Facet joint degeneration in adolescent idiopathic scoliosis

Adolescent idiopathic scoliosis (AIS) is a poorly understood deformity of the thoracolumbar spine which affects the intervertebral discs (IVDs) and the articular facet joints. The knowledge concerning facet joints in this context is very limited, although facet joint degeneration is a known contributor of back pain. In this study, a comprehensive investigation was performed to characterize the facet joint chondrocytes and extracellular matrix within the scoliotic spine. Surgically removed articular facet joint tissues were collected from patients undergoing spinal corrective surgery for AIS deformities, while non‐scoliotic articular facet joint tissues were obtained from cadaveric organ donors. Alterations in cartilage tissue structure were evaluated histologically with safranin‐O fast green and a modified OARSI grading scale. Pro‐inflammatory cytokines, matrix‐degrading proteases, and fragmented matrix molecules associated with cartilage degradation were analyzed by immunohistochemistry and western blotting. Safranin‐O fast green staining revealed that young scoliotic facet joints show clear signs of degeneration with substantial proteoglycan loss, similar to osteoarthritis (OA). The proteoglycan levels were significantly lower than in healthy asymptomatic non‐scoliotic control individuals. In comparison to controls, scoliotic articular facets showed increased cell density, increased expression of the proliferation marker Ki‐67, and higher expression of MMP‐3, MMP‐13, and IL‐1β. Expression and fragmentation of the small leucine‐rich proteins (SLRPs) chondroadherin, decorin, biglycan, lumican, and fibromodulin were analyzed with western blot. Chondroadherin and decorin were fragmented in cartilage from patients with a curve greater than 70°, whereas biglycan and fibromodulin did not show curve‐related fragmentation. AIS facet joint cartilage shows hallmarks of OA including proteoglycan loss, overexpression of pro‐inflammatory mediators, increased synthesis of matrix‐degrading proteases and fragmentation of SLRPs. As with patients with age‐related OA, the premature joint degeneration seen in scoliotic patients is likely to contribute to the pain perceived in some individuals.


