DEVELOPMENT OF AN EQUINE MRI OSTEOARTHRITIS SCORING SYSTEM

Transcription

1 DEVELOPMENT OF AN EQUINE MRI OSTEOARTHRITIS SCORING SYSTEM By ANDREW DAVIDSON SMITH A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2017

2 2017 Andrew Davidson Smith

3 To my loving and supportive wife, Dr. Amanda J Ardente

4 ACKNOWLEDGMENTS I have had the opportunity to work with an extraordinary group of mentors, technicians, family and friends who have supported me throughout my career. I would first like to thank Dr. Alison Morton who was both my resident advisor during my surgery residency and my primary advisor for my graduate program. She not only encouraged me to continue on and pursue a PhD but also secured funding for me throughout the graduate program through the UF Alumni Graduate Studies Fellowship Award. In addition, Dr. Morton and Drs. Matt Winter and Erin Porter spent countless hours with me scoring MRI s. This work was a big part of my dissertation research, without which I would not have been able to finish my graduate program. I would also like to thank Drs. Patrick Colahan and Steve Ghivizzani for allowing me the opportunity to work on two major research projects. The first project provided the MRI and arthroscopy images needed to perform my dissertation research, and the second assisted in funding my graduate program. These experiences provided me with an invaluable experience obtaining and interpreting MR images as well as performing diagnostic arthroscopy in the horse. Furthermore, Dr. Jorge Hernandez was instrumental in helping with study design and statistical analysis of my dissertation research projects. I would especially like to thank Brett Rice and the staff in the treadmill room for taking great care of the horses during research project and for putting up with my schedule. In addition, I would like to thank Mary Wilson, Bobby Davis, and Matt Bolin for helping me take radiographs and perform MRI examination on all of my research horses. Their knowledge and experience also proved to be invaluable when we were having trouble obtaining MRI images of diagnostic quality. I would also like to thank the 4

5 UF Large Animal Surgery Technicians and the UF Anesthesia Service for helping perform MRI and arthroscopic examinations on my horses and making sure they recovered safely. Furthermore, I would like to thank Dr. Ammon Peck and Mrs. Sally O Connell in the Office of Graduate Studies and Research at the University of Florida College of Veterinary Medicine for working to negotiate the logistics of my program and for providing me with the support needed to see my dissertation research through to completion. I am very grateful for the efforts and support of these two individuals. Lastly I would like to thank my family and my beautiful wife Dr. Amanda Ardente for supporting me throughout my career: to my mom who sent down spaghetti sauce and croutons whenever the previously supply ran out, to my dad who waited until the end of my graduate program to tell me I needed to get a real job, to my brother and his family who were always willing to FaceTime with me to see how I was doing, and to my loving wife for keeping me sane and for always providing me with a date on Friday nights. 5

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS... 4 LIST OF TABLES... 9 LIST OF FIGURES ABSTRACT CHAPTER 1 INTRODUCTION Problem Statement Aims and Objectives LITERATURE REVIEW Synovial Joint Anatomy Articular Cartilage Collagen Proteoglycans Chondrocyte Subchondral Bone Joint Capsule and Periarticular Structures Synovium Synovial Fluid Traumatic Osteoarthritis Pathophysiology Articular Cartilage Subchondral Bone Acute injury Chronic injury Synovium and Joint Capsule Periarticular Structures Equine Models of Osteoarthritis Medical Treatment of Osteoarthritis Nonsteroidal Anti-Inflammatory Drugs Glucocorticoids Chondroprotective Agents Hyaluronic acid Polysulfated glucosaminoglycans Chondroitin sulfate Biological Therapies Potential for the Use of Gene Therapy Osteoarthritis Imaging

7 Radiography Ultrasound Nuclear Scintigraphy Computed Tomography Magnetic Resonance Imaging Image acquisition Sequence selection Artifacts Field strength Comparison with other imaging modalities Cartilage imaging Arthroscopy MAGNETIC RESONANCE IMAGING SCORING OF AN EXPERIMENTAL MODEL OF POST-TRAUMATIC OSTEOARTHRITIS IN THE CARPUS Introduction Materials and Methods Experimental Design Magnetic Resonance Imaging Assessment Equine MRI Osteoarthritis Scoring System Statistical Analysis Results Discussion VALIDATION OF AN EQUINE MRI OSTEOARTHRITIS SCORING SYSTEM USING AN EQUINE MODEL OF POST-TRAUMATIC OSTEOARTHRITIS Introduction Materials and Methods Equine MRI Osteoarthritis Scoring System Arthroscopic Scoring System Statistical Analysis Results Reliability Validity Discussion USE OF AN EQUINE MRI OSTEOARHTRITIS SCORING SYSTEM TO EVALUATE THE EFFICACY OF A NOVEL THERAPY IN AN EQUINE MODEL OF POST-TRAUMATIC OSTEOARTHRITIS Introduction Materials and Methods Experimental Design Magnetic Resonance Imaging Equine MRI Osteoarthritis Scoring System

8 Statistical Analysis Results Discussion CONCLUSION LIST OF REFERENCES BIOGRAPHICAL SKETCH

9 LIST OF TABLES Table page 3-1 Magnetic Resonance Imaging Parameters Aligned ranks split plot in time (repeated measures) ANOVA testing for group (G) and day (D) effects Aligned ranks split plot in time (repeated measures) ANOVA testing for group and day Median Scores (Min, Max) For Pathological Changes Observed Between 10 OCI Joints and 10 CON Joints On Day Median Scores (Min, Max) For Pathological Changes Observed Between 10 OCI Joints and 10 CON Joints On Day Median Scores (Min, Max) For Pathological Changes Observed Between 10 OCI Joints and 10 CON Joints On Day Total Individual Articular And Whole-Organ Scores Observed In 10 OCI Joints And 10 CON Joints On Days 0, 14, And Median Scores (Min, Max) For Pathological Changes Observed Within 10 CON And 10 OCI Joints On Days 0, 14, And Magnetic Resonance Imaging Parameters Equine MRI Osteoarthritis Scoring System Arthroscopic Scoring System Percent Involvement (scores > 0) of each articular feature from consensus 1 on days 0, 14, and Reliability of the Equine MRI Osteoarthritis Scoring System Mean Total MRI and Arthroscopic Semiquantitative Scores Sensitivity and Specificity of MRI to Detect Intra-Articular Pathology Equine MRI Osteoarthritis Scoring System Semiquantitative MRI Scores For 10 CON-OCI Joints and 10 TX-OCI Joints On Day Semiquantitative MRI Scores For 10 CON-OCI Joints and 10 TX-OCI Joints On Day

10 5-4 Total Individual Articular And Whole-Organ Semiquantitative MRI Scores In 10 CON-OCI Joints And 10 TX-OCI Joints On Days 14 And Observed Change In Semiquantitative MRI Scores Between Day 14 and 98 For 10 CON-OCI Joints and 10 TX-OCI Joints Observed Change in Total Individual Articular And Whole-Organ Semiquantitative Scores In 10 CON-OCI Joints And 10 TX-OCI Joints Between Day 14 and

11 LIST OF FIGURES Figure page 2-1 Schematic, cross-sectional diagram of healthy articular cartilage Histological representation of articular cartilage and underlying subchondral bone in a normal joint Schematic representation of the interaction of collagen, hyaluronan, and aggrecan in cartilage Factors involved in enzymatic degradation of articular cartilage matrix Diagram of the 9 sub-regions that were evaluated using EMOSS Schematic representation of the scoring system used for high-signal bone lesion intensity Schematic representation of the scoring system used for synovial thickening Sagittal PD-FS and PD MRI images of an OCI joint at day 0, 14, and 98 showing progression of signal changes at the osteochondral fragment site Dorsal PD-FS and Axial T2-STIR MRI images of an OCI joint at day 0, 14, and 98 showing the progression of high-signal bone lesions and synovial thickening Sagittal PD-FS MRI images of an OCI joint at day 0 and day 14 showing the increase in joint effusion identified at day Dorsal T2-weighted MRI images of an OCI joint at day 0 and 98 demonstrating the progression of low-signal bone lesions and marginal osteophyte formation Dorsal SPGR-FS MRI images of an OCI joint demonstrating the progression of subchondral bone irregularity and cartilage signal abnormality Dorsal SPGR-FS image showing normal and abnormal cartilage signal intensity Schematic representation of the MRI scoring system used for high-signal bone lesion intensity Schematic representation of the MRI scoring system used for subchondral bone irregularity Schematic representation of the MRI scoring system used for low-signal bone lesions

12 4-5 Schematic representation of the MRI scoring system used for marginal osteophytes Schematic representation of the MRI scoring system used for the dorsomedial intercarpal ligament Schematic representation of the MRI scoring system used for the medial palmar intercarpal ligament and lateral palmar intercarpal ligament Schematic representation of the MRI scoring system used for synovial thickening Schematic representation of the MRI scoring system used for joint effusion Diagram of the 9 sub-regions that were evaluated using EMOSS Schematic representation of the arthroscopic scoring system used for grading articular cartilage lesions Schematic representation of the arthroscopic scoring system used for grading marginal osteophytes Schematic representation of the arthroscopic scoring system used for grading the dorsomedial intercarpal ligament Schematic representation of the arthroscopic scoring system used for grading the medial and lateral palmar intercarpal ligament Schematic representation of the arthroscopic scoring system used for grading synovial thickening Diagram of the 9 sub-regions that were evaluated using EMOSS

13 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy DEVELOPMENT OF AN EQUINE MRI OSTEOARTHRITIS SCORING SYSTEM Chair: Alison Morton Major: Veterinary Medical Sciences By Andrew Davidson Smith August 2017 Osteoarthritis is a common cause of lameness in the equine athlete. Conventional radiography is commonly used to evaluate structural changes, however, it lacks the sensitivity necessary to detect early changes and correlates poorly with clinical outcomes. Magnetic resonance imaging (MRI) is a multiplanar imaging modality that allows assessment of all structures of the joint and has the ability to detect pathology at an earlier stage of the disease process. Arthroscopic examination of both middle carpal joints were performed on twenty horses and an osteochondral fragment was created on the distodorsal aspect of the radial carpal bone in one randomly selected joint (OCI) and the contralateral joint served as the sham-operated control (SHAM). On day 14, 10 of the osteochondral injured joints were treated with a novel therapy (OCI-TX) and the other 10 were treated with saline (OCI-CON). The horses then underwent treadmill exercise until day 98. An equine MRI osteoarthritis scoring system (EMOSS) was developed and used to score 15 articular features on MRI exams performed on day 0, 14, and 98. High-signal bone lesions and osteochondral fragment scores were higher on day 14 and high-signal bone lesion, low-signal bone lesion, osteophyte formation, cartilage signal abnormality, subchondral bone irregularity, joint effusion, and synovial 13

14 thickening scores were higher on day 98 in the OCI-CON joints compared to SHAM- CON joints. Intraexaminer agreement (kappa range: to 0.780) of the EMOSS was inferior to the interexaminer agreement (kappa range: to 1.000). When semiquantitative MRI scores were compared to arthroscopy, articular cartilage lesion scores were higher on arthroscopy but marginal osteophyte and total joint scores were higher on MRI. The overall sensitivity of MRI to detect cartilage lesions was minimal (0.23), particularly if the size of the lesion was < 5 mm, but the specificity was high (0.99). The EMOSS was unable to detect a difference between OCI-TX and OCI-CON joints. The high variability in semiquantitative MRI scores, however, suggests that the inability to detect a difference between treatments may also be related to the variability in the equine model of osteoarthritis used and not due to the insensitivity of MRI. 14

15 CHAPTER 1 INTRODUCTION Problem Statement Osteoarthritis is a common cause of lameness and decreased performance in equine athletes. 1-3 The equine osteochondral fragment-exercise model of post-traumatic osteoarthritis mimics clinical disease and has been used to assess the efficacy of numerous therapies for the treatment of osteoarthritis. 4 Conventional radiography has traditionally been used to characterize the progression of structural changes associated with this model of osteoarthritis, however, disease-modifying effects have yet to be identified regardless of treatment Although there are several plausible explanations for treatment failure, conventional radiography may be an unresponsive measure of disease progression as it lacks the sensitivity necessary to detect early structural changes associated with osteoarthritis, correlates poorly with clinical outcomes, and cannot directly evaluate all structures of the joint thought to be involved in the pathogenesis of osteoarthritis Magnetic resonance imaging (MRI) is increasingly being used for the assessment of musculoskeletal disorders in horses as it enables direct visualization of all structures believed to be involved in pathogenesis osteoarthritis and has the ability to detect pathology at a much earlier stage of the disease process. 23,24 Semiquantitative MRI scoring systems have recently been developed in humans and have been used to evaluate the disease-modifying effects of drug treatment by assessing how individual articular features change over time Thus the development of an equine MRI osteoarthritis scoring system may be a more sensitive outcome measure to which to 15

16 assess the disease-modifying effects of novel therapies using the carpal osteochondral fragment-exercise model of post-traumatic osteoarthritis. Aims and Objectives The aim of this research was to develop an equine MRI osteoarthritis scoring system that could be used as an outcome measure for which to assess the disease modifying effects of novel therapies using the carpal osteochondral fragment-exercise model of post-traumatic osteoarthritis. Thus, the first objective was to characterize the progression of MRI changes in a carpal osteochondral injury-exercise model of posttraumatic osteoarthritis using a semi-quantitative scoring system for evaluation of the equine middle carpal joint. We hypothesized that MRI would be able to detect osteoarthritis-associated structural changes 14 days following osteochondral injury and scores obtained from a semi-quantitative equine MRI osteoarthritis scoring system would be higher in the osteochondral injured joints at both 14 and 98 days following osteochondral injury compared to baseline examination, as well as between osteochondral injured and control joints. The second aim was to assess the diagnostic value of an equine MRI osteoarthritis scoring system to identify pathologic changes over time in an experimental model of osteoarthritis. Therefore, the first objective of this study was to test the reliability of the equine MRI osteoarthritis scoring system in detection of pathology. The second objective of this study was to validate the equine MRI osteoarthritis scoring system by comparing MRI to the structures that can be visualized by arthroscopy, using arthroscopy as the gold standard. We hypothesize that semiquantitative MRI scores will show good reliability between observers and that MRI scores will also correlate to intraarticular pathology as detected by arthroscopy. 16

17 The third aim was to evaluate the ability of an equine MRI osteoarthritis scoring system to determine treatment efficacy of a novel therapy using an equine model of post-traumatic osteoarthritis. The objective of this study was to compare the observed change in semiquantitative MRI scores from days 14 and 98 between treated and control joints. We hypothesize that there will be a significant difference in observed change in semiquantitative MRI scores between treated and control joints. 17

18 CHAPTER 2 LITERATURE REVIEW Synovial Joint Anatomy Synovial joints are composed of multiple components, including bone, articular cartilage, and periarticular supporting structures, all of which contribute to the joint s normal function. Anatomic integrity and proper cellular function of each component are critical for these structures to work in unison, and thus synovial joints are often referred to as an organ system. 33 Joint homeostasis is achieved when the anabolic and catabolic processes of each component are equilibrated and match the biomechanical challenges posed to the joint. Each component has its own metabolic characteristics and turnover rate which ultimately affects its capacity for repair. 34 Inability to repair injury to any components may lead to dysfunction of other corresponding components, which can eventually lead to organ failure. Multiple terms have been used to describe the pathophysiological events that may lead to organ failure including osteoarthritis, osteoarthrosis, osteoarthropathy, and degenerative joint disease. 35 The author, however, feels that osteoarthritis is the more correct and currently accepted term to describe pathological changes that occur in the injured joint and therefore this is the term that will be used throughout this manuscript. Articular Cartilage Articular cartilage is the central structure of the joints that allows for simultaneous weight-bearing and motion with minimal friction. Chondrocytes account for 1-12% of the total volume of articular cartilage with the remaining termed extracellular matrix. 34 The extracellular matrix of articular cartilage consists of collagen, proteoglycans, and water. Water constitutes 70-80% of the extracellular matrix. Collagen, proteoglycans, and 18

19 glycoproteins constitute 50%, 35%, and 10% of the extracellular matrix on a dry weight basis, respectively. 34 Four microscopic layers have been classically described: the superficial, intermediate, and deep layers together form the hyaline portion of cartilage, and the calcified layer forms the interface between hyaline cartilage and subchondral bone. 34 (Figure 1-1) The superficial layer is characterized by a relatively small amount of proteoglycans, high water content and a higher percentage of flattened chondrocytes, densely packed and smaller diameter type II collagen fibrils that are aligned parallel to the articular surface. 36 Moving through the intermediate layer to the deep zone, there is a gradual transition to an increased amount of proteoglycans, less water content, and less frequent but larger diameter collagen fibrils oriented perpendicular to the articular surface. 36 The chondrocytes are also less frequent but increase in size and are round in shape. 36 These variations in tissue composition result in differences in biomechanical properties. Articular cartilage lacks vascular, lymphatic, and neural supply, so nutrition and waste removal occur through synovial fluid diffusion. The critical distance over which this type of diffusion can take place is limited and likely explains why cartilage thickness does not increase isometrically with increasing body mass when comparing various species. 37 The calcified layer forms an important interface between the non-calcified hyaline cartilage and subchondral bone. It functions not only as a transitional zone for the hyaline cartilage to attach to subchondral bone but it may also facilitate transmission of forces between the two biomechanically dissimilar layers. 38 Calcified cartilage is believed to minimize stress between hyaline cartilage and subchondral bone because it has mechanical properties that are intermediate to the two structures and because it 19

20 also has undulating insertions into the subchondral bone, spreading the force of impact over a larger surface area. 39 Nutrients diffuse from subchondral bone to the deeper cartilage layers. 38 Histologically, the calcified layer is separated from the hyaline cartilage by the tidemark (Figure 1-2), which is a single, hematoxyphil line up to 10 μm in thickness. 40 It has a complex three-dimensional structure that can dip through the entire calcified layer in places to abut adjacent bone or marrow spaces. 40 Articular cartilage s specific architecture is pivotal to its function. When the articular cartilage is not under load, extracellular matrix will swells due to the hydrophilic nature of proteoglycans. Collagen fibers, however, restrict the expansion of the extracellular matrix generated by proteoglycans due to the formation of a stiff, threedimensional scaffold. 34 As articular cartilage is loaded and unloaded, water is squeezed out and reabsorbed, respectively, making this structure a viscoelastic tissue. These properties allow cartilage to absorb and distribute compressive load across the joint during exercise. 34 But because of the differences in geometry of opposing joint surfaces, the joint is not loaded evenly over its surface and different areas of the joint are subject to different types and magnitude of loading. Consequently, the biomechanical composition of articular cartilage at different sites within the joint will reflect the type of load applied to it. This was shown in the equine metacarpophalangeal joint where high intermittent loading was related to high collagen content with a high level of crosslinking and less intense but more consistent loading was associated with higher proteoglycan content. 41 The biochemical and biomechanical properties of articular cartilage is not genetically determined but develops during the first months of life, and is influenced by the biomechanical loads experienced by the joint during that time

21 Collagen Collagen forms the framework, or lattice, in which all other matrix molecules are constrained. Its function is to counteract the tensile stresses at the joint surface. There are at least 16 types of collagen that have been described in mammalian species and can be categorized into either fibril-forming (types I, II, III, V, and XI) or non-fibril-forming (types IV, VI, VII, VIII, IX, X, XII, XIII, and XIV) collagen. 43 The principal collagen type of articular cartilage is the fibrillar type II collagen constituting 90-95% of the total collagen content. 44 This type of collagen consists of three identical α1 protein chains arranged in the form of a right-handed triple helix. 34 Type II collagen is produced intracellularly by the chondrocytes in its proform with propeptides on both the N (amino) and C (carboxy) terminals. 34 Once outside the cell, proteinases cleave the propeptides leaving behind telopeptides on each end. 34 The N-telopeptide and C-telopeptide ends are nonhelical and are important sites for formation of crosslinks to adjacent collagen molecules. 34 The collagen network of articular cartilage is a copolymer of collagen type II, IX, and XI with other collagen types playing minor roles. (Figure 1-3) Chondrocytes are responsible for the assembly of this collagen network (fibrillogenesis) and directs its remodeling when prompted by changing biomechanical challenges or damage. Crosslinking within the same collagen molecule, between different collagen molecules, and different collagen fibrils contributes to the structural coherence of the collagen network and is critical to the mechanical stability of articular cartilage. 34 Proteoglycans Within the extracellular matrix, proteoglycans can be found interspersed between the collagen fibrils and are connected to them either directly or through hyaluronan molecules. (Figure 1-3) Aggrecan is the largest and most abundant proteoglycan known 21

22 and is made up of a linear protein core with several hundred glycosaminoglycan side chains attached. 35 The protein core is divided into three domains. The G1 globular domain exists at the amino terminal end of the protein and attaches to hyaluronan via a linker protein. 34 Currently, there is no known function to the highly conserved G2 globular domain, but it is unique to aggrecan. The G3 domain forms the C-terminal end of the protein and it functions to link the proteoglycan aggregates to the extracellular matrix. 45 The glycosaminoglycan side chains, such as chondroitin sulfate and keratan sulfate, attach between the G2 and G3 domains. 46 The sulfate groups of these glycosaminoglycan side chains are negatively charged which makes them hydrophilic, which causes them to repel each other. 34 Multiple aggrecan molecules will attach to the same hyaluronan chain. Strongly water-binding aggrecan aggregates form a fan-shaped structure surrounding a central hyaluronan molecule. 34 This highly hydrated matrix gives cartilage its compressive stiffness and ability to dissipate load. A small percentage of proteoglycans present in the extracellular matrix are known as small leucine-rich proteins/proteoglycans (SLRPs). 47 The most well-known SLRPs in articular cartilage are decorin, byglycan, fibromodulin, lumican, and chondroadherin. They have an important role in the regulation of fibrillogenesis. 47 Chondrocyte Chondrocytes make up a small fraction of the total volume of articular cartilage (approximately 1 to 12% of total cartilage volume) but are responsible for the production of the major components of the extracellular matrix. 36 Signaling mechanisms cue chondrocytes to participate in tissue homeostasis, however, these mechanisms are not fully understood. 34 Physiological mechanical loading may be the primary stimuli for tissue homeostasis. 34 When articular cartilage is placed under load, there is deformation 22

23 of both the chondrocyte and its nucleus, as well as changes in the osmotic and hydrostatic pressure, resulting in extracellular flow of interstitial fluid. 48 These biomechanical stimuli are believed to stimulate the release of growth factors that promote the production of the extracellular matrix, as well as have effects on the nutrient supply and waste removal within the articular cartilage. 49 Transforming growth factor-β and insulin-like growth factor-1 are the two major growth factors that have been shown to have an anabolic effect on articular cartilage through stimulation of matrix production and inhibition of degradation. 34 Chondrocytes also possess a cytoplasmic process, otherwise known as a primary cilium, that is also believed to sense changes in the biomechanical environment and may play a role in mechanotransduction. 50 Subchondral Bone Subchondral bone consists of the subchondral bone plate and underlying trabecular bone which together function to support the articular cartilage, to maintain the shape of the weight-bearing surface, and to transmit loads from the joint to the rest of the bone. The subchondral bone plate consists of a layer of compact bone that is composed of an inorganic component of mineral salts and an organic component of predominately type I collagen, proteoglycan, glycosaminoglycan, and water. The inorganic phase imparts a hardness and rigidity to the subchondral bone plate, whereas the organic phase provides flexibility and resiliency. Histologically and biomechanically, the subchondral bone plate is similar to bone in other locations, however, the bone plate is much thinner and its Haversian systems are oriented parallel to the articular surface instead of perpendicular. 35 Additionally, it has been shown to be approximately ten times more deformable than the cortical shaft of long bones. 51 These differences allow the subchondral bone to perform its role in attenuating the axial loads generated by 23

24 locomotion and spare the overlying cartilage from damage. The subchondral bone plate is supported by trabecular bone. When the joint is loaded, a small degree of joint deformation is allowed by trabecular bone under normal circumstances. This assists in the dissipation of energy and increases the congruency of the joint. In accordance with Wolf s law, however, repeat loading can result in subchondral bone thickening and sclerosis, which can be amplified by trabecular microfractures and subsequent subchondral bone healing. This change in rigidity may have repercussions for the mechanical loading of the articular cartilage. 39,52-57 The subchondral bone plate is penetrated by channels that contain vessels and nerves. These channels provide a direct link between calcified cartilage and trabecular bone 58, and are thought to provide nutrients to the deeper cartilage layers that cannot be nourished by synovial fluid. 2 This permeability is also thought to allow crossover communication between the two structures. 59 Alterations in either tissue can modulate the properties and functions of the other through cellular signaling pathways and may play a role in development and progression of osteoarthritis. 59 Due to the high number of sensory nerve fibers, the subchondral bone is also one of the main sources of pain in osteoarthritis. 34 Joint Capsule and Periarticular Structures Tendons, ligaments, and joint capsule are structurally similar in that they are all sparsely cellular but contain an abundant extracellular matrix composed of predominately types I and III collagen, elastin, proteoglycan, glycosaminoglycans, and water. 34 Tissue differences exist, however, in their individual elastin/collagen ratio and collagen arrangement which relates to the amount of force each tissue is able to withstand. 34 Periarticular ligaments, muscles, tendons, and joint capsule function to 24

25 provide mechanical stability to the joint. Large numbers of proprioceptive nerve fibers provide information to the brain regarding joint position. 35 Each of their individual contributions depends on the anatomic location of the joint. More proximal joints obtain more stability from muscles whereas more distal joints are stabilized primarily by the joint capsule and ligaments. 35 Intra-articular structures, such as the menisci, function to improve joint congruency to better distribute load across the joint surface. 34 Menisci collagen fibers of the femorotibial joint are radial on the surface and circumferential in the interior, which converts compressive forces applied during loading into circumferential tensile forces. 60 All of these structures are innervated and may give rise to substantial pain when damaged. 34 Additionally, it is hypothesized that damage to any of these structures can result in destabilization of the joint and increase the strain placed on other structures of the joint. 61 Synovium The synovium, the inner layer of the joint capsule, consists of two layers: the intimal and subintimal layers. 34 The intimal layer is directly adjacent to the joint cavity and is typically 1 to 4 cell layers thick. Lack of a basement membrane in the intimal layer and the presence of a rich blood supply in the subintimal layer facilitates the passage of plasma components from the blood to the synovial cavity. 34 Components of the plasma such as glucose, oxygen, carbon dioxide, and proteins (typically less than 10 kda in size) diffuse through the membrane and contribute to the composition of synovial fluid, where as larger molecules are excluded. 35 This suggests a largely passive process, however, the synoviocytes which compromise the intimal layer have a high excretory capacity. Exchange of molecules between the synovial cavity and blood 25

26 plasma is also driven by both locomotion-induced hydrostatic pressure differences and colloid osmotic pressure differences. 34,62 Classically, synoviocytes have been divided into two categories. 63 Type A synoviocytes resemble macrophages and are have phagocytic activity. Type B synoviocytes are fibroblast-like and are responsible for production and excretion of proteins and other macromolecules such as hyaluronan and nonsulfated glycosaminoglycans into the synovial fluid, which determines the viscosity of this fluid. Type C synoviocytes are a third type of cell that may be a transitional form between types A and B. The synovium covers all intra-articular surfaces, excluding articular cartilage and localized areas of bone. 36 It is not uniform throughout the joint, with areas that are flat intermixed with areas of loose collections of villi, which can be described as seaweedlike as viewed through a fluid medium. 36 The difference between flat and villous synovium may be related to the biomechanical characteristics of the specific joint area. Villi may proliferate in areas of the joint associated with increased injury. 36 The synovial membrane may respond to mechanical injury through increased in collagen production, alterations in transsynovial diffusion, and/or changes in synoviocyte metabolism. 64,65 All synoviocytes have the capacity to produce a large array of cytokines, growth factors, and inflammatory mediators that play a role in the maintenance of joint homeostasis and pathology. 34 These changes can affect all of the structures within the joint, including the articular cartilage, which may alter the ability of the joint to respond appropriately to loading during motion

27 Synovial Fluid The function of synovial fluid is to provide an efficient supply of nutrients and waste management for the avascular articular cartilage and to act as a lubricant between structures within the synovial cavity. 61 The rheological properties of synovial fluid are much different than blood plasma due to the hyaluronic acid content, which gives it its viscous character. Lubrication within the joint involves two mechanisms: boundary lubrication and fluid-film lubrication. Boundary lubrication functions to resist shearing stresses between the two cartilage surfaces and between the articular cartilage and synovial membrane. Synovial fluid is directly responsible for boundary lubrication as it provides specific lubricating molecules, such as hyaluronan, proteoglycan-4 (lubricin), and surface active phospholipids that adhere to the articular surfaces and forms a protective layer, preventing excessive wear. 66 Hyaluronan is the principle boundary lubricant for the synovial membrane and hyaluronan and lubricin play an important role in the lubrication of articular cartilage. 67 Additionally, hyaluronan secretion from type B synoviocytes is thought to aid in excluding leukocytes and other solutes from the joint space through steric hindrance. 35 Boundary lubrication, however, is less effective at physiological loads where it is more supported by fluid-film lubrication. 68 The elastohydrodynamic model of fluid-film lubrication is the most widely excepted model to represent fluid-film lubrication and states that when load or pressure is applied to a joint, water is squeezed from the articular cartilage and interposed between the two articular surfaces separating the two structures. In reality, multiple mechanisms come into play during joint loading and is dependent upon the type of activity and speed of motion of the particular joint. 34 Additionally, variations in how the 27

28 joint is loaded can influence the behavior of the extracellular matrix of cartilage and the rheologic properties of synovial fluid. 69 Traumatic Osteoarthritis Pathophysiology Osteoarthritis is a common clinical problem in horses. Surveys estimate it to be involved in up to 60% of equine cases presenting for lameness. 3 Osteoarthritis has been defined as a disorder of moveable joints characterized by degeneration and loss of articular cartilage. 35 Although articular cartilage damage is still the primary feature used to define the severity of osteoarthritis, this definition is not completely correct, as osteoarthritis is no longer considered a single disease entity. The diarthrodal joint is a complex organ with all joint components including articular cartilage, subchondral bone, and soft tissue supporting structures involved in maintaining function. If any one or combination of these components lose biomechanical function, the resulting tissue damage will further impair joint function. This leads to a cycle of relative overloading and joint deterioration, a process similar to organ failure in other body systems. 36 Abnormal joint loading has been shown to cause a decrease in cartilage proteoglycan content, an increase in degradative enzymes, and cause chondrocyte necrosis and apoptosis. 70,71 This has been termed a chrondropenic response, and involves the loss of articular cartilage volume and stiffness, elevation of contact pressures, and development and progression of articular cartilage defects. 72 Furthermore, pathologic change in any other structures of the joint can alter the mechanical loading of articular cartilage and thus stimulate a chondropenic cascade. 1 Equine osteoarthritis should be considered a group of disorders characterized by a common end stage: progressive deterioration of the articular cartilage accompanied by changes in the bone and soft tissues of the joint

29 Two biomechanical mechanisms proposed as causes for joint injuries include abnormal loads applied to normal tissue and normal loads applied to abnormal tissue. 74 This is a rather simplistic approach to a disease that involves the complex biomechanical and biochemical interactions between all components of the joint. 75 As multiple joint components are commonly involved in traumatic joint injuries, it is now considered more accurate to describe how the different components of the joint are involved in the pathogenesis of osteoarthritis. 73 Post-traumatic osteoarthritis is a specific form of osteoarthritis that develops following a known injurious event. 76 The extent of mechanical damage to the joint is associated with the intensity of the impact. 77 Traumatic joint injuries are common in equine athletes and can occur by either a single episode or more commonly following repetitive episodes of trauma. 1 The majority of joint injuries in horses are acute clinically, they are secondary to chronic fatique injuries. 78 True acute traumatic injuries are a relatively uncommon cause of joint injury and are the result of an acute traumatic force to the joint above the biomechanical threshold of tissue failure. 78 These typically occur in areas of the joint that are not commonly injured and through tissue that lack any sign of healing and/or adaptive responses that are characteristic of chronic fatigue injury. 78 Such a superphysiological force can result in gross damage to one or more components of the joint including osteochondral damage (fracture, fragmentation, or contusion), primary articular cartilage damage, intraarticular or extraarticular ligament injury, joint capsule injury, meniscal injury, and/or synovitis. 78 In contrast, chronic fatigue injuries are the result of cyclic loading of tissue below the biomechanical threshold of tissue failure where the pathological process starts at the molecular and biochemical level and may 29

30 eventually progress to tissue failure. The progression of damage can occur by (1) microdamage formation within tissues, resulting in molecular and biochemical processes that either repair the injury or result in its progression; (2) microdamage may be absent, but the biochemical and tissue responses to cyclic loading can create an area of weakness in the tissues, predisposing it to damage; (3) adaptive tissue responses that fatigue the tissues, resulting in chronic progressive changes to its material properties, leading ultimately to injury. 78 Clinically, the location and pathological manifestation of these types of injuries are consistent but can differ by age, breed, and discipline. 79 In the equine athlete, chronic fatigue injuries consistently occur in the subchondral bone with some evidence to suggest that it also occurs in the articular cartilage and is worsened by degradative processes associated with synovitis. 78,80 Articular Cartilage The ability of the articular cartilage to respond to joint trauma depends largely on its ability to mount a reparative response. There is normally a low turnover of extracellular matrix to repair damage from physiological daily activity. There is a balance between catabolic (matrix metalloproteinases and aggrecanases, prostaglandins, free radicals, cytokines, interleukin-1, and tumor necrosis factor-alpha) and anabolic processes (transforming growth factor-beta and insulin-like growth factor-1), which is critical to maintaining healthy articular cartilage. 81 Physiological exercise benefits articular cartilage by enhancing the chondrocytic response to insulin-like growth factor In the equine metacarpophalangeal joint, proteoglycan synthesis was increased in horses given moderate exercise compared to non-exercised controls. 83 Volume, thickness, and glycosaminoglycan content of articular cartilage was increased in human knees following exercise and there is a positive relationship between repetitive loading 30

