenuation was elevated with injury in all regions of the growth plate at day 21 and day 42 (p < 0.05; Fig. 3(b)). Similar to growth plate thickness, injury mitigated the temporal decrease in CECT attenuation. Whereas attenuation progressively decreased from day 7 to 21 and 42 in the contralateral VOIs (p < 0.0001), no temporal changes were seen in the injured near VOI (p = 0.12). In the injured far VOI, attenuation did not differ between days 7 and 21 (p = 0.65), but was lower at day 42 versus days 7 and 21 (p < 0.0001). None of the measures of attenuation were correlated with the differences in femur length between injured and contralateral sides (p > 0.439).
Bone Bridges at Injury Site Have a Composition Distinct From Those Outside the Injury Site.
Bone bridges were evident by histology and μCT in all but two samples by day 7, and in all samples at later time points (Figs. 4(a) and 4(b)). While BV/TV and TMD increased over time within both injured and contralateral growth plates (p < 0.0001), the TMD of the bone within the injury site was higher than in regions far from the injury site and in the entire contralateral growth plate by day 21 (Fig. 4(c)). BV/TV was higher near the injury site compared to far from the injury site and to the entire contralateral growth plate at day 21 (p < 0.01). At all three time points, BV/TV was lower in the injured growth plate far from the injury site compared to the contralateral growth plate (p < 0.01). None of the BV/TV or TMD measures were correlated with the differences in femur length between injured and contralateral sides (p > 0.060).
In day-7 samples with no evident bone bridging, CECT indicated a lack of cartilage at the injury site (Fig. 4(d)). Histological analysis confirmed the presence of fibrous tissue at this site. At later time points, some samples showed safranin-O staining at the injury site, while CECT showed the presence of mineralized tissue (Fig. 4(e)). At days 21 and 42, bone bridges outside the injury site were observed via μCT. However, in matched histological sections, these bone bridges were stained with safranin-O rather than Fast Green, as was generally observed at the injury site (Fig. 4(f)). This staining was similar to that of calcified cartilage within the trabeculae immediately proximal to the growth plate.
Injury Disrupts Zonal Organization Within the Growth Plate.
Disorganized chondrocyte columns were observed adjacent to the injury site at all time points. Large, round, hypertrophic chondrocytes were prevalent at days 7 and 21 (Fig. 3(b)). In injured growth plates, the height of the resting zone was increased relative to the contralateral at all time-points (p < 0.01; Fig. 5). Near the injury site, the hypertrophic zone height was increased relative to the region far from the injury site (p < 0.01) and to the contralateral growth plate (p < 0.05). While the absolute height of the proliferative zone was not affected by injury, the proliferative zone made up a smaller proportion of the total growth plate height relative to the contralateral at all time points (p < 0.0001). This reduced proportional proliferative zone height was due to a combination of increased proportional resting (p < 0.01) and hypertrophic (p < 0.05) zone heights. None of the measures of zone height were correlated with the differences in femur length between injured and contralateral sides (p > 0.203).
Collagen X Staining was Enhanced Near the Injury Site.
The increased height of the hypertrophic zone was probed further using a molecular marker for hypertrophic chondrocytes. In the contralateral growth plate and the injured growth plate far from the injury site, collagen X staining was localized to a band specific to the hypertrophic zone (defined by cellular morphology) (Figs. 6(b) and 6(c)). In contrast, at the injury site, collagen X staining extended throughout the entire height of the growth plate (Fig. 6(a)). Small amounts of collagen X staining were also evident within the injury site.
In this study, the spatially varying response of the growth plate to injury was examined to determine how injury impacts the local versus more remote regions of the growth plate. We identified thickening of the growth plate and elevation of CECT attenuation near the injury site relative to far from the injury site and to the contralateral growth plate. Additionally, both the volume and the mineral density of bone formed at the injury site were higher than those of the bone bridges formed outside the injury site and in contralateral growth plates. These local changes in the cartilage and bone were accompanied by increased hypertrophic zone height immediately adjacent to the injury site and decreased proliferative zone height throughout the entire injured growth plate. These alterations in growth plate structure, composition, and cellular activity indicate a local dysregulation of endochondral ossification, which may explain the overall global disturbance in growth.