| INTRODUCTION
Adolescent idiopathic scoliosis (AIS) is a structural deformity which manifests as a gradually increasing curvature and rotation of the spine. 1 Patients with AIS can experience mild discomfort which later progresses as impairment in spinal function and impaired breathing due to reduced area for lungs to expand. 2 There is a worldwide prevalence of 3% of children aged 10-16 years who suffer from AIS, 3 and currently there are two treatment options: (1) bracing if the curve is below a 50 Cobb angle (major curve angle) and (2) spinal fusion with instrumentation when the curvature is greater than a 50 Cobb angle. 2 Although the treatment options are highly beneficial for managing progression, the long-term effects on patient's quality of life are still unclear. 4 ABBREVIATIONS: AIS, adolescent idiopathic scoliosis; IVD, intervertebral discs; MMP, matrix metalloproteases; OA, osteoarthritis; SLRP, small leucine-rich proteins The clinical system used to classify the wide range of curve types in AIS is called the Lenke system, developed by Lawrence Lenke and colleagues. 5 It provides reliable two-dimensional classification using upright coronal and sagittal radiographs to determine (1) Curve type (1)(2)(3)(4)(5)(6) (2) Lumbar spine modifier (A, B, C), and (3) Sagittal thoracic modifier (−, N, +). This system allows for an objective classification of every possible curve pattern in scoliosis and helps clinicians choose the best treatment accordingly.
In non-scoliotic individuals, the symmetry of the spine allows its load-bearing to be balanced between the intervertebral discs (IVD) and the corresponding facet (zygapophyseal) joints. 6 Due to the midline location of IVDs and their large size, 70%-80% of the load is transmitted through the IVDs and vertebral bodies. 7 The facet joints compensate for the remaining 20%-30% of load distribution. The bilateral position of facet joints gives them the role of limiting flexion and torsion of the spine. 8 In patients with AIS, this equilibrium is shifted and consequently the facet joints become subjected to increased and unbalanced loads because of the abnormal torsion and rotation of the spinal column. It is well-known that load magnitude affects bone and cartilage development and growth as well as proteoglycan production by chondrocytes. 9 Thus, inadequate or excessive loading has detrimental effects on cartilage matrix metabolism. 10 Apart from traumatic injury, which is the most obvious form of abnormal biomechanics, slight shifts in ambulatory mechanics, 11 abnormal tension, shear forces, and malalignment of articular joints 12 are known to contribute toward cartilage degeneration. In fact, load-induced changes are one of the triads of primary osteoarthritis, along with genetic influences and age-related changes. 10 In consequence of abnormal spinal curvatures and loading, osteoarthritis (OA) (defined as destruction of articular cartilage and subchondral bone) could be present in the facet joints of younger AIS patients. Osteoarthritis is also known to be prevalent in 15%-45% of the facet joints in patients with chronic low back pain. 13 Cell proliferation and formation of cell clusters can often be observed in OA. The chondrocytes overexpress pro-inflammatory cytokines, such as IL-6, IL-1β, and matrix-degrading enzymes, such as MMP-3 and MMP-13. 14 Increases in these factors shift matrix homeostasis toward catabolism, thereby promoting tissue destruction.
Consequentially, the cartilage loses proteoglycan content early along with its collagen network later in the disease process. 15 The small leucine-rich proteins (SLRPs), which include biglycan, decorin, chondroadherin, fibromodulin, and lumican, 15 have also been reported to be proteolytically fragmented in OA. 16 SLRPs are found in the pericellular or territorial matrix , 17 and their fragments may potentially act as alarmins 18 by activating toll-like receptors TLR 2 and 4 in articular cartilage. TLR activation triggers the secretion of pro-inflammatory mediators. 19 The aim of this study is to investigate the effects of scoliotic curvature on the articular facet joint homeostasis in patients with AIS. Our underlying hypothesis is that abnormal curvature of the scoliotic spine results in degenerative changes and increased secretion of cytokines and proteases in facet joints, as reported in osteoarthritis.

| Sample collection
Multiple superior articular facet joints were collected from AIS patients undergoing spinal fusion corrective surgery after consent.
The facet joints collected span multiple spinal levels with the levels collected depending on the surgical requirement. The patient's spinal curves were classified by the orthopedic surgeons according to their major curve types using the Lenke system, which categorizes all patients in 6 major curve types with subclassifications for lumbar and sagittal thoracic modifiers using bi-planar radiographs (Table 1). 5

| Histology
Dissected facet cartilage was fixed in 4% paraformaldehyde overnight, followed by sequential immersion in 10%, 20%, and 30% sucrose for 12 hours each before OCT embedding. Cryosectioning was performed

| Immunohistochemistry
The 12-μm-thick sections were immunostained for proliferative and

| Histology staining and quantification
A modified OARSI grading scheme (Grade 0-4) was used and grading of the safranin-O fast green-stained sections was performed blinded by 3 independent evaluators. 21 Healthy cartilage is given a grade 0, and OA starts at grade 1, where mild abrasion and proliferation can occur. At grade 2, there is discontinuity in the cartilage surface that can transform into fissures at grade 3. At grades 4 and 4.5, erosion and excavation takes place, respectively. OARSI grades 5 and 6 (characterized by denudation, deformation, and bone remodeling) were excluded since the cartilage was removed from the bone before analysis. For cell counts, DAPI (Vectashield) and Mayer's hematoxylin (Sigma) stained cell nuclei were counted and normalized to cartilage area using ImageJ software (NIH). All slides were examined using the ×5 objective. A MATLAB script was developed with background normalization, and the area selection tool was used to assess the proteoglycan content through average red pixel intensity in a semi-quantitative manner. 22 In the MATLAB script, the red staining intensity of the entire section was quantified by normalizing to the white background and extracting the RGB (red-green-blue) intensities of each pixel in the region of interest, which was drawn over the cartilage and then isolating the red channel only which was finally averaged over the entire region of interest. prior to SDS-PAGE and western blotting. 16

| Statistics
Unpaired and paired parametric student t test were performed in GraphPad (Prism) to assess significance valued at P < .05. Confidence intervals of 95% were used to plot error bars.