31 and articular cartilage function in human knees. 72,84 However, there is a critical level of exercise after which the beneficial effects of exercise are predominated by catabolic factors and ability for repair is impaired. Depending on the severity of injury, pathological changes to the articular cartilage can range from biochemical alterations to gross articular cartilage lesions. Repetitive loading may cause biochemical changes, which cause either no significant lesions or clinical signs such as mild synovitis. True acute, gross articular cartilage damage, however, is induced by acute superphysiological loading and often involves subchondral bone and surrounding soft tissue structures. 78 Pathological processes that occur in articular cartilage following joint injury include direct mechanical damage to the extracellular matrix and chondrocytes and/or enzymatic destruction of the extracellular matrix through the production of inflammatory mediators by chondrocytes, synoviocytes, and/or osteoblasts. 74 Abnormally high mechanical loads on articular cartilage can lead to the dysfunction of chondrocytes, breakdown of extracellular matrix, and ultimately cartilage degeneration. 85 Magnitude, orientation, and rate of loading is important to the extent of the injury. 86,87 If the load applied to the articular cartilage is sufficient, chondrocyte death can occur by either cell necrosis or apoptosis. 88,89 Damage to extracellular matrix and chondrocytes is highest in the superficial layer and decreases with cartilage depth. 86,87 Collagen fibers in the superficial layer are parallel to the articular surface, maximizing resistance to lateral deformation at the surface where tension in this plane is the greatest. When articular cartilage fissures penetrate the superficial layer, this resistance is lost due to the more perpendicular orientation of collagen fibers in the deeper regions 31

32 and increased in chondrocyte death are seen adjacent to these matrix fissures. 90 Chondrocyte necrosis at the sight of superphysiological load occurs immediately following impact. 76 Cell death by apoptotic mechanisms subsequently occurs in the areas surrounding the impacted region leading to expansion of the original lesion. 76 Mitochondrial dysfunction play a central role in both processes. This was evident when blocking of mitochondrial depolarization significantly improved chondrocyte viability in an equine articular cartilage explant model. 91 Chondrocytes are pivotal to maintaining the function of the articular cartilage through the production of the extracellular matrix. As chondrocytes die, they are no longer able to maintain or repair the extracellular matrix surrounding them, which puts additional mechanical and metabolic stress on the remaining chondrocytes. It has been hypothesized that once a critical number of chondrocytes are lost, more chondrocyte death would continue until degeneration of the entire cartilage surface has occurred. 85 Following joint injury, there is a significant increase in expression of inflammatory cytokines (interleukin-1, tumor necrosis factor-alpha) and matrix-degrading enzymes (matrix metalloproteinases and aggrecanases) and exhaustion of matrix-degrading enzyme inhibitors (tissue inhibitors of matrix metalloproteinases) that lead to the breakdown of the extracellular matrix shortly following injury. Chondrocytes and synoviocytes are the more commonly known sources, however, there is recent evidence to show that multiple joint tissues can contribute to inflammation after joint injury. 92 The effect of either direct mechanical and/or enzymatic degradation of the extracellular matrix can cause both the disruption of the collagen framework and a decrease in aggrecan concentrations. Disruption of the collagen fiber network is associated with 32

33 softening or swelling of the articular cartilage. 93 These morphological changes in the articular cartilage are associated with concurrent changes in the mechanical properties including a decrease in the tensile, dynamic shear, and compressive module and failure stress in a canine anterior cruciate transection model. 94 Continued loading to this mechanically-compromised cartilage will cause disease progression in the form of superficial fibrillation, fibrillation down to the middle zone, followed by full-thickness erosions if reparative mechanisms are not recruited. 1 Specifically in the horse, one of the more common findings is the presence of wear lines in the metacarpo- /metatarsophalangeal joints that are histologically associated with chondrocyte death, proteoglycan loss, and collagen fiber disruption in the superficial cartilage. 80 Once injured, articular cartilage has a limited ability to repair or regenerate. There are currently three proposed mechanisms that have been recognized as possible contributors to articular cartilage repair. 1 Chondrocytes have a limited capacity for intrinsic repair by increasing collagen and proteoglycan production. Subchondral bone participates in extrinsic repair of articular cartilage injury by providing mesenchymal elements that initially form fibrous tissue but may undergo metaplastic change to form cartilage elements. The third mechanism involves a process called matrix flow, whereby lips of cartilage form around the perimeter of the lesion and eventually migrate toward the center of the defect. 95,96 The capacity of these mechanisms to repair articular cartilage is dependent upon the depth of the injury, size of the defect, location and relation to the weight-bearing or non-weight-bearing surface, and age of the animal. 95,97,98 Unfortunately, partial thickness lesions have very little, if any capacity to repair and the fibrocartilaginous repair tissue that forms after full-thickness injuries is 33

34 biomechanically inferior to articular cartilage and is likely to undergo mechanical failure with use. 1 Subchondral Bone The subchondral bone and articular cartilage are dynamic stress bearing structures that play complementary roles in load-bearing. Articular cartilage provides a smooth, lubricated surface for articulation and facilitates the transmission of loads with a low frictional coefficient, where as the subchondral bone supports the overlying cartilage and distributes mechanical loads across joint surfaces. 52 Their close intimate contact has been described as a closely composited functional unit called the osteochondral junction. 99 Although osteoarthritis has long been considered as a primary disorder of the articular cartilage, there is increasingly more focus on the role of subchondral bone in the pathophysiology of osteoarthritis. 58 There is controversy as to whether subchondral bone alterations precede cartilage degeneration or vice versa. 100 Depending on the particular clinical disease entity or model of osteoarthritis, evidence suggests that subchondral bone alterations can occur before, during, or after cartilage damage. 101 Gaps in knowledge stem from the lack of understanding the exact nature and range of pathology that occurs in the subchondral bone in response to injury, the significance of that pathology in relation to the function of the joint as a whole, and its temporal relationship to pathology of other tissues in the development of osteoarthritis. 74 In the equine athlete, subchondral bone is susceptible to both acute traumatic and chronic fatigue-type injuries, which likely have different pathological manifestations. Acute injury As stated previously, true acute subchondral bone lesions occur in bone without evidence of healing or adaptive bone responses to prior injury. In the equine athlete, 34

35 acute subchondral bone injury is a rare or poorly recognized cause of lameness. The lesions can be caused by compressive, torsional and/or shear forces leading to either a complete fracture through the bone (involving two joint surfaces) or fragmentation either within or outside the joint capsule. 78 These types of fractures are difficult to prognosticate due to their low occurrence and unusual configuration. However, a good prognosis is generally given if the subchondral bone can be realigned and there is minimal articular cartilage damage. An acute impact injury to the subchondral bone can also result bone contusion (an area of microscopic damage and bruising) or high signal intensity as detected by magnetic resonance imaging (MRI). 20,102 Despite being recognized in multiple case studies using MRI 19,20, , the importance of bone contusions in the pathogenesis of osteoarthritis is unknown due to inability to detect these lesions on any imaging modality other than MRI and the lack of controlled studies to map the progression of these lesions over time. In people, however, it has been shown that high signal bone lesions that are large, persistent, or adjacent to the articular surface are associated with articular cartilage degeneration over time Chronic injury In contrast to acute subchondral bone injury where a monotonic force results in catastrophic failure, most subchondral bone injury in the equine athlete is due to repetitive, high-intensity loading of the appendicular skeleton. 74,79 Repetitive cyclic loading of bone during intense exercise results in bone fatigue and the term chronic fatigue injury is being used to describe these pathological processes. 78 In general, fatigue refers to degradation of material properties when a structure is repetitively loaded with forces that are less than the monotonic force required to cause catastrophic failure. 79 When loaded repetitively, however, damage accumulates at the molecular and 35

36 biochemical levels over time until it progresses to histological evidence of tissue damage and ultimately tissue failure. 78 The fetlock and carpal joints are most commonly affected, especially in racehorses where subchondral bone damage can lead to intraarticular fractures (carpal fractures, metacarpal/metatarsal condylar fractures, or fractures of the proximal phalanx) or focal joint surface injuries such as palmar osteochondral disease and lytic lesions of the subchondral bone The fact that these injuries are consistently associated with repetitive, high-intensity exercise without evidence of external trauma 116,117, occur in predictable locations within the joint depending on the breed, discipline and age of the horse 79, and are identified within or adjacent to areas of subchondral bone that show evidence of an adaptive or healing bone response ,118 supports the etiopathogenesis of fatigue failure. 79 Subchondral bone adapts to changes in exercise intensity by increasing the bone volume fraction of the subchondral trabecular bone in response to the increases in stresses applied to it The process by which this occurs is uncertain, however, it is generally believed that bone remodeling is inhibited by strenuous exercise through the inhibition of osteoclast recruitment 123 and that the increase in bone volume fraction is through modeling. 124,125 Modeling refers to either the formation or resorption of bone at a given site, without the coupling of the two processes. 126 Increases in trabecular bone volume fraction have been shown to improve both the monotonic mechanical properties and fatigue life of bone. 79 The importance of bone adaptive modeling is evident by the increased risk of lower limb fractures in horses that do not gallop prior to racing. 127,128 The amount of modeling is dependent on mode, rate, intensity, and duration of activity. 129 In contrast, the term remodeling refers to the replacement of bone with new 36

37 bone. It is an important process in removing damaged bone through a specific sequence of events that involve a series of osteoclasts and osteoblasts working together as a basic multicenter unit. Activation of the basic multicenter unit is the first step in remodeling, followed by resorption and later formation (A-R-F sequence). 126 In contrast to bone modeling, bone resorption and formation are coupled in bone remodeling. 126 The process of bone formation is slower in comparison to resorption and the newly formed bone is mineralized to a level of about 65-70% fairly quickly after formation. However, the time needed to achieve maximum density and stiffness takes much longer and can increase in diseased states. 126 Bone modeling is interpreted radiographically as subchondral bone sclerosis and is especially evident in the proximal sesamoid bones 130, bones of the carpus 53, , and the distal condyles of the third metacarpal bone 53,120 of racehorses. It has hypothesized that subchondral bone sclerosis causes reduction in the shock-absorbing ability of the joint and thereby places cartilage at risk of shear-induced tensile failure of collagen cross-links, particularly under repetitive impulse loading conditions. 39 This hypothesis was based on the premise that thickening of the subchondral bone leads to increased joint stiffness leading to increased strain incurred by the articular cartilage. 39 Attempts at providing experimental evidence to substantiate these hypotheses have largely been unsuccessful. 74 Another dichotomy is that the subchondral bone thickening may occur with or without articular cartilage changes. 74 Cause and effect relationship between subchondral bone sclerosis and cartilage degeneration remains vague because of the lack of understanding of pathophysiological changes that occur within the subchondral bone in response to injury and how these changes affect other 37

38 structures of the joint. It has also been hypothesized that marked modeling or subchondral bone sclerosis of the metacarpal/metatarsal condyles in response to training results in a density gradient at the parasagittal groove the predisposes this site to fracture. 120, Recent work, however, presents contradictory evidence to this hypothesis as the majority of condylar fractures evaluated occurred at sites deep within areas of high bone volume fraction with uniform density gradients rather than sites with high density gradients at the margin of these areas. 124 In addition to bone formation, repetitive cyclic loading also inhibits bone remodeling by a process thought to be due to inhibition of osteoclast recruitment. 123,138 Initially these effects may be negligible, but overtime bone remodeling inhibition limits the replacement of fatigued bone and the accumulation of microdamage. 139 Microdamage has also been shown to coalesce in areas of the joint subject to high loads during galloping exercise. 140,141 These areas of focal porosity in the calcified cartilage and subchondral bone may contribute to the development of fatigue fractures or subchondral bone injuries if the microdamage accumulates faster than can be repaired by remodeling. 125, At some point, however, bone remodeling is no longer inhibited as subchondral bone resorption, an indicator of bone remodeling, is lowest during the early training period but can increase the longer the horse stays in training. 142 It is hypothesized that the accumulation of microdamage can directly stimulate focal remodeling in the subchondral bone, even under high loading conditions 124,125, as microdamage has been found in association with focal areas of bone resorption in previous studies. 56,111,134,140,143,144 Although the mechanism is not completely understood, microdamage or microcracks are thought to stimulate bone remodeling 38

39 through damage to the osteocyte canalicular processes and cellular apoptosis, although other pathophysiological mechanisms may also contribute as well. 145 Remodeling repairs microdamage initially by resorption of the fatigued bone and the associated porosity causes a transient reduction in stiffness and yield strength, which is greatest in the areas of high strain. 124 Bone resorption is also more prominent in the subchondral bone of horses that are taken out of training, as remodeling is no longer inhibited by high-intensity loading. 141,142 Although the process of bone remodeling is important for maintenance of mechanical properties of bone, horses with highly porous subchondral bone are likely at an increased risk of subchondral bone injury if they are returned to or remain in high intensity exercise prior to the porous bone being filled in. 141,142 Microdamage or microcracks in the calcified cartilage have been correlated to the severity of cartilage degeneration in the third carpal bone of racehorses. 140 Similar findings have also been identified in experimental models of impact trauma. 146,147 Combined, these studies suggest that the appearance of microcracks may represent a key step in the pathway to post-traumatic osteoarthritis. Microcracks not only stimulate microdamage repair but also stimulate vascular invasion by angiogenic factors and create a conduit for which the subchondral bone and articular cartilage can communicate via subchondral pores. 148 Mechanisms by which microdamage stimulates remodeling are not completely understood. However, receptor anatagonist of nuclear factor-κβ (RANKL) is essential for osteoclast differentiation and regulates osteoclastic bone resorption and prevents osteoclast death. 149 RANKL is produced in greatest abundance by osteocytes but also by osteoblasts and chondrocytes. 150 Investigation of post-traumatic osteoarthritis in horses identified a stronger relationship between the 39

40 hyaline cartilage RANKL score and osteoclast density than either the subchondral bone or calcified cartilage RANKL. 150 Expression of RANKL by hyaline cartilage correlated with the degree of cartilage degradation. 150 These findings suggest that not only is RANKL a key signaling molecule for osteoclast recruitment, but may also be a key signaling molecule in the crosstalk between the articular cartilage and subchondral bone and that hyaline cartilage may play an important role in recruitment of osteoclasts. 150 Repair of microdamage is also associated with vascular invasion, which may extend to the deep layers of articular cartilage. 148 Angiogenic factors that stimulate vascular invasion induce chondrocytes to synthesize and secrete matrix-degrading enzymes and prevents their association with their inhibitors. 148 This creates ideal conditions for degradation of cartilage and creates a positive feedback loop for crosstalk between subchondral bone and the articular cartilage. There is both biomechanical and biochemical interplay between articular cartilage and subchondral bone implying that the two structures represent an authentic functional unit. 100 Synovium and Joint Capsule Synovitis (and/or capsulitis) is the most common problem in high motion joints of equine athletes result from cyclic trauma to the synovial membrane and/or joint capsule. In equine osteoarthritis, it may also be present secondary to injury to any other joint component. 73,74,151 Synovitis causes increased blood flow and increased permeability of the endothelium and synovial intima, which results in net accumulation of synovial fluid within the joint and is termed synovial effusion. Synovial effusion can have several pathological consequences. The functional stability of the joint is enhanced when the intra-articular pressures are subatmospheric (-1.25mmHg). 152 The intra-articular pressure, however, becomes positive when synovial effusion is present that decrease 40

41 the stability of the joint. 152 Increase pressure may also cause pain, impede vascular perfusion of joint tissues, or even rupture the joint capsule. 74 Net increases in synovial fluid volume also increases the distance between the synovial membrane and articular cartilage. effectively increasing the distance for diffusion of molecules and ultimately reduces the nutritional exchange between the two structures. Hyaluronan concentrations also decreases as synovial fluid volume increases and can affect the boundary lubrication properties of synovial fluid. 74 Hyaluronan concentrations were found to be lower in synovial fluid samples taken from injured joints compared to normal joints, and even lower in acutely injured joints compared to chronic. 66 Kinetic friction coefficient of the synovial fluid was inversely related to the hyaluronan concentration in that study, and perhaps articular cartilage may be more vulnerable to deterioration when boundary lubrication is deficient, especially in the acute stage of injury. 66 Synovium is a major source of inflammatory mediators that can have deleterious effects on the whole joint, especially articular cartilage. These inflammatory mediators include: matrix metalloproteinases and aggrecanases, prostaglandins, free radicals, and cytokines such as interleukin-1 (IL-1) and tumor necrosis factor alpha (TNF-α). 151, (Figure 1-4) The importance of synovitis and the production of these inflammatory mediators in the pathogenesis of equine osteoarthritis was first shown in the synovitis model using filipin 157, where cartilage degradation occurred in the absence of instability or traumatic disruption of the tissues. Since then, research has show that various mediators alter articular cartilage metabolism by disrupting the normal balance of anabolic and catabolic factors crucial for maintenance and repair of articular cartilage s extracellular matrix. 73,81 41

42 Interleukin-1 beta (IL-1β) is thought to be the most important factor responsible for the catabolism in osteoarthritis 9, and the principle mediator of the acute inflammatory response following joint trauma. 158 IL-1β induces proteoglycan depletion in articular cartilage by either increasing degradation or decreasing synthesis associated with the release of cytokines from chondrocytes or synovial cells. 1 By binding to its transmembrane interleukin 1 receptor (IL-1R) on chondrocytes and synovial cells, IL-1β activates intracellular signaling pathways that involve transcription factors NFκ-β and activating protein-1 (AP-1) which results in the expression of metalloproteinases, aggrecanases, and prostaglandin E2. 81 TNF-α is the second key cytokine involved in the osteoarthritis process, however, its exact role in equine osteoarthritis is less certain. 1 It is part of the acute inflammatory response in equine osteoarthritis 159 and upregulates the expression of degradative enzymes similar to IL-1β. 81 It is generally accepted that IL-1β is the principle cytokine involved in the progression of cartilage degradation and is involved in both early and late stages of osteoarthritis. In contrast, TNF-α contributes more to clinical morbidity and pain during the early stages of osteoarthritis. 81 Matrix metalloproteinases are a group of zinc-dependent endopeptidase enzymes involved in degradation of the extracellular matrix. Matrix metalloproteinase-13 (MMP-13) is the primary collagenase involved in the degradation of type II collagen, whereas matrix metalloproteinase-3 (MMP-3) has a variety of substrates including proteoglycans and type IV, V, VII, IX, and XI collagen. 160 Once collagen is fragmented and unwound, it is susceptible to further cleavage by matrix metalloproteinases-2 (MMP-2) and -9 (MMP-9). 154,155 Aggrecanase-1 and -2 (also known as ADAMTS-4 and - 5) are the principal degradative enzymes of aggrecan in equine joint disease. 1 Tissue 42

43 inhibitors of metalloproteinases (TIMP) are involved in the regulation of these matrix metalloproteinases(timp-1 and -2) and aggrecanases (TIMP-3). 161,162 Prostaglandin E2 is one of the main mediators of pain and inflammation in osteoarthritis. It is involved in decreasing proteoglycan content of articular cartilage by stimulating its degradation and inhibiting its synthesis. It also stimulates vasodilation, enhances pain perception, increases bone demineralization, and promotes secretion of plasminogen activator in the joint. 1 Blocking the effects of prostaglandins through cyclooxygenase inhibitors relieves pain, but does not ameliorate cartilage loss in people with osteoarthritis and may potentiate the antianabolic effects of IL-1β. 81,163 Oxygen-derived free radicals, like nitric oxide, have been shown to have many detrimental effects on articular cartilage including cleavage of hyaluronic acid, degradation of collagen alpha chains, and proteoglycan cleavage. 164 Damage-associated molecular patterns (DAMPs) have also been implicated in the pathogenesis of post-traumatic osteoarthritis. 165 These metabolites are generated through the mechanical or proteolytic degeneration of joint tissues such as extracellular matrix or DNA from stressed or dying cells, and serve to stimulate an innate immune inflammatory response. 165 This in conjunction with the continued degradation of extracellular matrix from inflammatory mediators could result in heightening the proinflammatory joint tissue response to injury. 166 Although catabolic factors dominate during osteoarthritis, anabolic factors play a role in osteoarthritis. Transforming growth factor beta (TGF-β) promotes expression of collagen type II and aggrecan and downregulation of matrix-degradative enzymes, thus counteracting IL-1β-induced matrix degradation. 81 Bone morphogenetic proteins-2 (BMP-2) and -7 (BMP-7) are members of the TGF-β superfamily of signaling proteins 43

44 and play a major role in maintenance of articular cartilage. Although BMP-7 is downregulated by IL-1β and TNF-α, BMP-2 is upregulated and increases expression of collagen type II. 81 BMP-7 has the strongest anti-degnerative effects including inhibiting degradative enzymes and providing positive effects on matrix biosynthesis. 167 Insulinlike growth factor-1 (IGF-1) plays a critical role in cartilage homeostasis through stimulation of extracellular matrix production and inhibition of matrix degradation. Additionally, IGF-1 protects cartilage against IL-1β-induced catabolic effects by upregulation of IL-1 decoy receptor (IL-1RU). 168 In contrast to the NFκ-B signaling pathway seen in catabolic factors, the anabolic factors IGF-1, TGF-β, and BMP-7 act through induction of Sox Sox-9 is a key regulator of mesenchymal chondrogenesis during embryonic development and enhances expression of cartilage-specific genes for type II collagen and aggrecan. 169,170 Development of post-traumatic osteoarthritis involves more than just the initial mechanical insult; several factors contribute to it sdevelopment. Experimental models and clinical evidence indicate there is an acute inflammatory phase following injury that involves the entire joint. There is significant contribution from the synovium due to its role in maintaining synovial fluid and its capacity for production of inflammatory mediators. The balance between proinflammatory and anti-inflammatory mediators and the resulting downstream activities influence whether or not the joint develops posttraumatic osteoarthritis or returns to cartilage homeostasis. Periarticular Structures Intra-articular and periarticular soft tissue structures of the joint, including the joint capsule, muscles, tendons, ligaments and menisci, are vital to providing mechanical stability and geometry of the joint through distribution of forces applied during 44

45 locomotion. 34 Injury to any of these structures can disturb joint homeostasis and result in severe pathology. Unfortunately, their exact role in equine joint disease is poorly understood. When muscles contract, a force is transmitted across joints in order to stimulate locomotion. The magnitude of force created and thus the stress applied to the joint is proportional to the muscle s cross-sectional area, muscle fiber composition, and timing of firing. 171,172 Proper muscle function is critical for good response to proprioceptive information obtained principally by the tendons, ligaments, and joint capsule during locomotion to ensure proper joint loading. 34 Any abnormality that would impede muscle function, such as damage to the muscle or tendon, physiological abnormalities, or disruptions in its neural input could affect the timing of muscle activation and thus increase or alter the stresses applied to the joint. 61 Muscle fatigue occurs commonly during strenuous exercise and may negatively affect this mechanism, resulting in loss of coordination and accuracy in the direction of joint loading and can result in increased risk of joint injury. 173 Tendons, ligaments, and the joint capsule are structurally and biomechanically similar qualitatively, however, differences exist between different tissues in the organization and quantity of collagen and elastin fibers between the different tissues. 61 Because these tissues are similar qualitatively, they all exhibit viscoelastic behavior in that their mechanical properties vary as they are stretched. On the other hand, their quantitative differences in elastin and collagen can explain the disparity in the amount of force each tissue is able to withstand and the variations of the slope of their respective stress-strain curves or tissue modulus. 61 Similar to other joint structures, injury can 45

46 result from either acute overloading or cumulative failure overtime. True acute injuries uncommonly occur during high intensity exercise, whereby acute failure of a tendon or ligament can cause marked destabilization of the joint resulting in increased stress applied to the other structures of the joint. 61 It is also reasonable to assume that chronic or clinically silent injuries to tendons and ligaments may also alter the stability of the joint during loading but to a lesser degree. Over time, this may result in excessive stretching of the joint capsule and synovial lining, or abnormal loads applied to articular cartilage, subchondral bone, or other soft tissue structures of the joint leading to release of inflammatory mediators. 1 Although synovial membrane can reform after an acute injury, synovial intima may become fibrotic following chronic, repetitive injury or chronic inflammation. 78 Finally, injury to the menisci and subsequent resection dramatically alters the normal biomechanics of the stifle by decreasing the contact area between the femur and tibia and increasing joint capsule and cruciate ligament laxity in horses. 174 Equine Models of Osteoarthritis Osteoarthritis is the most common joint disease and among the most important causes of pain, disability and economic loss in people. 75 A substantial fraction (~12%) of the overall burden of disease of osteoarthritis arises secondary to joint trauma. 175 This fraction represents a specific subcategory of osteoarthritis known as post-traumatic osteoarthritis. 175 Although the end-stage pathophysiology of post-traumatic osteoarthritis is similar to that of primary osteoarthritis, patients with post-traumatic osteoarthritis normally have a well-defined precipitating injury, such as an articular fracture or a chondral, ligament, or meniscal injury. 175 Pathophysiological events following injury are precipitated by either the initial injury itself or residual joint instability or incongruity caused by the acute injury. The time course over which clinically measurable 46

47 osteoarthritis develops is highly variable, ranging from two to five years in the case of certain articular fractures, to decades for less severe joint injuries. 75 It has been estimated that even with the best care currently available, risk of post-traumatic osteoarthritis increases as much as 10-fold following a significant ligamentous or capsular injury and as much as 20-fold following most articular fractures. 75 Compared to primary osteoarthritis, post-traumatic osteoarthritis represents an opportunity for early intervention since it occurs following a known precipitating event. Anderson et al. 75 proposed that there are three overlapping phases of cartilage injury and response to injury that occur over the first two week after trauma: (1) an early phase is characterized by cell death/apoptosis and inflammation (e.g. elevation of caspases, pro-inflammatory cytokines, nitric oxide, reactive oxygen species, basic fibroblast growth factor, matrix metalloproteinases, aggrecanases, and release of matrix fragments); (2) an intermediate phase follows, when a potential balance between catabolic and anabolic responses may exist (catabolic responses subside and anabolic responses are initiated); and (3) a late phase with limited repair/remodeling/matrix formation (activation of anabolic growth factors). An understanding of these overlapping phases will provide a basis for identifying potential biologic targets for intervention to prevent later joint degeneration. Animal models, in a controllable fasion, provide a means to study the pathogenesis of post-traumatic osteoarthritis and its potential treatment. The ideal animal model would precisely mimic the disease identified in people and the horse offers several advantages over other animals used in human translational research. Horses are unique in that they are not solely translational models as they are also 47

48 vulnerable to the onset of osteoarthritis as a consequence of excessive joint loading. 140 Because osteoarthritis is such a common disease in horses, the pathophysiology and treatment of traumatic joint injuries are better understood than in any other animal model. Anatomically speaking, the equine joint is the closest approximation to a human joint in terms of cartilage and subchondral bone thickness compared to other species. 176 The similarity in relative joint size makes the equine joint amendable to the same diagnostic imaging modalities available to people such as digital radiography, computed tomography, magnetic resonance imaging, nuclear scintigraphy, and ultrasonography. Outcome parameters such as serial synovial fluid analysis 177 and sequential arthroscopy 178,179 can also be assessed throughout a study. Compared to other species, larger tissue samples are also available for both ante-mortem and post-mortem examination. 4 Recommendations for macroscopic and histological evaluation of the equine joint have been reported. 180 Horses can be evaluated for lameness, both subjectively and objectively 181, allowing for evaluation of the symptom modifying effects of treatment. They can also be readily trained for supervised exercise, such as that on a high-speed treadmill. This is especially advantageous when a specified rehabilitation program is required. Finally, the equine genome was recently sequenced and the similarity between the horse and human genomes is closer than for other animals used for translational research. 182 There are also disadvantages in using the horse as a translational model of posttraumatic osteoarthritis. First, the horse raises significant ethical concerns because it is considered a companion animal. Second, it is difficult to obtain large numbers for a homogenous group because horses are not typically bred for biomedical research. 48

49 Because of the first two issues, the typical horses that are obtained for research are from the racetrack and must be screened for musculoskeletal issues prior to inclusion in the study. Horses are also expensive to purchase and house than other animals available for research, and require special facilities to both house and perform surgical procedures. The surgical procedures are often more technically challenging and require more advanced equipment compared to other species. Nevertheless, factors such as joint size, naturally occurring disease in this species, and the ability to evaluate them similarly to humans makes the horse an attractive model for post-traumatic osteoarthritis. There are three variations of the carpal osteochondral fragment model of osteoarthritis that have been tested which include (1) autogenous implantation of an osteochondral fragment from the lateral trochlear ridge into the middle carpal joint 183 ; (2) creation of an 8mm intra-articular fracture on the dorsodistal aspect of the radial carpal bone 5 ; and (3) creation of an 8mm intra-articular fracture on the dorsodistal aspect of the radial carpal bone with debridement of the subchondral bone using a motorized bur. 6 Results of the first study showed that the free-floating autogenous osteochondral fragments caused mechanical damage to the articular cartilage and that the fragment itself elicited severe synovitis. 183 Although this model causes extensive damage to the articular cartilage throughout the middle carpal joint, the pathological changes are not likely to be consistent between subjects if the autogenous osteochondral fragment is the cause of the mechanical damage to the articular cartilage. In contrast, it is also possible that the autogenous osteochondral fragment elicited an inflammatory response as the degree of synovial inflammation was similar to other chemically-induced models of 49

50 osteoarthritis. 157,184 Although this model gives evidence to remove free-floating osteochondral fragments (which rarely occur naturally), the model has no other clinical relevance to naturally occurring disease and is likely too severe to show any benefit from therapeutic intervention. The second osteochondral fragment model involves the creation of an osteochondral fragment on the distodorsal aspect of the radial carpal bone using an 8mm curved bone gouge. 5 This model of osteoarthritis is more clinically relevant as the fracture is created in a similar location to where fractures commonly occur in racehorses. 185 These fractures are thought to occur secondary to overextension of the carpus. The fracture is also created where an intra-articular ligament or plicae attaches to the radial carpal bone so that it prevents the osteochondral fragment from freefloating in the joint and causing secondary mechanical damage. 5 Variations in fracture dimension and number of fragments created likely contributes to the risk of developing post-traumatic osteoarthritis, as it will affect the contact stresses applied to the opposing joint surface through changes in incongruity of the articular surfaces. 74 Even though this model is not supposed to create instability, there is likely microinstability created by the reported 8mm osteochondral fragment and any increase in size of that fragment likely increases the degree of microinstability in that joint. Variability also exists in the time it takes for the osteochondral fragment to reattach to the parent bone. Longer healing times may increase the amount of time there is microinstability within the joint. Any instability created in the joint increases the forces on the rest of the articular cartilage, subchondral bone, and soft tissue supporting structures. It is unclear, however, which dimensions of the fracture are most important in the overall risk of post-traumatic 50

51 osteoarthritis and it is more likely associated with the overall surface area of the osteochondral defect. There is also variability in the energy needed to create these fragments and is likely dependent on the degree of subchondral bone sclerosis present within the radial carpal bone. It has been shown that the duration and magnitude of load that is applied to the joint surface is directly proportional to chondrocyte cell death and thus the risk of developing post-traumatic osteoarthritis. 86,186 The third osteochondral fragment model of osteoarthritis varies from the second only by the addition of mechanical debridement of the osteochondral fracture bed with a motorized burr. Debris created during creation of the fracture is not flushed from the joint with either model, however, by debriding the parent bone with a motorized burr, the later model increases the amount of debris present in the joint and increases the surface area of the joint defect. The purpose of mechanical debridement is to increase the time it takes for the osteochondral fragment to heal back to the parent bone and thus increase the severity of pathology seen. 6 However, in addition to the increase in pathology identified surrounding the fracture bed, articular cartilage pathology is also identified throughout the joint. 180 The purpose of this model is to not induce instability, with the intention of keeping most of the pathological changes adjacent to the intraarticular fracture. The cause of the widespread articular cartilage pathology is likely secondary to mechanical damage or greater degree of synovial inflammation caused by the increase in free-floating osteochondral debris. Although less likely, another potential cause is that the larger osteochondral defect in the parent bone increased the degree of microinstability of the jointm, thereby predisposing certain areas of the joint to be overloaded. It is likely that all three causes combine to increase the severity of 51

52 pathological change within the joint. The number of osteochondral fragments that have reattached has not changed significantly on post-mortem examination when comparing the two models. Medical Treatment of Osteoarthritis Joint injuries are the most common cause of lameness in sport horses and represent a significant portion of the caseload that is presented to the equine clinician. 36 Although osteoarthritis can originate from various causes, trauma and associated synovitis are the most common. 1 Treatments for osteoarthritis are intended to minimize pain or lameness, to inhibit or minimize the pathological progression of disease, and return joint function as quickly as possible. 73 This section focuses on the medical management of osteoarthritis, however, it should be emphasized that timely removal of osteochondral chip fragments, appropriate reduction and fixation of intra-articular fractures, accurate diagnosis of ligamentous and meniscal injuries, and appropriate treatment of osteochondrosis dissecans are also critical to preventing osteoarthritis. 73 Nonsteroidal Anti-Inflammatory Drugs Anti-inflammatory agents that inhibit a component of the enzyme system that converts arachidonic acid into prostaglandins and thromboxanes are called nonsteroidal anti-inflammatory drugs (NSAID). All cells, including chondrocytes and synoviocytes, possess arachidonic acid as a fatty acid constituent of phospholipids. Once released, arachidonic acid is oxidized by either cyclooxygenase or 5-lipoxygenase leading to production of prostaglandin or leukotrienes, respectively. NSAIDS effect the production of prostaglandin primarily by inhibiting cyclooxygenase (COX). 187 Two isoenzymes of COX have been identified. COX 1 enzyme is constitutively produced and has been termed the housekeeping enzyme because it is responsible for producing 52

