There are limited treatment options available today to slow down progression of osteoarthritis in its early stages and most interventions, such as highly invasive partial and total joint replacement surgeries, are performed only at the late stages of the disease. Understanding the mechanism of early articular cartilage stress-mediated wear and failure can aid in the design of new treatment options that are introduced at earlier stages of the disease, presenting the potential to slow down osteoarthritis progression and thus significantly improve patient outcomes. This dissertation aims to provide a basic science understanding of wear propagation and repair of articular cartilage in the absence of traumatic events under the normal reciprocal sliding motion of the articular layers at physiologic load magnitudes. In this dissertation there are three main thrusts: (1) characterize cartilage delamination wear under normal sliding (2) define a chemical environment that promotes cartilage explant homeostasis to enable long-term wear-and-repair studies (3) investigate a practical treatment modality capable of stopping or slowing down structural degeneration of articular cartilage in OA. We hypothesize that the mode of cartilage damage is delamination wear that progresses by fatigue failure of the extracellular matrix (ECM) under physiologic sliding, even when cartilage layers are subject to physiologic load magnitudes and contact stresses and even when the friction coefficient μ remains low (H1a). Based on prior literature findings regarding the role of synovial fluid (SF) boundary lubricants on the reduction of friction and wear, we also test the hypothesis (H1b) that SF delays the onset of cartilage delamination when compared to physiological buffered saline (PBS). We then test a third hypothesis (H1c) that loading cartilage against cartilage delays the onset of delamination wear compared to testing glass on cartilage, since contacting porous cartilage layers exhibit a much smaller solid-on-solid contact area fraction than impermeable glass contacting porous cartilage. Next, we hypothesize that the homeostatic dysregulation previously observed in cultured immature cartilage explants results from the presence of non-physiologic levels of important metabolic mediators in the culture medium. To this end, we hypothesize that: (H2a) immature bovine cartilage explants cultured in native synovial fluid will maintain homeostasis as characterized by maintenance of their mechanical properties and ECM contents at initial (post- explantation) levels, and (H2b) explants cultured in a physiologic-based medium, consisting of physiologic levels of key metabolic mediators, will maintain a similar homeostasis over long- term culture. Finally, a laser treatment strategy is explored that has the capability to reform collagen crosslinks, replacing those lost during OA progression. This novel therapy acts without injuring the cells and without any chemical additive or thermal ablation. The laser treatment protocol used in this application can specifically target the subsurface region, located 200 μm of the articular surface. By strengthening this region with enhanced crosslinking, we hypothesize (H3a) that cartilage will demonstrate greater resistance to fatigue failure than untreated controls. We then hypothesized (H3b) that this treatment protocol would also be effective on devitalized fibrillated human articular cartilage from OA joints with overall Outerbridge score OS1-3. We find that for both cartilage-on-cartilage and glass-on-cartilage sliding configurations at physiologic applied loads, long-term sliding with a low friction coefficient causes wear in the form of delamination. We show that the use of synovial fluid as a lubricant delays the onset of wear; and, similarly, that sliding with a cartilage counterface also reduces the incidence of wear. In subsequent studies we fully characterize a homeostatic culture medium to emulate cartilage in vitro behavior in synovial fluid. We show that explants cultured in this medium can maintain their properties for at least one month and have no loss in cell viability. Laser treatment is then tested on both living and devitalized bovine and devitalized human cartilage and the treatment is shown to improve the wear resistance of the tissue without harming embedded cells. Overall this work has led to novel insights that have clinical applicability. One strength of the in vitro investigations described in this body of work is the ability to separate out mechanically-mediated events from biochemically-mediated events, which would be impossible in vivo. Parsing out such specific mechanisms of cartilage wear can help guide better understanding of disease progression and drive therapeutic intervention. Intervening during the early stages of OA offers the promise of preventive care that currently does not exist and could provide significant benefits to a patient’s quality of life. This dissertation asserts that focusing on delaying or preventing wear by improving the resiliency of the extant intact cartilage in early OA is a viable strategy to improve patient outcomes and offers an innovative approach over existing regenerative techniques.