| Degenerative factor expression
Immunohistochemical staining showed a significant (P < .001) increase in positive staining for MMP-3, −13, and IL-1ß in the scoliotic tissue ( Figure 4). Positive cells were defined by strong intracellular and pericellular brown stain, which was normalized to all cells counterstained with hematoxylin. IL-6 showed the weakest immune reactivity with no difference between control and scoliotic groups.

| Fragmentation of SLRPs
Protein extracts of facet joint pairs (left and right facets) at the apex of the curve were pooled for 10 patients. The 10 scoliotic and 1 agematched non-scoliotic patient samples were analyzed on a weight per volume basis by western blotting using antibodies for chondroadherin, decorin, biglycan and lumican ( Figure 5).  As the articular cartilage matures after epiphyseal closure around 20 years of age, its proteoglycan content decreases, and the tissue slowly degrades with aging, a process that is strongly enhanced in OA. 24 In young scoliotic patients who were less than 20 years of age, loss of proteoglycan from the superficial zone was evident, and distinctive fragmentation of SLRPs was apparent confirming an early onset of facet degradation in AIS. This is somewhat similar to IVD matrix disruption as we have previously reported for chondroadherin fragmentation in AIS IVDs 25,26 and the recently reported aggrecan fragmentation in AIS IVDs. 27 In articular cartilage, the histological progression of OA is evidenced by surface discontinuity, fibrillation, and erosion ( Figure 1E).