53 prostaglandins involved in regulating normal cellular processes, such as maintaining normal gastric and renal function, vascular homeostasis, and coordinating the actions of circulating hormones. 188 It also plays a lesser role in the inflammatory cascade. COX 2 enzyme has been primarily associated with the inflammatory response and may play a smaller role in normal physiological processes. 188 This may explain, in part, the variability of efficacy, as well as toxicity, of different NSAIDs. These differences likely explain the variability in efficacy and toxicity of different NSAIDs as each NSAID shows variability in its COX 1 or COX 2 selectivity. 187 Because prostaglandins are believed to be involved in pain, altered cartilage metabolism, and ongoing inflammation in damaged joints, NSAIDs have been the mainstay of treatment of joint disease in horses for decades. 188 They are typically used in cases of acute joint injury in conjunction with other treatments. 187 Toxic side effects of NSAIDS (renal and gastrointestinal side effects), however, have impeded their longterm use in the treatment of joint disease. These side effects are most commonly associated with COX 1 inhibition, although there is some evidence to suggest that complete inhibition of COX 2 expression may not be optimal for the horse or the joint either. COX 2 has been shown to play a function in bone resorption and there is evidence to suggest that the brain, kidney, and pancreatic islet cells constitutively express COX 2 as well. 188 With that being said, the beneficial effects of COX 2 inhibition are still considered ideal for treatment of osteoarthritis. In addition to gastrointestinal and renal side effects, there is also evidence to suggest that prostaglandin inhibition may have detrimental effects on cartilage metabolism. Serum from horses treated for 14 days with phenylbutazone was tested on articular cartilage explants ex vivo and showed 53

54 decreased proteoglycan synthesis to a degree similar to the effects with human recombinant interleukin-1β. 189 Effects of phenylbutazone on bone formation in horses has also been evaluated by obtaining unicortical bone biopsies from the tibia in 1 to 2 year old Thoroughbred horses. 190 There was a decrease in mineral apposition rate in phenylbutazone treated horses and significantly more filling of the cortical defect in the control horses. Horses in previous two studies were treated for a minimum of 2 weeks with a dose of phenylbutazone (4.4 mg/kg twice a day) that exceeds the use in clinical practice (4.4mg/kg once a day or 2.2 mg/kg twice a day) and so a direct comparison should not be made. Oral administration of phenylbutazone using clinically acceptable doses (4.4 mg/kg twice a day for 3 days followed by 2.2 mg/kg once a day for 7 days) showed no significant differences between control and treated horses when evaluating biomarkers of cartilage and bone metabolism. 191 Effects of long-term administration of NSAIDs on cartilage and bone metabolism, however, should be considered when used in the athletic horses over extended periods of time. Phenylbutazone remains the most widely used NSAID in the treatment of joint disease due to its efficacy, availability, and affordability. Classified as an enolic acid, it is highly protein-bound in plasma (>99%), metabolized in the liver, and excreted in urine. 187 It can be administered orally or intravenously and is irritating if injected perivascularly. Phenylbutazone is relatively non-toxic at a dose of 2.2 mg/kg twice daily, but toxic effects can be seen with higher doses include neutropenia, hypoproteinenia and eventually death due to gastrointestinal ulceration, renal papillary necrosis and vascular thrombosis. 192 Oral administration of phenylbutazone may also cause oral ulcerations. 192 Plasma half-life following a 4.4 mg/kg intravenous dose of 54

55 phenylbutazone is 5.5 hours but increases in horses greater than 10 years old and decreases in ponies and younger horses. 193 Prostaglandin production is also reduced in inflammatory exudate due to a prolonged half-life and is thought to be due to its high degree of protein binding. 194 Oral absorption of phenylbutazone is affected by presence of feed in the stomach. Plasma levels peak at 13 hours with feed in the stomach compared to 6 hours following fasting. 187 Effects of phenylbutazone on lameness were evaluated using an experimentally induced synovitis model. Following pretreatment with phenylbutazone, a significant reduction in lameness was evident for 8 to 10 hours as well as a reduction in joint temperature, synovial fluid volume, and synovial concentrations of prostaglandin E Similar findings were found following phenylbutazone administration in an experimentally induced lameness model using an adjustable heart bar shoe. 196 There is no benefit in doses > 4.4 mg/kg in reducing pain in horses with chronic lameness; reductions in lameness scores were not seen until 24 hours in horses administered 8.8 mg/kg compared to 6 to 8 hours in hours administered 4.4 mg/kg. 197 Mechanism of pain reduction was evaluated by looking at thermal invoked hoof withdrawal reflexes to determine whether analgesic effects were due to decreasing nociceptive pain or secondary to anti-inflammatory effects. At a dose of 7.3 mg/kg, phenylbutazone did not alter normal cutaneous pain perception and so the authors concluded that phenylbutazone s analgesic effects were secondary to its anti-inflammatory effects. 198 NSAIDs do not block pain but primarily reduce the hyperalgesia associated with pain by mediating peripheral afferent discharge from hyperalgesic areas as well as having central antihyperalgesic effects. 188 Local effects of phenylbutazone in the horse were 55

56 identified in an acute lipopolysaccharide synovitis model. 199 In the placebo treated group, there was upregulation of the μ-opiod receptor in the synovial membrane, which was not apparent in phenylbutazone treated horses. Phenylbutazone is also indicated in post-operative management of pain as it has been shown to decrease total postoperative pain severity indexes compared to saline. 200 There were, however, no differences in plasma β-endorphin, catecholamine concentrations, or anesthetic recovery times. Flunixin meglumine (Banamine ) is the second most commonly used NSAID to treat musculoskeletal disorders in horses. It can be administered orally or intravenously using the same dose (1.1 mg/kg) because of its very efficient gastrointestinal absorption. 187 It can also be administered intramuscularly, however, the method of administration has been associated with myonecrosis. Due to its relatively short half-life, peak plasma levels are reached within 2 hours and can persist for up to 30 hours. Similar to phenylbutazone, rapid accumulation of the drug at inflammatory foci is responsible for a longer duration of action versus its short half-life. Flunixin meglumine was shown to alleviate 74% of lameness in horses suffering from various musculoskeletal conditions. 201 In a study comparing the effects of flunixin meglumine, phenylbutazone, and saline in horses with navicular syndrome, both subjective and objective lameness scores decreased in the flunixin meglumine and phenylbutazone treated horses and no significant differences were seen between these two groups. 202 Concurrent administration of flunixin meglumine and phenylbutazone was compared with phenylbutazone administration alone in horses with varying types of naturally occurring fore and hind limb lameness. 203 Treatment with both NSAIDs alleviated 56

57 lameness more effectively than phenylbutazone alone, however, one horse did die of necrotizing colitis. Duration of post-operative analgesic effect was longer following flunixin meglumine (12.8 hours) administration compared to phenylbutazone (8.4 hours). 204 Firocoxib is currently the only selective COX 2 inhibitor that has been licensed in U.S to control pain and inflammation associated with osteoarthritis. Pharmacokinetic studies showed a plasma half-life of approximately 44 hours following intravenous administration and 36 hours after oral administration. It is primarily eliminated via hepatic clearance similar to other NSAIDs. 205 Effects of orally administered firocoxib (0.1 mg/kg) in the treatment of musculoskeletal pain or lameness associated with osteoarthritis (50.5%) were evaluated in 390 horses. 206 Lameness improved in 80% of horses by day 14 of administration and improvement in lameness was most rapid in the first 7 days following treatment, then slowly improved to day 14. Firocoxib was also compared to phenylbutazone in a randomized control trial in 253 horses with naturally occurring osteoarthritis. 207 Although there were no significant differences in subjective lameness scores, firocoxib treated horses showed greater improvement for pain on manipulation or palpation, joint circumference scores, and range of motion scores. A safety study evaluating the concurrent administration of firocoxib (0.1mg/kg) and phenylbutazone (2.2 mg/kg) showed significant changes in serum creatinine and total protein suggesting co-administration may cause renal disease. 208 A topical NSAID preparation (1% diclofenac sodium cream, Surpass) for use in horses in the U.S has clinical benefits but without the side effects associated with systemic NSAID administration. Anti-inflammatory effects were shown in an induced 57

58 subcutaneous inflammation model in which diclofenac was readily absorbed transdermally and reduced transudate concentrations of PGE2 at 6 and 30 hours. 209 Diclofenac and phenylbutazone were compared using the carpal osteochondral fragment-exercise model. 13 There were no significant differences between phenylbutazone and diclofenac sodium cream in reducing lameness scores on osteoarthritis-affected limbs, however, there was a significant decrease in lameness in sham-operated limb in the phenylbutazone-treated group. Phenylbutazone decreased synovial concentrations of PGE2 compared to control and diclofenac sodium cream treatment, however, diclofenac sodium cream showed significant improvements in Safarin O-Fast Green (SOFG) staining and cartilage glycosaminoglycan (GAG) content compared to the other two groups. Glucocorticoids Glucocorticoids are powerful anti-inflammatory agents that strongly suppress immune response. Glucocorticoid cell receptors are present in neutrophils, lymphocytes, monocytes, and eosinophils and possibly mediate all glucocorticoid effects. 210 Specifically, neutrophil function and, to a lesser degree, movement are effected by glucocorticoids. 210 Lysosomal enzymes and neutrophilic phagocytosis can be negatively affected by higher doses of glucocorticoids but more physiological steroid doses may produce a less profound negative a response by the cellular immune system. Prostaglandin inhibition by takes place at the beginning of the inflammatory cascade by inhibition of phospholipase A2 via the steroid-inducible group of proteins called lipocortin. 210,211 These proteins inhibit phospholipid hydrolysis by phospholipase A2, which prevents membrane phospholipids from mobilizing arachidonic acid. As a result, fatty acids, such as arachidonic acid, can no longer be oxygenated by 58

59 cyclooxygenase and lipoxygenase enzymes to form eicosanoids. 211 Glucocorticoids may also play a role in inhibiting NF-κB, an intracellular transcription factor involved in the production of inflammatory cytokine production. 211 Through the use of high-performance liquid chromatography, glucocorticoids can be detected within the joint following administration. Concentrations of methylprednisolone acetate can be detected in plasma for 6 to 8 days and in synovial fluid for up to 39 days. 212 Triamcinolone acetonide can detected in plasma for up to 48 hours and in synovial fluid for 15 days. 213 Pharmacokinetics, however, should not be confused with pharmacodynamics. It was originally believed that duration of response after corticosteroid injection correlates inversely with the solubility of the corticosteroid preparation in water. 214 However, rate of hydrolysis of the drug by synovial membrane enzymes or binding affinity of the corticosteroid to the steroid receptor in the cytoplasm of target cells may play a role. 214 A combination of different factors including the dose administered, duration of action and treatment, rate of conversion to the biologically active metabolites, crystal size of the suspension, and other unknown factors may play a role. 214 Based on clinical observation, triamcinolone acetonide has a medium duration of action, betamethasone acetate has a medium to long duration of action, and methylprednisolone has a long duration of action. 187 Early reports high and repeated use of glucocorticoids in horses described deleterious effects on articular cartilage. However, the dosage and frequency of administration were higher than typically used clinically The high concentration and extended exposure used in these studies may have resulted in the detrimental cartilage effects. Since then, more positive reports regarding glucocorticoid effects on 59

60 articular cartilage have been published using the carpal osteochondral fragmentexercise model and will be reviewed below. Three detrimental side effects of intraarticular corticosteroid therapy include steroid arthropathy, postinjection flare, and joint infection. 187 The concept of steroid arthropathy originated in human medicine following reports of accelerated osteoarthritis after glucocorticoid administration. It is more likely, however, that this condition represents a normal or occasional accelerated progression of osteoarthritis that is unrelated to glucocorticoids. Postinjection flare is an inflammatory reaction after intra-synovial injection of glucocorticoid and may be caused by the microcrystalline characteristics of the preparations. Incidence is > 2% and is usually self-limiting with appropriate therapy. Potential for infection following corticosteroid injections is of great concern for practitioners. Although incidence is low, joint infections can be devastating and expensive to treat. Strict asepsis should be followed with all intra-articular injections. Methylprednisolone, betamethasone, and triamcinolone are the three most commonly used glucocorticoids used for treatment of joint disease in the horse and have been evaluated using the osteochondral fragment-exercise model. The original model used an osteochondral fragment on the distodorsal aspect of the radial carpal bone. 5 Betamethasone was the first drug examined and showed no detrimental effects either in control or exercised horses. 5 The dose of betamethasone used in this study was considered more physiologic (15.9 mg two weeks apart). In vitro work suggests, however, that betamethasone suppresses proteoglycan synthesis at medium-to-low physiological doses

61 The degree of damage caused by the osteochondral fragment alone may not be severe enough to see a difference between treated and control horses. Therefore, the original model was modified by using a motorized burr to debride subchondral bone adjacent to the osteochondral fragment leaving bone and cartilage debris within the joint. 8 Methylprednisolone acetate yielded a nonsignificant reduction in the degree of lameness as well as a significant reduction in PGE2 concentrations in the synovial fluid and lower scores for intimal hyperplasia and vascularity. 8 Modified Mankin scores, however, were worse compared to control horses indicating a detrimental affect on cartilage. In a study of repetitive administration of methylprednisolone acetate to exercising horses, mechanical integrity of articular cartilage was altered, however the subchondral bone was unaffected. 218 It has not been substantiated that the negative effects of methylprednisolone acetate are dose-related. 187 Triamcinalone acetonide showed different results in that treated horses were significantly less lame than control horses, had lower protein and higher hyaluronan and GAG concentrations in their synovial fluid, and modified Mankin scores were significantly better than control horses. 6 Interestingly, beneficial effects were also noted in the contralateral joint as well. Based on these findings and in vitro studies also showing a protective effect, triamcinolone acetonide is recommended for administration in high-motion joints. Chondroprotective Agents Chondroprotective agents used in the horse include hyaluronic acid, polysulfated glycosaminoglycan, and chondroitin sulfate. 61

62 Hyaluronic acid Hyaluronic acid, or hyaluronan, is a large unbranched nonsulphated glucosaminoglycan composed of repeating units of D-glucuronic acid and N-acetyl glucosamine and as such can have varying lengths. 219 It is a significant component of both the synovial membrane and articular cartilage and is synthesized by the Type B synoviocytes of the synovial membrane and chondrocytes in the articular cartilage. 219 Within the synovial fluid, hyaluronan serves various important functions, such as viscoelasticity to the joint fluid and boundary lubrication of the intra-articular soft tissues. 66 Hyaluronan may also influence the composition of synovial fluid through steric hindrance of active plasma components and leukocytes from the joint cavity. 219 Hyaluronan appears to modulate chemotactic response within the synovial membrane by reducing cell migration and decreasing rates of diffusion of solutes. 219 Hyaluronan is also an important component in articular cartilage matrix and provides the backbone of the aggrecating proteoglycan aggrecan. Compressive stiffness of articular cartilage depends on the integrity of matrix proteoglycans. Mechanisms by which treatment with hyaluronan is beneficial in the treatment of osteoarthritis are still speculative. Although studies differ in results, it is generally believed that both the concentration and molecular weight of hyaluronan is decreased in the synovial fluid of horses with joint disease. 219 It is theorized that exogenously administered sodium hyaluronan supplements, or even replaces, the actions of depleted or depolymerized endogenous hyaluronan in the synovial fluid, thereby restoring viscoelasticity, steric hindrance, and lubrication of articular soft tissues. 219 The half-life of exogenous, intra-articularly administered hyaluronan is 96 hours in the horse and has been shown to be reduced in joints with osteoarthritis. 219 Despite rapid clearance, a 62

63 portion of hyaluronan is believed to be localized in the intracellular space surrounding synoviocytes, influencing the metabolic activity of those cells. 219 Hyaluronan has also shown to have anti-inflammatory properties by inhibition of macrophage chemotaxis and reduction in lymphocytic proliferation and migration. 219 Hyaluronan may reduce the interaction of inflammatory mediators with target cells through steric hindrance. 219 Reduced chemotaxis and phagocytosis of activated neutrophils may be mediated through the interaction of hyaluronan with CD 44 cell receptors present on the membrane of neutrophils. 219 Hyaluronan has also been shown to decrease the formation and release of prostaglandins from macrophages during phagocytosis and to scavenge oxygen-derived free radicals. 219 There are also conflicting reports regarding the most effective molecular weight of hyaluronan to use, however, reports consistently show benefit with molecular weights greater than 500,000 Daltons. 219 Injecting hyaluronan into an osteoarthritic joint may result in increased synthesis of high molecular weight hyaluronan by the synoviocytes. This was supported by in vitro work in which hyaluronan preparations of molecular mass > 500,000 Daltons stimulated synthesis of hyaluronan by the synoviocyte in a concentration dependent manner. 219 Hyaluronan preparations that were < 500,000 Daltons had little or no effect. Most commercial products are between 1 and 3 million Daltons, but newer synthetic cross-linked exogenous products that are > 6 million Daltons are coming on the market. These products are believed to improve synovial fluid viscoelasticity, increases retention time, and are more resistant to free radicals. 219 The minimum effective dose has been determined in the horse (20 mg), however, frequency of administration has not. In humans, it has been suggested that a 63

64 minimum of 3 doses is required to see clinical improvement in pain and effusion. Numerous human studies using different formulations and frequency of hyaluronan administration show reduction in pain and effusion, preservation of cartilage volume, improvement in cartilage function, and greater improvement over corticosteroid therapy. 219 Numerous clinical reports have supported use of hyaluronan in equine joint disease with an overall favourable clinical impression. Hyaluronan was also evaluated using the osteochondral fragment-exercise model in which horses were administered three doses of mid-molecular weight hyaluronan at one-week intervals. 7 Treated horses had reduced lameness, less thinning of cartilage, and reduced surface fibrillation compared to saline-injected joints. It is uncertain whether hyaluronan acts only by inhibitory effects on inflammation or also has a secondary chondroprotective mechanism. Hyaluronan is also commonly injected with glucocorticoids in equine practice. A recent survey showed equine practitioners combined the two therapies 59% of the time. 220 In vitro studies of hyaluronan with triamcinolone acetonide showed a beneficial effect on proteoglycan matrix metabolism in the presence of IL-1 compared to either treatment alone. 221 However, a recent multicenter clinical trial in horses with various joint involvement showed no clinical benefit. 222 Intravenous hyaluronan is also popular in equine practice. Benefits may be due to its localization in the synovial membrane since plasma half-life is no greater than 5 min. The synovial membrane is highly vascularized and IV administration may allow greater exposure to synoviocytes than IA administered HA. 7 The effects of intravenous hyaluronan were also evaluated using the osteochondral fragment-exercise model. Treated horses had reduced lameness, better 64

65 synovial membrane scores, significantly lower synovial fluid total protein concentrations, and reduced synovial fluid prostaglandin concentrations compared to horses receiving saline. Polysulfated glucosaminoglycans Polysulfated glycosaminoglycans are glycosaminoglycans, primarily chondroitin sulfate, that are made from an extract of bovine lung and trachea modified by sulfate esterification. 223 The commercial product, Adequan, is marketed for intramuscular (500 mg) and intra-articular (250mg) use. Peak serum and synovial fluid concentrations occurred 2 hours after a single intramuscular dose (500 mg) in horses with measurable amounts found in articular cartilage. 224 In this study, there was no greater accumulation of polysulfated glycosaminoglycans in joints with an experimentally created articular cartilage defect compared to control joints. It has been theorized that polysulfated glycosaminoglycans form stable complexes with fibronectin and collagen fibers and are deposited in cartilage. 223 Although beneficial effects have been seen in synoviocytes and chondrocytes, the exact mechanism of action is largely unknown. Polysulfated glycosaminoglycans inhibit a plethora of degradative enzymes that contribute to osteoarthritic process including lysosomal elastase, cathepsins, lysosomal hydrolases, serine proteinases, neutral metalloproteinases, plasminogen activators, and inducible nitric oxide expression. 223 While the exact cellular mechanism of action remains unclear, inhibition of either synthesis or action of these enzymes would benefit osteoarthritic cartilage. Inhibitory effects on leukocyte migration and interleukin levels suggests an anti-inflammatory role. 223 Stimulation of hyaluronan, glycosaminoglycan, and collagen synthesis has been documented. 223 The only reported negative side effect of polysulfated glycosaminglycans is a slightly increased risk of joint infections following 65

66 intra-articular use compared to glucocorticoids and may be less likely to occur if administered with amikacin. 225 Polysulfated glycosaminoglycans are traditionally used when cartilage damage is presumed present, however, synovitis may also show good response. Intra-articular administration may be more beneficial than the intramuscular route. Intra-articular administration significantly reduces articular cartilage fibrillation, erosion, and chondrocyte death, and markedly improves glycosaminoglycan staining in a sodium monoiodoacetate induce carpal synovitis model. 226 It has also been evaluated using the osteochondral fragment-exercise model. 12 Treated horses showed significant reduction in synovial effusion, degree of vascularity, and subintimal fibrosis of the synovium compared to controls and a trend towards decreased cartilage fibrillation. Two studies, however, failed to show an effect of intra-articular polysulfated glycosaminoglycans on articular cartilage. 227,228 It has, however, shown positive effects clinically in horses that responded positively to intra-articular analgesia of the distal interphalangeal joint using 3 treatments at 8 day intervals. 229 There is less evidence for intramuscular administration of polysulfated glyclosaminoglycan. 230 The only significant difference identified when it was evaluated using the osteochondral fragment-exercise model was a reduction in serum levels of glycosaminoglycan levels 14 days following treatment. Despite these findings, it is still heavily used as a prophylactic treatment for joint disease. Polysulfated glycosaminoglycan and pentosan polysulfate are polysulfated polysaccharides that were previously used as anti-thrombotic-antilipidemic agents. Unlike polysulfated glycosaminoglycans, the use of pentosan polysulfate to treat joint 66

67 disease in the horse has just recently gained popularity. Pentosan polysulfate has been developed into two different salt derivatives: sodium pentosan polysulfate (IM) and calcium pentosan polysulfate (PO or IM). It has been licensed in Australia as Pentosan Equine Injection (250 mg/ml) and plasma concentrations peak 2 hours after intramuscular and subcutaneous administration. 231 Similar to the other chondroprotective drugs, the mechanism of action is not fully understood. Proposed mechanisms of action include preservation of proteoglycan content by promoting synthesis of large proteoglycans, inhibiting enzymes responsible for proteoglycan and collagen degradation, and increasing synthesis of tissue inhibitor of metalloproteinase-3 by synoviocytes and chondrocytes. 231 Because of its anticoagulant activity, pentosan polysulfate is believed to promote mobilization of vascular thrombi and lipids and improve blood flow. 231 This may be beneficial in compromised subchondral bone, where improved blood flow could improve osteocyte nutrition and reduce pain resulting from vascular hypoperfusion and increased intraosseous pressure attendant with vascular occlusion. 231 Due to the dose dependent increase in partial prothrombin time, doses of pentosan polysulfate up to 3 mg/kg should not be administered within 24 hours of highstress activities or where there is a risk of physical injury. 232 In vitro studies using equine chondrocyte monolayer cultures showed that treatment with pentosan polysulfate stimulated proteoglycan synthesis, however, an effect was not seen when used in cartilage explants. 233 Pentosan polysulfate improved but did not eliminate clinical signs of joint disease when used in racing Thoroughbreds with chronic osteoarthritis (2 to 3 mg/kg IM once a week for four treatments, then as required). 231 Pentosan polysufate was also evaluated using the osteochondral fragment- 67

68 exercise model of osteoarthritis. 15 In treated horses, cartilage fibrillation was significantly reduced with a trend for improved cartilage histology scores (articular fibrillation, chondrocyte necrosis, chondrocyte clustering, and focal cell loss). Chondroitin sulfate 846 epitope, a marker of chondroitin sulfate synthesis, was increased in both treated and sham-operated joints suggesting systemic upregulation of aggrecan synthesis. These results were favorable to intramuscular polysulfated glycosaminoglycan. 234 A combination of pentosan polysulfate with N-acetyl glucosamine, and hyaluronan was also evaluated using the osteochondral fragment-exercise model and showed improvements in total radiographic scores, total macroscopic scores, and microscopic cartilage pathology scores. 235 Although no effect was seen on lameness scores in either study, a clinical trial comparing pentosan polysulfate and polysulfated glycosaminoglycan in horses with clinical osteoarthritis showed a reduction in lameness scores in horses treated with pentosan polysulfate at two and six weeks, where as no significant differences from baseline examination were seen in the horses treated polysulfated glycosaminoglycan. 231 Chondroitin sulfate Chondroitin sulfate is a long-chain polymer of repeating disaccharide units containing N-acetyl galactosamine sulphate and glucuronic acid and is the most abundant glycosaminoglycan in aggrecan. It is most commonly isolated from bovine trachea and nasal septum, shark skeleton, and sea mussel. Glucosamine is an amino monosaccharide, a principle component of O-linked and N-linked glycosaminoglycans and is produced in the body by the addition of amino groups to glucose. Hyaluronan, keratin sulfate, and heparin sulfate are composed acetyl glucosamine. Studies evaluating the effects of chondroitin sulfate and glucosamine vary widely in study design 68

69 and hence there results. Despite this, chondroitin sulfate and glucosamine are believed to counteract cartilage degradation by inhibiting degradative enzymes such as collagenase, aggrecanase, and intermediary mediators such as nitric oxide, PGE2, and nuclear factor kappa B. 236 They may stimulate synthesis of extracellular matrix components by provision of substrates, upregulation of gene expression, stimulation of cellular receptors and cell signaling pathways, and inhibition negative intermediary messengers. 236 Beneficial effects on cartilage metabolism were observed with in vitro doses of 10 ug/ml to 25mg/mL. Chondroitin sulfate and glucosamine in combination were most effective. An in vitro study using equine cartilage showed no substantial negative effects when used in combination or alone. The most significant positive effect was reduction in glycosaminoglycan degradation in both normal and IL-1 conditioned equine cartilage. 237 The highest dose of chondroitin sulfate (250 ug/ml) and glucosamine (250 ug/ml) in combination was most effective. Both chondroitin sulfate and glucosamine exhibit a high tropism for cartilage and therefore blood concentrations may not reflect tissue concentrations. 236 The oral bioavailability of chondroitin sulfate is 22 to 32%, depending on molecular weight. Oral glucosamine bioavailability is 2.5 to 5.9% with a large volume of distribution. 238 Based on these studies, it appears that chondroitin sulfate has an increased potential for efficacy because of greater oral absorption. It also appears the glucosamine sulfate is superior to glucosamine hydrochloride based on higher plasma and synovial concentrations. 239 Whether this equates to clinical efficacy is unclear. Joint inflammation increases glucosamine hydrochloride synovial concentrations (up to 4x) compared to 69

70 normal joints. 240 These synovial concentrations (1-18 um), however, are below concentrations that were shown to be efficacious in in vitro studies. 236 Oral nutraceuticals are fed to horses for two purposes: to treat lameness or for prophylaxis against joint problems. However, these products are unregulated and therefore of unproven efficacy or safety. A study of 23 commercially available equine oral joint supplements revealed that 9 of them contained less glucosamine than claimed by the manufacturer. 241,242 In addition, only 5 of the 23 supplements had recommended dosages previously suggested by researchers (10g/day). There is some, albeit small, evidence to support their clinical use. Cosequin (chondroitin sulfate and glucosamine) was administered to horses with clinical signs of osteoarthritis in the distal interphalangeal, metacarpophalangeal, tarsometatarsal, and carpal joints and resulted in improvements in lameness grade, flexion test grade, and stride length within two weeks. 243 Despite these initial findings, no further improvements were noted after two weeks. Also, no beneficial effects were identified in a chemically induced model of osteoarthritis. 228 Chondrosulf (glucosamine, glutamic acid, glycine, glucuronic acid) showed improvements in ground reaction forces in horses with osteoarthritis of the distal interphalangeal and tarsometatarsal joints. A third product, administered long-term and containing chondroitin sulfate and multiple forms of glucosamine showed decreased frequency of intra-articular injections in horses with clinical signs of osteoarthritis. 244 No significant effects were identified in geriatric horses administered chondroitin sulfate and glucosamine over a 3-month period using kinematic outcome criteria

71 Biological Therapies Autologous conditioned serum (ACS) is made by incubating whole blood with medical-grade glass beads exposed to chromium sulfate. It may have its main effect through up-regulation of interleukin-1 receptor antagonist protein (IRAP) as IL-1 is a major mediator of inflammation in joint disease. Its trade name, IRAP or IRAP II, however, is misleading because ACS contains many other components besides IRAP. Differential display electrophoresis and mass spectroscopy analysis of human and horse ACS reveals at least 35 different proteins were differentially regulated (greater than 2x) following processing. 246 Proteins such as IL-10, insulin-like growth factor-1, transforming growth factor beta, tumor necrosis factor alpha, and IL-1 have all been shown to be upregulated in equine conditioned serum using commercial kits. 246 IRAP is beneficial in the treatment of equine joint disease based on experimental evidence with gene therapy. 9 In humans, patients treated with ACS had a decreased pain scores both shortand long-term, decreased adverse events, and short-term decreases in IL-1β concentrations. 247 Equine reports are mostly anectodal with one controlled experimental study. In a recent survey, equine practitioners most commonly used ACS in cases unresponsive to corticosteroids. More stringent medication rules have resulted in increased use by show and racetrack veterinarians. Evaluation of ACS using the osteochondral fragment-exercise model (6 mls once a week for 4 treatments), showed a significant improvement in lameness, improved synovial membrane parameters, and significant reduction gross articular cartilage fibrillation compared to controls. 10 ACS increased endogenous IRAP throughout the study duration, suggesting prolonged beneficial effect. 71

72 Platelet rich plasma (PRP) is defined as a concentration of platelets that is significantly greater than in peripheral blood. A variety of preparation methods results in considerable variation in constituents of PRP. Platelet activation methods also vary. To compare different preparations, a classification scheme was proposed based on three factors: concentration of platelets, presence of platelet activators, and white blood cell concentration. 248 Although platelet numbers can be compared between preparations, however, this does not take activation into consideration. Increase concentration of platelets may be better, but concentrations > 6 times that in peripheral blood have negative effects on bone healing. 248 Platelets originate from division or fragmentation of megakaryocytes are devoid of a nucleus, and have a relatively short half-life (approximately 1 week). Beneficial growth factors (platelet-derived growth factor, transforming growth factor beta, endothelial growth factor) are released upon platelet activation and degranulation of alpha granules. Platelet contain over 200 proteins, all of which may not be beneficial. Platelets have been shown to synthesize proteins for about 2 days following activation, some of which are proinflammatory including IL-1, TNF, and IL Three methods of activation include: endogenous activation, calcium chloride, and thrombin. In an equine study comparing the effects of each activator, endogenous activation and calcium chloride were associated with less discomfort following administration of PRP and were considered safe to use. 250 PRP did incite a mild to moderate inflammatory response in synovial fluid in normal joints that lasted approximately 1 day. This effect was minimized by not using an activator. 250 Unwanted effects of thrombin included increased synovial fluid total protein and white blood cell counts, effusion, pain on flexion, reluctance to bear weight when the contralateral limb 72

73 was picked up, and periarticular heat and swelling. 250 Optimal concentration of white blood cells is unknown, however, deleterious effects have been identified with high numbers but not low numbers. 251 The ideal preparation of platelet rich plasma is not known. Different preparation methods results in different composition, which affects treatment efficacy. The use of PRP to treat joint disease is small based on very few reports in both humans and in horses. The conclusion from two systematic review in humans were that PRP significantly improved functional outcome, its effectiveness and duration of effectiveness is likely superior to hyaluronan, and there were a similar number of adverse reactions as hyaluronan (9.59%). 247 Number of injections did not indicate a clear dose-response relationship but the minimum of three treatments was suggested. An autologous protein solution in which blood was processed using a proprietary system resulted in a 12 fold increase in white blood cells, 1.6 fold increase in platelets, and an increase in various other proteins (3x). 252 Lameness scores were significantly reduced 2 weeks after injection and clients reported a satisfactory outcome at 12 and 52 weeks. 252 Mesenchymal stem cells are pluripotent, self-renewing cells with the potential for tissue regeneration and have been implicated in repair of bone, cartilage, tendon, ligament, skeletal muscle, and cardiac muscle. 253 They can be derived from several different adult mesenchymal tissues including amniotic fluid, umblical cord, bone marrow, adipose tissue, synovium, synovial fluid, and periodontal ligament. Mesenchymal stem cells have a bimodal effect on the immune system, including an anti-inflammatory and immune-enhancing response. Many different isolation and culture 73

74 methods have been used in their isolation. A consensus on the definition and properties of mesenchymal stem cells after in vivo expansion has recently been reached for human mesenchymal stem cells and includes ex vivo adherence to tissue culture plastic, multilineage differentiation, and specific immunophenotype. 254 The two most common sources of mesenchymal stem cells are bone marrow and adipose tissue. While adipose-derived mesenchymal stem cells have the ability to differentiate into musculoskeletal tissue they appear inferior to bone marrow-derived mesenchymal stem cells given the current understanding of different tissue Adipose tissue yields more mesenchymal stem cells than bone marrow, but culture expansion of bone marrow-derived cells greatly increases their numbers. Low volume (5mL) aspirates from the equine ilium appears to be a better source of mesenchymal stem cells than sternum. 256 At this time, the dose of mesenchymal stem cells is largely anecdotal. However, a study in humans revealed that 50 million cells performed as good or better than 150 million cells. 258 The timing and number of injections is also not clear. Mesenchymal stem cells injected into acutely traumatized or inflamed tissue fare badly, although it is possible that the anti-inflammatory effects might positively influence the outcome. Significantly better outcomes may be seen when treatment is given after the acute inflammatory phase of injury. 259 It is possible that stem cells therapy may be given too early or too later to participate in tissue regeneration. 260 Mesenchymal stem cell therapy is directed toward cartilage resurfacing, osteoarthritis, and intra-articular soft tissue injuries. This review will focus on the latter two. Bone marrow-derived cultured stem cells (1 million) were administered to rabbits 12 weeks following anterior cruciate ligament transection. 261 At twenty weeks, treated 74