Since CECT enables simultaneous, three-dimensional visualization of both cartilage and bone, this method is a powerful research tool to examine structural and compositional consequences of growth plate injury. Magnetic resonance imaging (MRI) has been used clinically to visualize the growth plate cartilage in three dimensions and to identify disruptions caused by bone bridges [26–28]. However, while MRI can visualize bony structures, this technique does not capture bone mineral density. Alternatively, standard computed tomography (CT) allows three-dimensional visualization of the structure and mineralization of bone bridges in the clinic but is not well suited for imaging cartilage . In contrast, using CECT, it was possible to identify a lack of cartilage at the injury site plate at day 7 in a murine model, even in cases where bone bridges had not yet formed. Histological examination confirmed the lack of cartilage and showed infiltration of the injury site with fibrous tissue. Identification of a defect in the early stage of the injury response before bone bridges form suggests that CECT may enable earlier evaluation not possible with conventional μCT. Furthermore, CECT yields thickness measurements that were highly correlated with the thickness measured using histology, indicating that this research method is a reliable means of measuring growth plate thickness, which could be useful in the analysis of other types of growth plate pathologies.
While local thickening of the growth plate has been observed qualitatively via histology , this study is the first to quantitatively demonstrate a local, persistent change in growth plate thickness near the injury site. Other studies where growth plate thickness was measured using μCT in the absence of a cartilage-specific contrast agent reported only the average thickness of the growth plate. The observation of a locally increased thickness was surprising because thinning of the growth plate is often associated with slowing of growth and because injury has been associated with an overall decrease in growth plate thickness relative to contralateral controls [11,30] or no change in growth plate thickness relative to age-matched controls . The present findings thus emphasize that averaging the thickness over the entire growth plate limits the ability to detect local changes which may be associated with more subtle growth disturbances. The increased local thickening of the growth plate, paired with overall decreased rate of growth of the limb, observed in this study indicates that the mechanisms of growth disturbance in the murine pinhole defect model involve more than simply an overall acceleration of closure of the growth plate.
Similarly, regions of the growth plate near the injury site maintained a high CECT attenuation, indicative of high GAG content, following injury, even while, in general, CECT attenuation of the growth plate decreased progressively with age. This persistently high CECT attenuation at the injury site over time suggests abnormal proteoglycan synthesis. We note that the increased attenuation is unlikely due to mineralization within the growth plate, since the image segmentation procedures are specifically designed to remove from consideration any voxels containing mineralized tissue. Although, to our knowledge, GAG content of the growth plate has not been investigated within the context of aging or injury, researchers have demonstrated that GAG content varies zonally, with elevated levels in the hypertrophic zone . This pattern is consistent with our observations of increased attenuation and hypertrophy near the injury site. Furthermore, GAGs in the growth plate undergo compositional and chemical modifications across the zones during endochondral ossification [32,33]. Further work is required to determine how these modifications as well as the types of GAGs present might be affected by injury and if the elevated CECT attenuation is merely a consequence of the increased hypertrophy or if it has an independent contribution to the slowing of longitudinal growth.
Injury also led to local changes in bone at the injury site. Consistent with previous studies, the amount of bone at the injury site increased from the early osteogenic phase at day 7 to the remodeling phase at day 42 [7,11,12]. Bone bridges also formed outside the injury site and in the contralateral growth plate, albeit with a lower bone volume fraction outside the injury site compared to the contralateral. In contrast, Coleman et al. noted a higher bone volume fraction outside the injury site compared to the contralateral , and Macsai et al. observed bone bridges only in the injured growth plate . These studies differ from the current study in that they used a rat model with a 2-mm drill hole injury. The lack of bone bridges observed in the contralateral by Macsai et al. may be due to different processes of bone bridge formation in the tibia compared to femur. The reduced bone volume fraction outside the injury site compared to controls in this study may be due to altered signaling inhibiting age-related bone bridge formation. The mechanisms of age-related bone formation are poorly understood, and they have not yet been studied in an injury model. However, in rachitic vitamin D receptor-deficient mice, decreased formation of bone bridges and increased height of the hypertrophic zone were noted . It would, therefore, be interesting to investigate whether hypertrophic differentiation is related to formation of age-related bone bridges. Bone bridges at the injury site had increased TMD compared to both bone bridges outside the injury site and bone bridges in the contralateral physis, indicating that bone bridges at the injury site have a different composition from age-related bone bridges and may form via a different process. This idea is supported by histological evidence that age-related bone bridges tend to be stained with safranin-O, while bone bridges formed at the injury site do not. However, these observations may instead reflect the fact that bone bridges at the injury site formed earlier and therefore had more time to accrue mineral.