| Decorin and chondroadherin
These characteristic changes were evaluated by a modified OARSI grading 20 for the analysis of the AIS and non-scoliotic facet articular cartilage. This grading system shows that scoliotic tissues had the strongest OA-related phenotype with significantly higher OARSI grade compared to the control group. Furthermore, proteoglycan loss in facet joints of AIS patients was comparable to affected joints in osteoarthritic patients. In fact, the average age of the scoliotic donors was 15 and that of cadaveric organ donors was 34 years, which further reinforces the finding that known age-related OA-associated changes were more pronounced in younger AIS facets. Interestingly, the only age-matched non-scoliotic donor (17 years old) displayed similar amounts of proteoglycan content and OA-related changes to the older non-scoliotic donors, suggesting that the difference seen between control and scoliotic groups are not solely age-related.
There are many previous studies describing results in the context of a simple 2-dimensional concave or convex side of the spine. However, a simple 2-dimensional analysis by x-ray of spinal concavity and convexity cannot be used to assess differential biomechanical loads across facet joints for two reasons: (1) It does not account for spinal segment rotation which affects the perceived load on each side of the vertebrae and (2) Contrary to IVDs where the concave side gets wedged during scoliosis, the facet joints are not oriented perpendicularly to the spine, which alters the load-bearing as well. For example, the Lenke 1 facet joints showed advanced deterioration on the concave side (Figure 2). However, for more complex curves like the Lenke 3, a different pattern is observed (Figure 2). These observations are in accordance with recent studies that found the higher loading on facet joints in degenerative lumbar scoliosis is dependent on the curve intensity, the position of the apex, and spinal movements, which do not always correlate with the concavity of the curvature. 28 For these reasons, the scoliotic facet joints in this study were all considered as abnormally loaded, and they were compared to the cadaveric nonscoliotic facet joints which we deemed being exposed to physiological and balanced loading. An in-depth 3-dimensional analysis is needed to decipher the link between biomechanical forces and tissue deterioration.
Although osteoarthritis is an age-related disease, 29 there exist other characteristic features that separate osteoarthritis from normal age-related changes. These include cellular proliferation,  30 We found a significantly higher (P < .0001) cellularity in scoliotic cartilage, which was expected because of the age difference between groups and because cell density in articular cartilage decreases with age. However, since the cell density decrease is not linear during development and the bulk of the cell loss is seen before adolescence, 31 we believe the high cell density in scoliotic cartilage is abnormal. To support this argument, the only truly age- driven to proliferate and fill up the larger chondrons, which could be a potential cause seen in the scoliotic samples we studied ( Figure 3C). 33 Tissue homeostasis is critical in maintaining healthy cartilage. If the balance shifts toward matrix catabolism, the cartilage will quickly lose proteoglycans, water content, and consequentially its loadbearing function. This is seen in osteoarthritis when hypertrophic chondrocytes start overproducing matrix-degrading enzymes and proinflammatory factors in response to an inflammatory event. Here, we focused on 4 predominant degenerative factors that are overexpressed in cartilage degradation: proteases MMP3, MMP13, and inflammatory cytokines IL-1ß and IL-6. 34,35 Immunohistochemistry revealed an upregulation of the two proteases of interest, MMP3 and MMP13, compared to the non-scoliotic control group. The biggest difference was seen in MMP13, which had on average 2-fold more positive cell staining. These results follow the previous findings in which MMP-13 was found to be the most upregulated protease in OA. 36 In accordance to the previous findings, the pro-inflammatory cytokine IL-1ß was upregulated which is known to induce MMP secretion in articular cartilage. 37 IL-6, however, remained unchanged between the two groups, but its role in OA is also less understood.
Matrix-degrading enzymes MMP -1,2,3,5,7,10,13 and HTRA1 effectively cleave most SLRPs, fibronectin, collagen, and proteoglycans . 16,26,38 Overloaded cartilage loses its water content as a consequence of proteoglycan loss and synthesizes complex catabolic and degradative molecules which lead to increased fragmentation of its matrix molecules. 39 In this study, we report an extensive proteolysis of the SLRPs chondroadherin and decorin, which specifically appeared only when the scoliotic curves progressed toward higher Cobb angles with greater complexities. The appearances of fragmented chondroadherin and decorin were not detected in tissue sample from an agematched control individual. Extensive fragments of biglycan appeared in all the tissues examined and did not associate with scoliotic curve severities. This suggests that biglycan may be more sensitive to tissue loading in comparison to the others SLRPs. 40 The molecular mass of the chondroadherin, decorin, biglycan, and fibromodulin fragments were between 22 and 35 kDa, and based on previous reports, it can be suggested that proteolytic cleavage and abnormal loading in the scoliotic spine may have jointly contributed to the generation of these fragments. 26,40,41 Lumican, on the contrary, was not fragmented in any sample.
As very little tissue-specific details exist on the spinal curves in AIS patients, limitations are associated with this study: first, all the scoliotic facets examined in this study were subjected to distinct abnormal loading with variable patterns, which might explain the variable results in the scoliotic group. To assess this, biomechanical analysis would be required to better visualize and correlate the 3-dimensional deformity to actual loading and cartilage degeneration.
Second, the scarcity of young cadaveric donors influenced the age difference between the scoliotic and non-scoliotic groups. However, this can be used to reinforce our conclusion by considering that cartilage degenerates with age, and the scoliotic tissues were already more degenerate than the older control group.
In conclusion, this study shows the degenerative effects of chronic abnormal loading on the articular facet joints of patients with AIS. In these tissues, we observed hallmarks of OA, such as proteoglycan loss, overexpression of pro-inflammatory mediators, increased synthesis of matrix-degrading proteases, and fragmentation of the SLRPs. As with patients with age-related OA, the premature joint degeneration seen in scoliotic patients is likely to contribute to the pain perceived in some of these individuals.