75 rabbits had significantly less cartilage degeneration, osteophyte formation, and subchondral bone sclerosis compared to controls. Similar beneficial results have been identified in other laboratory models of osteoarthritis. 258 Recently, a comparison between adipose-derived stem cells, or stromal vascular fraction (16.3 million), and bone marrow-derived culture expanded stem cells (10.5 million) was performed using the osteochondral fragment-exercise model. 14 Synovial PGE2 concentrations were reduced in horses treated with bone marrow-derived culture expanded cells. The stromal vascular fraction treatment, however, induced an increase in tumor necrosis factor alpha. Poor results of this study may have been due to a suboptimal dose, frequency, and/or timing after induction of disease. 14 Intra-articular administration of mesenchymal stem cells has caused moderate inflammatory reactions which resolved by 48 hours after injection. 262 Treatment of soft tissue injuries within the joint with mesenchymal stem cells has been more encouraging. In a goat model in which the anterior cruciate ligament was transected and the medial meniscus was resected, autologous bone marrow-derived mesenchymal stem cells (6 million) were administered 6 weeks after induction of osteoarthritis. 263 Indices of articular cartilage degeneration and subchondral bone sclerosis were reduced in the treated group 6 weeks post treatment. Regeneration of a neomeniscus-like tissue was seen in 7 of 9 treated joints at 20 weeks. Effects of synovial-derived stem cells (10 million cells injected at 2 weeks) on partial medial menisectomy was studied in rabbits. 264 Labelled stem cells were found in the area of the removed meniscus at 14 days. Neomeniscal tissue was identified filling the defect at 4 and 12 weeks in the treated rabbits and also at 16 and 24 weeks in the nontreated 75

76 groups. Neomeniscal tissue in the untreated rabbits, however, was inferior to the treated group at all time points. The cartilage score of treated rabbits was also 5x better than the nontreated group. A report in horses with a broad range of stifle injuries also showed promising results. In horses treated with 15 to 20 million bone marrow-derived culture expanded stem cells, 76% returned to work with 43% returning to full work. 265 In addition, horses treated for primarily meniscal injuries (75%) were significantly more likely to return to work. Although there were no controls in this study, comparison to previous reports suggests superior results. 266 Potential for the Use of Gene Therapy Musculoskeletal injuries are common in equine athletes. Severity of injury, including inflammation and damage to surround tissues, affects the degree of spontaneous repair and balance between anabolic and catabolic processes. 267 Therapeutic approaches to either enhance anabolic response or limit catabolic response are sought to restore tissue in a more natural, functional, and permanent manner. 268 Gene therapy allows delivery of gene products to sites of tissue injury to facilitate endogenous repair and regeneration of orthopedic injuries. Recombinant gent product therapy is limited in its biological influence on tissue repair due to rapid clearance, requiring repeated treatment. 269 In contrast, gene transfer offers the advantage of sustained and regulated local synthesis of one or more gene products. 267 In addition, gene transfer allows the local synthesis of nascent proteins that have undergone authentic post-translational modification, which leads to greater biological activity and reduces concerns over possible immunogenicity. 267 Gene transfer also facilitates delivery of products with an intracellular site of action, for example transcription factors, intracellular signaling molecules, and sirna

77 Although systemic gene delivery has been evaluated, local gene therapy for joint disease has been promising. 270 Local approach concentrates the therapeutic gene produced specifically where it is needed within the joint, limiting exposure of nonaffected tissue. Fibroblastic cells that line the synovium are readily manipulated by genetic modification and the dense collagenous matrix of the capsule limits diffusion of vector from the joint. 270 Proteins do diffuse in and out of the joint, but the semi-closed system results in accumulation of gene products within synovial fluid and ultimately diffusing into cartilage and capsular tissue. 270 Local gene therapy research has focused on: 1) evaluation of methods and vectors of gene delivery to articular tissue, and 2) evaluation of candidate gene products for efficacy in models of disease. 270 There are ex vivo and in vivo methods of gene transfer. In vivo methods introduce vectors directly into the body, where as ex vivo methods involve implanting previously modified cells. 267 In vivo methods are more economical and efficient in the genetic modification of articular cells, but safety concerns are greater because direct control over their activity is not possible after injection. 267 In healthy tissue, fibroblastic cells typically compromise the vast majority of cells that are genetically modified following intra-articular injection because the dense extracellular matrix effectively excludes most viral vector systems from chondrocytes. 270 Successful transfection of cells also requires the existence of a sufficient population of healthy cells within the damaged tissue in order to take up and express the transgene endogenously at appropriate levels. 267 In patients with advanced osteoarthritis where there is usually considerable cell death, this requirement is not always met. In ex vivo gene transfer methods obviates these problems because cell populations are removed from the body; 77

78 then expanded, modified, seeded on a scaffold, and incubated in a bioreactor; and then finally reimplanted. 267 Prior to reimplantation, these genetically-modified cells can be pre-screened for adventitious agents, levels of transgene expression and potential tumorgenicity. 270 Furthermore, no unbound viral particles or other vector systems are delivered to the recipient. 270 Such technologies are, however, more expensive as it typically requires two surgeries to harvest and then reimplant the genetically-modified autologous cells. Work is being done to try and expedite this process so that it could be performed during the course of one procedure. 267 The use of allogeneic cells is an alternative method, however, the joint is highly immune sensitive and these cells have not performed well in the research setting. 270 Vectors are the vehicles that facilitate the translocation of genes (usually cdna) into the target cell, with transport to the nucleus where gene expression is initiated. 269 Non-viral vectors can be as simple as DNA plasmids, but are usually associated with liposomes or various types of polymer to enhance uptake. 270 Transfection efficiency can also be enhanced by physical stimuli such as electropolation and sonication. 270 Compared to viral vectors, they are less expensive and easier to construct but the transgene expression that they produce is sub-therapeutic and persists for short periods of time. 270 They also induce an inflammatory reaction at doses required to achieve measurable protein expression. 270 In contrast, viral vectors are very capable of efficient gene delivery and provide biologically relevant levels of transgene expression. 267 Viral vectors take advantage of their innate ability to deliver their own genomes efficiently to the cells they infect. 267 The creation of viral vectors involves removing the sequences of the viral genome that contribute to disease and replacing them with genes of interest 78

79 and their regulatory sequences, while still maintaining infectivity. 267 Although viral vectors are more efficient, several considerations need to be made before their clinical use. Biological considerations include the carrying capacity of the vector, the length of time its genome will persist in the body to sustain transgene expression, and the degree to which it generates a neutralizing immune response. 267 Safety concerns have centered on whether or not the vector inserts (integrates) its genetic material into that of the host cell, a process that can lead to insertional mutagenesis and cancer. 267 Immune responses to the vector can limit the duration of transgene expression and prevent repeated dosing. 267 Other considerations include its ease of manufacture and costeffectiveness. There are several different viral vectors that have been used successfully in animal models for the treatment of osteoarthritis, including retroviral (oncoretrovirus) vectors in ex vivo fashion, adenovirus, adeno-associated virus, herpes simplex virus, and lentivirus for in vivo delivery. 271 Retroviruses are enveloped viruses that contain two identical copies of their RNA genome that can be reverse transcribed into doublestranded DNA but are subsequently randomly incorporated into the genome and represents a disadvantage to its use due to tumorgenicity. 272 Advantages of retrovirus include long-term stable expression and integration into the genome. 272 Because of the need for host-cell division, retroviruses are normally used in ex vivo fashion. 271 Adenoviruses are a non-encapsulate double-stranded DNA virus that quickly infects non-dividing cells and consequently produces high titers quickly. 272 This makes adenovirus attractive for use, however, they also stimulate antigen-specific cellular immunity which leads to short (1-2 week) transgene expression. 272 Herpes simplex virus 79

80 have an enveloped double-stranded, DNA genome that are very efficient at infecting both non-dividing and dividing cells but are cytotoxic which limits its long-term transduction. 271 The Lentivirus is another type of retrovirus but only infect non-dividing cells. 271 Because they are incorporated randomly into the host s genome, there is a serious risk of insertional mutagenesis and will likely only be used in animal models. 270 The adeno-associated virus is a promising gene transfer vector because it is not associated with any human diseases. 272 These non-enveloped, single-stranded DNA viruses deliver transgenes to a wide variety of tissues, have low immunogenicity, and mediate long-term gene expression. Although the wild-type single-stranded DNA viruses provide only marginal levels of transgene expression, double-stranded viral genome vectors have been designed that allow rapid and highly efficient transduction and > 20 fold enhancement of gene expression. 272 Because the synoviocyte is the most commonly used target of gene transfer, choices of gene products can be broken down into two categories: 1) those that lead to the synthesis of diffusible, chondroprotective products (IGF-1 and TGF-β), and 2) those that reduce the synthesis of diffusible damaging synovial products (IL-1Ra). 268 Sufficient amounts of IGF-1 were produced to increase matrix synthesis by articular cartilage in experiments using recombinant adenovirus encoding IGF-1 in normal and inflamed knee joints of rabbits. 273 Similar findings were found in experiments using TGF-β, however, its overexpression from the synovium caused massive fibrosis and the formation of cartilage nodules within the synovium and certain animals died. 274 Synovial hyperplasia and osteophyte formation were also identified in a mouse model. In the horse, full-thickness cartilage defects were created in the lateral trochlear ridge of the 80

81 femur and subsequently filled with chondrocytes transfected with equine IGF-1 cdna using an adeno-associated virus. 275 Transduced chondrocytes resulted in significantly better healing at 8 weeks arthroscopically and better histological scores at 8 months including increased chondrocytes predominance and collagen type II. Mediation of IL-1 has shown great potential for inhibiting extracellular matrix breakdown and potential mediators of inflammation and pain in joints with osteoarthritis. Transfer of human IL-1Ra cdna using the ex vivo method two days after surgery in the canine ACL-transection model suppressed early degenerative changes of the tibial plateau and femoral condyles but the synovial changes may have been exacerbated. 276 The effects of the synovium, however, were less obvious and complicated by probable immune reaction against the foreign proteins expressed by the genetically altered synovia. Using the equine osteochondral fragment-exercise model of osteoarthritis, equine IL-1Ra cdna was transferred to the synovial lining using an adenovirus as a vector by direct injection 2 weeks after surgery. 9 Despite large cartilage erosions in the untreated joints, treated horses did not show visible cartilage lesions and revealed greater retention of proteoglycan as seen by histology. Synovial changes and lameness scores were also reduced in the treated horses. A combination of both inhibition of cartilage breakdown and promotion of cartilage repair was performed in an in vivo model using equine chondrocytes in normal cartilage and cartilage whose matrix was depleted by IL In this model, both IL-1Ra and IGF-1 was expressed in synovial fibroblasts by adenoviral gene transfer and then co-cultured with equine articular cartilage. The synovial expression of IGF-1 markedly enhanced matrix synthesis by 81

82 normal cartilage and it partially reversed the depletion of proteoglycan from the matrix of cartilage. Although most initial gene therapy efficacy studies are performed on small animal models of osteoarthritis, the successful scale-up from rodents to people is often a difficult challenge due to the differences in physiology of their respective joints. In vivo, patterns of cellular transduction and ensuing transgenic expression are a function of the biophysical interaction of the vector with the specific target tissue. 278 The direct translational benefit of small animal models of osteoarthritis to humans is hampered by the marked differences in joint volume, composition and surface areas of target tissue, the vascularity of tissue, the volume and composition of extracellular fluids, and compressive forces generated during locomotion. 278 Large animal joints, like the horse, are closer to people in all of the aforementioned aspects, which make them better translational models for gene therapy. 278 In addition, diagnostic imaging and clinical monitoring of response to treatment of osteoarthritis are similar between people and horses. Patterns of therapeutic transgene expression can also be monitored overtime due to the ability to sample larger volumes of undiluted synovial fluid. There is also a direct clinical benefit to performing research in this species due to the high percentage of lameness in athletic horses being attributed to osteoarthritis. Osteoarthritis Imaging Osteoarthritis is characterized by degenerative changes in cartilage, bone, and intra-articular and periarticular structures. Because osteoarthritis is a complex disease of the whole joint, it is important to assess all structures of the joint to further understand disease pathogenesis and progression. Imaging plays a crucial role in the diagnosis and management of osteoarthritis in the horse as imaging findings are used in conjunction 82

83 with clinical findings to identify pathological changes associated with osteoarthritis within these structures. Early identification of the pathological findings associated with osteoarthritis is crucial to improving both our understanding of disease progression as well as case management. 181 Although equine joint disease has traditionally been diagnosed using radiographs, recent advancement in diagnostic imaging capabilities has increased the number of ways in which we can evaluate the equine joint through the use of computed tomography, magnetic resonance imaging, nuclear scintigraphy, ultrasound, and arthroscopy. When lameness is present, these advanced imaging modalities are commonly employed when routine imaging modalities cannot identify the source of lameness. With these advancements comes the need to understand the fundamental differences that exist between each imaging modality in order to appreciate the individual strengths and limitations associated with each modality so that the appropriate imaging modality can be chosen. The purpose of this review is to evaluate the various imaging modalities available for osteoarthritis imaging and assessment, focusing on their utility as tissue-specific diagnostic tools for equine joint disease. Radiography In clinical practice, radiography still remains the most commonly used imaging modality for diagnosing equine joint disease because of its relative availability and costeffectiveness. Radiographic assessment of the joint includes the evaluation of the joint margins, joint space, subchondral bone, joint alignment, and the insertions of ligaments, tendons, and joint capsule. Due to the two-dimensional nature of radiographs, routine examination of joints consists of at least 2 and up to 8 projections plus any additional projections that may be needed to better define suspected pathology such as incomplete fractures. The acquisition of high-quality images of the joint requires multiple 83

84 radiographic projections of the joint while using the correct radiographic technique with the limb properly positioned. Because radiographs are a two-dimensional projection of a three-dimensional structure, they are subject to variability based on patient positioning. Poorly positioned radiographs may result from the horse standing with the limb not perpendicular to the ground, the cassette not parallel with the limb, or the X-ray tube not perpendicular to the x-ray cassette. The ability to duplicate these three factors from one radiographic examination to the next will better allow you to compare imaging results over time as well. The radiographic exposure should also have enough film contrast latitude to allow observation of bone and soft tissue outlines, and the film detail should be sufficient to demonstrate bone trabeculae. This is less critical when using digital radiographic systems as the exposure latitude is much greater than with film-screen systems. Three systems exist for obtaining radiographs: conventional film-screen, computed radiography, and digital radiography. Due to the reduction in cost, digital radiography is becoming increasingly more common in veterinary medicine. 279 It offers several advantages over film-screen radiography including the ability to view images electronically, ability to manipulate images after acquisition, and potentially a reduction in patient and personnel exposure to radiation. 280 Although digital radiography has less spatial resolution than conventional film-screen systems, digital radiography is superior in contrast resolution. 280 Some feel that that the improvement in contrast makes digital radiography similar or even superior to film-screen radiography in overall diagnostic capability. 281 In experimental studies, digital systems were superior to film-screen radiography for detection of subtle osseous lesions in some musculoskeletal modeling 84

85 studies , but similar in others. 285 In a study of the equine stifle, computed radiography was found to be superior to film-screen radiography in contrast, bone structure, and diagnostic value. 286 Despite recent advancements in radiographic technology, radiography is still limited in its ability to evaluate the joint as a whole organ because of its lack of detailed soft tissue detail. Radiography does give information about bone, joint space width, and periarticular soft tissue insertions. Radiographic abnormalities associated with osteoarthritis are identified based on changes in radiopacity and shape, and include: marginal osteophyte and enthesophyte formation, joint space narrowing, subchondral bone radiolucency, increased subchondral bone opacity (sclerosis), osteochondral bodies, joint capsule distension, and advanced remodeling or ankylosis. 287 A 40% change in bone density, however, is required for radiography to be able to detect these boney abnormalities making radiography relatively insensitive to the early stages of the disease process. 287 These boney abnormalities may also persist without ongoing clinical disease and it is for these reasons that we have failed to show a correlation between radiographic abnormalities and either the degree of lameness or the degree of cartilage degeneration detected arthroscopically. 178, In addition, joint space width, a surrogate for cartilage thickness, is highly dependent on limb positioning and radiographic technique. Slight variations in either one of these variables can affect the quantitative assessment and comparison of radiographic parameters over time. 295 Due to these limitations, the absence or presence of positive radiographic findings should not be interpreted as confirmation of a complete absence of clinical disease or the source of lameness, respectively. However, because radiography is the simplest and 85

86 least expensive method for imaging joints, it is still the first modality used to identify pathological features of osteoarthritis in the horse. Ultrasound Valuable information can be obtained from the ultrasound evaluation of equine joints despite it being better known for its ability to evaluate tendon and ligaments. It is considered a complimentary imaging modality to radiography, as it is able to morphologically assess soft tissue structures that are poorly or unable to be assessed by radiographs. 296 Ultrasound images are produced by sending high-frequency sound waves into tissue and listening for the return signal. The sound waves emitted from the ultrasound probe will penetrate tissue until they reach a tissue interface with different tissue densities or acoustic impedance. If the tissue interface is perpendicular to the path of the sound wave, a portion of the sound waves will reflect back to the ultrasound probe and the rest will continue deeper into the tissue along the same path. The amount of sound waves that are reflected back to the ultrasound probe is determined by the incident angle and the acoustic impedances of tissues and correlates to their echogenicity (different shades of gray) that is displayed on the ultrasound image. 297 The depth of a specific structure is calculated based on the amount of time it takes the sound wave to transmit, reflect, and be received back by the ultrasound probe. The unique properties of the piezoelectric crystals within the ultrasound probe are capable of both generating sound waves when they are stimulated by an electrical signal as well as generating an electrical signal when they receive the returning sound waves which is subsequently converted to a digital signal. 297 A variety of probes are available for musculoskeletal imaging and vary by footprint configuration and frequency. In general, 86

87 the depth of the structure to be imaged is the biggest determining factor when choosing between probes because the image resolution and depth are related to the frequency used. 297 Higher frequency probes have better resolution, but poorer sound penetration, and lower frequency probes can image deeper structures but the resolution is not as good. Images should be obtained with the highest frequency probe possible to obtain the best resolution of the structure of interest. The linear probe is most commonly used for musculoskeletal imaging because of its superior resolution and high frequency range (6 to 14 MHz). The resolution on the microconvex probes is not as good as the linear probes but they do have a small, curved footprint that allows them to fit into small and irregular spaces. The macroconvex probes have the lowest frequency, but the largest field of view, and is beneficial when imaging deeper structures. Ultrasound has become increasing more accessible due to the machines becoming increasingly more portable and affordable. Several technological advances have also helped to improve image quality including harmonic imaging and spatial compounding. 181 Overall, these functions work to reduce artifact and increase the spatial resolution of the image. 298 Other technological advances have been developed to identify pathology that might be challenging with conventional B-mode ultrasound. Elastography is one of these technological advances and is used to measure tissue strain (displacement). 299 Differences in tissue strain can be identified by examining ultrasound images taken before and after tissue compression. These differences can be used to detect pathological lesions that appear isoechoic to normal tissue on conventional B-mode ultrasound exams, as well as assess its ability to heal. Fusion imaging is another technological advancement that superimposes cross-sectional 87

88 imaging (CT or MRI) on to real-time ultrasound images. 299 Fusion imaging is beneficial because it allows you to identify the pathology by ultrasound so that it can be monitored over time and it can also be used to guide therapeutic interventions. Ultrasound is advantageous over other imaging modalities because it has the ability to evaluate soft tissue structures and bone surfaces in multiple planes in real-time without exposure to radiation. 296 The diagnostic accuracy of ultrasound, however, relies heavily on adequate equipment, limb preparation, and scanning skills of the ultrasonographer. 296 Artifacts are easily produced that compromise image interpretation and most commonly involve operator error or sound-tissue interactions. Common operator-induced artifacts include off-incidence imaging (ultrasound beam is not 90 degrees to the target structure), and improperly adjusted machine settings (power, gain, and focal zone position), which can significantly reduce image quality. Artifacts created by sound-tissue interactions may or may not be controllable and include acoustic enhancement (fluid-filled structures), refractive scattering (curved surfaces), reverberation (two or more reflectors within the sound path), and acoustic shadowing (very high or low acoustic impedance). The physical properties of sound also limit ultrasound s ability to evaluate certain structures of the joint due to anatomic barriers such as certain portions of the articular surfaces and intra-articular ligaments. Nuclear Scintigraphy Since its introduction into veterinary medicine 300, nuclear scintigraphy has become an established imaging modality for the evaluation of bone and joint disease in most equine referral hospital. The principles of nuclear scintigraphy involve using a gamma camera to detect gamma rays being emitted from a radioactive substance. The most commonly used radioactive substance is the radioisotope, technetium-99m. It has 88

89 a short half-life (6 hours) and is excreted almost entirely by the kidneys. 301 A pharmaceutical compound is bound to the radioisotope so that it localizes and distributes into a specific tissue of interest. Polyphosphonates are ideal for bone scintigraphy because they preferentially localize within bone and the most common polyphosphonate used for detecting changes in bone metabolism is methylene diphosphonate. 301 A complete examination consists of three phases: vascular, soft tissue, and bone. The three phases corresponds to when the maximal radiopharmaceutical concentration exists within each tissue type following intravenous administration. The vascular or flow phase is obtained 1 to 3 minutes after injection and the soft tissue or pool phase is obtained 3 to 15 minutes after injection. 301 The short time interval within each of the first two phases encourages localization of lameness prior to examination as only a select number of sites can be imaged. Increased radiopharmaceutical uptake on soft tissue phase images is useful in detecting areas of hyperemia or acute inflammation in muscle, tendons, and ligaments; however, it can also represent early uptake in bone. Early bone uptake can start as soon as 5 minutes after injection so soft tissue phase images should be acquired rapidly and compared to bone phase images to differentiate between these two sources of pathology. 181,301 Maximum uptake in bone can be seen as early as 60 minutes, however, bone or delayed phase images are not obtained until 2 hours after injection when unbound radiopharmaceutical has been cleared by the kidneys. 302 The radiopharmaceutical binds to the inorganic portion of bone on exposed hydroxyapatite crystals. These bindings sites are not found on normal bone but created by osteoblast activity in areas of active bone turnover. 302 The amount of uptake within bone is determined by the rate and 89

90 extent of osteoblastic activity within the bone as well as blood flow to the area. 301 Although the relationship of blood flow and radiopharmaceutical uptake is not linear, a certain amount of blood flow is required to deliver the radiopharmaceutical to bone for it to be able to bind. 303 There are two ways in which scintigraphic examination can be performed. The first method uses a handheld probe to quantify the emission of gamma rays at specific points in order to generate an anatomic histogram of radiopharmaceutical uptake. Although the probe is can be placed directly against the skin, is less expensive and requires substantially lower radioisotope doses, only quantitative assessment can be obtained. The second method uses a gamma camera that enables the transformation of data into a two-dimensional image. To summarize, gamma rays that are emitted from the patient first pass through a collimator, which blocks non-perpendicular gamma rays from entering the camera and degrading the image. Perpendicular-oriented gamma rays come into contact with the sodium iodine crystals, causing them to light up or scintillate with an intensity that is proportional to the energy of the incident photon. 302 Photomultiplier tubes coupled to the fluorescent crystals convert the light to an electric signal that is used to create an anatomic image. The anatomic image can then be assessed both quantitatively and qualitatively. Image acquisition time is based on either a specified length of time or until a certain number of counts is obtained. Good quality images require a certain number of counts, and in general, increasing counts per image improves image quality. 301 Either method may prove to be more beneficial in certain circumstances. However, when one limb is contributing substantially more counts than the contralateral limb, a count-based approach would prematurely terminate image 90

91 acquisition. A time-based approach is more appropriate in this situation as the acquisition time can be standardized allowing for meaningful comparisons between images. The body part-to-camera distance and camera position relative to the body part should also be standardized for accurate image comparison. Advancements in nuclear scintigraphy have centered on improving image quality and lesion detection, most of which are upgrades in its computer software. Motion correction software removes the effects of excess motion by acquiring images at one to two second intervals instead of a single, static acquisition. The images are then superimposed on one another and the images with similar spatial registration are retained and the outlining images are rejected. This produces an image with better resolution which is particular useful when imaging the axial skeleton. 304 Region-ofinterest calculations can now also be performed and allows for quantitative comparison between a similar anatomic region in the contralateral limb. Even though qualitative and quantitative scoring schemes have shown good correlation 53, quantitative assessment may improve the detection of subtle differences. 304 In addition, most software programs now have the capability to increase signal and remove artifact or noise. The digital filter will make subtle lesions more obvious but care must be taken when counts are high as it may increase the number of false-positives. 305 Recently, solid-state gamma cameras have been developed that use semiconductors instead of crystals and photomultiplier tubes. Semiconductors combine the function of crystals and the photomultiplier tube by directly converting a photon in a digital signal. This results in an increased sensitivity and spatial resolution compared to older systems

92 Nuclear scintigraphy is a highly sensitive but nonspecific imaging modality for the detection of equine joint disease. 307 It provides physiological information regarding bone turnover but it does not differentiate between pathologic and nonpathologic conditions and the images have poor anatomic detail. 308 The sensitivity and specificity of nuclear scintigraphy is largely a function of the quality of the image and can be affected by multiple factors including time from radiopharmaceutical administration to image acquisition, body part-to-camera distance, shielding, motion, time from injury to image acquisition, ambient temperatures and peripheral perfusion, and amount of background radiation. 301 Accurate interpretation of the scintigraphic examination may be improved by qualitatively evaluating areas of increased or decreased radiopharmaceutical uptake for their location, intensity, and character. 301 Becoming familiar with the patient s history can also increase the accuracy of interpretation because the patient s age, use, and breed as well specific details regarding the lameness, results of previous examinations, and administered treatments can all affect image results. Nuclear scintigraphy does, however, provide a functional evaluation of bone activity at the time of imaging and can provide a method for early detection of bone injury. 301 This represents an advantage over radiography because radiographs represents bone activity that has occurred in the past and it may take several weeks for scintigraphic evidence of bone activity to become visible radiographically. Conversely, abnormalities on nuclear scintigraphy may subside long before radiographic changes do so nuclear scintigraphy can be useful in determining the clinical significance of existing radiographic abnormalities. Computed Tomography Similar to radiography, computed tomography (CT) uses x-ray attenuation as the fundamental premise for image formation. 309 The x-ray generator tube, however, is 92

93 positioned across from a row of digital x-ray detectors in a circular gantry that spin around the patient obtaining multiple radiographic projections at intervals around the patient. A cross-sectional image is then produced by mathematical reconstruction of the data obtained by the detectors based on the degree of attenuation of the x-ray beam, or tissue density, by a process called filtered back projection. 310 Movement of the patient through the gantry or the gantry around the patient allows for this process to be repeated at multiple locations along the area of interest creating multiple, twodimensional, cross-sectional images or slices. The acquisition of data in the initial CT systems was performed by a process called axial scanning. 309 Each slice was imaged before the table moved to scan the next slice because the x-ray generator and detector were powered by cords so they would have to retract to their original position before scanning the next slice. Current CT systems now incorporate two technological advances that have greatly increased the speed to which examination are now being performed. 311 Sling ring technology negates the need for cords and enables helical scanning, whereby the x-ray generator and detectors are now able to spin continuously around the patient as they move through the gantry. Multirow detectors allows for multiple slices to be imaged during each rotation, the number of which is determined by the number of detectors. The increased imaging efficiency reduces the chance for patient motion during examination and allows for a greater area of interest to be examined at one time. 311 Each image is made up of pixels that are actually a two-dimensional representation of a volume of tissue, termed a voxel, where slice thickness determines the third dimension. 310 Each pixel represents the density of tissue within each voxel and 93

94 is recorded in Hounsfield units (HU). The slice thickness, which typically ranges from 1 to 10 mm, contributes to the overall contrast and spatial resolution of the image. 309 Contrast resolution is improved when images are obtained with thicker slices because a greater number of x-ray photons can be detected, however, the spatial resolution of the image is reduced. The reverse is true when thin slices are obtained. Contrast and spatial resolution can also be optimized at the time of data processing by using algorithms specific for bone (high-frequency) or soft tissue (low-frequency) postprocessing by the user through the manipulation of the window width and level. 311 The window width is the number of shades of gray that are portrayed on an image and the window level refers to the Hounsfield number upon which the window width is centered upon. 310 The number of shades of grey in an image is minimized because the human eye can only visualize shades of grey and in turn increases the likelihood of detecting a difference when there are subtle differences in contrast. 310 Therefore, if the user is interested in the evaluation of tissue with very little variation in tissue density, such as tendons or ligaments, a narrow window width should be used, where as a wide window width would be appropriate for bone. The brightness of an image can be manipulated by altering the window level and should be performed during image interpretation. 311 Both pre- and post-processing manipulation is used to improve the appearance of an image in order for the user to better evaluate specific structures more effectively. Images can be reconstructed in multiple two-dimensional planes or a threedimensional reconstruction despite only scanning perpendicular to direction in which the anatomy passes through the gantry. The quality of image reconstruction is dependent 94

95 upon the slice thickness used for acquisition and whether the axial or helical scanning mode was used. 310 Multislice scanners and helical scanning have allowed the acquisition of slices that are less than 1mm thick and overlapping tissue slices, respectively, which together increase the amount of information available for multiplanar re-formatting, reslicing images in planes other than the plane of acquisition, and threedimensional reconstruction. 311 Three-dimensional images are reconstructed by assigning a density value (HU) to each voxel. The three-dimensional image is then reconstructed by assigning a minimum density value to which can be displayed on the image. If surface rendering is used, only values above the threshold are displayed and only the surface is displayed. Volume rendering assigns a different color or shade to a density range, which allows for a different perspective to differentiate tissue with different tissue densities. The tomographic viewing prospective of CT offers an obvious advantage over radiography by providing cross-sectional images that are free of the projectional limitations. The bone detail achieved by CT is also superior to any other imaging modality due to its superior spatial resolution and is considered the imaging modality of choice for depicting cortical bone, intra-articular bone fragments, and soft tissue calcifications. 312 Valuable information can also be obtained about the subchondral bone, such as volumetric bone mineral density, which can provide information regarding the functional adaptation of the subchondral bone. 313 It does not, however, provide information regarding bone activity or bone fluid accumulation that can be obtained with nuclear scintigraphy and fluid-sensitive magnetic resonance imaging sequences. The soft tissue contrast provided by CT is also inferior to magnetic resonance imaging 95

96 making it difficult to identify subtle soft tissue abnormalities. Intravascular or intrasynovial contrast medium can improve CT s ability to detect soft tissue abnormalities by identifying tissues with abnormally increased or decreased blood flow or regions with altered vascular permeability. 311 The accumulation of contrast media in tissues by extravasation from abnormally permeable blood vessels can be identified in tissues that are injured or inflamed and as the lesion repairs itself, new blood vessels develop and contrast media is seen within small-caliber blood vessels. 314 When injected within a joint, contrast medium will outline the structures within the joint, which can aid in the marginal visibility of intra-articular soft tissue structures and permit an indirect evaluation of the overlying cartilage. When deciding to use CT as an imaging modality on equine patients, special considerations need to be made as the ability of CT scanner to exam a particular area of interest is limited by the size of the opening of the gantry and table configuration, which is particularly important when trying to evaluate the equine stifle. 311 The table must support the weight of the horse, and in systems where the CT gantry is stationary, the table must move at a predetermined speed with great precision so that the anatomy is accurately acquired. Computed tomography also requires special facilities, trained personnel, and in most cases the patient needs to anesthetized. The rapidity to which the CT scan can be performed compared to magnetic resonance imaging, however, makes it a more attractive imaging modality especially when a surgical procedure is to follow the examination. In addition, there are no safety concerns when using this modality to image patients with ferrous implants as there are with magnetic resonance imaging, although both modalities will result in metal artifacts

97 Magnetic Resonance Imaging Magnetic resonance imaging (MRI) is a unique, multi-planar, imaging modality that provides detailed anatomical and physiological information of both soft tissue and bone. 24 Over the past 2 decades, MRI has become a popular imaging modality for the evaluation of equine joint disease. The difference between MRI and other imaging modalities is related to the underlying physics behind the generation of MRI signal and the formation of the image. The simplest explanation of MRI image acquisition is one in which the patient is placed in a magnetic field and becomes temporarily magnetized. Resonance is achieved through the application of specific pulses of electromagnetic radiation, which is absorbed by the patient. Subsequently, the excess energy is liberated and measured. The captured signal is processed by a computer and converted to a gray scale image. In reality, however, the acquisition of MRI images is much more complicated and a full description of quantum physics is beyond the scope of this text. The purpose of this review is to provide a basic understanding of how the MRI image is produced, how the scanner settings and the environment can influence the image, and common artifacts encountered in equine orthopedic imaging. Image acquisition To begin explaining the acquisition of an MRI image, we must first start at the nucleus of an atom, where it is made up of protons and electrons. 315 Each proton carries a positive electric charge and spins around their axis. Because a moving electrical charge creates its own magnetic field, each proton generates its own magnetic field and behaves as magnetic dipoles. The magnetic dipoles are vector quantities that can be displayed graphically as a magnetic dipole moment (μ) and can be defined by two parameters: the strength or amplitude of their magnetic field and their orientation or 97