Articular cartilage is the connective tissue lining the ends of long bones, providing a dynamic surface that bears load while providing a smooth surface for articulation. When damaged, however, this tissue exhibits a poor capacity for repair, lacking the lymphatics and vasculature necessary for remodeling. Osteoarthritis (OA), a growing health and economic burden, is the most common disease afflicting the knee joint. Impacting nearly thirty million Americans and responsible for approximately $90 billion in total annual costs, this disease is characterized by a progressive loss of cartilage accompanied by joint pain and dysfunction. Moreover, while generally considered to be a disease of the elderly (65 years and up), evidence suggests the disease may be traced to joint injuries in young, active individuals, of whom nearly 50% will develop signs of OA within 20 years of the injury. For these reasons, significant research efforts are directed at developing tissue-engineered cartilage as a cell-based approach to articular cartilage repair. Clinical success, however, will depend on the ability of tissue-engineered cartilage to survive and thrive in a milieu of harsh mechanical and chemical agents. To this end, previous work in our laboratory has focused on growing tissues appropriate for repair of focal defects and entire articular surfaces, thereby investigating the role of mechanical and chemical stimuli in tissue development. While we have had success at producing replacement tissues with certain qualities appropriate for clinical function, engineered cartilage capable of withstanding the full range of insults in vivo has yet to be developed. For this reason, and in an effort to address this shortcoming, the work described in this dissertation aims to (1) further characterize and (2) optimize the response of tissue-engineered cartilage to physical loading and the concomitant chemical insult found in the injured or diseased diarthrodial joint, as well as (3) provide a clinically relevant strategy for joint resurfacing. Together, this holistic approach maximizes the chances for in vivo success of tissue-engineered cartilage. Regular joint movement and dynamic loads are important for the maintenance of healthy articular cartilage. Extensive work has been done demonstrating the impact of mechanical load on the composition of the extracellular matrix and the biosynthetic activity of resident chondrocytes in explant cultures as well as in tissue-engineered cartilage. In further characterizing the response of tissue-engineered cartilage to mechanical load, the work in this dissertation demonstrated the impact of displacement-controlled and load-controlled stimulation on the mechanical and biochemical properties of engineered cartilage. Additionally, these studies captured tension-compression nonlinearity in tissue-engineered cartilage, highlighting the role of the proteoglycan-collagen network in the ability to withstand dynamic loads in vivo, and optimized a commercial bioreactor for use with engineered cartilage. The deleterious chemical environment of the diseased joint is also well investigated. It is therefore essential to consider the impact of pro-inflammatory cytokines on the mechanical and biochemical development of tissue-engineered cartilage, as chemical injury is known to promote degradation of extracellular matrix constituents and ultimately failure of the tissue. Combining expertise in interleukin-1\alpha, dexamethasone, and drug delivery systems, a dexamethasone drug delivery system was developed and demonstrated to provide chondroprotection for tissue-engineered cartilage in the presence of supraphysiologic doses of pro-inflammatory cytokines. These results highlight the clinical relevance of this approach and indicate potential success as a therapeutic strategy. Clinical success, however, will not only be determined by the mechanical and biochemical properties of tissue-engineered cartilage. For engineered cartilage to bear loads in vivo, it is necessary to match the natural topology of the articular surface, recapitulating normal contact geometries and load distribution across the joint. To ensure success, a method for fabricating a bilayered engineered construct with biofidelic cartilage and subchondral bone curvatures was developed. This approach aims to create a cell-based cartilage replacement that restores joint congruencies, normalizes load distributions across the joint, and serves as a potential platform for the repair of both focal defects and full joint surfaces. The research described in this dissertation more fully characterizes the benefits of mechanical stimulation, prescribes a method for chondroprotection in vivo, and provides a strategy for creating a cartilage replacement that perfectly matches the native architecture of the knee, thus laying the groundwork for clinical success of tissue-engineered cartilage.