The increased thickness near the injury site was accompanied by increased hypertrophy at day 7. Disorganized growth plate columns and accumulation of hypertrophic cells near the injury site have been observed in other models of growth plate injury, particularly at early time points [6,11,13–15]. This local increase in hypertrophy is consistent with the collagen X staining—a marker of hypertrophic chondrocytes—observed throughout the entire height of the growth plate near the injury site. In contrast to a local, transient increase in hypertrophy, proliferation was reduced in the entire injured growth plate. Although technical difficulties with carrying out immunohistochemistry following CECT prevented successful immunostaining with a proliferative marker, our morphology-based zone measurements are in agreement with prior observations of a decrease in proliferative cells identified by proliferating cell nuclear antigen (PCNA) following injury . This global reduction in proliferation may account for the overall slower growth of injured femora, in spite of local thickening near the injury site. These local differences in hypertrophy and growth plate thickness may be indicative of irregular endochondral ossification and impaired longitudinal growth.
We did not find any correlations between the amount of shortening (defined as the percentage difference in femur length between injured and contralateral side) and measures of thickness or attenuation of the growth plate, whether in regions near or far from the injury site. This lack of correlations may be due to the relatively small defect used in this study and the correspondingly mild amount of growth disruption. However, it is also possible that such correlations would only be apparent with a longitudinal study design. The growth disruption would be expected to be most directly related to how much bone is formed at the proximal surface of the growth plate, which in turn is related to the activity of the chondrocytes and cartilage matrix in the growth plate. Our assessments are reflective of the latter, and changes in this activity observed at a given time point may not manifest as an impairment in bone formation and femur length until a later timepoint.
Since growth plate injury is associated with irregular chondrocyte proliferation and hypertrophy, both of which may contribute to disturbance of longitudinal growth, one approach to treating growth plate injury would be to target factors that regulate chondrocyte proliferation and hypertrophy. Thus far, research has focused on identifying therapeutic targets to prevent formation of bone bridges without considering how these targets regulate chondrocytes. For example, since vascular endothelial growth factor (VEGF) is present at the injury site and is critical for the process of angiogenesis that accompanies bone formation, it was hypothesized that inhibition of VEGF would prevent bone bridge formation at the injury [35,36]. However, due to VEGF's role in the resorption of calcified cartilage of the growth plate, VEGF inhibition reduced bone bridge formation but led to an expanded hypertrophic zone and reduction in limb length [36,37]. Additional factors identified during growth plate injury, which are also regulators of chondrocyte proliferation and hypertrophy in the growth plate, include the inflammatory factors cycloxyenase-2 and interleukin-1β, growth factors fibroblast growth factor and insulin-like growth factor, and bone morphogenetic proteins [38–46]. Thus, it is important to consider the growth plate as a whole when developing therapies. By enabling local and global assessments of both bone and cartilage at the growth plate, CECT is a useful tool for these efforts.
The findings in this study are also relevant to the biomechanics of the growth plate. The features that were observed to change following injury are those that can affect the mechanical behavior of this cartilaginous structure. Physeal thickness is inversely correlated with shear strength and shear stiffness , and while CECT attenuation is strongly associated with the compressive stiffness of cartilage, it is only very weakly correlated with shear stiffness . Axial strains induced within the growth plate during compressive loading of the epiphysis have been found to be highest in the hypertrophic zone . This evidence suggests that the local increase in thickness and hypertrophy observed with the pinhole injury may increase the susceptibility of the growth plate to further damage, particularly under shear forces.