98 phase. 316 If an even number of protons is present within the nucleus, the spins of each proton will pair up and cancel each other out. However, if an odd number of protons exist within the nucleus, a positive charge is maintained and the nucleus is considered magnetically active. 316 The hydrogen ion is the most useful nucleus for clinical MRI due to its abundance in biological tissue (both water and fat) and strong MRI signal. 316 In biological tissue, the magnetization vector of each proton (μ) is distributed in various orientations making the net macroscopic magnetization vector (M) zero. However, when the same tissue is placed within a static magnetic field (B), the tissue acquires a total magnetization moment (M) parallel to the magnetic field B. 315 This is because the angle of spin of each proton is forced to be either parallel or antiparallel to B. Because μ of each proton favors the more stable state, there is a slight excess that stays parallel with B so that the collective bulk magnetization M is aligned parallel to B (z-axis) and begin to rotate at a precise rate or frequency (Larmor frequency). 316 This process is not instantaneous but rather is exponential with respect to time up to a plateau (Mo). 316 Each individual vectors, μ, will rotate clockwise or counterclockwise around the z-axis with a rotation frequency that is proportional to the strength of the static field B and can be determined by the Larmor equation: (ω= γb). Each nuclear species has their unique resonant frequency for any B and is determined by the gyromagnetic ratio, γ, and is dependent on their local microenvironment. 316 Different nuclei have different gyromagnetic ratios; for protons the value is MhzT -1. For high- (1.5T) and low- (0.26 T) field magnets, this corresponds to 64 MHz and 11 MHz, respectively

99 As stated previously, the patient acquires a temporary magnetization vector that is parallel to the main magnetic field B that is centered on the z-axis (called longitudinal magnetization). At this point, the magnetization vector (Mo) cannot be quantified directly. In order to do so, radiofrequency pulses, or short-lived external magnetic fields that rotate around in the xy plane, are superimposed on the static field B to induce resonance. 315 The radiofrequency pulse (β) deflect the bulk magnetization vector M away from the z-axis (Mz) and towards the xy-axis (Mxy) by the angle theta (θ) and is known as the flip angle. 316 Resonance is accomplished only if the frequency of radiation matches the natural oscillation frequency of the protons (ie. the Larmor frequency). 316 The radiofrequencey pulse causes excitation of the spins through absorption of energy and induces phase coherence of them as well. If the flip angle equals 90 0, the bulk magnetization vector M is transferred entirely onto the xy-axis making Mz equal zero and the Mxy equal to Mz prior to the radiofrequency or excitation pulse. 316 Once the radiofrequency pulse is withdrawn, the spins are free to return to equilibrium or relax. The emission of radiofrequency energy as the nuclei relax is the source of the MRI signal. The MRI signal that is produced is a time-oscillating waveform because we measure it in the x-y plane as the magnetization vector rotates about the longitudinal axis. This rotating signal is translated into a sinusoidal time-varying voltage. The MRI signal produced is dependent upon the number of nuclei present (spin density) and the time it takes for the nuclei to relax (T1 and T2). T1 relaxation measures the rate of return of the nuclei to align with the static magnetic field Mz. The maximum value of Mz is proportional to the number of spins per voxel or spin density. 316 It is also known as spin-lattice relaxation to reflect the fact that there is transfer of energy from 99

100 the spins to their environment, and longitudinal relaxation to reflect the fact that the nuclear magnetization aligns along the axis of the main magnetic field B. 315 The ability to transfer energy from the spins to the environment is dependent upon the strength with which an individual nucleus is bound to its chemical backbone and can affect the oscillating frequency of the protons. Simply speaking, if the environment causes little to no change in the oscillating frequency of the protons, resonance becomes possible and there will be efficient T1 relaxation. However, environments that cause the frequency to significantly increase (free water) or decrease (collagen) comparatively to the Larmor frequency will result in delayed T1 interaction and thus inefficient relaxation and long T1 rates. Because the Larmor frequency is dependent on field strength, T1 relaxation times are also field strength dependent. 317 In contrast, T2 relaxation measures the loss of transverse magnetization (Mxy). As stated previously, the radiofrequency pulse induces phase coherence of all the spins so that they are all in the same phase as they rotate around the z-axis and the vector sum of μ has its greatest value. Once the radiofrequency pulse is removed, there is a progressive loss of phase coherence, or dephasing of spins due to the forces that develop between dipoles moving in close proximity of each other known as spin-spin interaction. 316 A second cause of spin dephasing is due to local magnetic field inhomogeneity. Variations in the local magnetic field cause the spins to dephase quicker than T2 and is indicated by the time constant T2*. In summary, T1 gains signal exponentially and T2 loses signal exponentially when the radiofrequency pulse is withdrawn. As time elapses from the cessation of the radiofrequency pulse, different tissues gain or lose signal at different rates creating T1 contrast and T2 contrast, respectively. A third variable of tissue is based on the number 100

101 of protons per unit of tissue and is called proton density. As stated previously, the number of protons per unit tissue determines the maximum signal each tissue can produce and is represented by differences in the plateau of the T1 curve. To produce MRI images, additional magnetic coils (gradient coils) are necessary to encode the signal and thus allow its origin to be determined. Gradient coils produce magnetic field gradients within the bore of the magnet and the key to understanding how this allows for spatial encoding goes back to the Larmor equation. 317 The Larmor frequency or the precessional frequency is directly related to the magnetic field strength. If a spatially varying magnetic field is superimposed on the main magnetic field, then each voxel along that plane will be related to a unique precessional frequency. 317 The referred gradient coil is actually a construct of three gradient coils that are perpendicular to each other and will be turned on and off at different times in order to localize a point in space. 316 MRI signals will be localized by selective application of slice selection, phase encoding, and frequency encoding gradients. The slice of interest is first selected by applying a gradient along an axis (for this example, z-axis) while applying an RF pulse with a narrow frequency range. This will cause all the nuclei in the slice of interest to precess at the same frequency and phase. A second gradient is then applied in a plane perpendicular to the first (y-axis), which will cause the nuclei to precess at different frequencies along that plane. Once the gradient is turned off, the nuclei will again precess at the same frequency but their phase will be different. Phase is difficult to measure quantitatively because it can attain positive and negative values with overlap for each complete cycle. To obtain discrimination in the phase encoding direction, this process must be repeated many times to more accurately reconstruct the 101

102 object. The third gradient (x-axis) causes the nuclei to precess at different frequencies. It is also called the readout gradient because it is typically at the same time the MRI signal is being collected. 317 The plane in which each of these gradients are applied will be determined by the desired plane to be imaged in (sagittal, frontal, or transverse). The MRI signal is digitized and stored on the acquisition workstation ( k-space ) for subsequent reconstruction to form the clinical image. Each point in K-space represents a different spatial frequency in the object being image. The strength of the MRI signal indicates the degree to which the spatial frequency is represented in the object. 315 Sequence selection MRI image formation involves one or more stimulating radiofrequency pulses, field gradients in all three directions, and acquisition of the resulting radiofrequency signal. The specific arrangement and variations in these steps are what differentiate pulse sequences. There are two major classes of pulse sequences based on the way in which they refocus the MRI signal Mxy or echo (explained below): Spin Echo (radiofrequency pulses) or Gradient Echo (gradients). 316 It was previously stated that the MRI signal detected is generated from the transverse component of magnetization Mxy (termed free induction decay before computer processing) and because Mxy rotates around the xy-plane, the MRI signal that is detected is a time-oscillating waveform. The sinusoidal time voltage that is determined by this data has a time constant T2* due to inhomogeneity in the static magnetic field B that is inherent to any MRI machine. T2 values, however, are of greater interest as they are independent of the static magnetic field and are an intrinsic property of each tissue. 316 Spin echo sequences negate the effects of local magnetic field inhomogeneities by reversing the direction of spin precession, without affecting the 102

103 precession rate. To explain this further, remember that transverse relaxation or dephasing of the spins is affected by spin-spin interaction and variations in the external field B. After termination of the initial 90 0 radiofrequency pulse, there is progressive loss of phase coherence that is due to both spin-spin interaction and local magnetic field variations so that some spins will move faster or slower based on these factors. Extrinsic variations, however, are constant with respect to time and space. If a refocusing pulse of is applied, the direction of the spins will change directions so that the fast spins will soon catch back up with the slower ones and the macroscopic Mxy will reappear as a signal, called the spin echo. The refocusing pulse can be repeated multiple times, generating a train of echoes. The effects of spin-spin interaction cannot be negated so that each successive echo will be of lower intensity than the first. The T2 curve can then be determined by connecting the peaks of the successive echoes. 316 Because T1 and T2 relaxation occur simultaneously, the final value of the MRI signal is a mixture of spin density and both T1 and T2 relaxation. The relative contribution of the above factors is modified by changes in repetition time (TR) and echo time (TE). Repetition time is the amount of time between successive radiofrequency or excitation pulses and echo time is the time between the radiofrequency or excitation pulse and data collection. 315 Improper selection of parameters can result in signal overlap with low or no tissue contrast. To obtain images with T1 weighting, a short TR and TE are chosen so that T1 contrast is maximized and T2 contrast is minimized. For proton density weighted images, a long TR is used in order to maximize the differences in signal intensities and a short TE in used to minimize T2 impurities. A long TR and TE 103

104 is used to obtain T2 weighted images in order to escape from T1 influences and maximize T2 contrast. A variation of the spin echo sequence is the fast or turbo spin echo sequence. In this sequence, multiple echoes are generated per TR cycle, each one with a different value of the phase encoding gradient. The time saved between these sequences and the conventional spin echo sequences is proportional to the number of echoes and is maximal for long TE sequences. 317 All echoes contribute to the contrast in the image, despite the fact that they are shifted in time and phase with respect to each other. The relative weighting of each echo depends on the time distance between echoes and the number of echoes. 317 The data can then be arranged in such a way as to produce the desired tissue contrast. The fast and turbo spin echo sequences, however, will cause blurring of tissue edges with short/moderate TE values and can cause the persistence of fat hyperintensities even with heavily T2-weighted images. 316 Gradient echo sequences differ significantly from spin echo sequences. They use an initial excitation pulse or flip angle that is less than 90 0 so that only a fraction of the longitudinal magnetization Mz is converted to transverse magnetization Mxy. 315 By doing so, it shifts the starting point of the T1 curve to the right. This allows for faster imaging compared to spin echo sequences because a considerable portion of the macroscopic vector is kept on the z-axis and available for successive TR cycles. 316 Unlike spin echo sequences, bipolar gradients used to refocus the spins to obtain a MRI signal. The first frequency-encoding gradient dephases the spins with negative polarity and the second gradient negates the first with positive polarity and causes the spins to rephase which forms the gradient echo. Because they do not use a 180 radiofrequency pulse to create 104

105 an echo, gradient echo sequences, are susceptible to local magnetic field inhomogeneity and their signal decay is characterized by T2*. 317 In addition to manipulating TR and TE, tissue contrast can also be manipulating by changing the flip angle. 315 Large flip angles ( ) coupled with short TE (5-10 ms) result in T1- weighted images. A small flip angle ( ) disturbs Mz to a smaller degree and so spin density becomes more important. 315 The combination of a small flip angle with a long TE (20-35 ms) will produce a T2*-weighted image. A variation of these two sequences involves the selective suppression of the signal from fat. Fat suppression is beneficial due to the presence of fat in bone marrow which may obscure the assessment of lesions in T2-weighted images or contrast enhancement in T1-weighted images. 317 Selective fat suppression relies on the differences in Larmor frequency between fat and water. Frequency differences exist because the oxygen atom in the water molecule draws the electrons of the hydrogen atom closer than the carbon-bound hydrogen of fat. 317 This deshields the proton of the water molecule and changes the local magnetic field it experiences. As a result, the Larmor frequency of mobile fatty acids are below the water frequency and allows for the selective suppression of fat. Magnetic field homogeneity, however, is critical for successful fat suppression as any change will result in poor or non-uniform fat suppression. 316 Fat can also be suppressed by using an inversion pulse (180 0 ) prior to starting the spin echo sequences. This creates a new variable, called the inversion time, which is the delay between the inversion and the excitation pulse. 316 Similar to fat selective fat suppression techniques, if a 90 0 excitation pulse is transmitted at the null point of fat, that tissue will return no signal and the lesion will stand out against a muted 105

106 background. This sequence is called short tau inversion recovery (STIR). STIR images, however, are not specific to fat suppression because it inverts the longitudinal magnetization of all contents of a given slice. Thus, any tissue or process with a short T1 will be suppressed, including gadolinium enhancement and some hematomas. 316 Selective fat suppression of PD-weighted and T2 weighted sequences enjoy a stronger signal and higher spatial resolution than STIR sequences. 317 Artifacts Artifacts are misrepresentation of tissue structures seen in medical images produced by patient motion and tissue heterogeneity, by intrinsic limitations of utilized methods, by inappropriate user interactions, and sometimes by system insufficiencies. 317 These artifacts are caused by a variety of mechanism, such as underlying physics, data acquisition errors, undersampling, truncation or compromised reconstruction algorithms. 317 Artifacts can be broken down into patient-related, methodrelated, and user-related categories. Patient-related artifacts introduced by the heterogeneous tissue composition of the patient are sometimes unavoidable but can sometimes be reduced with the utilization of appropriate technique. Susceptibility artifacts are exaggerated signal voids due to artificial shortening of T2* relaxation time as a consequence of significant local magnetic field inhomogeneities. 317 Nail fragments, metal implants, dirt, blue Play-Doh and blood-break down products commonly influence the local magnetic field in the horse. 315 Changes in the local magnetic field will affect the Larmor frequency of the tissue, and because the Larmor frequency is use to identify the position of the signal, the signal will be misassigned to another location causing a hyperintense area in the direction of frequency encoding as well as leaving a signal void. 317 In gradient echo 106

107 imaging, the size of the artifact will also be proportional to the T2* sensitivity of the imaging protocol and the selected spatial resolution. Flow or motion artifacts is another patient-related artifact and are commonly associated with blood vessels. Motion during image acquisition will cause displacement of the signal origin. The pulsatile nature of arteries causes a series of displaced signal in the direction of phase encoding due to its repetitive nature. 315 Magic angle artifact is also a commonly seen artifact seen in tendon that course at or near 55 0 with respect to the direction of the main magnetic field. 317 In contrast to tendon s normally short T2 relaxation time, this artifact will artificially increase the signal of tendon where the fiber orientation is close to this angle with respect the magnetic field. The magic angle effect is obvious on T1-weighted images, but becomes less apparent as TE increases. The chemical shift artifact is one of the most common method-related artifacts and is due to differences between the Larmor frequency of hydrogen nuclei in fat and water. 315 Because frequency information is used for spatial encoding, the fat image is shifted relative to the water image and is a function of the image bandwidth. 317 This will cause a bright rim to appear on one side of the image where fat is assigned to a location that already contains signal contributions from the water image and a dark rim on the opposite side of the image will indicate the opposite. The chemical shift phenomenon can also be used to your advantage, as it enables spectral suppression of the signal from fat as well as utilization of in-phase vs. opposed-phase imaging for tumor characterization. 317 The Gibbs artifact, otherwise known as a truncation artifact, refers to a series of bright lines within a dark region or the image that is parallel to sharp changes in signal intensity. This occurs because image reconstruction from k-space is a 107

108 finite sampling of the signal that is subjected to inverse Fourier transform in order to obtain the final image. At high-contrast boundaries, the Fourier transform corresponds to an infinite number of frequencies producing a series of lines that can occur in both the phase and frequency encoding directions. 317 Two commonly encountered user-related artifacts include wrap-around and external interference artifacts. Wrap-around artifacts occur when the field of view does not cover the entire patient and the portion of the patient that is outside the field of view appears on the opposite side of the image. This occurs because the frequency of a signal source outside the field of view is crossing the identical sampling spots of a lower frequency (at the opposite end) within the field of view. 317 Doubling the sample rate will allow the correct identification of the signal source of twice the field of view and can be done in the direction of the frequency encoding without untoward effects. 317 External interferences such as radio transmission or frequencies emitted from an electronic device that is not shielded can cause artifacts due to the high sensitivity of the radiofrequency coils. 317 These artifacts appear as streak artifacts that are perpendicular to the direction of frequency encoding. Field strength Several different MRI systems are available for use in the equine patient and can be divided into either high- (field strength of 1.0 tesla or greater) or low-field MRI systems (field strengths generally ranging 0.15 to 0.5 tesla), each having their own associated strengths and weaknesses. 24 High-field systems are comparatively more expensive than low-field systems due not only to the initial purchase price, but also for annual maintenance costs due to the cryogens required to maintain the magnetic field in the superconducting magnet. Additionally, high-field systems require the horse to be 108

109 anesthetized to perform the MRI exam, where as there are low-field systems that are capable of performing MRI exams on standing, sedated horse. This represents an obvious advantage of low-field systems, however, there is usually some degree of motion in standing horses creating a higher degree of motion artifact that can degrade image quality. It is difficult to compare examination times between high- and low-field systems because imaging parameters usually need to be set differently between the two systems in order to optimize image quality, but in general, similar sequences take on average 2 to 5 minutes longer using low-field systems. 318 This may sound like a small difference at first, but this difference will be multiplied by the number of sequences used to image each anatomical region, which is typically 6 to 8 sequences. There may also be a need to repeat imaging protocols because of poor patient compliance or incomplete examination of the area of interest due to the smaller field of view of lowfield systems. 318 It is generally considered that high-field systems produce images with superior image quality because the strength of the signal obtained during image acquisition increases with higher magnetic field strengths. 24 As the signal strength increases, the more information that is able to be obtained from tissue, which translates into producing images with higher spatial and contrast resolution compared to low-field systems. 319 Its reasonable to assume that MRI systems that produce images of higher quality will also increase the ability to identify small or subtle lesions and increase the overall diagnostic accuracy of the imaging modality. Recent work has shown that high-field systems more accurately identified experimentally-created osteochondral defects compared to lowfield systems and the largest differences were observed when evaluating defects that 109

110 only involved the articular cartilage. 320 In addition, the frequency differences between fat and water protons are smaller on low-field systems making fat saturation; a technique which exploits these frequency differences to identify abnormal fluid with bone and soft tissue, less reliable on low-field systems. 320 Comparison with other imaging modalities The ability to assess all tissue structures of the joint simultaneously represents the greatest advantage of MRI compared to other imaging modalities. Similar to CT, MRI is a cross-sectional imaging modality that can be obtained in multiple planes, which allows for identification of all the components of the joint directly. The difference between the modalities lies in the fact that MRI not only provides information about bone, but also provides unparalleled information regarding the soft tissue structures of the joint due to its superior contrast resolution. 321 Additionally, MRI can reveal important clinical data regarding lesions and their underlying pathophysiological mechanisms by its unique ability to identify changes in fluid content in tissue. 24 This not only allows for the identification of lesions not obtainable by other imaging modalities but more importantly allows them to be identified at an earlier time point which can be beneficial when determining appropriate treatment and management strategies. 321 It is able to do this through the acquisition of different pulse sequences in multiple planes in order to maximize the contrast differences between different tissues. Depending on the pulse sequence used, different tissue types will be associated with its own signal intensity, and any alteration in signal change within that tissue type can provide information about lesion content and severity. 24 The ability to evaluate the joint as a whole organ has also helped researchers realize that osteoarthritis is not just a disease of articular cartilage, but rather the result of the structural and functional failure of the synovial joint organ 110

111 with loss and erosion of cartilage, subchondral bone alterations, meniscal damage, synovitis, and overgrowth of bone (osteophytes). 312 Magnetic resonance imaging, however, does have its limitations. It has reduced spatial resolution compared to ultrasound, CT, or radiography, so small fractures and certain osseous lesions may be missed. In addition, mature scar tissue and mineralization are inherently of low signal intensity on MRI due to the low level of mobile protons in each tissue and can be difficult to identify in structures that are also of low signal intensity, such as tendons and ligaments. There are also numerous artifacts that commonly occur using this modality that result from the imperfections in the design of any MRI system and can affect image quality. 322 A thorough understanding of MRI physics and anatomy is necessary for artifact detection and accurate image interpretation. They can occur in both high- or low-field systems, yet each system varies in their relative susceptibility depending on the artifact. High-field systems are more susceptible to magnetic susceptibility and chemical shift artifacts, where as, low-field systems are more susceptible to motion and magnetic field artifacts and partial volume averaging. 318 In comparison to other imaging modalities, MRI examinations are also more expensive, require more time to acquire, and may require general anesthesia depending on the MRI system used. This increases the importance of lameness localization prior to examination because the number of regions that can be imaged during a particular study is limited. Similar to CT, MRI is also limited to areas that can be placed within the magnetic field so that MRI examination can only be performed on the carpus, tarsus, and regions of the limb distal to these regions. In some large bore magnets, the stifle can also be imaged depending on the size of the horse. There is also 111

112 a limited ability to scan patients with metal implants because the metal will create large magnetic field homogeneities causing an area of signal void surrounding the implant as well as causing the implant to heat up and potentially burn the patient. Cartilage imaging Magnetic resonance imaging (MRI) has emerged as the imaging modality of choice for evaluating articular cartilage given its superior soft tissue contrast and multiplanar capabilities. Due to the structural properties of articular cartilage, MRI acquisition parameters should be optimized to acquire images with high spatial resolution and signal-to-noise ratio while ensuring optimal contrast differentiation between the articular cartilage and the adjacent subchondral bone, synovial fluid, and soft tissue structures of the joint. 323 Given the small thickness of articular cartilage and the need for high image contrast, effective MRI requires high-signal-to-noise ratio that is best accomplished using high-field MRI systems (1.5 T or 3.0T). 324 A multitude of different conventional pulse sequences have been used for evaluation of articular cartilage, however, they can largely be classified into two fundamental groups of pulses sequences based on the way they refocus the MRI signal: Spin Echo and Gradient Echo. These techniques allow for the morphological assessment of articular cartilage and can be combined with fat suppression techniques to increase the contrast between lipid and non-lipid surfaces, add dynamic range, and reduce chemical shift artifacts. 325 Spin echo sequences used in the evaluation of articular cartilage can be divided by their respective tissue weighting and include: T1-weighted, proton density-weighted, and T2-weighted imaging sequences. Fast spin-echo is a variation of the spin echo sequence that incorporates the use of multiple echoes per repetition time and allows faster acquisition of data. The ability to acquire more echoes in a shorter period of time 112

113 also increases the signal-to-noise ratio, thus improving the spatial resolution of the image. 325 This decrease in scanning time will also decrease motion artifact. In addition, fast spin echo sequences are relatively insensitive to magnetic susceptibility artifacts often encountered in patients who have undergone previous surgery, which can result in loss of cartilage signal. 323 Moreover, fast spin echo sequences improve the contrast between articular cartilage and adjacent tissue and between normal and abnormal cartilage due to the magnetization transfer effect. 323 Because of these advantages, twodimensional fast spin echo sequences are used more commonly over conventional spin echo sequences in clinical practice. These sequences are also well suited for the morphological assessment of the other soft tissue structures of the joint, as well as, bone marrow edema on fat-suppressed sequences. 326 Articular cartilage has an intermediate signal intensity on T1-weighted sequences, which gives it good contrast with subchondral bone but poor contrast with synovial fluid. The poor delineation between articular cartilage and synovial fluid makes identification of surface abnormalities difficult, especially in the presence of synovial effusion. 323 This sequence may, however, be beneficial in the evaluation of the articular cartilage and subchondral interface. 323 Contrast between the articular cartilage and synovial fluid is improved on T2-weighted sequences. Articular cartilage has low to intermediate signal intensity, which contrast with the high signal intensity of synovial fluid, which allows for the accentuation of surface irregularities, giving it an arthogram-like effect. 323 This sequence, however, has very poor contrast between the articular cartilage and subchondral bone. The normal articular cartilage signal intensity on proton densityweighted sequences is higher than on standard T2-weighted sequences, which 113

114 improves the contrast between the articular cartilage and subchondral bone. The increased contrast between articular cartilage and synovial fluid evident on T2-weighted and proton density-spin echo sequences allows for identification of surface abnormalities and the low to intermediate signal of the articular cartilage allows identification of pathology within the cartilage matrix. 323 The main limitations of fast spin echo sequences are their poor spatial resolution and susceptibility to partial volume averaging because they suffer from anisotropic voxels, limited slice thickness, and section gaps due to their two-dimensional nature. 325 The most commonly used gradient echo cartilage-sensitive sequences include three-dimensional spoiled gradient-recalled acquisition in the steady state (SPGR) and fast low-angle shot (FLASH) and show improved detection of cartilage lesions with fatsuppressed sequences. In contrast to spin echo sequences, the acquisition of nearly isotropic voxels allows for high-resolution three-dimensional data that is ideal for avoiding partial volume averaging artifacts and achieving superior in plane resolution resulting in more accurate quantitative assessment of cartilage thickness and volume. 325 Gradient echo sequences, however, have several disadvantages. 323 Normal articular cartilage has high signal intensity in gradient echo fat-suppressed sequences compared to the low signal intensity of subchondral bone and intermediate signal intensity of synovial fluid. The high signal intensity of articular cartilage limits visualization of pathology within the cartilage matrix. In addition, it is difficult to identify surface abnormalities on gradient echo sequences given the intermediate signal of synovial fluid next the high signal intensity of articular cartilage. 326 The lack of reliable contrast between the articular cartilage and synovial fluid may also obscure small focal lesions 114

115 and is related to their sensitivity to particular types of artifacts. 325 Because of their long acquisition times, gradient echo sequences are also prone to motion artifacts which can lead to inaccurate measurements. These sequences are also more sensitive to magnetic susceptibility and truncation artifacts than spin echo sequences. In addition, they are susceptible to magic angle effects, which can lead to false-positive findings in ligaments and menisci. Given these disadvantages, gradient echo sequences seem to be more suited to measure cartilage volume and in some instances depict the cartilage surface but less suited to identify subtle cartilage signal abnormalities compared to spin echo sequences In the horse, the proton density fat-suppressed fast spin echo and fat-suppressed SPGR imaging sequences are more commonly used to evaluate articular cartilage. 320 A previous report validated the use of MRI to measure the thickness of equine cartilage using the fat-suppressed SPGR sequence as it showed excellent correlation with histological measurements. 327 The use of fast spin echo sequences to measure equine cartilage has not been validated. Both sequences, however, have shown to provide reasonable accuracy in detecting the presence of articular cartilage defects using a 1.5T MRI in equine cadaver studies, although the fast spin echo sequences seemed to be preferred over the gradient echo sequences. 320,328 However, these studies also showed that MRI is not very accurate in its ability to determine the size of the articular defect using these sequences, especially its thickness. 320,328 To improve the accuracy in detecting articular pathology in the horse, an increase in spatial resolution would be needed. In contrast to the cartilage in the knee of people, equine articular cartilage is much thinner, making assessment more difficult in the horse. 180 As the thickness of the 115

116 cartilage decreases, the more important spatial resolution is for delineating cartilage interfaces and minimizing volume averaging. 324 Increasing the spatial resolution of these sequences, however, would lower the signal-to-noise ratio and result in decreased image quality and longer acquisition times would be required to maintain similar image quality. 324 Scan times, however, have to be kept short to complications associated with prolonged anesthesia in horses undergoing high-field MRI examination and to limit motion artifact when using low-field MRI systems. It is clear that both spin echo and gradient echo sequences have their own advantages and disadvantages, however, the advantages of one sequences complements the disadvantages of the other, and vice versa. Thus, these modalities should be used together in such a complementary fashion when evaluating articular cartilage abnormalities. 326 Compared to arthroscopy, MRI offers several advantages in the evaluation of articular cartilage including the ability to provide in vivo morphologic information regarding thickness, volume, and three-dimensional configuration of cartilage as well as provide in vivo biochemical information regarding collagen and extracellular matrix composition of hyaline cartilage, including the free water, proteoglycan, and sodium content. 323 In addition, MRI is highly accurate in measuring cartilage volume, providing a quantitative and global marker for the progression of cartilage loss; is noninvasive in nature, allowing for serial assessment of articular cartilage abnormalities without risk of injury and for follow-up of abnormalities in order to determine the natural history of the response to therapy; and is cost-effective. 323 Arthroscopy Although more commonly known as a treatment, diagnostic arthroscopy is still considered the gold standard for grading cartilage damage, for a definitive intra-articular 116

117 diagnosis, and for decisions regarding therapeutic options in patients with osteoarthritis. 329 It is the only imaging modality that allows direct visualization of the intra-synovial structures of the joint and is invaluable when it comes to assessing the synovial membrane and articular cartilage, as well as portions of the intra-articular ligaments and menisci where present. The evaluation of the intrasynovial structures, however, is limited to what can be directed visualized by the arthroscope, which is limited in most joint. Knowledge of arthroscopic anatomy and the use of other imaging modalities prior to examination will help in determining whether certain areas of the joint can be evaluated. In addition, there is also a low risk of both intraoperative and postoperative complications associated with arthroscopy. 330,

118 Figure 2-1: Schematic, cross-sectional diagram of healthy articular cartilage: A) cellular organization in the zones of articular cartilage; B) collagen fiber architecture. (From: Buckwalter JA, Mow VC, Ratcliffe A. Restoration of injured or degenerated articular cartilage. Journal of American Academy of Orthopaedic Surgeons. 1994; 2: ) 118

119 Figure 2-2: Histological representation of articular cartilage and underlying subchondral bone in a normal joint. NCC, non-calcified cartilage; CC, calcified cartilage; SBP, subchondral bone plate; STB, subchondral trabecular bone. Arrows denote the tidemark; the dotted line indicates the cement line. (From: Li G, Yin J, Gao J, et al. Subchondral bone in osteoarthritis: insight into risk factors and microstructural changes. Arthritis Research and Therapy. 2013; 15: ) 119

120 Figure 2-3: Schematic representation of the interaction of collagen, hyaluronan, and aggrecan in cartilage. A) Organization of Type II collagen fibrils, containing Type IX and XI collagens and the multiple proteoglycan molecules that bind hyaluronan to form aggrecan in the extracellular matrix of articular cartilage. B) Structure of proteoglycan molecule demonstrating the protein core and glycosaminoglycan side chains. (From: McIlwraith C. From Arthroscopy to Gene Therapy. AAEP Proceedings. 2005; 51: ) 120

121 Figure 2-4: Factors involved in enzymatic degradation of articular cartilage matrix. FGF, fibroblast growth factor; PLA2, plasminogen activator; Dotted lines indicate factors that may inhibit degradation. (From: McIlwraith C. From Arthroscopy to Gene Therapy. AAEP Proceedings. 2005; 51: ) 121

122 CHAPTER 3 MAGNETIC RESONANCE IMAGING SCORING OF AN EXPERIMENTAL MODEL OF POST-TRAUMATIC OSTEOARTHRITIS IN THE CARPUS Introduction Osteoarthritis is a multifactorial process characterized by changes in structure and function of the entire joint, and is the most common cause of lameness and decreased performance in the equine athlete. 3 A diagnosis of osteoarthritis is commonly made based on patient history, clinical examination, and diagnostic imaging findings. Conventional radiography is the most common imaging modality used to characterize the progression of structural changes associated with osteoarthritis, however, it lacks the sensitivity necessary to detect early changes associated with osteoarthritis and correlates poorly with clinical outcomes These limitations likely play a role in why it has been difficult to demonstrate treatment efficacy in osteoarthritis research as radiographically detectable osteoarthritis may represent a stage of osteoarthritis that is already too advanced for pharmacological intervention. 22,333 Identifying imaging modalities that are able to recognize changes associated with osteoarthritis starting at the onset of the disease process and correlating these imaging findings to clinical symptoms is critical to improving our understanding of osteoarthritis and for developing effective therapeutic interventions. Magnetic resonance imaging (MRI) is a non-invasive, cross-sectional, multiplanar imaging modality that is increasingly being used for the assessment of musculoskeletal disorders in horses. 23,24 Recently, several case series have shown the ability of MRI to identify osteoarthritis-related structural changes related to equine lameness that could not be identified by radiography, 19,20, however, very little information is known regarding the significance and pathological progression of these imaging findings in the 122

123 horse. Such information would be beneficial in determining the prognosis, as well as for case management. 321,334 In people, semi-quantitative MRI whole-organ scoring systems have been shown to provide a reliable method to quantify the structural changes that occur in joints over time. 25,335 They have also been shown to relate clinical symptoms to pathological features and to identify risk factors for structural changes Several equine cadaver studies have evaluated the ability of MRI to assess pathological change in cartilage, bone and soft tissue structures, 17,320,327, however, studies using MRI to evaluate the joint as a whole-organ are lacking and could improve our understanding of the progression of this complex disease. The carpal osteochondral injury-exercise model is a commonly utilized model of post-traumatic osteoarthritis in the horse, and several studies have evaluated imaging and pathological findings at the completion of these studies. 180,345,346 To the authors knowledge, there are no studies evaluating the use of MRI to identify the progression of osteoarthritis-associated structural changes that occur over time following osteochondral injury using this model. The objective of this study was to characterize the progression of MRI changes in a carpal osteochondral injury-exercise model of posttraumatic osteoarthritis using a semi-quantitative scoring system for evaluation of the equine middle carpal joint. We hypothesized that MRI would be able to detect osteoarthritis-associated structural changes 14 days following osteochondral injury and scores obtained from a semi-quantitative equine MRI osteoarthritis scoring system (EMOSS) would be higher in the osteochondral injured joints at both 14 and 98 days 123