Osteoarthritis (OA) is a debilitating degenerative joint disease affecting 27 million Americans over the age of 25. Whereas OA is a disease of the entire joint organ, the contribution of the synovium, a specialized lining that envelops the knee joint, to cartilage degeneration and disease progression has been underappreciated. Synovial inflammation often precedes the development of cartilage damage and is observed in early and late stage OA. The onset of synovitis is driven by both elevated concentrations of pro-inflammatory cytokines and tissue debris in the joint space. Accordingly, surgeons have observed cartilaginous debris embedded within the synovium of OA patients presenting with severe synovial hyperplasia. It has been hypothesized that the fibrotic shortening of the synovial capsule results in OA pain and joint stiffness and contributes to further joint destruction through the release of degradative enzymes. Current strategies to treat synovial inflammation and joint pain, such as intra-articular injections and synovectomy, have had limited and variable success. To this end, cell and tissue engineering culture models provide a versatile platform to study the tissues and cells involved in OA. Our lab has typically employed mechanical overload or cytokine insult of chondrocytes and cartilage explants to study cartilage degradation. Similarly, to isolate the role of synovium in OA, synovial explants or fibroblast-like synoviocytes (FLS) can be exposed to chemical or physical OA stimuli. Although often overlooked as an instigator of OA, cartilage wear particles have been reported to induce synovial inflammation and OA-like joint changes in various animal models. As opposed to non-biologic (metal or plastic) wear particles, small (sub-10um) cartilage wear particles are comprised of extracellular matrix constituents that are degradable and may interact with cells beyond phagocytosis. Using cells derived from the pathologic joint provides the opportunity to study inherent changes to OA cells (both FLS and chondrocytes) within their own de novo extracellular matrix. The work presented in this dissertation aims to combine knowledge from basic science and pre-clinical culture models of OA to develop a clinically relevant disease model using cells derived from clinical samples.

L'obésité est associée à un risque de fragilité musculo-squelettique, en particulier d’arthrose (OA). Notre but est d’explorer leurs contributions des conditions métaboliques et du surpoids. L’obésité MM (mécanique et métabolique) est obtenue par un régime alimentaire. L’obésité M (mécanique) est mimée par hypergravité à 2g. L’OA est induite par acte chirurgicale. Nous avons exploré les effets des obésités MM et M sur le système musculo-squelettique de souris non OA. Les MM ont un os trabéculaire préservé, un os cortical détérioré et des muscles fragilisés. Chez les M, l’os est préservé et les muscles sont renforcés. Les troubles métaboliques sont responsables de la fragilisation de l’os cortical et du muscle. Dans une 2ème partie, les conséquences de l’OA sont évaluées chez des souris non obèses, MM ou M. L’OA chez les non obèses fragilise uniquement l’os trabéculaire. L’OA chez les MM accentue la diminution de l’épaisseur corticale. L’OA chez les souris M fragilise encore plus l’os cortical et le muscle que chez les souris MM. Cependant si on ne considère que les souris OA, la composante MM est toujours plus délétère que la composante M. Finalement, nous avons testé les vibrations corps entiers pendant les 4 dernières semaines comme traitement potentiel des détériorations musculo-squelettiques des MM couplée ou non à l’OA. Les vibrations n’impactent pas l’obésité et l’OA. Un effet musculaire est observé au niveau moléculaire, ces diminutions étant plus importantes chez les OA. Aucun changement de masse musculaire n’est observé. Le tissu osseux n’est pas influencé. / Obesity is associated with a risk of musculoskeletal fragility, especially osteoarthritis (OA). Our goal is to explore their contributions of metabolic and overweight conditions. MM obesity (mechanical and metabolic) is obtained through a diet. Obesity M (mechanical) is mimed by hypergravity at 2g. OA is induced by surgery. We explored the effects of MM and M obesity on the non-OA mouse musculoskeletal system. MMs have preserved trabecular bone, deteriorated cortical bone and weakened muscles. In M, bone is preserved and muscles are strengthened. Metabolic disorders are responsible for the weakening of cortical bone and muscle. In a second part, the consequences of OA are evaluated in non-obese mice, MM or M. OA in non-obese only weakens the trabecular bone. OA in MM accentuates the decrease in cortical thickness. OA in M mice further weakens cortical bone and muscle than in MM mice. However, if we consider only the OA mice, the MM component is always more deleterious than the M component. Finally, we tested entire body vibrations during the last 4 weeks as a potential treatment for musculoskeletal deterioration of MM, whether or not coupled to OA. Vibrations do not affect obesity and OA. A muscular effect is observed at the molecular level, these decreases being greater in OA. No change in muscle mass is observed. The bone tissue is not influenced.