One limitation of this study is that all histology followed CECT scans. The extended time at room temperature before formalin fixation may have contributed to proteoglycan loss, while incomplete clearance of the contrast agent may have interfered with staining. However, a preliminary study did indicate that 24 h were adequate to clear contrast agent and that safranin O labeled growth plate cartilage following this clearance period. In 30/76 samples, safranin-O staining within the growth plate (but not within calcified cartilage) was faint, limiting the ability to detect correlations between CECT attenuation and red content. While the spatial variation in CECT attenuation agreed qualitatively with that in the safranin-O staining in matched 2D sections, CECT attenuation was not correlated with the safranin O staining intensity (measured as red content) measured quantitatively in digital images of the histology sections when the data from all samples were pooled (R = 0.23, p = 0.10). Although methods for quantification of safranin-O staining have been developed based on the stoichiometric relationship between safranin-O and proteoglycans [24,50–52], sensitivity of safranin-O to pH, temperature, and exposure time, coupled with low sensitivity to GAGs in proteoglycan-depleted cartilage, limits the utility of quantification of safranin-O staining . In fact, many of the samples with unsuccessful staining were from later time points, where aging may contribute to depletion of proteoglycans. However, since CECT attenuation has been shown to be correlated to GAG content in articular cartilage, the limitations of quantitative measurement of safranin-O staining do not obviate interpretation of relative changes in CECT attenuation as reflective of relative changes in GAG content [17,18]. Furthermore, the sample processing may have prevented successful immunohistochemical staining. In addition to staining for collagen X, we were also unsuccessful in staining for markers of proliferation (proliferating cell nuclear antigen (PCNA) and phosphor-histone H3), apoptosis (cleaved-caspase 3), and chondrocyte differentiation (parathyroid-related protein (PTHrP)).
In addition to these limitations with regard to histology, there are limitations with regard to the surgical model. The defect created in our study does not closely recapitulate the most common types of injury seen in the clinic, yet it does bear similarities to Salter Harris Type IV physeal injuries and to the disruption to the growth plate and epiphyseal bone caused during transphyseal fixation methods for surgical reconstruction of the anterior cruciate ligament and for tibial eminence fractures. As with physeal injuries, this surgical approach raises concern for disruption of bone growth . Moreover, hole defects have been used often to study formation of bone bridges in animal models [7,14]. One difference between these animal models and the current study is the method for introducing the defect: in other studies, the defect is introduced by a drill via a cortical window in the epiphysis, while in this study, the defect is introduced by a pin between the femoral condyles. In spite of these differences, the injury response was similar to that seen with drill defects . Finally, the size of the injury does affect the injury response and growth disturbance. This variable was investigated in a study of New Zealand white rabbits, where an injury affecting 13% of the growth plate resulted in no growth disturbance, while an injury affecting 20% of the growth plate resulted in shortening of the femur . In this study, the injury site was less than 5% of the distal femoral growth plate area, whereas in a prior murine study, the defect was approximately four times this size . Our relatively small defect size may explain why injury slowed, but did not arrest, growth.
In summary, these results indicate that growth plate injury leads to locally increased thickness, altered composition, and increased hypertrophy of the cartilage near the injury site and increased mineral density within the bone that forms in the injury site. These changes were not the same as those occurring more remote to the injury site. This spatial variation in the effects of injury may be responsible for nonuniform growth disturbances. Therefore, when developing therapies to reduce bone bridge formation, it is important to consider the effects of these treatments on the tissues of the remainder of the growth plate.
The authors thank Dr. Mark Grinstaff and Dr. Jonathan Freedman for the contrast agent used in this study.
This study was supported by National Science Foundation Graduate Research Fellowship under Grant No. DGE-1247312 and by the Osteosynthesis and Trauma Care Foundation Grant No. EMTE2013.