124 following osteochondral injury compared to baseline examination, as well as between osteochondral injured and control joints. Materials and Methods Experimental Design The horses used in this experimental study were part of an independent project evaluating the therapeutic effects of a novel treatment for osteoarthritis. The MRI images used were from ten control (untreated) horses in that project. The study protocol was approved by the University of Florida Institutional Animal Care and Use Committee. Study horses were obtained from the University of Florida College of Veterinary Medicine Research Herd and included in the study if they were free of lameness and lacked clinical abnormalities in their carpi including effusion, periarticular soft tissue swelling, or decreased range of motion. Subjective gait analyses were performed by two evaluators using the American Association of Equine Practitioners lameness grading scale 347, and confirmed using the Lameness Locator (Equinosis, LLC, St. Louis, MO) for objective gait analysis. Complete radiographic examination of each middle carpal joint [dorsopalmar, dorsal 45 lateral-palmaromedial oblique, dorsal 45 medialpalmarolateral oblique, lateromedial, lateromedial (flexed), dorsal 55 proximaldorsodistal oblique (flexed), and dorsal 35 proximal-dorsodistal oblique (flexed) views] were obtained and horses were excluded from the study if there was presence of soft tissue swelling, osteochondral fragmentation, marginal osteophytosis, enthesophytosis, subchondral bone sclerosis, or subchondral bone lysis. Prior to the start of the study, all horses were exercised 5 days/week on a high-speed treadmill for 21 days. For each exercise day, horses underwent trotting (16-18 km/h) for 2 min, then galloping (30-124

125 32km/h) for 2 min, followed by trotting (16-18 km/h) for two additional minutes to simulate the exercise of race training. Magnetic resonance imaging examinations were performed under general anesthesia with the horse placed in left lateral recumbency on days 0, 14, and 98. Both carpi were scanned individually with each carpus placed in partial flexion (15-25 ) using a 1.5 Tesla high-field unit and a quadrature transmit/receive knee coil (Toshiba America Medical Systems,Tustin, CA) with one receiver channel. Magnetic resonance imaging sequences were selected to be clinically applicable and to allow timely completion to avoid a prolonged period of general anesthesia. Sequences included sagittal proton density (PD), axial PD, dorsal T2-weighted (T2), axial T2 short-tau inversion recovery (T2-STIR), sagittal proton density with fat suppression (PD-FS), dorsal PD-FS, and dorsal T2*-weighted spoiled gradient echo with fat suppression (SPGR-FS). (Table 3-1) Total acquisition time was approximately 80 minutes for all sequences on both limbs. On day 0, immediately following the initial MRI examination, all horses were placed in dorsal recumbency and underwent bilateral arthroscopic examination of the middle carpal joints using the standard arthroscopic approach. 348 An osteochondral fragment was created at the distodorsal aspect of the radial carpal bone in one randomly selected middle carpal joint of each horse to simulate osteochondral injury. The osteochondral fragment was made by using an 8 mm curved bone gouge placed through the medial arthroscopic portal and directed perpendicular to the articular surface of the distal aspect of the radial carpal bone at the level of the medial synovial plica. The osteochondral fragment was allowed to remain adhered to the medial synovial plica and any osteochondral debris resulting from creation of the osteochondral 125

126 fragment was left within the joint as previously described. 5 This joint was designated the osteochondral injured joint (OCI), the other middle carpal joint was sham-operated and designated the control joint (CON). The arthroscopic portals were closed with 2-0 polydioxanone in a simple, interrupted pattern and the limbs were bandaged routinely for recovery. Each horse received one dose of intravenous potassium penicillin (22,000 U/kg), gentamicin (6.6 mg/kg), and phenylbutazone (4.4 mg/kg) at the time of surgery, and phenylbutazone (4.4 mg/kg) intravenously once daily for 5 days following. All horses were housed in a 3.65 m x 3.65 m stall from day 0 to 10. During this time, the bandages were changed and horses were monitored daily for vital parameters and any complications following surgery. On day 10, sutures were removed and all horses were returned to paddock turnout. On day 14, all horses resumed exercise on a high-speed treadmill using the protocol described above until the end of the study on day 98. Magnetic Resonance Imaging Assessment All images were transferred and stored in Digital Imaging and Communications in Medicine (DICOM) format on a Picture Archive and Communication System (Merge Healthcare, Chicago, IL). Images were analyzed at the same time by three observers including a board-certified veterinary radiologist, a board-certified large animal surgeon, and a large animal surgery resident using a dedicated DICOM workstation equipped with three megapixel greyscale monitors. The observers were not aware of the identity of experimental groups. Images were scored by consensus agreement in which images with different scores assigned by one or more observers were reviewed again and scores discussed. A final score was then assigned by consensus of all three observers. All images were used to evaluate each feature, although certain sequences were more 126

127 suitable for grading individual articular features than others and were reported accordingly. Equine MRI Osteoarthritis Scoring System A semi-quantitative whole-organ scoring system for use in the equine middle carpal joint was adapted from a previously described human knee osteoarthritis scoring system. 25 Images were scored ordinally based on fifteen independent articular features including presence of cartilage signal abnormalities, size and intensity of high-signal bone lesions; subchondral bone irregularity; low-signal bone lesions; marginal osteophytes; medial and lateral collateral, medial and lateral palmar intercarpal, and dorsomedial intercarpal ligament integrity; number and size of osteochondral fragment(s); synovial thickening; and joint effusion. The six cuboidal bones that make up the middle carpal joint were divided into 9 sub-regions. (Figure 3-1) The radial carpal, intermediate carpal, and third carpal bones were each divided into two sub-regions based on the bone on the opposing articular surface with which it articulates. The radial carpal bone was divided into the second and third facets, the intermediate carpal bone was divided into the third and fourth facets, and the third carpal bone was divided into the radial and intermediate facets. The margins of each sub-region division were extended along a line perpendicular to the articular surface of the middle carpal joint to either the corresponding radiocarpal or carpometacarpal joint. The ulnar, second, and fourth carpal bones were evaluated as their own undivided sub-regions. Six of the articular features (cartilage signal abnormalities, high-signal bone lesion size and intensity, low-signal bone lesion, subchondral bone irregularity, and marginal osteophytes) were evaluated independently at the 9 different sub-regions in the middle carpal joint. Scoring was based solely on the 127

128 middle carpal joint such that scored abnormalities originated and extended only from the articular surface of the middle carpal joint. Abnormalities within the bone originating from either the radiocarpal or carpometacarpal joint were not scored. Cartilage signal abnormalities were scored based on changes in signal and morphology on SPGR-FS and PD-FS images and graded as: 0 = no abnormalities, 1 = single lesion < 5 mm, 2 = single lesion 5-10 mm or multiple lesions < 5 mm, 3 = multiple lesions with at least one lesion > 10 mm. High-signal bone lesions were defined as poorly marginated areas of increased signal intensity in the normally hypointense trabecular or subchondral bone on both the PD-FS and T2-STIR images. The size of the high-signal bone lesion was based on the extent of regional involvement and graded as: 0 = none, 1 = < 25%, 2 = 25-50%, and 3 = > 50%. The intensity of increased signal for each high-signal bone lesion was subjectively graded based on severity of signal increase (0 = none, 1 = mild, 2 = moderate, 3 = marked) compared to the surrounding subchondral or trabecular bone on PD-FS and T2-STIR images. (Figure 3-2) Subchondral bone irregularity was scored based on severity and graded as: 0 = smooth and regular chondro-osseous junction, 1 = mild subchondral plate irregularity, 2 = marked subchondral plate irregularity with intact trabecular bone, 3 = bone irregularity extending to the trabecular bone with preservation of some bone trabecular pattern, 4 = cyst-like formation. Subchondral bone irregularity was differentiated from high-signal bone lesions by the presence of subchondral bone plate irregularity with progressive involvement of trabecular bone. Low-signal bone lesions were defined as reduced signal intensity on PD and T2-weighted images and graded as: 0 = <10% of region, 1 = 10-25%, 2 = 26-50%, 3 = >50%. Marginal osteophytes were graded according to size (0 128

129 = none, 1 = < 1 mm, 2 = 1-2 mm, 3 = >2 mm), and based on the largest osteophyte identified within each sub-region. The soft tissue supporting structures associated with the middle carpal joint, including: the lateral and medial collateral ligament, dorsomedial intercarpal ligament, and lateral and medial palmar intercarpal ligament, were assessed independently and graded together as normal (0) or abnormal (1) based on changes in size and signal intensity. Synovial thickening and joint effusion were scored independently using T2- STIR and PD-FS images. Synovial thickening was recognized as tissue of moderate signal intensity protruding into the hyperintense synovial fluid and subjectively graded (0 = normal, 1 = mild, 2 = moderate, 3 = marked) based on the intensity and distribution of synovium. (Figure 3-3) Joint effusion was graded from 0-3 and based on the estimated maximum distension of the synovial cavity: 0 = normal, 1 = <33% of maximum potential distention, 2: 33% - 66% of maximum potential distension, 3 = > 66% of maximum potential distension. Each joint was scored according to the number of osteochondral fragments present and by the size of the largest osteochondral fragment. The number of osteochondral fragments were graded as: 0 = none, 1 = 1 fragment, 2 = 2 fragments, 3 = > 2 fragments. The dorsal to palmar depth and lateral to medial width of the fragment was measured using the longest dimensions on axial images, and the proximal to distal height was measured using the largest dimension on the sagittal images. The size of the largest osteochondral fragment was then graded based on the estimated volume of the osteochondral fragment (0 = none, 1 = <250 mm 3, 2 = mm 3, 3 = >500mm 3 ). Final scores were totaled to determine a whole-organ score for the entire middle carpal joint (0-188). 129

130 Statistical Analysis Statistical analysis was performed by the primary author (AS) using statistical analysis software. Since the response was ordinal scale and non-normal, a nonparametric approach was considered. The experimental design was a split plot in time 349 with group (CON vs OCI joints within a horse) as the main plot factor and day as the sub-plot factor. The aligned-rank transform method 350 was applied to a split plot in time design (SAS Institute Inc, Cary, NC). Initially, a group-day interaction was tested followed by main effect tests for group (CON vs. OCI) and day (0, 14, and 98). These tests were followed with the nonparametric Wilcoxon Signed Rank Tests to determine which differences were significant between CON and OCI groups fixing the day or pairs of days (O vs 14, 0 vs. 98, and 14 vs. 98) and fixing the group. Outcome ordinal data (scores) were reported as median (minimum, maximum) because the data were not normally distributed. The null hypotheses that median scores for pathological changes observed between groups (CON vs. OCI) on days 0, 14, and 98 were not different were tested by using the Wilcoxon Signed Rank Test (Statistix 9 for Windows, 2008, Analytical Software, Tallahassee, FL). The null hypotheses that median scores for total individual articular and total joint scores observed between groups (CON vs OCI) on days 0, 14, and 98 were not different were tested by using the Wilcoxon Signed Rank Test (Statistix 9 for Windows, 2008, Analytical Software, Tallahassee, FL). For these two hypotheses, values of p < 0.05 were considered statistically significant as only one comparison was made between each outcome variable. Among OCI joints, the null hypotheses that median scores were not different between days 0, 14, and 98 was tested using the Wilcoxon Signed Rank Test (Statistix 9 for Windows, 2008, Analytical Software, Tallahassee, FL) for paired data (i.e., day 0 130

131 vs 14, day 0 vs 98, day 14 vs 98). Similarly, among CON joints, the null hypothesis that median scores were not different between days 0, 14, and 98 was tested by using the Wilcoxon Signed Rank Test (Statistix 9 for Windows, 2008, Analytical Software, Tallahassee, FL) for paired data. Because each outcome variable was tested against three time points (0, 14, and 98) in these two hypotheses, a Bonferroni-adjusted significance level of was calculated to account for the increased possibility of type-1 error. Results Ten thoroughbred horses (5 mares and 5 geldings) between 2 and 9 years of age (median: 4.0, min: 3.0, max: 9.0) that were free of lameness, and clinical and radiographic evidence of middle carpal joint disease were included in the study. Tables 4-2 and 4-3 contain the aligned rank split plot in time analyses. In the majority of cases among the totals, there was a significant interaction between group and day indicating that the difference between CON and OCI scores changed over time. Furthermore, in every case there was a significant main effect for day and in all but one case a significant main effect for group. This meant that there were significant differences between days averaged over groups and significant differences between groups averaged over days. Tables 4-4 to 4-7 represent the median (min, max) scores for whole-organ assessment of the CON and OCI joints and their comparisons between group on days 0, 14, and 98. On baseline MRI examination (day 0), no significant differences were seen between CON and OCI joints in any of the 15 articular features evaluated or whole-organ scores. (Tables 4-4 and 4-7) Low-grade pathology, however, was variably identified in each articular feature evaluated despite strict inclusion criteria excluding 131

132 horses with radiographic abnormalities. (Tables 4-4 and 4-7) When evaluating individual articular features, the third and fourth facets of the intermediate carpal bone, the intermediate facet of the third carpal bone, and the fourth carpal bone (Tables 4-4 to 4-6); the lateral and medial collateral ligaments and the medial and lateral palmar intercarpal ligaments (Table 4-7) showed no significant osteoarthritis-associated structural changes throughout the study period in either the CON or OCI joints. On day 14, osteochondral fragments of varying size (p = 0.001) and number (p = 0.001) were identified on all sequences in the OCI joints. (Figure 4-4) The osteochondral fracture planes were best identified on PD-FS, SPGR-FS, and T2-STIR sequences in the dorsal or axial images as distinct linear patterns of increased signal intensity that was continuous with the synovial fluid and communicated with the joint distally and exited proximally in the dorsomedial aspect of the third facet of the radial carpal bone. High signal bone lesions were also identified on PD-FS, SPGR-FS, and T2-STIR sequences and extended radially from the fracture plane in the third facet of the radial carpal bone in all of the OCI joints, and extended into the second facet of the radial carpal bone in 9 of 10 OCI joints on day 14. (Figures 2-4 & 2-5) Additionally, median dorsomedial intercarpal ligament scores (p = 0.031) were also increased in the OCI joints compared to CON joints on day 14. On day 98, there was no longer a significant difference between osteochondral number or size between CON and OCI joints as a discrete osteochondral fragment was identified in only 3 out of the 10 OCI joints. The osteochondral fragments in the other 7 OCI joints were identified adjacent to their fracture bed on the third facet of the radial carpal bone and the previous areas of increased signal intensity was no longer 132

133 continuous with the synovial fluid nor did it span across two cortices. (Figure 3-4) In 4 out of these 7 OCI joints, the trabecular bone along the previous fracture plane was of uniform low signal intensity with complete resolution of high signal intensity. In the other 3 OCI joints, a portion of the previous fracture line was still evident along a variable portion of its length as evidenced by a thin line of intermediate to high signal intensity on PD-FS, T2-STIR and SPGR-FS sequences within the trabecular bone of the third facet of the radial carpal bone. The size and signal intensity of the high-signal bone lesions were also no longer significantly different between groups on day 98, however, median low-signal bone lesions (p = 0.016), cartilage signal abnormality (p = 0.020), and subchondral bone irregularity (p = 0.004) scores for the third facet of the radial carpal bone were significantly increased in the OCI joints. Median synovial thickening scores (p = 0.028) were also increased in the OCI joints compared CON joints on day 98. Finally, total joint scores for the OCI joints were increased on days 14 (p = 0.014) and 98 (p = 0.014) compared to CON joints. (Table 4-7) Table 4-8 represents the median (min, max) scores for the second and third facet of the radial carpal bone, the second carpal bone, and the radial facet of the third carpal bone for both the CON and OCI joints and their corresponding within group comparisons on days 0, 14, and 98. On day 14, increases in high-signal bone lesion scores from baseline examination were identified in the second and third facets of the radial carpal bone of the OCI joints for both size (RC-2: p = 0.010, RC-3: p = 0.001) and signal intensity (RC-2: p = 0.004, RC-3: p = 0.001) (Figures 2-4 & 2-4). In both the CON (p = 0.008) and OCI (p = 0.016) joints, median joint effusion scores were also increased on day 14 compared to baseline examination. (Figures 2-4 to 2-6) 133

134 At the end of the study (day 98), no significant differences were found in any of the articular features evaluated in the CON joints compared to days 0 or 14. However, significant osteoarthritis-associated structural changes were identified in the OCI joints, the majority of which were in the third facet of the radial carpal bone. These changes included increases in median scores for high-signal bone lesion size (p = 0.016) and intensity (p = 0.016) (Figures 2-4 & 2-5), marginal osteophyte formation (p = 0.008) (Figure 3-7), cartilage signal abnormalities (p = 0.004) (Figure 3-8), and subchondral bone irregularity (p = 0.004). (Figure 3-8) Although the high-signal bone lesion median scores of the third facet of the radial carpal bone remained increased compared to day 0 in the OCI joints, they were decreased in both size (p = 0.002) and signal intensity (p = 0.002) compared to day 14. In addition to the osteoarthritis-related structural changes identified in the third facet of the radial carpal bone, median cartilage signal abnormalities scores (p = 0.008) were increased in the radial facet of the third carpal bone on day 98 compared to baseline exam. Marginal osteophyte formation was also identified on the second facet of the radial carpal bone (p = 0.008). Median joint effusion scores remained increased compared to day 0 in the OCI joints (p = 0.023), although there were no differences compared to day 14. Median synovial thickening scores were also increased in the OCI joints on day 98 compared to day 0 (p = 0.008). (Figure 3-5) Total joint scores for the OCI joints were increased compared to baseline on days 14 (p = 0.001) and 98 (p = 0.002), although no difference was noted between days 14 and 98. Discussion The Equine MRI Osteoarthritis Scoring System used in the current study allowed temporal assessment of the osteoarthritis-associated structural changes that occur following osteochondral injury in the equine middle carpal joint. This semi-quantitative 134

135 whole-organ scoring system was adapted from the Whole-Organ Magnetic Resonance Imaging Score (WORMS) 25 knee scoring system used in people for use in the equine middle carpal joint. The articular features included are considered to be important for whole-organ assessment of the joint in people 351 and considered clinically relevant to traumatic joint injuries in the equine athlete. The sequences selected for use in this study were similar to those used in the clinical evaluation of traumatic joint injuries in the horse and provided an efficient means of evaluating all structures of the joint in order to minimize anesthetic time. The equine carpal osteochondral fragment model was easily performed and provided a suitable model to study post-traumatic osteoarthritis pathology from the onset of osteochondral injury to the development of osteoarthritis. The osteochondral fragments created on the third facet of the radial carpal bone were easily identified on all imaging sequences on day 14 in the OCI joints. In contrast, a distinct osteochondral fragment was no longer observed in 7 out of 10 OCI joints on day 98. In 3 of these 7 OCI joints, there was an intermediate to high signal intensity on PD-FS, T2-STIR and SPGR-FS sequences still present along a variable portion the fracture plane within the trabecular bone of the third facet of the radial carpal bone, which was not identified in the remaining 4 OCI joints. The differences in these findings likely represent variability in healing of the fractures. This is similar to what is seen clinically in horses, where resolution of high signal intensity on fat saturated sequences is associated with resolution of the fracture plane and the persistence of high signal intensity on PD-FS and T2-STIR images likely represents continued boney remodeling

136 In addition, there was also evidence of injury to the subchondral and trabecular bone of the second and third facet of the radial carpal bone identified on MRI as large, ill-defined areas of increased signal intensity on fat-suppressed sequences on day 14. These lesions have been referred to as occult, as they are not recognized by other imaging modalities. 20 Because MRI has the ability to provide both anatomic and physiological information, these lesions can be identified and characterized. 20 Lesions with a similar appearance have been identified on MRI examination of horses following a known traumatic joint injury and are typically associated with lameness. 19,20,352 Although the mechanism by which subchondral bone injury leads to osteoarthritis is controversial, 145 it is recognized that injury to the subchondral bone predispose horses to osteoarthritis. 353 In the current study, all of the high-signal bone lesions present in the third facet of the radial carpal bone on day 14 showed at least 50% resolution in size on day 98 in the OCI joints. Of the 6 OCI joints with persistent high-signal bone lesions on day 98, the pattern of resolution was from the trabecular bone towards the articular surface with a focal, discrete area of mild to moderate signal intensity remaining within the subchondral bone on fat-suppressed sequences. Additionally, cartilage signal abnormalities were identified adjacent to these areas of high-signal bone lesion in all (6) of the OCI joints with persistent high-signal bone lesion and in 2 of the OCI joints without persistent high-signal bone lesions on day 98. Prior to this study, the presence and progression of these occult lesions were not well documented in the horse because of the lack of chronological, comparative MRI studies of these types of injuries. 19,103,105,343,352,354 In people, similar appearing high-signal bone lesions have 136

137 been identified on MRI initially following traumatic knee injuries with similar patterns of resolution overtime associated with cartilage degeneration The findings in this current study suggest that cartilage degeneration may be associated with these patterns of high-signal bone lesions following osteochondral injury in the horse as well. The use of MRI to evaluate these changes longitudinally over time will help us determine their significance in the pathological progression of post-traumatic osteoarthritis. 355 In the current study, there is evidence that there is a continuum between lowand high-signal bone lesions. This was most evident in the third facet of the radial carpal bone where low-signal bone lesions scores developed in areas of previous highsignal bone lesions and adjacent to areas of resolving high signal bone lesions. In addition, there was a trend (p = 0.06) for an increase in high-signal bone lesion scores in the radial facet of the third carpal bone on day 98 within areas of persistent low-signal bone lesion. Although high-signal bone lesions are included in human knee semiquantitative MRI scoring systems, low-signal bone lesions, or more commonly termed subchondral bone sclerosis, have not been included, despite being considered one of the hallmarks of osteoarthritis. 148,356 Several studies have shown a correlation between subchondral bone sclerosis and cartilage degeneration in veterinary medicine, 17,140,339,357,358 however, it is unclear if subchondral bone sclerosis is a cause or a result of cartilage injury, 148 or how it relates temporally in the development of osteoarthritis. Although cartilage signal abnormalities were identified on both third facet of the radial carpal bone and the radial facet of the third carpal bone, the difference in appearance and manifestation of high-signal bone lesions and low-signal bone lesions between these two sub-regions likely reflects differences in mechanism of injury. 359 If 137

138 we consider that the differences in signal intensities between these two time points in both sub-regions likely represents a continuum of boney remodeling following a traumatic injury, it is likely that a relationship exist between high- and low-signal bone lesions identified on MRI and cartilage degeneration. In addition to differences in size of the high-signal bone lesions, the intensity, or severity of signal also changed over time. This was most evident in the third facet of the radial carpal bone of the OCI joints between days 14 and 98. (Figures 2-4 & 2-5) Previous studies in people and dogs have documented that changes in signal intensity correlated with differences in histopathological findings. 360,361 Although histopathology was not performed in the present study, we speculate that the changes in signal intensity of high-signal bone lesions over time in the current study represents different levels of severity of bone injury and remodeling. A limited number of studies have correlated MRI and histopathology in the horse; 327,340,362 however, to the authors knowledge there are no studies that have correlated MRI and pathological findings over time or at the early stages of disease. These comparisons would validate the findings identified on MRI in the current study and provide a better understanding of how these MRI findings relate to the pathological progression of post-traumatic osteoarthritis. Additionally, changes in the severity of high signal intensity over time may prove to be better associated with return to soundness than either the persistence or resolution of high signal intensity as neither have been shown to be an accurate predictor of return to soundness in a limited number of case reports. 103,352,363,364 In addition to cartilage signal abnormalities, subchondral bone irregularities were identified in similar locations on the third facet of the radial carpal bone on day 98 in the 138

139 OCI joints. We may have been biased when evaluating cartilage signal abnormalities because areas of subchondral bone irregularity were likely closely scrutinized for cartilage signal abnormalities, which could explain their association. However, subchondral bone irregularity scores on MRI have been correlated to the presence of gross cartilage damaged previously. 328 These subchondral bone irregularities were also associated with the site of osteochondral fragmentation and could represent incomplete healing of the osteochondral fragment. However, it is more likely that these represent true lesions as most of the subchondral bone irregularities identified extended along the articular margin well beyond the original fracture line. Additionally, persistent high-signal bone lesions were present in only 3 of the 10 OCI joints at the articular aspect of the previous fracture line. It is in these 3 OCI joints that we believe there was evidence of incomplete healing of the osteochondral fragment at the articular surface. Progressive marginal osteophyte formation was identified in the medial aspect of the OCI joints on the second and third facet of the radial carpal, surrounding the site of osteochondral injury. The formation of marginal osteophytes is commonly assessed when evaluating osteoarthritis-associated structural changes in the horse. 365 There are several proposed causes for osteophyte formation, but they all seem to result in progressive overgrowth of cartilage and subchondral bone along the borders of articulation. 366 In the current study, marginal osteophytes appeared isointense to cortical bone on all sequences, regardless of the time period, which would suggest that they represent mineralized tissue. In a canine anterior cruciate ligament transection model of osteoarthritis, 18 however, osteophytes were identified as early as 4 weeks and were characterized as being marginally hyperintense to cortical bone on T1-weighted 139

140 gradient echo sequences which would be consistent with cartilage and incomplete mineralization (i.e. chondrophyte). The differences in osteophyte appearance between the two studies may be explained by the different pulse sequence parameters used or by the timing of MRI examination. Osteophyte appearance may also have been affected by the study design and exercise regimens used or by species-specific pathologic changes occurring in osteoarthritis disease progression. Joint effusion scores were increased compared to baseline in both the CON and OCI joints 14 days following initial arthroscopy. Although synovial fluid analyses were not performed, we assume that the increases in joint effusion scores in the CON joints were related to acute synovitis secondary to arthroscopy as no other procedures were performed and effusion scores returned to baseline by day 98. A study using a similar model of osteoarthritis reported an increase in total white blood cells and total protein in the control joints for up to two weeks following arthroscopic surgery which subsequently returned to baseline levels after two weeks. 346 Joint effusion and synovial thickening scores in the OCI joints were increased compared to both CON and OCI joints at baseline exam and to CON joints on day 98. These increases are likely a result of ongoing synovitis in the OCI joints, as chronic synovitis may lead to both synovial thickening and concurrent increases in joint fluid production. 73 These results are similar to the previous study 346 that found a continued increase in both white blood cells and total protein for up to 10 weeks in experimental osteoarthritis joints. In another study using the carpal osteochondral injury-exercise model, MRI identified increases in synovial thickening but not joint effusion in the experimental osteoarthritis joints compared to controls at 10 weeks. 140

141 The PD-FS and SPGR-FS sequences were used in the current study to evaluate the articular cartilage because these sequences increase the contrast between articular cartilage and subchondral bone. Although, these sequences have demonstrated reasonably accurate detection of articular cartilage defects in cadaver studies, 320,327,328 they do not accurately determine the size of the articular defect, particularly with respect to thickness. 320,328 In contrast to the cartilage in the knee of people, equine articular cartilage is much thinner and more difficult to assess using MRI. 180 As cartilage thickness decreases, the more important spatial resolution becomes to be able to delineate cartilage interfaces and minimize volume averaging. Although the sequence parameters used in the current study were improved to provide better cartilage imaging (Table 4-1) over the previously mentioned studies, 320,328 spatial resolution was still limited. We chose not increase our spatial resolution further, as it would result in lowering the signal-to-noise ratio and result in decreased image quality. Longer acquisition times would be required to maintain similar image quality, 324 which were not chosen in this study to avoid potential complications associated with prolonged anesthesia. Because of this, the authors elected to score the articular cartilage based on changes in cartilage signal only. This differs from how the articular cartilage is scored in the whole-organ assessment in people, where the score is based on changes in both cartilage signal and thickness. 351 The identification of variable low-grade pathological abnormalities on baseline MRI examinations illustrates the inherent variability that exists in any study population, and highlights the limitations of radiography and the necessity for baseline MRI examinations. Every middle carpal joint in our study population showed some degree of 141

142 low-grade pathology despite all horses included in the study being clinically sound and free of radiographic signs of carpal disease. The increased sensitivity of MRI to detect osteoarthritis-associated structural changes compared to conventional radiography has been previously demonstrated in the equine metacarpophalangeal joint with natural osteoarthritis 17 and canine stifles with experimental osteoarthritis. 18 Because scores were compared over time, differentiation between pre-existing low-grade pathology and pathology that resulted from the induction of post-traumatic osteoarthritis was possible. A limitation of this study was the variability of the size and number of the osteochondral fragments that were created despite strictly adhering to the guidelines set forth by Foland et al. 5 The size of the osteochondral fragments created in the current study varied in both axial to abaxial dimensions as well as the proximal to distal dimensions. In addition, small, discrete osteochondral fragments were still present on MRI examination in 3 OCI joints on day 98. Differences in number, size and continued presence of a separate osteochondral fragment may contribute to the progression and severity of development of osteoarthritis in the OCI joints and may explain the variability in the osteoarthritis-associated structural features reported. Another limitation of this study was the use of consensus agreement to score the MRI images. Consensus agreement was used in this study to limit the subjectivity of the EMOSS scoring system and to apply the system accurately and consistently to all MRI evaluations. Although the number of times the three examiners disagreed was not recorded, it was not frequent. When a difference in score was encountered, the images were reviewed and discussed carefully until a consensus was achieved. 142

143 To the authors knowledge, this was the first study to evaluate the osteoarthritisassociated structural changes that occur over time using a semi-quantitative MRI wholeorgan scoring system in an established model of post-traumatic osteoarthritis in the horse. The EMOSS allowed temporal assessment of high-signal bone lesions, cartilage signal abnormalities, low-signal bone lesions, subchondral bone irregularity, marginal osteophytes, joint effusion, synovial thickening, and other soft tissue abnormalities. The ability of MRI to identify early osteoarthritis-associated structural changes, such as highsignal bone lesions, represents an advantage over other imaging modalities and should be further investigated to determine how these changes relate to clinical outcomes. In conclusion, this method of semi-quantitative analysis of osteoarthritis-associated structural changes may provide an effective and non-invasive way to study disease progression and measure the effects of future disease modifying drugs and therapies. 143

144 Table 3-1. Magnetic Resonance Imaging Parameters Sequence Acquisition Type ST/IG (mm) TE (msec) TR (msec) FA (deg.) TI (msec) NEX FOV (cm) Matrix (Phase x Frequency) Pixel Size (mm) Sagittal PD-FSE 2D 3.0/ n/a 1 19 x x x :48 Axial PD-FSE 2D 3.5/ n/a 1 17 x x x :19 Dorsal T2-FSE 2D 3.0/ n/a 2 16 x x x :37 Axial STIR- FSE 2D 3.5/ x x x :14 Sagittal PD-FS-FSE 2D 3.0/ n/a 1 19 x x x :23 Dorsal PD-FS-FSE 2D 3.0/ n/a 1 16 x x x :17 Dorsal SPGR-FS-GE 3D 2.0/ n/a 1 16 x x x :32 ST/IG = Slice Thickness/Interslice Gap; TE = Echo Time; TR = Repetition Time; FA = Flip Angle; TI = Inversion Time; NEX = Number Of Excitations; FOV = Field Of View; PD = Proton Density; FSE = Fast Spin Echo; STIR = Short Tau Inversion Recovery; FS = Fat Saturation; SPGR = T2*-Weighted Spoiled Gradient Recalled Echo; GE = Gradient Echo Time (min:s) 144

145 Table 3-2. Aligned ranks split plot in time (repeated measures) ANOVA testing for group (G) and day (D) effects. Articular Sub-Regions* Effect RC-2 RC-3 IC-3 IC-4 UC 2C 3C-R 3C-I 4C Total Cartilage SA G D G*D HSBL Intensity G D G*D HSBL Size G D G*D SCBI G D G*D LSBL G D G*D Marginal Osteophyte G D G*D RC-2 = Second Facet Of The Radial Carpal Bone; RC-3 = Third Facet Of The Radial Carpal Bone; IC-3 = Third Facet Of The Intermediate Carpal Bone; IC-4 = Fourth Facet Of The Intermediate Carpal Bone; UC = Ulnar Carpal Bone; 2C = Second Carpal Bone; 3C-R = Radial Facet Of Third Carpal Bone; 3C-I = Intermediate Facet Of Third Carpal Bone; 4C = Fourth Carpal Bone; SA = Signal Abnormality; G = Control, Osteochondral Injured; D = Day 0. 14, 98; HSBL = High Signal Bone Lesion; SCBI = Subchondral Bone Irregularity; LSBL = Low Signal Bone Lesion. P-values reported as may be interpreted as or less. A series of dashes (---) indicates insufficient variation in the scores to conduct the hypothesis test 145

146 Table 3-3. Aligned ranks split plot in time (repeated measures) ANOVA testing for group and day. Group Day Group*Day Total Cartilage Signal Abnormality Score Total HSBL Intensity Score Total HSBL Size Score Total Subchondral Bone Irregularity Score Total LSBL Scores Total Marginal Osteophyte Score Lateral Collateral Ligament Integrity Medial Collateral Ligament Integrity Medial Palmar Intercarpal Ligament Integrity Lateral Palmar Intercarpal Ligament Integrity Osteochondral Fragment Number Osteochondral Fragment Size Joint Effusion Synovial Thickening Whole-Organ Score Group = Control, Osteochondral Injured; Day = Day 0, 14, 98; HSBL = High Signal Bone Lesion; LSBL = Low Signal Bone Lesion. P-values reported as may be interpreted as or less. 146