INTRODUÇÃO: Estudo prévio do nosso grupo demonstrou que uso de palmilha em cunha medial associada ao estabilizador de tornozelo produz uma melhora clínica na osteoartrite do joelho valgo. No entanto, não existem dados na literatura sobre os efeitos biomecânicos destas órteses. OBJETIVO: Avaliação biomecânica dos pés, de maneira estática e dinâmica na osteoartrite do joelho valgo com a utilização de palmilha em cunha medial associada ao estabilizador de tornozelo. MÉTODO: Um total de 42 pés de 21 mulheres com osteoartrite de joelho bilateral (critérios ACR) e deformidade em valgo > 8 graus, foram avaliadas quanto a dados clínicos e biometria. As pacientes foram avaliadas em esteira ergométrica elétrica com: 1. calçado padrão sem a palmilha (controle), 2. palmilha em cunha medial (com 8 mm de elevação medial no retropé) e 3. com essas palmilhas e estabilizador de tornozelo em neoprene. O sistema FSCAN® versão 3.816, com palmilhas flexíveis e 960 sensores de carga na superfície foi utilizado para obter a força plantar vertical. RESULTADOS: Houve redução no pico de pressão plantar estático (PP) com a utilização da palmilha em cunha (P = 0,001) e com a palmilha e estabilizador (P < 0,001) vs. controle em ambos os lados. Além disso, o uso da palmilha associada ao estabilizador de tornozelo resultou em uma redução mais efetiva neste parâmetro em comparação ao uso somente da palmilha (P = 0,027). A avaliação dinâmica deste parâmetro revelou resultado similar no lado direito, com uma redução mais significativa com o uso da palmilha (P < 0,001) e com palmilha e estabilizador de tornozelo (P < 0,001) em relação ao controle. Não foi observada diferença no lado esquerdo (osteoartrite mais grave). A força vertical máxima estática (FVM) também demonstrou diminuição em ambos os lados com o uso da palmilha (P = 0,001) e palmilha associada ao estabilizador de tornozelo (P < 0,001) em relação ao controle. Além disso, o uso da palmilha associada ao estabilizador de tornozelo resultou em uma redução mais efetiva da força vertical máxima estática em comparação ao uso somente da palmilha (P = 0,041). Da mesma forma, na condição dinâmica, esse parâmetro foi significativamente reduzido com o uso da palmilha associada ao estabilizador de tornozelo em comparação à condição controle (P < 0,001). Também houve redução na FVM entre o uso de palmilha com e sem estabilizador (P = 0,003). A avaliação qualitativa revelou que a órtese altera significativamente a trajetória do vetor de força (P < 0,001). CONCLUSÃO: O uso da palmilha em cunha medial associada ao estabilizador de tornozelo promoveu uma redução no pico de pressão plantar e na força vertical máxima em condições estática e dinâmica, subjacente à melhora clínica na osteoartrite do joelho valgo / INTRODUCTION: We have previously demonstrated a significant clinical improvement in valgus knee osteoarthritis with the use of medial-wedge insole associated with ankle support. There is, however, no data regarding the foot plantar forces underlying this beneficial effect. Objective: Static and dynamic feet plantar biomechanical evaluation of medial-wedge insole associated with ankle support in valgus knee osteoarthritis. METHOD: A total of 42 feet of 21 women with bilateral knee osteoarthritis (ACR criteria), with valgus deformity were evaluated regarding clinical and biometric data. Patients were assessed with: 1. standard shoes without the insoles (control); 2. medial-wedge insole; 3. insoles/neoprene ankle support. The system FSCAN® 3816 version, with flexible soles and 960 load sensors on the surface was used to obtain the foot plantar vertical forces. RESULTS: A decreased peak plantar pressure was observed with insoles (P = 0,001) and insoles/ankle support (P < 0,001) vs. control while standing motionless in both sides. In addition, insoles/ankle support resulted in a more effective reduction in this parameter than solely insoles (P = 0,027). The dynamic evaluation of this parameter revealed a similar finding on the right side with a more significant reduction with use of insoles (P < 0,001) and with insoles/ankle support (P < 0,001) compared to controls. No difference was observed on the left side (more severe OA). The static maximum vertical force was also decreased in both sides with insoles (P = 0,001) and insoles/ankle support (P < 0,001) compared to control. In addition, the later condition provided a more significant reduction in the static maximum vertical force than solely insoles (P = 0,041). Likewise, in the dynamic condition this parameter was significantly reduced with insoles/ankle support compared with control condition (P < 0,001) and solely insoles (P = 0,003). The qualitative evaluation revealed that orthoses significantly changed the center of force shift (P < 0,001). CONCLUSION: We have identified that a reduction in the feet plantar peak and maximum vertical force in valgus knee osteoarthritis in static and dynamic conditions underlies the clinical improvement of medial-wedge insole associated with ankle support