147 Table 3-4. Median Scores (Min, Max) For Pathological Changes Observed Between 10 OCI Joints and 10 CON Joints On Day 0. Articular Sub-Regions RC-2 RC-3 IC-3 IC-4 UC 2C 3C-R 3C-I 4C Total Cartilage SA CON OCI HSBL-Intensity CON OCI HSBL-Size CON OCI SCBI CON OCI LSBL CON OCI Marginal Osteophyte CON OCI 0 (0,2) 0 (0,4) 0 (0,2) 0 (0,2) 1 (0,2) 1 (0,2) 1 (0,3) 1 (0,1) 0.5 (0,1) 1 (0,2) 1 (0,2) 1 (0,3) 1 (0,3) 0.5 (0,2) 1 (0,2) 0.5 (0,2) 0 (0,2) 0 (0,3) 0 (0,3) 0 (0,2) 2 (0,3) 1 (0,2) 1 (0,3) 1 (0,2) 1 (0,2) 1 (0,2) 0 (0,2) 0 (0,2) 0 (0,5) 0 (0,3) 1 (0,4) 0 (0,4) 6 (2,12) 6 (2,13) 3.5 (2,10) 4 (1,8) RC-2 = Second Facet Of The Radial Carpal Bone; RC-3 = Third Facet Of The Radial Carpal Bone; IC-3 = Third Facet Of The Intermediate Carpal Bone; IC-4 = Fourth Facet Of The Intermediate Carpal Bone; UC = Ulnar Carpal Bone; 2C = Second Carpal Bone; 3C-R = Radial Facet Of Third Carpal Bone; 3C-I = Intermediate Facet Of Third Carpal Bone; 4C = Fourth Carpal Bone; SA = Signal Abnormality; CON = Control Limb; OCI = Osteochondral Injured Limb; HSBL = High Signal Bone Lesion; SCBI = Subchondral Bone Irregularity; LSBL = Low Signal Bone Lesion. Data are reported as median (min, max). a,b,c Within each column, groups with different superscripts are significantly different (p < 0.05). 147

148 Table 3-5. Median Scores (Min, Max) For Pathological Changes Observed Between 10 OCI Joints and 10 CON Joints On Day 14. Articular Sub-Regions RC-2 RC-3 IC-3 IC-4 UC 2C 3C-R 3C-I 4C Total Cartilage SA CON OCI HSBL-Intensity CON OCI HSBL-Size CON OCI SCBI CON OCI LSBL CON OCI Marginal Osteophyte CON OCI 0 (0,2) a a 0 (0,3) 2 (0,3) b 2.5 (1,3) b 0 (0,2) a a 1 (0,3) b 2 (1,3) b 1 (0,2) 1 (0,1) 1 (0,2) 1 (0,2) 0 (0,2) 1 (1,2) 1 (0,3) 1 (0,2) 0 (0,3) 1 (0,2) a 1 (0,2) 0 (0,2) b 0 (0,2) 0 (0,2) 0 (0,2) 0 (0,3) 1 (0,2) 0.5 (0,2) 0 (0,2) 0 (0,2) 2 (1,2) 1 (0,2) 1 (0,2) 1 (0,2) 0 (0,4) 0 (0,2) 0.5 (0,4) a 6 (2,8) b 0.5 (0,6) a 4 (2,8) b 1.5 (0,3) 1 (0,4) 7.5 (3,12) 4.5 (2,15) RC-2 = Second Facet Of The Radial Carpal Bone; RC-3 = Third Facet Of The Radial Carpal Bone; IC-3 = Third Facet Of The Intermediate Carpal Bone; IC-4 = Fourth Facet Of The Intermediate Carpal Bone; UC = Ulnar Carpal Bone; 2C = Second Carpal Bone; 3C-R = Radial Facet Of Third Carpal Bone; 3C-I = Intermediate Facet Of Third Carpal Bone; 4C = Fourth Carpal Bone; SA = Signal Abnormality; CON = Control Limb; OCI = Osteochondral Injured Limb; HSBL = High Signal Bone Lesion; SCBI = Subchondral Bone Irregularity; LSBL = Low Signal Bone Lesion. Data are reported as median (min, max). a,b,c Within each column, groups with different superscripts are significantly different (p < 0.05). 0 (0,2) 0 (0,2) 4.5 (1,7) 4 (0,5) 148

149 Table 3-6. Median Scores (Min, Max) For Pathological Changes Observed Between 10 OCI Joints and 10 CON Joints On Day 98. Articular Sub-Regions RC-2 RC-3 IC-3 IC-4 UC 2C 3C-R 3C-I 4C Total Cartilage SA CON OCI HSBL-Intensity CON OCI HSBL-Size CON OCI SCBI CON OCI LSBL CON OCI Marginal Osteophyte CON OCI 0 (0,2) 0 (0,) 0 (0,2) 1 (0,2) 1 (0,2) 1 (0,3) 1.5 (1,3) a 2 (0,2) b 0 (0,2) 1 (0,2) 1 (0,1) a 2.5 (1,4) b 0 (0,2) 1 (0,2) a 1 (0,2) 2 (1,3) b 1 (0,2) 1.5 (1,3) 2 (1,3) 0 (0,2) 0.5 (0,1) 0.5 (0,1) 0.5 (0,2) 0 (0,2) 0 (0,2) 1 (0,2) 0.5 (0,1) 1 (0,2) 0 (0,2) 1 (0,1) 1 (0,3) 1.5 (0,2) 1 (0,2) 0.5 (0,2) 1 (0,2) 1 (0,3) 1 (0,2) a 3 (0,5) b a 1.5 (0,4) b a 1 (0,3) b 1 (0,3) a 4 (1,8) b 7 (2,12) 8 (2,15) 4.5 (2,7) 6 (2,13) RC-2 = Second Facet Of The Radial Carpal Bone; RC-3 = Third Facet Of The Radial Carpal Bone; IC-3 = Third Facet Of The Intermediate Carpal Bone; IC-4 = Fourth Facet Of The Intermediate Carpal Bone; UC = Ulnar Carpal Bone; 2C = Second Carpal Bone; 3C-R = Radial Facet Of Third Carpal Bone; 3C-I = Intermediate Facet Of Third Carpal Bone; 4C = Fourth Carpal Bone; SA = Signal Abnormality; CON = Control Limb; OCI = Osteochondral Injured Limb; HSBL = High Signal Bone Lesion; SCBI = Subchondral Bone Irregularity; LSBL = Low Signal Bone Lesion. Data are reported as median (min, max). a,b,c Within each column, groups with different superscripts are significantly different (p < 0.05). 149

150 Table 3-7. Total Individual Articular And Whole-Organ Scores Observed In 10 OCI Joints And 10 CON Joints On Days 0, 14, And 98. Groups Day 0 Day 14 Day 98 Total Cartilage Signal Abnormality Score CON OCI Total HSBL Intensity Score CON OCI Total HSBL Size Score CON OCI Total Subchondral Bone Irregularity Score CON OCI Total LSBL Scores CON OCI Total Marginal Osteophyte Score CON OCI Lateral Collateral Ligament Integrity CON OCI Medial Collateral Ligament Integrity CON OCI Dorsomedial Intercarpal Ligament Integrity CON OCI Medial Palmar Intercarpal Ligament Integrity CON OCI Lateral Palmar Intercarpal Ligament Integrity CON OCI Osteochondral Fragment Number CON OCI Osteochondral Fragment Size CON OCI Joint Effusion CON OCI Synovial Thickening CON OCI Whole-Organ Score CON OCI 0(0,4) 0 (0,2) 0 (0,2) 0 (0,5) 0 (0,3) 1.0 (1,4) 0 (0,4) 6 (2,12) 6 (2,13) 3.5 (2,10) 4 (1,8) 0 (0,0 0 (0,0 0, (0,0) 0 (0,2) 0 (0,2) 0.5 (0,2) 1 (0,2) 0 (0,4) 0 (0,2) 0.5 (0,4) a 6 (2,8) b 0.5 (0,6) a 4 (2,8) b 1.5 (0,3) 1 (0,4) 7.5 (3,12) 4.5 (2,15) 4.5 (1,7) 4 (0,5) 1 (0,2) a 3 (0,5) b a 1.5 (0,4) b a 1 (0,3) b 1 (0,3) a 4 (1,8) b 7 (2,12) 8 (2,15) 4.5 (2,7) 6 (2,13) a 0.5 (0,1) b a 2 (1,3) b a 2 (1,3) b 1 (0,2) 1 (0,2) 1 (0,2) 1 (0,2) 0 (0,4) 0 (0,3) 1 (0,2) 1 (1,3) 1 (0,2) a 1 (1,3) b 16.5 (5, 28) 21.5(9,31) a 16.5 (4,22) 29(17,45) b HSBL = High Signal Bone Lesion; LSBL = Low Signal Bone Lesion. Data are reported as median (min, max). 20 (5,28) a 31.5(14,50) b a,b,c Within each column, groups with different superscripts are significantly different (p < 0.05). 150

151 Table 3-8. Median Scores (Min, Max) For Pathological Changes Observed Within 10 CON And 10 OCI Joints On Days 0, 14, And 98. Cartilage SA Day 0 Day 14 Day 98 HSBL-Intensity Day 0 Day 14 Day 98 HSBL-Size Day 0 Day 14 Day 98 SCBI Day 0 Day 14 Day 98 LSBL Day 0 Day 14 Day 98 Marginal Osteophyte Day 0 Day 14 Control Limbs Osteochondral Injured Limbs RC-2 RC-3 2C 3C-R RC-2 RC-3 2C 3C-R 0 (0,2) 0 (0,2) 0 (0,2) 1 (0,2) 1 (0,2) 1 (0,2) 1 (0,3) 1 (0,2) 1 (0,3) 0.5 (0,1) 1 (0,2) 1 (1,2) 1 (0,2) 1 (0,3) 1 (0,2) 1.5 (1,3) 0 (0,3) 0 (0,3) 0 (0,3) 1 (0,2) 1 (0,2) 0(0,1) 0 (0,2) 0 (0,2) 0 (0,2) 0 (0,2) 1 (0,1) 2 (0,3) 2 (1,2) 1 (0,3) 1 (0,3) 0 (0,2) 1 (0,2) 0 (0,2) a 2 (0,3) b a 0 (0,2) a 1 (0,3) b a 0 (0,2) 1 (0,2) 1 (0,1) 1 (0,2) 1 (0,1) a 1 (0,2) b 1.5 (1,3) b a 0 (0,2) a 2 (0,2) b a 2.5 (1,3) b 1 (0,2) c a 2 (1,3) b 1 (0,1) c a 0 (0,2) a 2.5 (1,4) b 1 (0,2) 1 (0,3) 2 (1,3) 1 (0,3) a,b 0 (0,2) a 2 (1,3) b 0.5 (0,2) 0.5 (0,1) Day 98 RC-2 = Second Facet Of The Radial Carpal Bone; RC-3 = Third Facet Of The Radial Carpal Bone; IC-3=Third Facet Of The Intermediate Carpal Bone; IC- 4=Fourth Facet Of The Intermediate Carpal Bone; UC=Ulnar Carpal Bone; 2C=Second Carpal Bone; 3C-R=Radial Facet Of Third Carpal Bone; 3C- I=Intermediate Facet Of Third Carpal Bone; 4C=Fourth Carpal Bone; SA = Signal Abnormality; HSBL = High Signal Bone Lesion; SCBI = Subchondral Bone Irregularity; LSBL = Low Signal Bone Lesion. Data are reported as median (min, max). a,b,c Within each column, groups with different superscripts are significantly different (p < ). a a 1 (0,2) b 1 (0,1) 1 (0,2) 1 (0,2) 1.5 (0,2) 1 (0,2) 0 (0,2) 0.5 (0,2) 151

152 Figure 3-1. Diagram of the 9 sub-regions that were evaluated using EMOSS. RC- 2: Second Facet Of The Radial Carpal Bone; RC-3: Third Facet Of The Radial Carpal Bone; IC-3: Third Facet Of The Intermediate Carpal Bone; IC-4: Fourth Facet Of The Intermediate Carpal Bone; UC: Ulnar Carpal Bone; 2C: Second Carpal Bone; 3C-R: Radial Facet Of The Third Carpal Bone; 3C-I: Intermediate Facet Of The Third Carpal Bone; 4C: Fourth Carpal Bone. 152

153 Figure 3-2. Schematic representation of the scoring system used for high-signal bone lesion intensity using the sagittal PD-FS (A,D,G), dorsal PD-FS (B,E,H) and axial T2-STIR (C,F,I) sequences. The intensity of increased signal for each high-signal bone lesion were subjectively scored based on severity of signal increase compared to the surrounding trabecular and subchondral bone and graded as 0 = none, 1 = mild (A,B,C), 2 = moderate (D,E,F), 3 = marked (G,H,I). 153

154 Figure 3-3. Schematic representation of the scoring system used for synovial thickening using axial T2-STIR images. On T2-STIR sequence, synovium was recognized as tissue of moderate signal intensity protruding into the hyperintense synovial fluid. Synovial thickening was scored based on the intensity and distribution of synovium identified and graded as: 0 = normal (A), 1 = mild - mild increase in either intensity or distribution (B), 2 = moderate - mild increase in both intensity and distribution or moderate increase in either intensity or distribution (C), 3 = marked - moderate increase in both intensity and distribution, marked increase in either intensity or distribution, or presence of synovial adhesion(s) (D). 154

155 Figure 3-4. Sagittal PD-FS (A-C) and PD (D-F) MRI images of an OCI joint at day 0 (A,D), 14 (B,E), and 98 (C,F) showing the progression of signal changes at the osteochondral fragmentation site in third facet of the radial carpal bone. An osteochondral fragment is present on day 14 that is delineated from the parent bone by a linear pattern of marked increased signal intensity within the fracture line in the PD-FS image (B) and an intermediate to decreased signal intensity on the PD image (E). There is marked increased signal intensity (grade 3 high-signal bone lesion) adjacent to the fracture line in the RC-3 in the PD-FS image (B). A separate osteochondral fragment can no longer be appreciated on day 98 in either image (C, F). The high-signal bone lesions have significantly decreased in size and signal intensity (grade 1 high-signal bone lesion) in the third facet of the radial carpal bone adjacent to and within the healing fracture fragment on day 98 on the PD-FS images (C), however, there is markedly decreased signal intensity (grade 3 low-signal bone lesion) on the PD image (F) on day

156 Figure 3-5. Dorsal PD-FS (A-C) and Axial T2-STIR (D-F) MRI images of an OCI joint at day 0 (A,D), 14 (B,E), and 98 (C,F) showing the progression of high-signal bone lesions (A-F) and synovial thickening in the middle carpal joint (D-F). Note the focal, large region of increased signal intensity (grade 2 high-signal bone lesion - size, grade 3 high-signal bone lesion - signal intensity) on day 14 (B), with partial resolution at day 98 (C). Increase in synovial thickening is seen in the axial T2-STIR image on day 98 (F) dorsally as an increased amount of low and intermediate signal intensity synovial tissue within high signal intensity joint fluid. 156

157 Figure 3-6. Sagittal PD-FS MRI images of an OCI joint at day 0 (A) and day 14 (B) showing the increase in joint effusion identified on day 14 (B) where increased volume of hyperintense joint fluid and cranial displacement of the joint capsule is seen within the middle carpal joint. Figure 3-7. Dorsal T2-weighted MRI images of an OCI joint at day 0 (A) and 98 (B) demonstrating the progression of low-signal bone lesions (a) and marginal osteophyte formation (arrow) in the third facet of the radial carpal bone. A large angular osteophyte (grade 3 marginal osteophyte) is present in the distomedial aspect of the third facet of the radial carpal bone with an adjacent area of low signal intensity (grade 3 lowsignal bone lesion) (B). 157

158 Figure 3-8. Dorsal SPGR-FS MRI images of an OCI joint demonstrating the progression of subchondral bone irregularity and cartilage signal abnormality on the third facet of the radial carpal bone on day 0 A), 14 B), 98 C). On days 0 and 14, no subchondral bone irregularity or cartilage signal abnormalities are seen, however, the osteochondral fragment is evident on day 14 (B) by a linear intermediate signal intensity separating it from the parent bone. There is also increased signal intensity within the third facet of the radial carpal bone within and surrounding the fracture fragment. On day 98, a separate osteochondral fragment is no longer evident; however, there is a focal area of increased signal within a defect in the subchondral bone that is continuous with the joint fluid. This is an example of grade 3 subchondral bone irregularity and grade 2 cartilage signal abnormality. 158

159 CHAPTER 4 VALIDATION OF AN EQUINE MRI OSTEOARTHRITIS SCORING SYSTEM USING AN EQUINE MODEL OF POST-TRAUMATIC OSTEOARTHRITIS Introduction Although once considered a disease of articular cartilage, osteoarthritis is now considered a disease of the whole joint. 1 Injury to one component can lead to injury or loss of function to other components of the joints eventually leading to organ failure and clinical manifestations of osteoarthritis. Given this, diagnostic imaging of the joint should be able to directly visualize all structures of the joint including the bone, cartilage, menisci, ligaments and synovial tissue simultaneously, as this will provide a much more detailed picture of the pathological changes associated with osteoarthritis. 367 In addition, the imaging modality should be able to detect changes associated with osteoarthritis as early as possible, as well as be able to tract their progression throughout the course of the disease in order to improve our understanding of the disease process and to facilitate clinical decisions. 33,345 Conventional radiography is the most common imaging modality used in the clinic and research setting for evaluation of osteoarthritis, but is limited in its ability to evaluate all structures of the joint and is neither sensitive nor specific to the early changes of osteoarthritis. 17,18,287 Magnetic resonance imaging (MRI), however, allows unparalleled visualization of all structures believed to be involved in osteoarthritis joint pathology due to its tomographic viewing perspective and has the ability to detect pathology at a much earlier stage of the disease process. 17,18,345,368 Semi-quantitative scoring systems provide a standardized method to assess a variety of features identified on MRI that are currently believed to be 159

160 involved in the pathogenesis of osteoarthritis. 369 Each score can then be used to generate a whole-organ score that can be compared over time or used as an outcome measure. They have proven to be invaluable in the study of the natural history of osteoarthritis, in relating clinical symptoms to pathological features, to identify risk factors for structural changes, and to assess the effectiveness of interventions for osteoarthritis. 367,369 In the evaluation of equine joint disease, grading of similar articular features using semiquantitative scoring systems have shown to be reliable, correlate well to gross articular cartilage scores, and therefore deserve inclusion in a whole-organ scoring system. 17,328 Recently, a semi-quantitative whole-organ MRI scoring system was developed for use in the equine middle carpal joint and was used to describe the pathological changes that occur over time using the equine carpal osteochondral fragment model of post-traumatic osteoarthritis. 370 The reliability and validity of this scoring system should be assessed prior to its use in assessing the effectiveness of interventions in clinical osteoarthritis research. 367 The equine middle carpal joint provides advantages over the metacarpophalangeal joint as it allows near complete visualization of its articular surfaces by arthroscopy 371,372, which allows for the assessment of articular cartilage over time and enables validation of the equine MRI osteoarthritis scoring system in the live horse. The purpose of this study was to assess the diagnostic value of a semiquantitative MRI scoring system to identify pathologic changes over time in an experimental model of osteoarthritis. The first objective of this study was to test the reliability of the equine MRI osteoarthritis scoring system in detection of pathology. The 160

161 second objective of this study was to validate the equine MRI osteoarthritis scoring system by comparing MRI to the structures that can visualized by arthroscopy, using arthroscopy as the gold standard. We hypothesize that semiquantitative MRI scores will show good reliability between observers and that MRI scores will also correlate to intra-articular pathology as detected by arthroscopy. Materials and Methods The protocol for this experimental study was approved by the University of Florida Institutional Animal Care and Use Committee. Horses were part of the University of Florida Research Herd and consisted of a control (untreated) group from a seperate study evaluating the therapeutic effects of a novel treatment for osteoarthritis. Study horses were included in the study if they were free of lameness and lacked clinical and radiographic abnormalities in either carpus. 370 Prior to the start of the study, all horses were exercised 5 days/week on a highspeed treadmill for 21 days. For each exercise day, the horses underwent trotting (16-18 km/h) for 2 min, galloping (30-32 km/h) for 2 min, followed by trotting (16-18 km/h) for two additional minutes to simulate race training. On day 0, magnetic resonance images of both carpi were obtained with the horses anesthetized and placed in left lateral recumbency using a 1.5 T highfield magnet (Titan, Toshiba America Medical Systems, Tustin, CA). During examination, each carpus was placed in partial flexion (15-25 ) and scanned individually using a quadrature transmit/receive knee coil (Toshiba America Medical Systems, Tustin, CA). The following sequences were acquired: sagittal proton density (PD), axial PD, dorsal T2-weighted (T2), axial T2 short-tau 161

162 inversion recovery (T2-STIR), sagittal proton density with fat suppression (PD- FS), dorsal PD-FS, and dorsal T2*-weighted spoiled gradient echo with fat suppression (SPGR-FS). These sequences are the same sequences that are used clinically in our hospital and optimized for ease of evaluation in a clinical setting based on the authors experience with this system. (Table 4-1) Total acquisition time was approximately 80 minutes for all sequences on both limbs. Immediately following MRI examination, the horses were switched to dorsal recumbency and complete arthroscopic examinations of both middle carpal joints were performed with a 4mm, 30 arthroscope (Smith and Nephew, Andover, MA) using the standard arthroscopic approach to the middle carpal joint. 348 All arthroscopic examinations were video recorded for review at a later date. An osteochondral fragment was then created on the dorsodistal aspect of the third facet of the radial carpal bone in one randomly assigned joint using an 8mm curved bone gouge directed perpendicular to the articular surface. The osteochondral fragment remained adhered to the medial synovial plicae proximally and any joint debris that resulted from creation of the osteochondral fragment was left within the joint as previously described. 5 The contralateral middle carpal joint was sham-operated. The arthroscopic portals were closed using 2-0 polydioxanone in a simple, interrupted pattern. Each horse was administered intravenous potassium penicillin (22,000 IU/kg), gentamicin (6.6 mg/kg) and phenylbutazone (4.4mg/kg) pre-operatively and then intravenous phenylbutazone (4.4mg/kg) once a day for an additional 4 days. Horses were housed in 3.65 m x 3.65 m stalls from day 0 to 10 during which time the 162

163 bandages were changed and the horse s vital parameters were monitored daily. Sutures were removed on day 10 and the horses returned to paddock turnout. On day 14, a second MRI was performed on all horses before resuming exercise on a high-speed treadmill until the completion of the study as previously described. Arthroscopic and MRI examinations were repeated at the end of the study (day 98) for both carpi as described above. Equine MRI Osteoarthritis Scoring System All MR images were transferred and stored in Digital Imaging and Communications in Medicine (DICOM) format on a Picture Archive and Communication System (Merge Healthcare, Chicago, IL). Images were assessed by three observers including a board-eligible large animal surgeon in postgraduate studies (observer A), a board-certified surgeon (observer B), and a board-certified radiologist (observer C) using a dedicated DICOM workstation equipped with three megapixel greyscale monitors. A semi-quantitative wholeorgan scoring system developed for use in the equine middle carpal joint was used to score each MRI examination as previously described. 370 (Table 4-2) Images were scored ordinally based on fifteen independent articular features including presence of cartilage signal abnormalities (0-3) (Figure 4-1); size (0-3) and intensity (0-3) of high-signal bone lesions (Figure 4-2); subchondral bone irregularity (0-4) (Figure 4-3); low-signal bone lesions (0-3) (Figure 4-4); marginal osteophytes (0-3) (Figure 4-5); medial (0-1) and lateral collateral ligament (0-1), dorsomedial intercarpal ligament (0-1) (Figure 4-6) and medial (0-1) and lateral palmar intercarpal ligament (0-1) (Figure 4-7) integrity; number (0-3) and size (0-3) of osteochondral fragment(s); synovial thickening (0-3) (Figure 4-8); and joint 163

164 effusion (0-3) (Figure 4-9). Six of the articular features (cartilage signal abnormalities, high-signal bone lesion size and intensity, low-signal bone lesion, subchondral bone irregularity, and marginal osteophytes) were evaluated independently at the 9 different sub-regions in the middle carpal joint. (Figure 4-10) Only abnormalities originating from the middle carpal joint were scored and any abnormalities involving or extending from the radiocarpal or carpometacarpal joint were ignored. Final scores were totaled to determine a whole-organ score for the entire middle carpal joint (0-188). All images were used to evaluate each feature, although certain sequences were more suitable for grading individual articular features than others and were reported accordingly. All 60 MR joint image sets (10 OCI and 10 CON limbs on days 0, 14, and 98) were initially scored by consensus agreement of three observers (consensus 1). Any discrepancies in scores by one or more observers were reviewed again until a final score was assigned by consensus of all three observers. A subset of these images were rescored to determine the reliability of the equine MRI osteoarthritis scoring system. A sample size was determined by comparison of means (OCI mean + standard deviation: , CON: ) from data from a previous study 370 performed by the authors using OpenEpi ( To obtain a confidence interval of 95% and a power of 80%, a sample size of 9 OCI joints and 9 CON joints was obtained. Assuming a 10% less difference in mean (OCI: , CON: ), 13 MRI images were randomly picked from both the OCI and CON joints over the three time periods (days 0, 14, and 98) using research randomizer ( 164

165 Images were rescored two years after the initial observation. Twenty-six (13 OCI and 13 CON limbs) of the sixty total MRI examinations were randomly chosen from the three different time periods (days 0, 14, and 98) and scored independently by three observers (observers A, B, C) to assess inter-reader reliability. The same twenty-six MRI examinations were subsequently rescored by the same three observers using consensus agreement (consensus 2) as described above and compared to the initial consensus agreement scores (consensus 1) to assess intra-reader reliability. At all time points the observers were unaware of the identity of the experimental group. Arthroscopic Scoring System Arthroscopic examination was defined as the gold standard for detecting intra-articular pathology and was used for comparison with MRI examination. Videos of each arthroscopic examination were scored by consensus agreement by a board-eligble large animal surgeon in postgraduate studies (observer A) and a board-certified surgeon (observer D) using a high-resolution monitor (Apple Thunderbolt Display Monitor, Apple, Cupertino, CA). An ordinal arthroscopic grading scale was developed in parallel with the equine MRI osteoarthritis scoring system and used to score 7 articular features, which included articular cartilage lesion size; marginal osteophyte; dorsomedial intercarpal ligament and medial and lateral palmar intercarpal ligament integrity; synovial thickening; and osteochondral fragment number. Similar to the MRI semi-quantitative scoring system, the articular surface of the middle carpal joint was divided into 9 subregions. (Figure 4-10) Two of the 8 articular features (articular cartilage lesion 165

166 size and marginal osteophyte formation) were evaluated independently at the 9 different sub-regions. Articular cartilage lesions were scored based on the size of the largest dimension of the lesion within each sub-region and graded as: 0 = no abnormalities, 1 = single lesion < 5 mm, 2 = single lesion 5-10 mm or multiple lesions < 5 mm, or 3 = multiple lesions with at least one lesion > 10 mm. (Figure 4-11) Marginal osteophytes were scored subjectively within each sub-region based on their size and graded as 0 = no abnormalities, 1 = small (< 1 mm), 2 = medium (1-2 mm), or 3 = large (> 2 mm) (Figure 4-12). The dorsomedial intercarpal ligament was scored based on its size and graded as: 0 = less than 2 mm in width (dorsolateral-dorsomedial) and blends into the joint capsule or 1 = greater than 2 mm in width or easily differentiated from the joint capsule over its entire length. (Figure 4-13) The medial and lateral palmar intercarpal ligaments were independently assessed by careful probing the ligament to determine the degree of fiber disruption and graded as: 0 = normal or mild mild surface fibrillation or 1 = obvious evidence of ligament disruption. (Figure 4-14) Synovial thickening was subjectively scored based on both the overall thickness of the synovial villi and the presence of new villi formation and graded as 0 = normal, 1 = mild increase in either the thickness or new villi formation, 2 = mild increase in thickness and new villi formation or moderate increase in thickness or new villi formation, or 3 = moderate increase in thickness and new villi formation or marked increase in thickness or new villi formation or the presence of synovial adhesion(s). (Figure 4-15) Each joint was scored according to the number of 166

167 osteochondral fragments present and graded as: 0 = none, 1 = 1 fragment, 2 = 2 fragments, or 3 = > 2 fragments. Final scores were totaled to determine a wholeorgan score for the entire middle carpal joint (0-63). Separate from the arthroscopic scoring system, the most severe articular cartilage lesion within each sub-region was further categorized by its shape and graded as: 0 = no abnormalities, 1 = focal lesion with all dimensions < 5 mm, 2 = linear lesion with one dimension < 5 mm and the opposite dimension > 5 mm, or 3 = circular lesion with two dimension > 5 mm. The most severe lesion within each sub-region was also categorized by the depth of the lesion and graded as: 0 = smooth and regular cartilage, 1 = superficial (superficial softening, fissure, fibrillation and/or crack), 2 = partial thickness (partial thickness erosion or deep fibrillation not extending to subchondral bone), 3 = full thickness (full thickness defect with visualization of the subchondral bone plate), or 4 = full thickness defect filled with fibrocartilage. (Table 4-3) Statistical Analysis The semi-quantitative MRI scores were summarized by percent involvement (grade > 0, inclusively) within each articular feature and within each sub-region. Inter-reader reliability for semiquantitative MRI scores was first assessed by calculating a weighted-kappa coefficient between each of the three observers (A vs. B, A vs. C, B vs. C) for each of the articular features. 373 The three weighted-kappa coefficients were then averaged to obtain one weightedkappa coefficient (A vs. B vs. C) for each articular feature. 374 Intra-reader reliability for semiquantitative MRI scores was assessed by calculating the weighted-kappa coefficients between consensus agreements (consensus 1 vs. 167

168 consensus 2) for each articular feature. 373 Agreements for both inter- and intrareader reliability were described as slight (0 0.2), fair ( ), moderate ( ), substantial ( ), or almost perfect (0.81 1). 375 For intermodality comparison of intra-articular features (articular cartilage lesion size, marginal osteophyte, dorsomedial intercarpal and medial and lateral palmar intercarpal ligament integrity, osteochondral fragment, and synovial thickening), consensus data (consensus 1) for MRI semiquantitative scores and arthroscopic semiquantitative scores on days 0 and 98 were used. Total articular feature scores for MRI and arthroscopic semiquantitative scores were determined for each joint by adding the scores from each sub-region. Total joint MRI and arthroscopy scores were determined by adding up semiquantitative scores from each articular feature and sub-region. Mean + standard deviation [range] total articular feature scores were calculated for articular cartilage lesion size, high signal bone lesion, subchondral bone irregularity, low signal bone lesion, marginal osteophyte, and total joint scores. Median [minimum, maximum] scores were calculated for lateral and medial collateral ligaments, dorsomedial intercarpal ligament, medial palmar intercarpal ligament, lateral palmar intercarpal ligament, synovial thickening, and joint effusion scores. Total articular cartilage, marginal osteophyte, synovial thickening, and total joint scores were compared between MRI and arthroscopy using Wilcoxon s signed-rank test. Sensitivity, specificity, and predictive values for detecting intra-articular pathology (grades 1-3 inclusively) on MRI compared with arthroscopic examination were determined. Spearman s rank tests were used to compare total 168

169 articular cartilage, marginal osteophyte, synovial thickening and total joint scores between MRI and arthroscopy semiquantitative scoring systems. An r value lower than 0.3 indicated little or no association, between 0.3 and 0.7 indicated moderate association, and above 0.7 indicated strong association. A value of p < 0.05 was considered significant for all tests. Statistical analysis was performed by using MedCalc for Windows, Version 15.1 (MedCalc Software, Ostend, Belgium). Results Ten, skeletally mature, Thoroughbred horses (mean age: 4.5 years, min: 3.0, max: 9.0) were included in the study with an equal number of left and right thoracic limbs in the OCI and CON groups. Table 4-4 shows the frequency of involvement (scores > 0) of each articular feature using pooled data from consensus 1 of both OCI and CON limbs on days 0, 14 and 98. The second and third facet of the radial carpal bone and the radial facet of the third carpal bone showed abnormalities more than ninety percent of the time. Low-signal bone lesions were present in 100% of the middle carpal joints, with the radial facet of the third carpal bone being the most frequent site present (91.7%). Ninety-eight percent of the middle carpal joints showed marginal osteophytes on the second and third facet of the radial carpal bone being the most common sites present (93.3 and 98.3%, respectively). Forty-eight percent of the middle carpal joints showed high-signal bone lesions, with the third facet of the radial carpal bone being most common sites present (35%). Subchondral bone irregularities and cartilage signal abnormalities were present in 15.9% and 7.4% of middle carpal joints, respectively, with the third facet of the radial carpal bone and radial facet of the third carpal bone being most common sites present. Ligamentous 169

170 abnormalities were identified in 55% of the middle carpal joints with the dorsomedial intercarpal (26.7%), the medial collateral (20.0%), and the medial palmar intercarpal (20.0%) ligaments being the most common sites present. Joint effusion was seen in 70% of the middle carpal joints and synovial thickening in 66% of the middle carpal joints. Osteochondral fragment(s) were present in 21% percent of middle carpal joints. Reliability The assessment of inter-reader and intra-reader reliability of the equine MRI osteoarthritis scoring system is shown in Table 4-5. When the assessment of agreement was averaged across the three observers, there was substantial to almost perfect inter-reader agreement (kappa range: to 1.000) for all of the articular features evaluated. Intra-reader agreement was moderate to substantial (kappa range: to 0.780) at scoring all articular features, with the exceptions of scoring the medial collateral ligament (kappa: 0.843) and osteochondral number (kappa: 1.000) and size (kappa: 0.924), where agreement was almost perfect. Validity Articular cartilage lesions were detected in 117 of the 360 sub-regions evaluated by arthroscopy. Their prevalence from all 9 sub-regions was 67.5%, 21.1%, 8.3%, and 3.1% for grades 0, 1, 2, and 3; respectively. Marginal osteophytes were detected in 69 of 360 sub-regions with a prevalence of 80.8% 12.8%, 5.8%, and 1.1% for grades 0, 1, 2, and 3; respectively. Fourteen dorsomedial intercarpal ligaments, 11 medial palmar intercarpal ligaments, and 7 lateral palmar intercarpal ligaments were considered abnormal on arthroscopy 170

171 out of 40 limbs. Synovial thickening was present in 34 out of 40 limbs where 15% were graded as 0, 62.5% grade 1, 17.5% grade 2, and 5% grade 3. Osteochondral fragment(s) were identified in 2.5% (1/40) of limbs on arthroscopy. Total semiquantitative articular feature scores for the equine MRI osteoarthritis scoring system and arthroscopic scoring system are presented in Table 4-6. Total articular cartilage lesion scores were significantly higher on arthroscopy than they were on MRI (p < ). Total marginal osteophyte (p < ) and total joint scores (p < ) were significantly higher on MRI than they were on arthroscopy. There was no significant difference between MRI and arthroscopy when scoring synovial thickening. Table 4-7 shows the sensitivity, specificity, and predictive values of MRI to detect intra-articular pathology as determined by arthroscopy. Of the 117 subregions where articular cartilage lesions were identified by arthroscopy, only 28 were correctly identified with MRI. Using MRI, observers identified 15.8% (12/76), 23.3% (7/30), and 81.8% (9/11) of the arthroscopic grade 1, 2, and 3 articular cartilage lesions, respectively. When the articular cartilage lesions were categorized by depth, observers identified 5.8% (1/17) of superficial lesions, 23.3% (10/43) of partial thickness lesions, 20.0% (3/15) of full thickness lesions, and 33.3% (14/42) of full thickness lesions partially filled with fibrocartilage using MRI. Lastly, observers identified 16.9% (13/77) of focal lesions, 28.6% (10/35) of linear lesions, and 100% (5/5) of circular lesions using MRI when articular lesions were categorized by shape. 171

172 Total articular cartilage (r = , p < ) and synovial thickening (r = , p = ) scores were moderately correlated between MRI and arthroscopy, but total marginal osteophytes did not show a significant relationship (r = , p = ). Total joint MRI scores were also moderately correlated to total arthroscopic articular cartilage scores (r = , p = ) and total joint arthroscopic scores (r = , p = ). Discussion The diagnostic value of any outcome measure used in osteoarthritis research is determined by its ability to identify the progression of osteoarthritisassociated structural changes over time and its ability to do so in a repeatable fashion. 335 The equine MRI osteoarthritis scoring system used in this study has previously been shown to be able to assess these structural changes over time using a model of post-traumatic osteoarthritis. 370 The results of the current study build on previous findings and provide an objective assessment of the reliability and validity of this semiquantitative whole-organ scoring system. The equine MRI osteoarthritis scoring system was shown to be a reliable scoring method and to be positively correlated with arthroscopic findings, however, it was variably sensitive and specific in detecting individual articular features when using arthroscopy as the gold standard. The first objective of this study was to test the reliability of the equine MRI osteoarthritis scoring system in the detection of articular pathology. The average agreement between three different observers was substantial to perfect for all articular features tested (kappa range: to 1.000). These results are similar to inter-reader reliability results obtained in human semiquantitative MRI scoring 172

173 systems studies 335, and would seem to be superior to previous results obtained from studies looking at individual articular features in equine cadaver fetlocks. 17,328 Direct comparison between studies, however, should be performed with caution. The severity of osteoarthritis in previous studies 17,328 was greater and covered a much wider range of structural changes than that of the current study. The disparity in disease prevalence between study populations will make direct comparison of the kappa coefficient difficult. 376 The intra-reader reliability was not as impressive as the inter-reader reliability, although there was still moderate to substantial agreement. This is contradictory to what commonly occurs and it differs from previous studies using semiquantitative scoring systems. 17,28,328 Increased experience with the use of the scoring system is the most likely explanation for a lower intra-reader reliability when compared to inter-reader reliability. The first time the equine MRI osteoarthritis scoring system was used by the three observers was during the first consensus agreement (consensus 1) two years earlier. These scores were then compared to consensus 2, which was actually the third time the three observers had scored the MRI images as they had scored them individually (observer A, B, C) prior to the second consensus agreement. It is common for scoring systems to become more reliable or scores to become more repeatable the more times an observer uses the scoring system and that is likely the case in this instance. Despite all three observers having experience in reading MR images, this type of scoring system was a new way of evaluating osteoarthritisrelated structural changes. A longer training session prior to scoring the first set 173

174 of images for the study may have increased the level of intra-reader reliability to a level similar or better then inter-reader reliability. The second objective of this study was to determine the validity of the equine MRI osteoarthritis scoring system by comparing it to arthroscopy. The sensitivity of MRI to detect articular cartilage lesions was poor (23.9%) and is inferior to results found in previous studies in people 377 and horses. 328 The articular cartilage lesions in the study by Olive et al. 328 were more advanced than those found in the current study. It is reasonable to assume that the sensitivity of MRI to detect articular cartilage lesions will increase with increasing severity as has been shown in MRI of the knees of people. 378 To better identify the cause of poor sensitivity in the current study, the arthroscopically-identified articular cartilage lesions were further categorized by their shape and depth in addition to the size of their largest dimension. Contrary to what the authors expected, there was subjectively very little difference in MRI s ability to detect articular cartilage lesions when they were categorized according to the depth of the lesion, although MRI was mildly more sensitive at detecting partial and full-thickness lesions as compared to superficial lesions. This is likely because the normal thickness of cartilage in this region in horses is very thin (1-2 mm). There was, however, a noticeable difference when articular lesions were categorized by their size and shape. Magnetic resonance imaging detected 81.8% of grade 3 lesions and 100% of circular lesions compared to less than 25% of grade 1 or 2 lesions and less than 30% of focal or linear lesions. As a majority of lesions were either grade 1 and 2 lesions (106/117) when categorized by size and were focal or 174

175 linear lesions (112/117) when categorized by shape, it stands to reason that the poor sensitivity is more related to the smaller size and shape of the majority of lesions identified in this study, although all three factors likely play a role. Decreasing the voxel size by decreasing the MRI slice thickness or matrix may have increased the sensitivity of detecting some of these small articular cartilage lesions by increasing the spatial resolution of the images and by minimizing the partial volume-averaging artifact. These changes in MRI sequence parameters, however, would result in longer scan times as well as a corresponding decrease in signal-to-noise ratio that can also limit lesion detection. 379,380 The current study graded articular cartilage lesions on MRI based solely on its largest dimension, as opposed to previous equine studies where articular cartilage lesions were scored based on depth. 328 In human MRI osteoarthritis semiquantitative scoring systems, articular cartilage lesions are given a single score based on both the size and depth of the lesion. 25 The authors agree that scoring articular cartilage lesions by both their size and depth would better identify the progressive loss of articular cartilage that occurs over time. However, there is conflicting evidence regarding the ability to accurately measure the thickness of normal equine hyaline cartilage using MRI 327,328,381 and less than satisfactory evidence that MRI can be used to accurately measure the depth of articular cartilage lesions in horses. 320,328 In people, MRI is able to accurately measure the thickness of articular cartilage because the articular cartilage in human s knees is much thicker than it is in the equine middle carpal joint. It is also possible in people to manipulate imaging parameters in order to increase 175

176 the spatial resolution of the image while maintaining a similar signal-to-noise ratio without the fear of anesthetic complications associated with longer scan times. 176 It is because of these limitations that we chose not to include the depth of the lesion in our MRI scoring system. 370 The specificity and predictive values for detection of articular cartilage lesions on MRI in the current study were similar, if not better than Olive et al. 328 The observers in the current study tried to make a conscientious effort not to over interpret MRI images, especially in areas of the joint where a partial volumeaveraging artifact existed. For instance, cartilage signal loss occasionally occurred when articular surfaces were in close contact with each other, especially in areas of curved articular surfaces, or along the dorsal rim of the joint where synovium overlaid the articular surface. This approach resulted in a high positive predictive value where there were only 2 instances where MRI identified cartilage signal loss and there wasn t a corresponding articular cartilage reported by Olive et al 328, but the results of the current study may be skewed because there was a high number of sub-regions where an articular cartilage lesion was not present. Lastly, MRI articular cartilage lesions scores were found to be significantly lower than arthroscopic scores. MRI consistently underestimated the size of the articular cartilage lesion, which is consistent with previous studies. 320,328 Despite this, total articular cartilage lesion scores were still moderately correlated between MRI and arthroscopy. To the authors knowledge, this is the first study to compare marginal osteophyte, intra-articular ligament integrity, synovial thickening, and 176

177 osteochondral fragments between MRI and arthroscopy in the horse. For the purposes of this study, arthroscopy was used as the gold standard diagnostic test by which these articular features were to be compared to because they are readily identifiable on arthroscopic examination. One limitation of this study is that arthroscopy has never been validated as a gold standard for the evaluation of these articular features. Magnetic resonance imaging has previously been shown to be the most sensitive diagnostic imaging modality for the detection of marginal osteophyte due to its tomographic nature. 17,18 Our results suggest that early marginal osteophyte formation may be detected with MRI before it is identified arthroscopically. There was a large number of grade I osteophytes detected by MRI that were not identified arthroscopically and total marginal osteophyte scores were also significantly higher on MRI than arthroscopy. Most of the articular surface of the middle carpal joint can be evaluated arthroscopically for the presence of marginal osteophytes with the exception of the most palmar aspect of the joint. 371 This may explain, in part, the poor positive predictive value of MRI when compared to arthroscopy as only MRI is able to identify marginal osteophytes along the palmar margins of the joint. It is also possible, however, that the appearance of small osteophytes may have been increased on MRI, as it has been shown that MRI sequences with long TE GRE sequences can artifactually enhance bone contours that can resemble marginal osteophyte formation. 17 Magnetic resonance imaging is considered the gold standard for detecting soft tissue injuries, especially injuries involving the central region of tendons or 177

178 ligaments. 382 Superficial lesions are not as easy to identify on MRI, especially within a synovial structure where partial volume-averaging artifact may hinder their detection. In contrast, arthroscopy can be used to directly visualize superficial tearing or complete rupture of intra-synovial tendons or ligaments, but cannot be used to visualize core injuries. 383 The differences in each imaging modalities ability to detect ligament injuries likely explains why the sensitivity, specificity, and predictive values for these articular features were not impressive. Clinical experience would suggest that the most information is gained when intrasynovial soft tissue structures, like the intercapal ligaments, are evaluated by both imaging modalities. Histological evaluation of these articular features is likely a better way to validate these imaging modalities, as well as how the overall sensitivity improves when the two are used in combination. Changes in signal intensity within the intercarpal ligaments may have also represented partial volume-averaging with synovial fluid or magic angle artifacts. 384 The observers in this study attempted to differentiate true artifact by evaluating potential MRI abnormalities on more than one imaging plane before identifying a ligament as abnormal. Our results agree with previous reports in humans, which have found that MRI can be a sensitive indicator of synovitis. 385,386 However, contrast-enhanced MR images have been shown to be better correlated to the synovial inflammation than non-contrast-enhanced MRI images when histology is used as the reference standard. 387,388 The addition of intravenous contrast to our MRI protocol would likely have improved the sensitivity of our MRI examination for diagnosing 178

179 synovitis by improving the observers ability to delineate inflamed synovium from joint effusion. 389 Despite these advantages, non-contrast-enhanced MRI is routinely used to assess synovitis in human osteoarthritis studies because of cost and risk of reaction. 28 Contrast was also not used in the current study because of the increased risk of anesthetic complications associated with the increased scan time. Although it may not be as accurate as synovial biopsies, arthroscopy is commonly used to evaluate the level of synovial inflammation clinically in horses 383 and has been shown to correlate well with synovial biopsies and contrast-enhanced MRI in people. 387,388 In this study, non-contrast-enhanced MRI proved to be a moderately sensitive and specific measure of synovitis and was moderately correlated with arthroscopic scores. An additional limitation of this study is that direct comparison of total joint MRI and arthroscopic semiquantitative scores is difficult. (Table 4-6) Many articular features that can be assessed by MRI, such as alterations in bone signal intensity, subchondral bone irregularity, collateral ligament integrity, and joint effusion could not be validated arthroscopically, as there was no antemortem gold standard for comparison. Despite this, total joint MRI and arthroscopy scores were still moderately correlated when scores were compared. Olive et al. 17 has previously correlated some of these articular features included subchondral bone irregularity, low signal bone lesions, and marginal osteophyte formation to total macroscopic articular cartilage lesion scores. Although this type of comparison showed an association, it did not identify a cause and effect relationship and did little to improve our understanding of the etiopathogenesis of 179

180 this disease process. We do believe, however, that correlation of abnormalities identified on MRI to those identified arthroscopic examination of the joint is clinically important. In addition to the limitations previously discussed, the scores used for validation of the equine MRI osteoarthritis scoring system were based on consensus agreement and so the results of this study may not be referable to a single observer. However, since the original description of the semiquantitative scoring system used consensus agreement in order to describe the structural changes that occurred over time 370, we sought to use the same scores to validate those findings. We found agreement between observer A and observer B or C was lower than agreement between observers B and C. (Table 4-5) A possible explanation for this finding is that observers B and C were more experienced clinicians than observer A. Also, observer A was involved both MRI and arthroscopic examination and scoring, so it is possible observer A was more familiar with the location of articular lesions that typically occurred in this study. Despite blinding observers by removing all identifiers from the MR images and randomizing their order, this form of systemic bias could not be prevented given the design of the study. In conclusion, our results support that the equine MRI osteoarthritis scoring system provided an effective and non-invasive method to identify the early osteoarthritis-related structural changes and disease progression seen in this equine model of post-traumatic osteoarthritis. The articular features included in the scoring system are considered important in the pathophysiology of 180

181 osteoarthritis, as well as clinically relevant in traumatic joint injuries in horses. 370 Further work is necessary to determine how these structural changes relate to clinical symptoms such as pain and lameness scores, as this comparison will improve the diagnostic value of this scoring system. 181

182 Table 4-1. Magnetic Resonance Imaging Parameters Sequence Acquisition type ST/IG (mm) TE (msec) TR (msec) FA (deg.) TI (msec) NEX FOV (cm) Matrix (Phase x Frequency) Pixel Size (mm) Sagittal PD-FSE 2D 3.0/ n/a 1 19 x x x :48 Axial PD-FSE 2D 3.5/ n/a 1 17 x x x :19 Dorsal T2-FSE 2D 3.0/ n/a 2 16 x x x :37 Axial STIR- FSE 2D 3.5/ x x x :14 Sagittal PD-FS-FSE 2D 3.0/ n/a 1 19 x x x :23 Dorsal PD-FS-FSE 2D 3.0/ n/a 1 16 x x x :17 Dorsal SPGR-FS-GE 3D 2.0/ n/a 1 16 x x x :32 ST/IG = Slice Thickness/Interslice Gap; TE = Echo Time; TR = Repetition Time; FA = Flip Angle; TI = Inversion Time; NEX = Number Of Excitations; FOV = Field Of View; PD = Proton Density; FSE = Fast Spin Echo; STIR = Short Tau Inversion Recovery; FS = Fat Saturation; SPGR = T2*-Weighted Spoiled Gradient Recalled Echo; GE = Gradient Echo Time (min:s) 182

183 Table 4-2. Equine MRI Osteoarthritis Scoring System Total Possible Articular Feature Cartilage Signal Abnormalities No abnormalities Single lesion < 5 mm Single lesion 5-10 mm or multiple lesion < 5 mm Multiple lesions with at least one lesion > 10 mm HSBL-Size No abnormalities < 25% 25-50% > 50% 27 HSBL-Intensity No abnormalities Mild Moderate Severe 27 SCBI Smooth, regular chondro-osseous junction Mild Subchondral bone plate irregularity Cyst-like formation 36 Marked subchondral plate irregularity with intact bone marrow Bone lysis extending to the medullary bone with preservation of some bone trabecular pattern LSBL <10% 10-25% 25-50% > 50% 27 Osteophyte No abnormalities Small Medium Large 27 (< 1 mm) (1 2 mm) (> 2 mm) Ligament Integrity (LCL,MCL,DMIL, LPIL, MPIL) No abnormalities Increased thickness or signal 5 Synovial Thickening Normal Mild Moderate Severe 3 Joint Effusion 0-10% max < 33% of max 33-66% of max > 66% of max 3 distension distension distension distension OCF-Size None 1 fragment 2 fragments >2 fragments 3 OCF-Number None Small Medium Large 3 (< 250 mm 3 ) ( mm 3 ) ( > 500 mm 3 ) Total Possible Joint Score 188 HSBL = High Signal Bone Lesion; SCBI = Subchondral Bone Irregularity; LSBL = Low Signal Bone Lesion; LCL = Lateral Collateral Ligament; MCL = Medial Collateral Ligament; DMIL = Dorsomedial Intercarpal Ligament; LPIL = Lateral Palmar Intercarpal Ligament; MPIL = Medial Palmer Intercarpal Ligament; OCF = Osteochondral Fragment Score

184 Table 4-3. Arthroscopic Scoring System Total Possible Articular Feature Cartilage Lesion- Size No abnormalities Single lesion < 5 mm Single lesion 5-10 mm or multiple lesion < 5 mm Multiple lesions with at least one lesion > 10 mm Marginal Osteophyte No abnormalities Small (< 1 mm) DMIL Integrity < 2 mm in width and > 2mm in width or Medium (1 2 mm) Large (> 2 mm) 27 1 blends with joint easily differentiated capsule from joint capsule MPIL Integrity LPIL Integrity Normal or only mild surface fibrillation Normal or only mild surface fibrillation along entire length Overt evidence of ligamentous disruption Overt evidence of ligamentous disruption OCF Number None 1 fragment Synovial Thickening Normal Mild: mild increase in 2 fragments > 2 fragments 3 3 either the thickness or new villi formation Moderate: mild increase in both thickness and the presence of new villi formation or moderate increase in either thickness or new villi formation Marked: moderate increase in both thickness and new villi formation or marked increase in either thickness or new villi formation, or the presence of synovial adhesion(s) Total Possible Joint 63 Score DMIL = Dorsomedial Intercarpal Ligament; LPIL = Lateral Palmar Intercarpal Ligament; MPIL = Medial Palmer Intercarpal Ligament; OCF = Osteochondral Fragment Score

185 Table 4-4. Percent Involvement (scores > 0) of each articular feature from consensus 1 on days 0, 14, and 98. Sub-Region RC-2 RC-3 IC-3 IC-4 UC 2C 3C-R 3C-I 4C Total Cartilage SA 5.0% 26.7% 6.7% 0.0% 0.0% 0.0% 25.0% 3.3% 0.0% 42.7% HSBL 25.0% 35.0% 8.3% 0.0% 6.7% 11.7% 13.3% 1.7% 1.7% 48.3% SCBI 26.7% 46.7% 11.7% 5.0% 0.0% 5.0% 41.7% 6.7% 0.0% 71.7% LSBL 78.3% 46.6% 58.3% 45.0% 30.0% 48.3% 91.7% 73.3% 33.3% 100.0% Osteophyte 75.0% 83.3% 36.7% 33.3% 11.7% 23.3% 53.3% 8.3% 1.7% 98.3% Sub-Region Total 93.3% 98.3% 75.0% 70.0% 40.0% 60.0% 95.0% 73.3% 35.0% 100.0% LCL 5.0% MCL 20.0% DMIL 26.7% MPIL 20.0% LPIL 5.0% TLA 55.0% Synovium 66.7% Effusion 70.0% OCF 21.7% Total Joint 100.0% RC-2 =Second Facet Of The Radial Carpal Bone; RC-3 = Third Facet Of The Radial Carpal Bone; IC-3 = Third Facet Of The Intermediate Carpal Bone; IC-4 = Fourth Facet Of The Intermediate Carpal Bone; UC = Ulnar Carpal Bone; 2C = Second Carpal Bone; 3C-R = Radial Facet Of The Third Carpal Bone; 3C-I = Intermediate Facet Of Third Carpal Bone; 4C = Fourth Carpal Bone; SA = Signal Abnormality; HSBL = High Signal Bone Lesion; SCBI = Subchondral Bone Irregularity; LSBL = Low Signal Bone Lesion; LCL = Lateral Collateral Ligament; MCL = Medial Collateral Ligament; DMIL = Dorsal Medial Intercarpal Ligament; MPIL = Medial Palmar Intercarpal Ligament; LPIL = Lateral Palmar Intercarpal Ligament; OCF = Osteochondral Fragment 185

186 Table 4-5. Reliability of the Equine MRI Osteoarthritis Scoring System Inter-Reader Agreement Intra-Reader Agreement Weighted κ (95% CI) Average Weighted κ (95% CI) Weighted κ (95% CI) A vs B A vs C B vs C A vs B vs C Consensus 1 vs 2 Cartilage SA (0.510 to 0.849) (0.606 to 0.911) (0.836 to 0.986) (0.644 to 0.922) (0.501 to 0.824) HSBL-Intensity (0.859 to 0.999) (0.798 to 0.967) (0.905 to 1.000) (0.854 to 0.990) (0.395 to 0.705) HSBL-Size (0.866 to 0.982) (0.845 to 0.976) (0.882 to 0.991) (0.864 to 0.982) (0.352 to 0.678) SCBI (0.781 to 0.949) (0.703 to 0.903) (0.845 to 0.979) (0.775 to 0.945) (0.261 to 0.601) LSBL (0.952 to 0.997) (0.906 to 0.982) to 0.990) (0.928 to 0.990) (0.397 to 0.562) Osteophyte (0.893 to 0.971) (0.878 to 0.963) (0.935 to 0.992) (0.902 to 0.976) (0.412 to 0.614) LCL (1.000 to 1.000) (1.000 to 1.000) (1.000 to 1.000) (1.000 to 1.000) (0.367 to 1.000) MCL (0.401 to 0.969) (0.519 to 1.000) (0.773 to 1.000) (0.533 to 1.000) (0.637 to 1.000) DMIL (0.776 to 1.000) (1.000 to 1.000) (0.776 to 1.000) (0.829 to 1.000) (0.530 to 1.000) MPIL ( to 1.000) ( to 1.000) (1.000 to 1.000) (0.158 to 1.000) (0.169 to 1.000) LPIL (1.000 to 1.000) (1.000 to 1.000) (1.000 to 1.000) (1.000 to 1.000) ( to 1.000) Synovium (0.661 to 0.995) (0.590 to 0.970) (0.863 to 1.000) (0.585 to 1.000) (0.169 to 0.791) Effusion ( ) (1.000 to 1.000) (0.724 to 1.000) (0.792 to 1.000) (0.24 to 0.844) 186

187 Table 4-5 Continued. Inter-Reader Agreement Intra-Reader Agreement Weighted κ Average Weighted κ Weighted κ (95% CI) (95% CI) (95% CI) A vs B A vs C B vs C A vs B vs C Consensus 1 vs 2 OCF-Number (1.000 to 1.000) (1.000 to 1.000) (1.000 to 1.000) (1.000 to 1.000) (1.000 to 1.000) OCF- Size (1.000 to 1.000) (1.000 to 1.000) (1.000 to 1.000) (1.000 to 1.000) (0.785 to 1.000) Total Joint (0.900 to 0.938) (0.888 to 0.903) (0.939 to 0.968) (0.913 to 0.941) (0.571 to 0.649) κ = kappa; CI = Confidence Interval; A = Observer A; B = Observer B; C = Observer C; SA = Signal Abnormality; HSBL = High Signal Bone Lesion; SCBI = Subchondral Bone Irregularity; LCL = Lateral Collateral Ligament; MCL = Medial Collateral Ligament; DMIL = Dorsomedial Intercarpal Ligament; MPIL = Medial Palmar Intercarpal Ligament; LPIL = Lateral Palmar Intercarpal Ligament; Synovium = Synovial Thickening; OCF = Osteochondral Fragment 187

188 Table 4-6. Mean Total MRI and Arthroscopic Semiquantitative Scores MRI Score Arthroscopy Score Cartilage Lesions-Size* [0-5] a [0-11] b High Signal Bone Lesion-Intensity* [0-8] High Signal Bone Lesion-Size* [0-8] Subchondral Bone Irregularity* [0-8] Low Signal Bone Lesion* [2-15] Marginal Osteophyte* [0-13] a [0-11] b Lateral Collateral Ligament** 0 [0,1] Medial Collateral Ligament** 0 [0,1] Dorsal Medial Collateral Ligament** 0 [0,1] 0 [0,1] Medial Palmar Intercarpal Ligament** 0 [0,1] 0 [0,1] Lateral Palmar Intercarpal Ligament** 0 [0,1] 0 [0,1] Synovial Thickening** 1 [0,3] 1 [0,3] Joint Effusion** 1 [0,3] Osteochondral Fragment-Size** 0 [0,3] Osteochondral Fragment-Number** 0 [0,3] 0 [0,2] Total Joint Score* [4-46] a [1-32] b *Data reported as mean + standard deviation [range] **Data reported as median (min, max) a,b Within each row, groups with different superscripts are significantly different (p < 0.05). 188

189 Table 4-7. Sensitivity and Specificity of MRI to Detect Intra-Articular Pathology Sensitivity (95% CI) Specificity (95% CI) PPV (95% CI) NPV (95% CI) Cartilage Lesion- Size (n = 360) 23.9% (17.1 to 32.4) 99.2% (97.1 to 99.8) 93.3% (78.7 to 98.2) 73.0% (68.0 to 77.5) Marginal Osteophyte (n = 360) 76.8 % (65.6 to 85.2) 72.2% (66.8 to 77.0) 39.6% (31.7 to 48.0) 92.9% (88.8 to 95.6) Synovial Thickening (n = 40) 73.5% (56.9 to 85.4) 66.7% (30.0 to 90.3) 92.6% (76.6 to 97.9) 30.8% (12.7 to 57.6) DMIL (n = 40) 57.1% (32.6 to 78.6) 88.5% (71.0 to 96.0) 72.7% (43.4 to 90.3) 79.3% (61.6 to 90.2) MPIL (n = 40) 0.09% (1.6 to 37.7) 75.9% (57.9 to 87.8) 12.5% (2.2 to 47.1) 68.6% (51.5 to 82.1) LPIL (n + 40) 0.0% (0.0 to 35.4) 93.9% (80.4 to 98.3) 0.0% (0.0 to 65.8) 81.6% (66.6 to 90.8) Soft Tissue Total (n = 120) 28.1% (15.6 to 45.4) 86.4% (77.7 to 92.0) 42.9% (24.5 to 63.5) 76.8% (67.5 to 84.0) Osteochondral Fragment (n = 40) 100.0% (20.6 to 100.0) 94.8% (83.1 to 98.6) 33.3% (6.2 to 79.2) 100.0% (90.6 to 100.0) PPV = Positive Predictive Value; CI = Confidence Interval; NPV = Negative Predictive Value; DMIL = Dorsomedial Intercarpal Ligament; MPIL = Medial Palmar Intercarpal Ligament; LPIL = Lateral Palmar Intercarpal Ligament 189

190 Figure 4-1. Dorsal SPGR-FS image showing normal (A) and abnormal (B) cartilage signal intensity of the third facet of the radial carpal bone. Using this sequence, cartilage signal is high compared to the low signal of the subchondral bone and intermediate signal of synovial fluid. The arrow in B is pointing to an area of cartilage signal abnormality. Cartilage signal abnormalities were scored based on the size of their largest dimension and graded as 0 = no abnormality, 1 = single lesion < 5 mm, 2 = single lesion 5-10 mm or multiple lesions < 5 mm, 3 = single lesion > 10 mm or multiple lesions < 10 mm. 190

191 Figure 4-2. Schematic representation of the MRI scoring system used for high-signal bone lesion intensity using the sagittal PD-FS (A,D,G), dorsal PD-FS (B,E,H) and axial T2-STIR (C,F,I) images. The intensity of increased signal for each high-signal bone lesion was subjectively scored based on severity of signal increase compared to the surrounding trabecular and subchondral bone and graded as 0 = none, 1 = mild (A,B,C), 2 = moderate (D,E,F), 3 = marked (G,H,I). 191

192 Figure 4-3. Schematic representation of the MRI scoring system used for subchondral bone irregularity using sagittal PD-FS images. Subchondral bone irregularity was scored based on severity and graded as: 0 = smooth and regular chondro-osseous junction, 1 = mild subchondral plate irregularity (A), 2 = marked subchondral plate irregularity with intact trabecular bone (B), 3 = bone irregularity extending to the trabecular bone with preservation of some bone trabecular pattern (C), 4 = cyst-like formation (D). The white arrowheads point to subchondral bone irregularities in the third facet of the radial carpal bone. Figure 4-4. Schematic representation of the MRI scoring system used for low-signal bone lesions using sagittal PD images. Low-signal bone lesions were subjectively scored based on their size and graded as: 0 = <10% of region (A), 1 = 10-25% (B), 2 = 26-50% (C), 3 = >50% (D). 192

193 Figure 4-5. Schematic representation of the MRI scoring system used for marginal osteophytes using dorsal T2-weighted images. Marginal osteophytes were scored based on size and graded as: 0 = normal (A), 1 = small (B), 2 = medium (C), 3 = large (D). The white arrowheads in B, C, and D point to the marginal osteophyte on third facet of the radial carpal bone. Figure 4-6. Schematic representation of the MRI scoring system used for the dorsomedial intercarpal ligament using axial T2-STIR images. The dorsomedial intercarpal ligament was scored based on size and graded as 0 = normal (A) and 1 = abnormal (B). The white arrow is pointing to a normal dorsomedial intercarpal ligament in A and the white arrowhead is pointing toward an abnormal dorsomedial intercarpal ligament in B. 193

194 Figure 4-7. Schematic representation of the MRI scoring system used for the medial palmar intercarpal ligament (A,B) and lateral palmar intercarpal ligament (C,D) using dorsal T2 weighted images. The medial and lateral palmar intercarpal ligaments were individually scored (0-1) based on changes in size and signal intensity and graded as 0 = normal (A,C) and 1 = abnormal (B,D). The white arrows point to a normal medial (A) and lateral (C) palmar intercarpal ligament and the white arrowheads points to the abnormal medial (B) and lateral (D) palmar intercarpal ligament. 194

195 Figure 4-8. Schematic representation of the MRI scoring system used for synovial thickening using axial T2-STIR images. On T2-STIR sequence, synovium was recognized as tissue of moderate signal intensity protruding into the hyperintense synovial fluid. Synovial thickening was scored based on the intensity and distribution of synovium identified and graded as: 0 = normal (A), 1 = mild - mild increase in either intensity or distribution (B), 2 = moderate - mild increase in both intensity and distribution or moderate increase in either intensity or distribution (C), 3 = marked - moderate increase in both intensity and distribution, marked increase in either intensity or distribution, or presence of synovial adhesion(s) (D). 195

196 Figure 4-9. Schematic representation of the MRI scoring system used for joint effusion using sagittal PD-FS images. Joint effusion was scored based on the estimated maximum distension of the synovial cavity and graded as: 0 = normal (A), 1 = <33% of maximum potential distention (B), 2: 33% - 66% of maximum potential distension (C), 3 = > 66% of maximum potential distension (D). 196

197 Figure Diagram of the 9 sub-regions that were evaluated using EMOSS. RC-2: Second Facet Of The Radial Carpal Bone; RC-3: Third Facet Of The Radial Carpal Bone; IC-3: Third Facet Of The Intermediate Carpal Bone; IC-4: Fourth Facet Of The Intermediate Carpal Bone; UC: Ulnar Carpal Bone; 2C: Second Carpal Bone; 3C-R: Radial Facet Of The Third Carpal Bone; 3C-I: Intermediate Facet Of The Third Carpal Bone; 4C: Fourth Carpal Bone. 197

198 Figure Schematic representation of the arthroscopic scoring system used for grading articular cartilage lesions (0-3). Articular cartilage lesions were scored based on the size of largest dimension and graded as 0 = no abnormality, 1 = single lesion < 5 mm, 2 = single lesion 5-10 mm or multiple lesions < 5 mm, 3 = single lesion > 10 mm or multiple lesions < 10 mm. Figure Schematic representation of the arthroscopic scoring system used for grading marginal osteophytes (0-3). Marginal osteophytes were scored based on size and graded as 0 = Normal (A), 1 = small (B), 2 = medium (C), 3 = large (D). Figure Schematic representation of the arthroscopic scoring system used for grading the dorsomedial intercarpal ligament (0-1). The dorsomedial intercarpal ligament was scored based on its size and graded as: 0 = less than 2 mm in width (dorsolateral-dorsomedial) and blends into the joint capsule (A) or 1 = greater than 2 mm in width or easily differentiated from the joint capsule over its entire length (B). 198

199 Figure Schematic representation of the arthroscopic scoring system used for grading the medial palmar (A,B,C) and lateral palmar intercarpal ligament (D,E,F). The medial and lateral palmar intercarpal ligaments were independently assessed by careful probing of the ligament to determine the degree of fiber disruption and graded as: 0 = normal (A,D) or 1 = evidence of ligament disruption (B,E). An arthroscopic probe was used to manipulate each ligament (C) or displace the two opposing joint surfaces to stretch the ligament (F) to determine the integrity of each ligament. 199

200 Figure Schematic representation of the arthroscopic scoring system used for grading synovial thickening (0-3). Synovium was scored based on both the overall thickness of the synovial villi and the presence of new villi formation and graded as 0 = normal (A), 1 = mild - mild increase in either the thickness or new villi formation (B), 2 = moderate - mild increase in both thickness and the presence of new villi formation or moderate increase in either thickness or new villi formation (C), or 3 = marked - moderate increase in both thickness and new villi formation or marked increase in either thickness or new villi formation (D), or the presence of synovial adhesion(s) (E). 200