Intraarticular fractures (IAFs) are a leading cause of posttraumatic osteoarthritis (PTOA). Despite the latest orthopaedic treatment techniques, the risk of PTOA after IAFs has remained unacceptably high. In order to progress in this field, a new mechanical insult technique to create a large animal survival model of human IAF was developed. Current IAF models report the initial gravitational potential energy as the fracture energy value. However, this model included a pendulum device that was instrumented to accurately measure the amount of energy absorbed during fracture insult. After validating the energy absorption measurement with a mechanical testing machine and motion capture system, an in vivo study was conducted. The range of energy absorption measurements during fracture of the eleven animals was 11.7 é31.8 joules, with a mean and standard deviation of 20.8 ± 5.7 joules. On average, the energy absorption measurements were approximately 52 percent of the pre – impact kinetic energy values. These data showed that there was a substantial difference between the energy absorbed during fracture insult and the pre éimpact energy, which provided novel information associated with the pathomechanics of the induced injury.

Tissue inhibitor of metalloproteinase-3 (TIMP-3) is a protein with multiple functions that include regulating the turnover of the extracellular matrix (ECM) by inhibiting members of the metzincin family. Extracellular levels of soluble TIMP-3 are low, reflecting its binding to components of the ECM including sulfated glycosaminoglycans (SGAGs) and its endocytosis by low density lipoprotein receptor-related protein 1. Because TIMP-3 inhibits ECM-degrading enzymes, the ability of SGAG mimetics to elevate extracellular concentrations of TIMP3 is of interest for osteoarthritis treatment. However, previous studies of such interactions have utilized immobilized forms of the protein or ligands. Here we have quantified the thermodynamics of the interactions of the inhibitory domain of TIMP-3 with chondroitin sulfate (CS), pentosan polysulfate (PPS) and suramin in solution using isothermal titration calorimetry. All three interactions are driven by a (favorable) negative enthalpy ychange combined with an unfavorable decrease in entropy. The heat capacity change (ΔCp) for the interaction of N-TIMP-3 with CS, PPS, or suramin is essentially zero, indicating an insignificant contribution from the hydrophobic effect. Based on the effects of ionic strength on the interaction of N-TIMP-3 with suramin, their interaction appears to be driven by electrostatic interactions. Modeling supports the view that the negatively charged sulfates of CS, PPS, and suramin interact with a cationic region on N-TIMP-3 that includes Lys -26, -27, -30, and -possibly 76 on the opposite face of TIMP-3 from its reactive site for metalloproteases. / Includes bibliography. / Dissertation (Ph.D.)--Florida Atlantic University, 2019. / FAU Electronic Theses and Dissertations Collection

Osteoarthritis is a chronic, deleterious disease of the joints. It currently affects nearly 25 million Americans. Clinically, osteoarthritis presents as joint pain and verified by radiographic evidence of joint space narrowing. Unfortunately, symptomatic osteoarthritis describes the later stages of disease, at which point irreversible cartilage and bone damage has occurred. Cross-sectional imaging modalities offer the promise of visualizing early features of disease, enabling the development and evaluation of interventions to forestall or prevent degenerative change. Modalities of clinical interest include magnetic resonance imaging (MRI) and multi-detector computed tomography (MDCT). The following work describes the efficacy of MRI-derived measures for the identification and accurate quantification of local and whole joint changes in articular cartilage thickness changes in vivo. This was performed as part of a study investigating the diagnostic potential of clinical morphometric and compositional MRI to identify early features of osteoarthritis in a large animal model of traumatic knee joint injury. Surgically induced trauma consisted of a partial medial meniscectomy and blunt impact of either 0 J, 0.6 J, or 1.2 J to the weight-bearing cartilage of the medial femur. The study was six months in duration. To evaluate the accuracy of MRI-derived measures of cartilage thickness, imaging acquired at time of euthanasia was compared to high-resolution contrast-enhanced micro-computed tomography (micro CT). 3-dimensional multimodal analysis demonstrated that morphometric MRI imaging is sensitive to sub-voxel changes in cartilage thickness. Therefore, MRI is a clinically relevant modality to quantify subtle cartilage damage, thereby presenting an opportunity to identify patients earlier in the disease process.

Cartilage lesion is a fairly common problem in orthopaedic practise. It is often a consequence of traumas, inflammatory conditions, and biomechanics alterations. However, as an avascular and aneural tissue, articular cartilage has minimal healing ability. Over the past decades, surgeons and scientists have proposed a nubmer of treatment strategies to promote restoration of articular cartilage, like arthroscopic lavage, microfracture surgery, osteochoncral autografts and allografts, autologous chondrocyte implantation, and other cell-based repairs. Nevertheless, these solutions often result in fibrocartilage, which has inferior mechanical and biochemical properties, with increased susceptibility to injury, which usually ultimately leads to osteoarthritis (OA). Stem cell therapy techniques are widely applied in treating disease or injury. Many medical researchers have proposed stem cell transplantation treatment for enhancing cartilage repair by using mesenchymal stem cells (MSCs) along with biocompatible scaffolds. In addition to that, chondrogenic progenitor cells (CPCs) have also been discovered in OA patients and healthy articular cartilage. However, neither the method for isolating CPCs is not well established, nor the origin and function is not fully understood. Stem cells may be measured in CFUs (Colony-forming units). Ideally, adult stem cells should be clonogenic. In other words, a single adult stem cell should be able to generate a line of genetically identical cells. Fully characteraization of stem/progenitor cell potential requires purified population. Single-cell cloned population maybe serve as a convincing source for study of stem/progenitor cells. Therefore, a single cell clonogenecity screening system was developed to identify and isolate putative stem/progenitor cells from cartilage based on fluorescence-activated cell sorting (FACS). Also, genetical and functional characterization of isolated cells was taken.

Post-traumatic osteoarthritis (PTOA) is a debilitating joint disease in which cartilage degenerates following joint trauma, including intra-articular fracture or ligament rupture. Acute damage and chronically altered joint loading have both been implicated in the development of PTOA, but the precise pathway leading from injury to cartilage degeneration is not yet known. A series of computational analyses were performed to gain insight into the initiation and progression of cartilage degeneration. Finite element models of in vitro drop-tower impacts were created to evaluate the local stress and strain distributions that cartilage experiences during such experiments. These distributions were compared with confocal imaging of cell viability and histologically apparent matrix damage. Shear strain and tensile strain both appear to correlate with the non-uniform percentage of cell death seen in the impact region. In order to objectively evaluate structural damage to the cartilage matrix, an automated image processing program was written to quantify morphologic characteristics of cartilage cracks, as seen in histology slides. This algorithm was used to compare the damage caused by different rabbit models of PTOA and to investigate the progression of matrix damage over time. Osteochondral defect insults resulted in more numerous and more severe cracks than ACL transection. Interestingly, no progression of structural damage was identified between 8 weeks and 16 weeks in these rabbit PTOA models. A finite element based optimization algorithm was developed to determine cartilage material properties based on the relaxation behavior of an indentation test. This was then used to evaluate the spatial and temporal progression of cartilage degeneration after impact. Impacting cartilage with 2.18 J/cm2 through a metal impactor caused an immediate increase in permeability and decrease in modulus, both of which recover to nearly pre-impact levels within two weeks. Biologic testing suggests that the modulus changes were due to collagen fibril damage that is then repaired. Impacting with higher energy caused material softening that did not return to normal, suggesting an impact injury threshold below which cartilage had some ability to repair itself. To evaluate the effects of cartilage cracks on local stress and strain environments, finite element models of cracked cartilage were created. A physiologically-relevant, depth-dependent cartilage material model was developed and used to ensure accurate strains throughout the cartilage depth. The presence of a single crack was highly disruptive to the strain fields, but the particular shape or size of that crack had little effect. The most detrimental perturbations included two cracks within close proximity. When two cracks were within 0.5 mm of one another, the strain field between them increased in an additive fashion, suggesting a threshold for the amount of structural damage cartilage can withstand without being severely overloaded. The finite element models of cracked cartilage were also incorporated into an iterative degeneration simulation to evaluate the ability of mechanical loading to cause localized cartilage damage to spread to full-joint osteoarthritis.